1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.15:
3 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
5 * bug fix: loading of default sysinit file works. (thanks to Leonid
8 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
9 * feature: implemented the READER-METHOD-CLASS and
10 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
12 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
14 * optimization: faster LOGCOUNT implementation on x86 and x86-64
15 (thanks to Lutz Euler)
16 * optimization: hashing of general arrays and vectors has been
17 improved. (reported by Any Fingerhut)
18 * enhancement: SB-INTROSPECT is now able to find definitions of
19 profiled functions. (thanks to Troels Henriksen)
20 * enhancement: compiler-macro expansion applies now to FUNCALL forms
21 as well, allowing compiler-macros for SETF-functions to expand.
22 * enhancement: step-instrumentation no longer wraps known
23 single-value functions with multiple-value context, allowing
24 better type inference.
25 * fixed bug #337: use of MAKE-METHOD in method combination now works
26 even in the presence of user-defined method classes. (reported by
27 Bruno Haible and Pascal Costanza)
28 * fixed bug #339(c): if there are applicable methods not part of any
29 long-form method-combination group, call INVALID-METHOD-ERROR.
30 (reported by Bruno Haible)
31 * fixed bug #361: the :FUNCTION initarg in the protocol for
32 initialization of methods can now be used to override
33 internally-produced optimized functions. (reported by Bruno
35 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
36 system-provided lambda expression no longer cause warnings about
37 unbound #:|pv-table| symbols.
38 * bug fix: improved the handling of type declarations and the
39 detection of violations for keyword arguments with non-constant
41 * bug fix: potentially erronous calls to PATHNAME and
42 MERGE-PATHNAMES were being flushed in some cases.
43 (reported by Richard Kreuter)
44 * bug fix: compiled calls to TYPEP were mishandling obsolete
45 instances. (reported by James Bielman and Attila Lendvai)
46 * bug fix: format strings with ~> without matching ~< no longer
47 trigger an AVER, but signal an understandable error instead.
48 (reported by Antonio Martinez)
49 * bug fix: specifying an output-file without a pathname-type for
50 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
51 of input-file instead of "fasl". (reported by Robert Dodier)
52 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
53 (reported by James Y Knight).
54 * bug fix: compiler-macros-function did not consider the environment
55 argument for shadowing by local functions.
56 * bug fix: compiler-macros expansion was inhibited by local INLINE
58 * bug fix: inline expansions of known functions were subject to
59 step-instrumentation in high debug policies, leading to problems
61 * bug fix: compiler failed to differentiate between different CONS
63 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
64 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
65 * bug fix: #368: incorrect use of expressed vs. upgraded array
67 * bug fix: #306a: more precise unions of array types.
68 * thread-safety improvements:
69 ** CONDITION-WAIT could return early on Linux, if the thread was
70 interrupted and subsequently continued with SIGCONT.
71 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
73 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
74 * added support for the ucs-2 external format. (contributed by Ivan
76 * minor incompatible change: pretty printing of objects of type
77 (cons symbol) is, in the default pprint-dispatch-table, now
78 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
80 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
81 called later in a class's lifetime, possibly as late as when the
82 first instance of the class is created. Previously,
83 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
84 class became finalizeable.
85 * fixed bug: FILE-POSITION sometimes returned inconsistent results
86 for multibyte external-format streams. (thanks to "vbzoli")
87 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
88 with :ALLOCATION :CLASS inherited from superclasses of the
90 * fixed bug: anonymous classes can now be created using the :NAME
91 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
92 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
93 * fixed bug: core-files saved with :EXECUTABLE T can again be
94 executed when SBCL_HOME isn't set. (reported by James Knight)
95 * fixed bug: toplevel LOCALLY forms with declarations could
96 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
97 * fixed bug: printing from several different threads using different
98 values of *print-case* could cause invalid output, due to
99 some internal special variables of the printer not being bound
100 thread-locally (reported by Max Mikhanosha)
101 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
102 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
103 (reported by Pascal Costanza)
104 * fixed bug: FINALIZE-INHERITANCE is called from
105 REINITIALIZE-INSTANCE on classes when the class has previously
106 been finalized, as required by AMOP.
107 * minor code generation optimizations:
108 ** better register allocation in CLOS dispatching functions
109 ** overflow detection when coercing signed bytes to fixnums on x86-64
110 ** is now implemented with one IMUL instruction instead of three shifts
111 ** more efficient bit-vector access on x86 and x86-64
112 ** more efficient access to raw structure slots on x86 and x86-64
113 * fixed some bugs revealed by Paul Dietz' test suite:
114 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
117 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
118 * feature: thread support on Solaris/x86, and experimental thread support
120 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
122 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
123 default method for SLOT-UNBOUND.
124 * minor incompatible change: prevent the user from specializing the
125 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
126 somewhat counter to the intent of the protocol, I (CSR) think, and
127 additionally it just doesn't work in SBCL as currently
128 implemented, thanks to optimizations (that are always valid for
129 the other three SLOT-VALUEish functions, but not for the setter).
130 * minor incompatible change: the :SB-LDB feature is now enabled by
131 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
132 the low-level debugger.
133 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
134 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
135 down to an SB-SYS:FD-STREAM, allowing direct communication in
136 more cases. Temporary files and pipes are still used as a fallback
138 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
139 on the non-reentrant C functions localtime(3) and gmtime(3).
140 * bug fix: global optimization policy was not visible in LOCALLY and
142 * bug fix: class objects can be used as specializers in methods.
143 (reported by Pascal Costanza)
144 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
145 with a type error. (reported by blitz_ on #lisp)
146 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
147 failed with a type error. (reported by Pascal Bourguignon)
148 * bug fix: merging pathnames against defaults with :DIRECTORY
149 starting with '(:RELATIVE :BACK) should preserve the :BACK.
150 (reported by James Y Knight)
151 * bug fix: saving large (>2GB) cores on x86-64 now works
152 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
153 constant index and a (simple-array (signed-byte 32)) array
154 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
155 error (patch by Robert J. Macomber)
156 * feature: TIME now displays the amount of run-time spent in GC
157 * bug fix: The debugger now does a better job of respecting
158 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
159 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
160 workaround for bug 403.)
161 * bug fix: floating point exception handling now works on FreeBSD
162 (thanks to NIIMI Satoshi)
163 * fixed some bugs revealed by Paul Dietz' test suite:
164 ** MISC.641: LET-conversion were not supposed to work in late
166 * improvements to the Win32/x86 port:
167 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
168 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
170 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
171 * new feature: source path information is generated for macro-expansion
172 errors for use in IDE's like Slime (thanks to Helmut Eller)
173 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
174 * bug fix: compiler does not loop forever on an invalid type in
176 * improvement: compilation of most CLOS applications is significantly
178 * optimization: added a limited bytecode compiler for simple toplevel
179 forms, speeding up compilation and FASL loading
181 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
182 * minor incompatible change: in sbcl-0.9.11 (but not earlier
183 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
184 into subclasses of the finalized class. Now user calls to
185 FINALIZE-INHERITANCE finalize just the one class, and calls by the
186 system return before any subclasses are finalized.
187 * minor incompatible change: The reader no longer ignores errors
188 regarding non-existent packages in #+ and #- feature tests.
189 * new feature: command line options --no-sysinit, --no-userinit to
190 inhibit loading the corresponding init files
191 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
192 for sending data through UDP sockets (thanks to François-René Rideau)
193 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
194 error is signalled instead (thanks to François-René Rideau)
195 * minor incompatible change: Error signalling behaviour of lexical
196 operations violating package locks has changed slightly. Refer to
197 documentation on package locks for details.
198 * enhancement: EVAL can process IF-expressions without resorting to the
200 * bug fix: Some operations involving SIN, COS, or TAN weren't being
201 constant-folded properly due to missing out-of-line functions.
202 (reported by Mika Pihlajamäki)
203 * bug fix: LISTEN sometimes returned T even in cases where no data was
204 immediately available from the stream
205 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
206 were reversed. (reported by Levente Mészáros)
207 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
208 when given a vector displaced to an adjustable array. (reported
210 * bug fix: derivation of float boundaries from numbers outside the
211 appropriate float range (reported by John Wiseman)
212 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
214 * bug fix: precomputing cache entries for generic functions with
215 some subclasses of specializers as yet invalid does not attempt to
216 fill a cache line with a negative offset. (reported by Levente
218 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
219 allow condition class objects as arguments to DOCUMENTATION and
220 (SETF DOCUMENTATION); only find and set documentation for
221 structure names for the STRUCTURE doc-type. (suggested by Gary
223 * improvements to the Win32/x86 port:
224 ** (user-homedir-pathname) and default initialization file
225 locations now know about the user's "Documents and Settings"
226 directory (thanks to Yaroslav Kavenchuk)
227 ** run-program is implemented (thanks to Mike Thomas)
228 ** sockets support (thanks to Timothy Ritchey)
229 ** better backtrace support (thanks to Alastair Bridgewater)
230 ** sb-grovel supported
231 ** asdf-install and sb-posix work somewhat
232 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
233 * improvements to the Solaris/x86 port:
234 ** works on Solaris 11/Solaris Express
235 ** floating-point exception handling support
236 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
237 * fixed some bugs revealed by Paul Dietz' test suite:
238 ** REMOVE-METHOD returns its generic function argument even when
239 the method is not one of the generic functions' methods.
240 ** objects declared as MEMBER types can be admissible arguments to
242 ** printing characters should simply be printed by the FORMAT ~:C
244 ** compiler failure when compiling functions with hairy constant
245 defaults for optional parameters.
246 ** compiler produces wrong code when MAYBE-INLINE-expanding a
247 function, which is already optimized.
249 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
250 * new platform: experimental support for SBCL x86/Darwin, including
251 MacOS X 10.4.5 on Intel.
252 * new feature: Unicode character names are now known to the system
253 (through CHAR-NAME and NAME-CHAR).
254 * new feature: the filesystem location of SBCL's core file is
255 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
256 * minor incompatible change: the contrib modules SB-POSIX and
257 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
258 this change is to make it easier to distribute
259 location-independent binaries.
260 * bug fix: as implied by AMOP, standardized classes no longer have
261 slots named by external symbols of public packages. (reported by
263 * optimization: calling functions via constant symbols -- as in
264 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
265 via the function object as in (FUNCALL #'FOO).
266 * optimization: on x86 and x86-64, the calling convention for the
267 case of dealing with an unknown number of values has been altered
268 to be friendlier to the prediction heuristics implemented,
269 particularly with reference to CALL and RET pairing. (thanks to
270 Alastair Bridgewater)
271 * optimization: on x86, the code for access to thread-local symbol
272 values and binding/unbinding of thread-local symbols is smaller.
273 (thanks to Alastair Bridgewater)
274 * enhancement: CONSTANTP is now able to determine constantness of
275 more complex forms, including calls to constant-foldable standardized
276 functions and some special forms beyond QUOTE.
277 * fixed bug: occasional GC crashes on Solaris/x86
278 * optimization: x86-64 supports stack allocation of results of simple
279 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
280 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
281 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
282 (thanks to James Knight)
283 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
284 by the standard (thanks to Stephen Compall)
286 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
287 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
288 be used for bundling the runtime and the core file into one
289 executable binary. This feature is not currently supported on all SBCL
290 platforms. (thanks to James Bielman and NIIMI Satoshi)
291 * new feature: a generational or ephemeral garbage collector is now
292 the default on the PowerPC platform (both Linux and Darwin). The
293 old Cheney (stop and copy) collector is a build-time option.
294 (thanks to Cyrus Harmon, after Raymond Toy)
295 * minor incompatible change: the method by which SBCL finds its
296 contributed modules has changed; it no longer relies on symbolic
297 links from an $SBCL_HOME/systems directory, but searches directly
298 in the subdirectories of $SBCL_HOME.
299 * enhancement: the dynamic heap size on the Linux/PPC platform is
300 markedly larger, even using the older Cheney garbage collector.
301 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
302 certain complicated string types. (reported by Gary King)
303 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
304 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
305 * fixed bug: DOCUMENTATION on structures no longer signals an
306 error if no structure type of the right name exists. (reported by
308 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
309 applicable non-standard primary or :AROUND method on
310 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
311 no longer cause unbound variable errors. (reported by Kevin Reid)
312 * fixed bug: metacircle resolution in cases where methods have slots
313 added before the slots from STANDARD-METHOD. (reported by Jean
315 * fixed bug: the Power PC assembly code for calling into lisp
316 sometimes computed its return address wrongly (depending on the
317 15th bit of the address, fixed at link time). This used to
318 manifest itself in a segmentation violation while building PCL.
319 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
320 many others over the years)
321 * fixed bug: ROOM no longer reports silly numbers for stack usage if
322 the stack pointer (treated as a signed integer) is negative.
323 (thanks to Peter van Eynde)
325 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
326 * new platform: experimental support for the Windows operating
327 system has been added. (thanks to Alastair Bridgewater)
328 * fixed several bugs in and robustified the PPC FFI (including
329 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
330 * bug fix: allow non-simple string symbol names (reported by Paul
332 * bug fix: interrupt handling on NetBSD (thanks to Richard M
334 * bug fix: saving a core corrupted callbacks on x86/x86-64
335 * bug fix: closed a loophole in metacircularity detection and
336 grounding in the PCL implementation of CLOS.
337 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
338 (reported by tomppa on #lisp)
339 * bug fix: clear the direction flag on Lisp -> C transitions, as
340 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
341 (reported by Andrej Grozin and Hendrik Maryns)
342 * optimization: major improvements to GC efficiency on GENCGC platforms
343 * optimization: faster implementation of EQUAL
344 * optimization: emit more efficient opcodes for some common
345 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
346 * optimization: several other minor code-generation improvements on x86-64
347 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
349 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
350 * minor incompatible change: (SETF CLASS-NAME) and (SETF
351 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
352 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
353 by ANSI as a generic function, and for consistency (SETF
354 GENERIC-FUNCTION-NAME) is treated likewise.
355 * fixed bug #233.b: make constraint propagation notice when a variable
356 value is changed after it is referenced but before it is used
357 * fixed bug #296: no more arbitrary behaviour from filenames with
358 odd characters as --load arguments. (reported by Adam Warner)
359 * fixed bug #390: :CHARACTER-SET pathname components now work as
360 expected. (reported by Tim Daly Jr)
361 * fixed bug #391: complicated :TYPE intersections in slot
362 definitions no longer cause an error in PCL internals.
363 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
364 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
365 returning the number of octets which would be written to the
366 file-stream. (thanks to Robert J. Macomber)
367 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
368 arguments to the internal stream functions. (thanks to David
370 * bug fix: stack exhaustion is now handled on the x86/SunOS
371 platform. (thanks to Daisuke Homma)
372 * bug fix: more accurate ROOM results on GENCGC platforms
373 * optimization: improved type inference for arithmetic-for
374 index variables in LOOP
375 * optimization: faster floating-point SQRT on x86-64
376 * fixed some bugs revealed by Paul Dietz' test suite:
377 ** DOCUMENTATION returns NIL instead of "" for method combinations
378 that don't have a docstring
380 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
381 * minor incompatible change: (SETF CLASS-NAME) and (SETF
382 GENERIC-FUNCTION-NAME) are no longer generic functions, and
383 therefore are not customizeable by user code (as seems to be at
384 least permitted and maybe required by AMOP). As a consolation,
385 however, the SBCL implementation of these functions now calls
386 REINITIALIZE-INSTANCE as specified by AMOP.
387 * bug fix: slot-definition documentation is propagated as per ANSI
388 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
389 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
390 argument, as do eslotd initialization methods. (from Pascal
391 Costanza's "Closer" project)
392 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
393 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
395 * bug fix: it is now possible to have more than one subclass of
396 STANDARD-GENERIC-FUNCTION without causing stack overflow.
397 (reported by Bruno Haible, Pascal Costanza and others)
398 * bug fix: the dependent update protocol now works for generic
399 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
401 * bug fix: condition-class instances corresponding to
402 DEFINE-CONDITION forms are now created eagerly. (reported by
403 Kalle Olavi Niemitalo on comp.lang.lisp)
404 * bug fix: floating point printing is more accurate in some
405 circumstances. (thanks to Simon Alexander)
406 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
407 merged with *DEFAULT-PATHNAME-DEFAULTS*.
408 * bug fix: callbacks on OS X now preserve stack-alignment.
409 (thanks to Cyrus Harmon)
410 * enhancement: the x86-64 disassembler is much better at
411 disassembling SSE instructions. (thanks to Lutz Euler)
412 * enhancement: CHECK-TYPE on a variable now allows type-inference
413 (thanks to Vincent Arkesteijn)
414 * optimization: improved performance of EUC-JP external format.
415 (thanks to NIIMI Satoshi)
416 * optimization: performance improvements to IO on file streams of
417 :ELEMENT-TYPE CHARACTER
418 * optimization: much faster memory allocation on x86-64
419 * optimization: faster garbage collections (latency and throughput)
421 * optimization: faster bignum subtraction and division on x86 and x86-64
422 (thanks to Lutz Euler)
423 * optimization: more accurate type-inference for loops that use a
424 floating point index variable or a negative step.
426 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
427 * bug fix: add a workaround to SBCL looping infinitely at startup on
428 Linux kernels with apparently buggy implementations of personality().
429 (thanks to Svein Ove Aas)
430 * bug fix: Unicode symbols are correctly printed in LDB backtraces
431 (thanks to David Lichteblau)
432 * bug fix: local bindings shadow global symbol macros for
433 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
434 * optimization: non-open coded uses of numeric comparison operators
435 (e.g. >) no longer cons when called with more than one parameter
436 on platforms supporting dynamic-extent allocation.
437 * enhancement: saving cores with foreign code loaded is now supported
438 on MIPS/Linux in addition to the previously supported platforms.
439 * bug fix: FIND-RESTART now tests for activity, not applicability when given
440 a restart object as identifier. (reported by Helmut Eller for CMUCL)
441 * bug fix: division by zero in sb-sprof when no samples were collected
442 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
443 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
444 (thanks to Alexander Kjeldaas)
445 * new feature: ignoring the return values of destructive functions
446 like NREVERSE signals a compile-time style-warning.
447 (thanks to Kevin Reid)
448 * new feature: the alignment of alien structure fields can be
449 explicitly specified. (thanks to Cyrus Harmon)
450 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
451 and dump core on SIGQUIT
453 ** incompatible change: threads do not inherit values of specials
454 from their parents (see manual)
455 ** bug fix: threads stacks belonging to dead threads are freed by the
456 next exiting thread, no need to gc to collect thread stacks anymore
457 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
458 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
459 ** bug fix: dynamic variable and thread start related gc lossage
460 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
462 * fixed some bugs revealed by Paul Dietz' test suite:
463 ** SUBTYPEP is slightly more accurate on heinously complicated
464 CONS types where some of the members have uncertain (in the
465 NIL, NIL sense) type relationships to each other.
467 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
468 no longer purified unless :PURIFY T is explicitly specified.
469 ** Non-purified cores are significantly smaller than before
471 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
472 * new feature: timers based on Zach Beane's excellent timer package
473 * added support for the following external formats: koi8-u,
474 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
475 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
476 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
477 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
478 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
479 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
480 * incompatible change: a threaded SBCL will no longer revert to
481 non-threaded mode on non-NPTL systems, but refuse to start entirely.
482 * bug fix: interrupts are disabled until startup is complete; no
483 more sigsegvs when receiving a signal to soon
484 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
486 * bug fix: add a workaround for the memory randomization features in
487 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
488 workaround will only be in effect on systems with the proc filesystem
490 * bug fix: printing objects of type HASH-TABLE signals a
491 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
493 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
494 have been running for over 50 days. (reported by Gilbert Baumann)
495 * bug fix: the logic for getting names of functions gets less
496 confused when confronded with alternate-metaclass
497 funcallable-instances. (reported by Cyrus Harmon)
498 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
499 consistent, even on internal alternate-metaclass objects.
500 * bug fix: flush closure information collected by physical
501 environment analysis prepass before the main pass. (bug reported
503 * bug fix: compiler pack phase does not modify a hashtable, which is
504 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
505 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
506 may be expanded. (reported by Alexander <malishev> on c.l.l)
507 * bug fix: no more style-warnings when compiling code using LAST
508 under high-SPEED low-SPACE settings. (reported by David Wallin
510 * fixed some incompatibilities between SBCL's MOP and the MOP
512 ** the METAOBJECT class is now implemented;
513 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
514 STANDARD-OBJECT, as required;
515 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
516 now compatible (as required by VALIDATE-SUPERCLASS); there
517 remains a consistency requirement over the presence or absence
518 of the FUNCTION class in the superclass list of finalized
519 classes; see the manual for more details;
520 ** the :AROUND method for COMPUTE-SLOTS on
521 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
522 requested slot ordering.
524 ** bug fix: parent thread now can be gc'ed even with a live
526 ** bug fix: binding a special with PROGV to no value is not
527 the same as that symbol not having been bound (thanks to
529 ** bug fix: binding specials is thread safe (thanks to
531 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
532 and SB-SPROF do not die with 'no handler for signal XX in
533 interrupt_handle_now(..)' anymore
534 ** bug fix: WITH-TIMEOUT works with multiple running threads
535 * fixed some bugs revealed by Paul Dietz' test suite:
536 ** ENSURE-GENERIC-FUNCTION should take a method class object for
537 the :method-class keyword argument.
539 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
540 * new port: the Solaris operating system on x86 processors is now
541 mostly supported, though some rough edges in the environment
542 remain. (thanks to Daisuke Homma)
543 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
544 space for its heap. (thanks to Thiemo Seufer)
545 * optimization: SBCL on MIPS platforms now supports dynamic-extent
546 closures. (thanks to Thiemo Seufer)
547 * minor incompatible change: eof selects abort in the debugger.
548 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
549 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
551 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
552 deprecated, and will go away in a future revision of SBCL.
553 * minor incompatible change: GC-ON and GC-OFF are no longer
554 implemented with a counter, it does not matter how many times gc
555 is switched on or off
556 * bug fix: discriminating functions for generic function classes
557 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
558 make invalid assumptions about method precedence order. (reported
560 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
561 platforms. (thanks to Thiemo Seufer)
562 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
563 * bug fix: correct pathname printing with printer escaping is on.
564 (thanks to Kevin Reid)
565 * bug fix: complex VOP definitions in "user-space" no longer trigger
566 package locks. (reported by Zach Beane)
567 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
568 for funcallable-instance functions, and is no different from
570 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
571 characters in the current readtable. (reported by Nicholas Neuss)
572 * bug fix: SUBTYPEP on various CONS types returns more a more
573 accurate acknowledgment of its certainty.
574 * optimizations: REMOVE-DUPLICATES now runs in linear time on
575 lists in some cases. This partially fixes bug 384.
576 * flush all standard streams before prompting in the REPL and the
578 * bug fix: signal handling and triggering gc do not conflict
579 directly anymore, in particular a high frequency sb-sprof does
580 not prevent gc from running
581 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
582 approximation for timezone and DST information between the
583 universal time epoch and the smallest negative 32-bit time_t.
584 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
585 year 1899 so long as the encoded time is non-negative.
586 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
587 the correct answer for non-constant rotate amounts when performing
588 an inline 32-bit rotation.
590 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
591 there is only one thread in the session
592 ** bug fix: memory leak for streams created in one thread and
593 written to in another
594 ** bug fix: lockup when compiled with gcc4
595 ** bug fix: race that allows the gc to be triggered when gc is
597 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
598 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
600 ** bug fix: finished threads are gc'ed properly
601 * fixed some bugs revealed by Paul Dietz' test suite:
602 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
603 has arguments with a different set of applicable methods from
604 the orignal arguments.
605 ** The type error thrown by MAP now has a correct expected-type
607 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
608 name a compiled function.
609 ** Three MISC tests where a large bignum was improperly coerced to
610 a float that couldn't represent that bignum during type
611 derivation were fixed.
612 ** SUBTYPEP can now handle types involving the negation of a
613 list-form FUNCTION type.
614 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
615 as (COMPLEX (AND RATIO (NOT FIXNUM))).
616 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
618 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
619 * New feature: Experimental support for bivalent streams: streams
620 opened with :element-type :default now allow character and binary
621 (unsigned-byte 8) I/O
622 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
623 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
625 * bug fix: no more highly sporadic "couldn't check whether ~S is
626 readable" when reading a stream and an interrupt hits in the middle
627 of a select system call
628 * compiler better recognizes complex arrays (reported by Eduardo
630 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
631 for variables on SBCL built with linkage-tables. (reported by Luis
633 * various error reporting improvements.
634 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
635 (Thanks to Thiemo Seufer)
636 * bug fix: nested reader invokations work correctly
637 * bug fix: it is possible to have simultaneous references to foreign
638 code and foreign data with the same name.
640 ** added x86-64 support
641 ** incompatible change: the threading api now works with thread
642 objects instead of thread ids
643 ** bug fix: threads are protected from signals and interruption when
644 starting up or going down
645 ** bug fix: a race where an exiting thread could lose its stack to gc
646 ** bug fix: don't halt on infinite error in threads if possible
647 ** fixed numerous gc deadlocks introduced in the pthread merge
648 ** bug fix: fixed thread safety issues in read and print
649 ** bug fix: debugger doesn't hang on session lock if interrupted at
650 an inappropriate moment
651 ** bug fix: run-program is now thread safe(r)
652 ** bug fix: inner with-recursive-lock no longer releases the mutex
653 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
654 * fixed some bugs revealed by Paul Dietz' test suite:
655 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
656 EXPECTED-TYPE slots filled.
657 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
659 ** MAKE-CONDITION accepts classes as type-designators.
660 ** COMPILE may never return NIL.
661 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
662 range before calling Unix time functions
664 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
665 * numerous signal handling fixes to increase stability
666 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
667 * minor incompatible change: we now correctly canonize default
668 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
669 as per the MOP, rather than the historical (INITARG INITFUNCTION
670 INITFORM). (reported by Bruno Haible)
671 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
672 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
673 for more information.
674 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
675 pathname is a directory pathname.
676 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
677 * dynamic space size on PPC has been increased to 768Mb. (thanks to
679 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
680 :METACLASS argument in addition to a class name. (reported by
681 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
682 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
683 (reported by Bruno Haible; patch from Gerd Moellmann)
684 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
686 * bug fix: more cleanups to the floating point exception handling on
687 x86-64 (thanks to James Knight)
688 * bug fix: the compiler does not try to constant fold calls of
689 COERCE and BIT-* functions when they return freshly constructed
690 objects. (reported by Nikodemus Siivola and Paul Dietz)
691 * optimization: improved the MIPS versions of generic (in the
692 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
693 * optimization: direct conversion of (unsigned-byte 32) to floats on
694 the PowerPC platform.
695 * optimization: structure instances with raw slots now use less
696 memory, and probably show better memory locality. (thanks to
698 * optimization: DYNAMIC-EXTENT declarations for lists and closures
699 are treated as requests for stack allocation on the x86-64,
700 Alpha-32, PPC and SPARC platforms.
701 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
702 the wrong file descriptor; implementation of SOCKET-OPEN-P.
703 (thanks to Tony Martinez)
705 ** gcing a dead thread can no longer lead to lockups
706 ** threads block signals until they are set up properly
707 ** errno is no longer shared by threads
708 ** interrupt-thread restores the eflags register on x86
709 ** fixed some lockups due to gc/thread interaction
710 * fixed some bugs revealed by Paul Dietz' test suite:
711 ** invalid dotted lists no longer raise a read error when
713 ** signal an error if a symbol that names a declaration is used
714 as the name of a type, or vice versa
715 ** allow using the (declare (typespec var*)) abbreviation for
716 (declare (type typespec var*)) with all type specifiers
717 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
718 ** always return NIL from PPRINT-POP when OBJECT is NIL
719 ** don't signal errors when pretty-printing malformed LABELS,
720 FLET or MACROLET forms
721 ** declarations in a DOLIST body are also in scope for the
723 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
725 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
728 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
729 * fixed cross-compiler leakages that prevented building a 32-bit
730 target with a 64-bit host compiler.
731 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
732 opened with :IF-EXISTS :OVERWRITE.
733 * fixed bug 281: error for an invalid qualifier in a short-form method
734 combination method is not signalled until the faulty method is called.
735 * bug fix: iteration variable type inferrer failed to deal with open
736 intervals. (reported by Alan Shields)
737 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
738 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
739 or more runtime options were provided to the sbcl binary.
740 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
741 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
742 * optimization: there's now a fast-path for fixnum arguments in the
743 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
744 * optimization: the code generated on x86-64 is more compact thanks
745 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
746 * bug fix: floating-point exception handling now partly works on
747 x86-64. (Thanks to James Knight)
748 * improvement to the MIPS backend from Thiemo Seufer: C-style
749 64-bit long long arguments and return values to alien functions
751 * fixed some bugs revealed by Paul Dietz' test suite:
752 ** the type-error signalled from WARN has a filled-in DATUM slot.
753 ** the type-error required when a stream is not associated with
754 a file has the stream as its datum.
755 ** type-errors on single-floats on x86-64 no longer have
756 :INVALID-OBJECT as the datum
757 ** the type-errors signalled for invalid function names now have
758 a correct expected type
759 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
760 for structures defined with a :TYPE. Documentation strings for
761 typed structures are no longer immediately discarded
762 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
763 broadcast streams and synonym streams. FILE-LENGTH now also works
764 on broadcast streams.
766 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
767 * incompatible change: the --noprogrammer option, deprecated since
768 version 0.7.5, has been removed. Please use the equivalent
769 --disable-debugger option instead.
770 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
772 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
773 inoperational for a while now) has been completely removed.
774 * null lexical environments are now printed as #<NULL-LEXENV>,
775 significantly reducing the amount of clutter in typical backtraces.
776 * documentation on weak pointers, finalization, and after GC hooks
777 has been added to the manual.
778 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
779 as other specialized array types, is much faster. SUBSEQ and
780 COPY-SEQ on such arrays have also been sped up.
781 * optimization: EQL is now more efficient when at least other argument
782 is known to be of type (OR FIXNUM (NOT NUMBER)).
783 * fixed bug: the runtime is now less vulnerable to changes in the
784 size of the SBCL object on OS X, and virtual memory is reserved for
785 all spaces that need to be at a fixed address.
786 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
787 * fixed bug: finalizers and after GC hooks that cause consing are now
789 * fixed bug: compiler error messages and summaries are now printed to
790 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
791 * fixed inference of the upper bound of an iteration variable.
792 (reported by Rajat Datta).
793 * fixed bug 211e: calling local functions with duplicated constant
794 keyword argument no longer causes a bogus style warning about an
796 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
797 ftype declaration to be disregarded. (reported by Dave Roberts)
798 * fixed bug 373: caused by erronous compilation of references to alien
799 variables in the runtime on ppc/darwin.
800 * fixed bug 376: CONJUGATE type deriver.
801 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
802 (reported by Baughn on #lisp)
803 * a cross-compiler bug on non-x86 platforms has been identified and
804 fixed. (thanks to Bruno Haible)
805 * improvements to the MIPS runtime code for increased stability.
806 (thanks to Thiemo Seufer)
807 * increased the maximimum compact environment size to allow
808 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
809 * partial workaround for bug 135: don't name the function we're
810 calling for hairy cases of EVAL, so as not to accumulate one environment
811 entry per call to EVAL. (reported by Kevin Reid)
812 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
813 * optimization: type testing for non-vector arrays should be faster.
814 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
816 * bug fix: setting 31st element of a bit vector to zero did not work
818 * bug fix: redefining a class definition which failed due to a
819 previous accessor / function clash now works (but see BUGS entry
820 #380 for more problems in this area). (thanks to Zach Beane)
821 * the long form of DEFINE-METHOD-COMBINATION disables method group
822 checking when given a single method group with pattern *. (thanks
824 * on x86 compiler supports stack allocation of results of simple
825 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
826 * support for building and running on Mac OS X 10.4 "Tiger" has been added
827 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
828 * fixed some bugs related to Unicode integration:
829 ** the restarts for recovering from input and output encoding
830 errors only appear when there is in fact such an error to
832 * fixed some bugs revealed by Paul Dietz' test suite:
833 ** MISC.549 and similar: late transformation of unsafe type
834 assertions into derived types caused unexpected code
836 ** SCALE-FLOAT type deriver is less wrong.
837 ** type derivers for EXP, LOG and similar functions compute result
838 types for complex arguments better.
839 ** (MISC.563) CONJUGATE type deriver works for very restricted
841 ** out-of-line type testers for character strings are available.
842 ** EQUAL compiler transform understands specialness of objects
844 ** accessing double-floats stored on the stack now works on x86-64.
845 ** debugger internals could sometimes create invalid lispobjs,
846 resulting in GC crashes.
847 ** MISC.548: type check weakening can convert required type into
849 ** initialization forms for bindings are not in scope of free special
852 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
853 * incompatible change: thread support for non-NPTL systems has
854 been removed - locking is buggy and unreliable. A threaded
855 SBCL build will now warn at startup and refuse to create
856 new threads, unless futex support is detected in the kernel
857 * incompatible change: the top level REPL now has only an ABORT
858 restart associated with it, not TOPLEVEL and ABORT as it used to.
859 TOP and TOPLEVEL are now available as debugger commands for
860 returning to the top level.
861 * incompatible change: forms evaluated in the REPL now use the
862 global optimization policy.
863 * incompatible change: user- and system-initialization files are
864 no longer processed with LOAD, but by READ and EVAL; hence the
865 global optimization policy, startup package, readtable, etc,
867 * The .fasl file format number has been incremented because of
868 various incompatible changes.
869 * internal entry point details and argument counts no longer appear
870 in backtraces unless explicitly requested by setting
871 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
872 * built-in and standard functions no longer have names like "top
873 level local call to FOO".
874 * fixed bug 32: functions defined in non-null lexical environments
875 now have more legible printed representation
876 * fixed bug 33: functions defined in non-null lexical environemnts
877 are now more amenable to inspection by INSPECT.
878 * workaround for bug 354: XEPs no longer appear in backtraces unless
879 explicitly requested.
880 * fixed bug: receiving the signal which results from attempting to
881 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
882 SIGBUS on other *BSDs) on architectures where the C stack is also
883 the Lisp stack (x86 and x86-64) from foreign code no longer leads
884 to debugger confusion or wild instability. (reported by Cheuksan
886 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
887 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
888 (reported by Lutz Euler)
889 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
890 compiled code. (reported by Paul Dietz)
891 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
892 specialized parameter to a method to something that is not TYPEP
893 the specializer is now possible.
894 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
895 face of package deletion.
896 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
897 pathname. (reported by David Tolpin on c.l.l)
898 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
899 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
900 STRING-STREAM and FILE-STREAM, respectively.
901 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
902 than previously; in addition, its test suite is now run on build.
903 (thanks to Luke Gorrie)
904 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
905 * optimization: added a immediate representation for single-floats
907 * fixed some bugs revealed by Paul Dietz' test suite:
908 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
909 ** attempting to create a package with a colliding nickname causes
910 correctable errors to be signalled.
911 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
912 ** several x86-64 backend bugs related to sign-extension of immediate
915 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
916 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
917 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
919 * fixed loading of multiply forward-referenced layouts.
920 (thanks to Cheuksan Wang)
921 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
922 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
923 (see COMPILE-FILE documentation for details of :PRINT option)
924 * fixed bugs 19 and 317: fixed-format floating point printing is
925 more accurate. This also fixes a bug reported by Adam Warner
926 related to the ~@F format directive.
927 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
929 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
930 dispatch table if the from-char is a dispatch macro character.
931 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
932 by the wordsize no longer ignore the last word. (reported by Lutz
934 * fixed bug in type checking of dynamic-extent variables. (reported
936 * optimization: sequence traversal functions use their freedom to
937 coerce function designators to functions.
938 * optimization: code with many calls to CLOS methods specialized on
939 CLOS classes has had redundant type checks removed.
940 * optimization: added declarations to speed up operations that access
941 the internal character database (for example STRING-UPCASE)
942 * optimization: comparison operations between floats and sufficiently small
943 fixnums no longer create extra rationals
944 * fixed some bugs related to Unicode integration:
945 ** portions of multibyte characters at the end of buffers for
946 character-based file input are correctly transferred to the
947 start of the buffer at the next read.
948 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
949 passing it through to OPEN.
950 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
951 argument given to internal calls to OPEN.
952 ** the built-in comment readers (introduced by character sequences
953 ";" and "#|") are more forgiving to encoding errors; they will
954 STYLE-WARN and then attempt to resync the stream at a character
955 boundary. (thanks to Teemu Kalvas)
956 * fixed some bugs in the x86-64 port:
957 ** Negative short int return values from c-calls are sign-extended
959 ** The stack is aligned to 16-bytes for c-calls, as required by
960 the ABI. (thanks to Cheuksan Wang)
961 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
962 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
963 for enums. (thanks to Vincent Arkesteijn)
964 ** Multiple small optimizations and bugfixes for floating point
966 * fixed some bugs revealed by Paul Dietz' test suite:
967 ** Space, Tab, Linefeed, Return and Page have the invalid
968 secondary constituent character trait.
969 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
971 ** WITH-INPUT-FROM-STRING should only update the index place on
973 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
975 ** Bit-array operations (BIT-AND and similar) worked incorrectly
976 with one-dimensional arrays with fill pointers.
977 ** TYPE-OF failed on a complex with an integer realpart and a
979 ** compiler failure during type inference for the code of form
980 (IF (EQL X (THE ...)) ...) (MISC.535).
982 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
983 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
984 platform. The port supports 61-bit fixnums, large memory spaces
985 and reloading shared object files.
986 * enhancement: saving cores with foreign code loaded is now
987 supported on ppc/Darwin in addition to the previously supported
989 * enhancement: the statistical profiler now walks deeper into the
990 call stack for more meaningful call-graphs and accrued time
991 reports (x86/x86-64 only). It also now reports time spent in
993 * enhancement: it is now possible to trace most individual methods
994 of a generic function in addition to tracing the generic function
996 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
998 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
999 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1000 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1001 produces an error. (thanks to Vincent Arkesteijn)
1002 * bug fix: NAMESTRING on pathnames with :WILD components in their
1003 directories works correctly. (thanks to Artem V. Andreev)
1004 * fixed bug 125: compiler preserves identity of closures. (reported
1006 * bug fixed: functions with &REST arguments sometimes failed with
1007 "Undefined function" when compiled with (DEBUG 3). (reported by
1009 * bug fix: overflow during compiling of setting element of a bit
1010 vector with constant index and value. (reported by Timmy Douglas)
1011 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1012 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1013 * on x86 compiler partially supports stack allocation of dynamic-extent
1015 * GO and RETURN-FROM do not check the extent of their exit points
1016 when compiled with SAFETY 0.
1017 * fixed some bugs related to Unicode integration:
1018 ** encoding and decoding errors are now much more robustly
1019 handled; it should now be possible to recover even from invalid
1020 input or output to the terminal. (thanks to Teemu Kalvas)
1021 ** provided a first cut at implementing STRING-TO-OCTETS and
1022 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1023 ** altered the SB-MD5 contributed module slightly, changing the
1024 interface just enough for it to be supportable for builds where
1025 lisp characters are not eight bits.
1026 * fixed some bugs revealed by Paul Dietz' test suite:
1027 ** the FORMATTER-generated functions for ~V[ conditionals require
1028 the correct number of arguments.
1029 ** READ-FROM-STRING returns the mandated second value when applied
1030 to displaced strings.
1031 ** the #\Rubout and #\Backspace characters are treated as invalid
1032 constituent characters by the tokenizer.
1034 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1035 * new feature: reloading changed shared object files with
1036 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1037 * new feature: references to foreign variables and functions
1038 can now be compiled and loaded before the corresponding shared
1039 object file is loaded, as long as the foreign definitions are
1040 available at runtime.
1041 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1042 supported. (thanks to Dan Debertin)
1043 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1044 just keywords. (thanks to Vincent Arkesteijn)
1045 * fixed bug #331: structure-class instances corresponding to
1046 DEFSTRUCT forms are now created eagerly.
1047 * fixed bug #345: backtraces from calls to undefined functions work
1048 on x86 as well. Related bug #61 is now also partially fixed on x86
1049 (backtraces from throws to unknown catch tags.)
1050 * bug fix: lambda-list parsing is now stricter vrt. order and number
1051 of lambda-list keywords.
1052 * bug fix: as specified by AMOP, an error is signalled if a
1053 class-option appears multiple times in a DEFCLASS form. (reported
1055 * bug fix: RANDOM can be compiled when the compiler derives the type
1056 of its numeric argument as a disjoint set of small integers.
1057 (reported by Paul Dietz)
1058 * bug fix: starting a core saved with shared objects loaded when
1059 those objects are not available no longer causes threaded SBCL to
1060 hang. (reported by Sean Ross)
1061 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1063 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1064 argument, which must always be NIL. (reported by Kalle Niemitalo)
1065 * bug fix: printing 1.0d+23 no longer results in an error.
1066 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1067 * bug fix: structure slot setters preserve evaluation order. (thanks
1069 * fixed some bugs related to Unicode integration:
1070 ** RUN-PROGRAM can allow its child to take input from a Lisp
1071 stream. (reported by Stefan Scholl)
1072 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1073 locales. (reported by Ken Causey)
1074 * fixed some bugs revealed by Paul Dietz' test suite:
1075 ** INCF, DECF and REMF evaluate their place form as specified in
1077 ** FORMATTER expands ~{ ~} iteration directives with V or #
1078 parameters correctly.
1079 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1080 consequent uses no arguments correctly.
1081 ** the system has a partial understanding of the (COMPLEX RATIO)
1083 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1086 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1087 * new feature: a build-time option (controlled by the :SB-UNICODE
1088 keyword feature, enabled by default) for building the system with
1089 support for the entire 21-bit character space defined by the
1091 * new feature: the system now has rudimentary external-format
1092 support; the primary user-visible change at this time is that
1093 characters with the high bit set (such as lower-case-e-acute) will
1094 print correctly to a terminal in a UTF-8 environment.
1095 * minor incompatible change: BASE-CHAR no longer names a class;
1096 however, CHARACTER continues to do so, as required by ANSI.
1097 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1098 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1099 has been moved to the SB-EXT package (temporarily re-exported from
1101 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1102 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1104 * fixed bug #351: better error-handlind and reporting for malformed
1106 * fixed bug #350: bignum-printing is now more memory-efficient,
1107 allowing printing of very large bignums, eg. (expt 2 10000000).
1108 (reported by Bruno Haible)
1109 * fixed bug #302: better primitive-type selection for intersection
1111 * fixed bug #308: non-graphic characters now all have names, as
1112 required. (reported by Bruno Haible)
1113 * bug fix: redefining a class with different superclasses now correctly
1114 removes it from the direct-subclasses of its previous superclasses.
1115 (reported by David Morse)
1116 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1117 new value now works. (reported by Bruno Haible)
1118 * bug fix: correct canonicalization of multiple non-standard slot
1119 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1120 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1121 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1122 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1123 now exists, an signals an error.
1124 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1125 during macroexpansion for non-iteration variables as well. (reported
1126 by Bruno Haible for CMUCL)
1127 * bug fix: Cyclic structures and unprintable objects in compiler
1128 messages no longer cause errors. (reported by Bruno Haible)
1129 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1130 and READ-FROM-STRING all now return a primary value of NIL if
1131 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1132 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1133 (reported by Bruno Haible for CMUCL)
1134 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1135 do not designate a condition. (reported by Bruno Haible for
1137 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1138 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1139 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1140 permitting resolution in favour of any of the conflicting symbols.
1141 (reported by Bruno Haible for CMUCL)
1142 * bug fix: EQUAL compiler optimizations is less aggressive on
1143 strings which can potentially compare true despite having distinct
1144 specialized array element types.
1145 * bug fix: unit enumerations can be defined without dividing by
1146 zero. (thanks to Vincent Arkesteijn)
1147 * FORMAT compile-time argument count checking has been enhanced.
1148 (report from Bruno Haible for CMUCL)
1149 * a partial workaround for the bug 262: the compiler does not try to
1150 inline-expand a local function doing RETURN-FROM from a deleted
1151 BLOCK. (thanks to Peter Denno for the bug report and to David
1152 Wragg for the simple test case)
1153 * fixed some bugs revealed by Paul Dietz' test suite:
1154 ** RENAME-PACKAGE allows all package designators as new package
1156 ** constraint propagation and lambda variable substitution are
1157 more cautious in dealing with partially deleted code.
1158 ** compiler sometimes lost reoptimization passes.
1159 ** CERROR, given a condition as condition designator, passes its
1160 remaining arguments to the continue format control without
1162 ** Case-altering FORMAT directives work correctly on non-ASCII
1164 ** The REAL type specifier handles bounds outside the floating
1165 point ranges without signalling FLOATING-POINT-OVERFLOW.
1166 ** Functions with IR1-transformations can create intercomponent
1167 references to global functions.
1168 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1170 ** FORMAT ~R treats a nil value for its first parameter correctly.
1172 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1173 * enhancement: saving cores with foreign code loaded is now
1174 supported on x86/NetBSD and sparc/Linux in addition to the previously
1175 supported platforms.
1176 * bug fix: on some platforms repeated installations caused multiple
1177 copies of HTML documentation to be installed -- should not happen
1178 any more. (reported by Stefan Scholl)
1179 * bug fix: parsing self-recursive alien record types multiple times
1180 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1181 original patch by Helmut Eller for CMUCL)
1182 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1183 (thanks to Richard Kreuter)
1184 * bug fix: defining classes whose accessors are methods on existing
1185 generic functions in other (locked) packages no longer signals
1186 bogus package lock violations. (reported by François-René Rideau)
1187 * bug fix: special variables as DEFMETHOD parameters no longer have
1188 associated bogus type declarations. (reported by David Wragg and
1190 * bug fix: read-write consistency on streams of element-type
1191 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1192 * bug fix: redefiniton of the only method of a generic function with
1193 no DEFGENERIC no longer emits a full WARNING. In addition,
1194 redefinition of generic functions with no DEFGENERIC to an
1195 incompatible lambda list now signals an error. (thanks to Zach
1197 * bug fix: DEFGENERIC now works even when there's a function of the
1198 same name in an enclosing lexical environment. (thanks to Zach
1200 * fixed compiler failure, caused by instrumenting code during
1201 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1202 * optimization: added loop analysis and improved register allocation
1203 * fixed some bugs revealed by Paul Dietz' test suite:
1204 ** POSITION on displaced vectors with non-zero displacement
1205 returns the right answer.
1206 ** (SIMPLE-STRING) is a valid type specifier for sequence
1208 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1210 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1211 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1213 ** The justification version of the FORMAT ~< directive treats
1214 non-zero minpad parameter correctly.
1216 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1217 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1218 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1219 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1220 the supported interface.
1221 * new feature: Single-stepping of code compiled with DEBUG 2 or
1222 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1223 * new feature: saving cores with foreign code loaded is now
1224 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1225 Timothy Moore's work for CMUCL)
1226 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1227 parameters to * instead of NIL if no initform is supplied.
1228 (reported by Johan Bockgaard)
1229 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1230 correctly when FUNCALL appears as the car of the form. Note:
1231 despite this FUNCALL forms are not currently subject to
1232 compiler-macro expansion. (port of Raymond Toy's fix for the
1233 same from CMUCL, reported by Johan Bockgaard)
1234 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1235 (thanks for Teemu Kalvas)
1236 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1237 case instead of terminating the iteration (thanks for Julian
1238 Squires, Sean Champ and Raymond Toy)
1239 * bug fix: incorrect expansion of defgeneric that caused a style
1240 warning. (thanks for Zach Beane)
1241 * bug fix: slot accessor effective method computation works properly
1242 for classes with multiple non-standard applicable methods on
1243 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1244 * on x86 compiler supports stack allocation of results of LIST and
1245 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1246 CMUCL implementation by Gerd Moellmann)
1247 * fixed some bugs revealed by Paul Dietz' test suite:
1248 ** FORMAT strings with both the ~<~:;~> form of the justification
1249 directive and pretty-printing directives cause an error.
1251 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1252 * incompatible change: the internal functions
1253 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1254 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1255 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1256 instead of the old functions.
1257 * new feature: on platforms where "dladdr" is available foreign
1258 function names now appear in backtraces. (based on Helmut Eller's
1260 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1261 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1263 * documentation: ASDF manual is now distributed with SBCL.
1264 * bug fix: SBCL can now load its contributed modules with REQUIRE
1265 even if the system-provided entries have been removed from
1266 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1268 * bug fix: backtraces involving undefined functions or assembly
1269 routines are more informative. (thanks to Brian Downing)
1270 * bug fix: mutually referent alien structures now work correctly.
1271 (reported by Rick Taube)
1272 * bug fix: structures defined by WITH-ALIEN can be referred to
1273 within other definitions in the same WITH-ALIEN.
1274 * bug fix: division operators (MOD, TRUNCATE and the like) with
1275 constant zero divisors and integer dividends no longer generate
1277 * bug fix: provide default methods for INPUT-STREAM-P and
1278 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1279 * bug fix: improve the detection and resolution of MOP metacycles
1280 (where effective method computation depends on the generic function
1281 having its effective method computed). (reported by Bruno Haible)
1282 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1283 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1284 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1285 * bug fix: pathnames with relative directory components can be
1286 represented relative to default pathnames.
1287 * optimization: in taking the GCD of bignums, reduce the two bignums
1288 to approximately the same size (using Euclid's algorithm) before
1289 applying the more sophisticated binary GCD. (thanks to Juho
1291 * optimization: COUNT on bitvectors now operates word-at-a-time.
1292 * optimization: ASH with a positive, but not necessarily constant,
1293 (leftwards) shift, when in a modular context, is compiled to a
1295 * fixed some bugs revealed by Paul Dietz' test suite:
1296 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1297 if the corresponding argument is NIL.
1298 ** FORMAT directives accepting a minpad parameter treat negative
1300 ** literal commas as character parameters to FORMAT directives are
1302 ** literal spaces directly after ~<Newline> directives within a
1303 format-logical-block (~:< ~@:>) do not induce :FILL-style
1304 conditional newlines.
1305 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1306 as specified: it no longer includes conditional newlines).
1307 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1309 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1310 :PER-LINE-PREFIX argument does not evaluate to a string.
1312 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1313 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1314 the manual for details; for now, package locks can be disabled by
1315 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1316 if no major problems are found then it is likely that they will be
1317 compiled in unconditionally.
1318 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1319 unsupported operators on all platforms. To load a shared library
1320 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1321 object file, link it into a shared library outside of SBCL and
1322 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1324 * bug workaround: ROOM T and the GC were not working together
1325 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1326 reporting facility aren't preserved. That reporting has been
1327 disabled, so now until and unless someone figures out how to make it
1328 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1329 * minor incompatible change: as threatened around sbcl-0.8.0, the
1330 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1331 The symbols are also exported from SB-PCL for backwards
1332 compatibility, but more so than before SB-PCL should be treated as
1333 an implementation-internal package.
1334 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1336 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1337 It is known as of this release not to work on the Alpha, however.
1338 * fixed bug #167: errors signalled due to illegal syntax in method
1339 bodies are now more legible.
1340 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1341 designators and can hence be used with TYPEP.
1342 * fixed bug #333: CHECK-TYPE now ensures that the type error
1343 signalled, if any, has the right object to be accessed by
1344 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1345 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1346 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1348 * fixed bug #334: programmatic addition of slots using specialized
1349 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1350 and :ALLOCATION :CLASS slots.
1351 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1352 float exponents. (rereported by Peter Seibel)
1353 * fixed a bug: #\Space (and other whitespace characters) are no
1354 longer considered to be macro characters in standard syntax by
1355 GET-MACRO-CHARACTER.
1356 * fixed bug: initialization of condition class metaobjects no longer
1357 causes an instance of the condition to be created. (reported by
1359 * fixed bug: it is now possible to have slots such that
1360 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1361 description is neither :INSTANCE nor :CLASS.
1362 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1363 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1364 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1365 non-standard SLOT-DEFINITION-ALLOCATION values.
1366 * fixed bugs: various race conditions handling exiting threads.
1367 CL-PPCRE's thread torture test now passes.
1368 * fixed arguably-a-bug: GC time proportional to number of threads in
1369 system even when most of them are idle
1370 * optimization: improved performance of BIT and SBIT on bit-vectors.
1371 * .fasl file incompatibility: The fasl file version number has
1372 been incremented because of changes associated with package locks.
1374 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1375 * minor incompatible change: the system no longer provides
1376 optimization hints (conditions which are TYPEP
1377 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1378 mode; these hints are still emitted when higher SPEED optimization
1380 * new contrib module: a sampling profiler (profiling by statistical
1381 sampling, rather than by instrumenting functions) is available as
1382 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1383 * the behaviour of the standard function ED is now customizeable by
1384 third parties through a hook variable: see ED's documentation
1385 string for information on the protocol.
1386 * the compiler no longer emits efficiency notes for (FUNCALL X)
1387 when the type of X is uncertain under default optimization
1389 * fixed bug 276: mutating a binding of a specialized parameter to a
1390 method to something that is not TYPEP the specializer is now
1392 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1393 exhibit the required behaviour on the x86 platform. (thanks to
1394 Peter van Eynde, Eric Marsden and Bruno Haible)
1395 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1396 even for difficult arguments. (reported by Peter Graves)
1397 * fixed bug 141a: the backquote printer now descends quoted
1399 * fixed another bug in backquote printing: no more destructive
1400 modification of the form's list structure. (reported by Brian
1402 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1403 move between its address being taken and the call to
1404 interrupt_thread, fixing a crashing race condition.
1405 * the SB-POSIX contrib implementation has been adjusted so that it
1406 no longer exhibits ridiculously poor performance when constructing
1407 instances corresponding to C structs.
1409 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1410 * minor incompatible change: the sb-grovel contrib now treats C
1411 structures as alien (in the sense of SB-ALIEN) objects rather than
1412 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1413 has implications for memory management of client code
1414 (sb-grovel-returned objects must now be manually managed) and for
1415 type safety (alien objects now have full types).
1416 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1417 used to control emission of compiler diagnostics, rather than the
1418 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1419 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1420 quality should be considered deprecated.
1421 * install.sh now installs the user manual as well
1422 * (not quite a new documentable feature, but worth considering in
1423 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1424 beginnings of a semantically meaningful condition hierarchy is
1425 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1427 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1428 designator as the defaults argument.
1429 * fixed bug: Displaced arrays whose displaced-to array has become
1430 too small now cause ARRAY-DIMENSION to signal an error, providing
1431 for safer bounds-checking. (reported by Bruno Haible)
1432 * fixed bug: DEFCLASS slot definitions with identical :READER and
1433 :WRITER names now signal a reasonable error. (reported by Thomas
1435 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1437 * fixed bug: Invalid slot specification errors now print correctly.
1438 (thanks to Zach Beane)
1439 * fixed bug 320: Shared to local slot value transfers in class
1440 redefinitions now happen corrently from superclasses as
1441 well. (reported by Bruno Haible)
1442 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1444 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1445 improper lists correctly. (reported by Bruno Haible)
1446 * fixed bug 313: source-transform for <fun-name> was erroneously
1447 applied to a call of a value of a variable with name <fun-name>.
1448 (reported by Antonio Menezes Leitao)
1449 * fixed bug 307: The obsolete instance protocol ensures that
1450 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1451 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1452 remove dead unknown-values globs from the middle of the stack.
1453 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1454 easier to resume long computations after using *BREAK-ON-SIGNALS*
1455 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1456 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1457 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1458 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1459 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1460 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1462 * fixed a (fairly theoretical) bug in string printing: if
1463 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1464 does not have array-element-type equal to the most general string
1466 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1467 function name. (thanks to Nikodemus Siivola)
1468 * fixed bug: compile-time format string checker failed on
1469 non-closed ~{. (reported by Thomas F Burdick)
1470 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1471 don't warn on legal loop code involving more than one aggregate
1472 boolean. (thanks to Nikodemus Siivola)
1473 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1474 in keyword position to MAKE-PACKAGE was not being reported as
1476 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1477 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1479 * fixed bugs 280 and 312: the checking for multiple definitions in a
1480 file is less likely to become confused by uses of inline
1482 * fixed bug: the #S reader macro performs the keyword coercion
1483 specified for slot names. (reported by Kalle Niemitalo)
1484 * fixed bug: lambda lists may contain symbols whose names start with
1485 & but are not lambda-list-keywords; their occurrence triggers a
1487 * fixed bug 321: define-method-combination argument lambda lists do
1488 not cause invalid code to be generated when &optional or &aux
1489 variables have default values. (reported by Bruno Haible)
1490 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1491 have CLOS classes; this appears also to have fixed reported
1492 instability in stack exhaustion detection.
1493 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1494 MUFFLE-WARNING when no associated restart is present is now
1496 * optimization: rearranged the expansion of various defining macros
1497 so that each expands into only one top-level form in a
1498 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1500 * optimization: used a previously little-used slot in symbols to
1501 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1503 * fixed some bugs revealed by Paul Dietz' test suite:
1504 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1506 ** Fixed an optimization bug related to inheritance of initforms
1507 from local to shared slots.
1508 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1509 ** CAST optimizer forgot to flush argument derived type.
1510 ** print/read consistency on floats is now orders of magnitude
1511 more likely. (thanks also to Bruno Haible for a similar report
1513 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1514 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1515 for objects of type REAL. Make it so.
1516 ** SXHASH is specified (infelicitously) to respect similarity,
1517 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1518 it so. (thanks to Markus Ziegler)
1519 ** on the Alpha, the compiler succeeds in compiling functions
1520 returning a known number of arguments greater than 63.
1521 ** fixed handling of invalid NIL arguments in keyword position
1523 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1525 ** provided a readably-printable representation for RANDOM-STATE
1527 ** ensured that pathnames, where they have a namestring, always
1528 print using #P"..." syntax.
1530 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1531 * Support for the forthcoming 2.0 version of the NetBSD kernel
1532 running on x86 hardware has been added. (thanks to Perry
1533 E. Metzger most immediately, and others for their past work)
1534 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1535 current release version; SBCL's OpenBSD support had been broken
1536 since about the time of OpenBSD's switch to ELF binary format.)
1537 * [placeholder for DX summary]
1538 ** user code with &REST lists declared dynamic-extent, under high
1539 speed or space and low safety and debug optimization policy.
1540 * The manual has been converted to Texinfo format and the debugger
1541 chapter from the cmucl manual has been added.
1542 * A facility has been added to extract documentation strings from
1543 sbcl and store them as Texinfo-formatted snippets for inclusion in
1544 the manual (via Texinfo's @include directive)
1545 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1546 the test case to Dave Roberts)
1547 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1548 pointers. (reported by Sean Ross)
1549 * bug fix: PROFILE output is printed nicely even for large numerical
1550 values. (thanks to Zach Beane)
1551 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1552 greater than 32 handle EOF correctly.
1553 * bug fix: on X86 an immediate argument of the IMUL instruction is
1554 correctly printed in disassembly. (thanks to Lutz Euler)
1555 * bug fix: class slots in redefined classes preserve their old
1556 values. (thanks to Bruno Haible and Nikodemus Siivola)
1557 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1558 succeeds. (reported by Marco Baringer)
1559 * fixed some bugs revealed by Paul Dietz' test suite:
1560 ** READ-SEQUENCE now works on ECHO-STREAMs.
1561 ** RATIONALIZE works more according to its specification. (thanks
1564 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1565 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1566 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1567 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1568 should matter to you only if you rebind the printer control
1569 variables and then find you want different bindings in the
1570 debugger than in the ordinary execution of your program.)
1571 * The runtime build system has been tweaked to support building
1572 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1573 assembler and linker. (thanks to Nikodemus Siivola)
1574 * Unbound, undefined, undeclared variables now trigger full
1575 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1576 more often programmer error than deliberate exploitation of undefined
1578 * optimization: the hash algorithm for strings has changed to one
1579 that is less vulnerable to spurious collisions. (thanks to Juho
1581 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1582 needless bounds checking. (thanks to Juho Snellman)
1583 * optimization: implemented multiplication as a modular
1584 (UNSIGNED-BYTE 32) operation on the PPC backend.
1585 * fixed some bugs revealed by Paul Dietz' test suite:
1586 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1588 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1589 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1590 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1592 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1594 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1596 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1597 their output stream on EOF from read.
1598 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1599 have been read to end-of-file.
1600 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1602 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1603 description of determination of which consecutive characters
1605 ** Printing the "Space" character with escaping on now yields "#\\ ",
1606 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1607 ** Reading floating-point numbers with *READ-BASE* set to a number
1608 less than 10 works correctly.
1609 ** Reading floating-point numbers with *READ-BASE* set to a number
1610 more than 10 works correctly.
1611 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1612 the readtable currently in effect.
1614 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1615 * minor incompatible change: parsing of namestrings on a physical
1616 (Unix) host has changed; numbers after the final #\. in a
1617 namestring are no longer interpreted as a version field. This is
1618 intented to be largely invisible to the user, except that the
1619 meaning of the namestring "*.*.*" has changed: it now refers to a
1620 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1621 should usually be replaced by
1622 (make-pathname :name :wild :type :wild :version :wild)
1623 with the added benefit that this is more likely to be portable.
1624 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1625 option to OPEN now signals an error if the file being opened
1626 exists; this may have an impact on existing code.
1627 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1628 signals received, on the PowerPC platforms (both Linux and
1629 Darwin). (thanks to Pierre Mai for pointing out the location of
1631 * several fixes on OS X: The system now builds and runs cleanly on
1632 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1633 (thanks to Brian Mastenbrook)
1634 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1635 second-resolution: integer multiples of 1/3600 between -24 and 24.
1636 (thanks to Vincent Arkesteijn)
1637 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1638 type when called with 1 argument; PEEK-CHAR checked type of
1639 PEEK-TYPE only after having read first character from a
1640 stream. (reported by Peter Graves)
1641 * bug fix: the garbage collector now has much better locality
1642 behaviour, and in particular no longer treats all memory as being
1643 exhausted when many small objects point to each other in a deeply
1645 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1646 recognized as being TYPEP their class.
1647 * bug fix: the PUSHNEW documentation string has been corrected.
1648 (thanks to Vincent Arkesteijn)
1649 * bug fix: defaulting of the value for the last of an atypically
1650 large number of multiple values being bound was not being
1651 performed correctly on the Alpha or PPC platforms
1652 * optimization: implemented multiplication as a modular
1653 (UNSIGNED-BYTE 32) operation on the x86 backend.
1654 * optimization: SEARCH on simple-base-strings can now be open-coded.
1655 (see also contrib/compiler-extras.lisp for inspiration for
1656 teaching the compiler about the Boyer-Moore algorithm).
1657 * value, returned by MAX (and MIN) called with several EQUALP, but
1658 not EQL, arguments now does not depend on compiler settings.
1659 * fixed some bugs revealed by Paul Dietz' test suite:
1660 ** in stack analysis liveness information is propagated from
1661 non-local entry points.
1662 ** pathwise CAST removing failed when the CAST node did not start
1664 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1665 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1667 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1668 host is already defined.
1669 ** RENAME-FILE works on streams instead of signalling an internal
1671 ** PEEK-CHAR uses the current readtable when determining whether
1672 or not a character is whitespace.
1673 ** MERGE-PATHNAMES handles the case when the pathname does not
1674 specify a name while the default-pathname specifies a version.
1675 ** Pathnames now stand a better chance of respecting print/read
1677 ** Attempting to use standardized file system operators with a
1678 pathname with invalid :DIRECTORY components signals a
1680 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1682 ** DIRECTORY on logical pathnames is more correct.
1683 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1684 signal a TYPE-ERROR if their argument is not a stream.
1685 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1686 designator argument does not designate a stream.
1687 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1688 examining the synonym.
1689 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1691 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1692 element-type larger than ([UN]SIGNED-BYTE 32).
1694 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1695 * When built with the :SB-FUTEX feature, threaded builds now take
1696 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1697 for faster/more reliable mutex and condition variable support.
1698 * Incompatible change (but one you probably shouldn't have been using
1699 anyway): the interface and code for arbitrating between multiple
1700 threads in the same user session has been redesigned.
1701 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1702 error when called without an explicit environment argument.
1703 (thanks to Nikodemus Siivola)
1704 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1705 confused about their position. (thanks to Adam Warner and Gerd
1707 * bug fix: Pretty printing backquoted forms with unquotations in the
1708 argument list position of various code constructs such as LAMBDA
1709 now works correctly. (reported by Paul Dietz)
1710 * bug fix: Pretty printing unquotations no longer loses all
1711 stream position information.
1712 * optimization: performance of string output streams is now less
1713 poor for multiple small sequence writes.
1714 * optimization: performance of CSUBTYPEP in the presence of complex
1715 expressions involving CONS and NOT many times has been improved.
1716 (reported by Paul Dietz)
1717 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1719 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1720 streams for the REPL, for the debugger, and so on. (thanks to
1722 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1723 * fixed some bugs revealed by Paul Dietz' test suite:
1724 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1725 is ignored if it is too small, rather than propagating through
1726 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1727 ** extremely complex negations of CONS types were not being
1728 sufficiently canonicalized, leading to inconsistencies in
1730 ** VALUES tranformer lost derived type.
1732 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1733 * fixed a bootstrapping bug: the build process no longer assumes
1734 that the various BOOLE-related constants have the same value in
1735 host and target lisps. (noted by Paul Dietz' test suite on an
1736 SBCL binary built from CLISP)
1737 * The system can now be dynamically linked on the MIPS platform,
1738 which enables dynamic loading of foreign code from Lisp. (thanks
1739 to Ralf Baechle for discussions on the MIPS ABI)
1740 * The system now records debugging information for its own source
1741 files in a filesystem-position-independent manner, relative to
1742 the "SYS" logical host.
1743 * fixed a compiler bug: MV-LET convertion did not check references
1744 to the "max args" entry point. (reported by Brian Downing)
1745 * tweaked disassembly notes to be less confident about proclaiming
1746 some instruction as an LRA. (thanks to Brian Downing)
1747 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1748 now each have their own history, command character, and other
1749 characteristics. (thanks to David Lichteblau)
1750 * fixed some bugs revealed by Paul Dietz' test suite:
1751 ** compiler failure in compiling LOGAND expressions including a
1753 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1754 shift greater than 32.
1755 ** FLUSH-DEST did not mark blocks for type check regeneration.
1756 ** HANDLER-CASE failed to accept declarations in handler clauses
1757 in some circumstances.
1759 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1760 * New code in contrib/sb-introspect (still probably not entirely
1761 stable yet) provides some support for smart Lisp development
1762 environments like SLIME.
1763 * The conditions signalled for errors occurring when loading .fasl
1764 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1765 in a way which should help ASDF recover gracefully.
1766 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1767 stuff has been cleaned up. If you code contrib/ stuff, this might
1768 affect you, and you can look at contrib/README, contrib/STANDARDS,
1769 and/or the 0.8.4.27 diff to check.
1770 * In full calls the compiler now does not generate checks for declared
1771 argument types for all arguments.
1772 * various threading fixes
1773 ** and some experimental patches which didn't make it into
1774 the main tree for this release, but which are shipped in
1775 contrib/experimental-thread.patch as a possible fix for some
1776 failures (deadlock, spinning...) in GC-intensive multithreaded
1778 * fixed PPC build problem (source code incompatibility of different
1779 library versions): added offsetof() hackery which attempts to divine
1780 where glibc maintainers put uc_mcontext today
1781 * fixed bug 282: compiler does not trust type assertions while passing
1782 arguments to a full call.
1783 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1784 &OPTIONAL VALUES type parameter.
1785 * fix bug 214: algorithm for noting rejected templates is now more
1786 similar to that of template seletion. (also reported by rydis on
1788 * fixed bug 141b: printing backquoted information readably and prettily
1789 inserts a space where necessary.
1790 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1791 (CHARACTER 10) are now reported as errors, rather than propagated
1792 as unknown types. (reported by piso on #lisp)
1793 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1794 with values NIL and :ERROR. (thanks to Milan Zamazal)
1795 * fixed bug 191c: CLOS now does proper keyword argument checking as
1796 described in CLHS 7.6.5 and 7.6.5.1.
1797 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1798 longer raise an error; further, using a list as a for-as-arithmetic
1799 counter now raises a meaningful error.
1800 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1801 now understood by sequence creation functions such as MAKE-SEQUENCE
1803 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1804 a STRING-INPUT-STREAM.
1805 * compiler enhancement: SIGNUM is now better able to derive the type
1807 * type declarations inside WITH-SLOTS are checked. (reported by
1809 * fixed some bugs revealed by Paul Dietz' test suite:
1810 ** incorrect optimization of TRUNCATE for a positive first
1811 argument and negative second.
1812 ** compiler failure in let-convertion during flushing dead code.
1813 ** compiler failure while deriving type of TRUNCATE on an
1814 interval, containing 0.
1815 ** ASH of a negative bignum by a negative bignum count now returns
1817 ** intersection of CONS types now canonicalizes properly, fixing
1818 inconsistencies in SUBTYPEP.
1820 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1821 * incompatible change: The --disable-debugger command line
1822 option now clobbers the debugger at a more fundamental
1823 level, by redefining #'INVOKE-DEBUGGER instead of by
1824 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1825 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1826 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1827 enter the debugger REPL and then suffer recursive errors
1828 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1829 while under the new system, BREAK in --disable-debugger mode
1830 terminates the system just as an unhandled error would.
1831 * fixed compiler performance when processing loops with a step >1;
1832 * bug fix: DOCUMENTATION now retrieves generic function
1833 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1834 support has been systematized, and now supports the methods
1835 specified by ANSI, along with a default method and a method for
1836 slot documentation. (reported by Nathan Froyd)
1837 * bug fix: effective methods associated with a generic function are
1838 no longer cached over a change of that generic function's method
1839 combination. (reported by Andreas Fuchs)
1840 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1841 names a non-existent directory. (thanks to Andreas Fuchs)
1842 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1843 the PPC platform, lead to stack corruption; this has been fixed.
1844 (reported by Rainer Joswig)
1845 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1846 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1847 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1848 arithmetic in general, is now much more reliable on the PPC
1850 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1851 platform now returns the right answer.
1852 * optimization: restored some effective method precomputation in
1853 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1854 precomputation is now tunable.
1855 * optimization: compiler-internal data structure use has been
1856 reviewed, and changes have been made that should improve the
1857 performance of the compiler by about 20%.
1858 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1859 simple-base-strings and simple-bit-vectors is improved.
1860 * optimization: the optimization of 32-bit logical and arithmetic
1861 functions introduced in version 0.8.3 on the x86 has been
1862 implemented on the mips, ppc and sparc platforms; an
1863 implementation of the same facility, but for 64-bit arithmetic,
1864 has been added for the alpha.
1865 * microoptimization: the compiler is better able to make use of the
1866 x86 LEA instruction for multiplication by constants.
1867 * bug fix: in some situations compiler did not report usage of
1868 generic arithmetic in (SPEED 3) policy.
1869 * bug 145b fix: compiler used wrong type specifier while converting
1870 MEMBER-types to numeric.
1871 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1873 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1874 index argument now works properly on the Alpha platform.
1875 * bug fix: floating point exception treatment on the Alpha platform
1877 * bug fix: FILE-POSITION works much better on string input and
1878 output streams. (thanks to Nikodemus Siivola)
1879 * bug fix: many threading/garbage collection symptoms sorted.
1880 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1881 might be pseudo-atomic.
1882 * internal change: Stopping for GC is now done with signals not
1883 ptrace. GC is now done in whichever thread wanted it, instead of
1885 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1887 * bug fix: result form in DO is not contained in the implicit
1889 * incompatible change: ICR structure is changed; the value part of
1890 CONTINUATION is now called LVAR; corresponding functions are
1891 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1892 * added type deriver for ISQRT (thanks to Robert E. Brown).
1893 * bug fix: better support for loading from the command line when an
1894 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1896 * fixed some bugs revealed by Paul Dietz' test suite:
1897 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1898 ** ROUND and FROUND now give the right answer when given very
1899 small float arguments.
1900 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1902 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1903 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1904 ** LCM should return a non-negative integer.
1905 ** PARSE-INTEGER returned the index of a terminator instead of the
1906 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1907 ** PARSE-INTEGER returned an incorrect index being applied to a
1909 ** LCM with two arguments of 0 returns 0 rather than signalling
1911 ** unsigned addition of a 32-bit constant with the high bit set no
1912 longer causes an internal compiler error on the x86.
1913 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1914 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1915 with negative last argument.
1916 ** byte specifiers with zero size and position no longer cause
1917 an error during type derivation.
1918 ** bignum multiplication on the Alpha platform now returns the
1920 * porting: The system now builds on SuSE AMD64, although it still
1921 generates a 32-bit binary.
1922 * .fasl file incompatibility: The fasl file version number has
1923 been incremented (because of the changes to internal compiler
1924 data structures referred to above).
1926 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1927 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1928 more accurately, on the Darwin kernel running on PowerPC hardware.
1929 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1930 * Compiler code deletion notes now signal a condition of type
1931 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1932 an associated MUFFLE-WARNING restart.
1933 * The compiler now performs limited argument count validation of
1934 constant format strings in FORMAT, and where appropriate in ERROR,
1935 CERROR and WARN. (thanks to Gerd Moellmann)
1936 * New ASDF-INSTALL contrib can be used for automatic download and
1937 installation of third-party Lisp code from CCLAN or other sites
1939 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1940 which forces another thread to execute a function supplied by the
1942 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1943 now accept and act upon their :ELEMENT-TYPE keyword argument.
1944 (reported by Martin Atzmueller, Edi Weitz)
1945 * bug fix: FILE-POSITION now accepts position designators up to
1946 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1947 is the greater. (thanks to Patrik Nordebo)
1948 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1949 Kalle Olavi Niemitalo)
1950 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1951 printable. (reported by Eric Marsden)
1952 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1953 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1954 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1955 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1956 (thanks to Stig Sandoe)
1957 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1958 UNREAD-CHAR now correctly includes the unread character in the
1959 target sequence. (thanks to Gerd Moellmann)
1960 * bug fix in threaded builds: the system can now be suspended and
1961 resumed by shell job control with minimal disruption.
1962 * bug fixes in times and timezones >2038AD
1963 * better handling of "where is GNU make?" problem in build scripts
1964 (thanks to Nikodemus Siivola)
1965 * new optimization: inside a named function any reference to a
1966 function with the same name is considered to be a self-reference;
1967 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1968 optimization quality.
1969 * new optimization on x86: logical functions and + now have
1970 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1971 used when the result is truncated to 32 bits.
1972 * VALUES declaration is partially enabled.
1973 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1974 * fixed some bugs revealed by Paul Dietz' test suite:
1975 ** The system now obeys the constraint imposed by
1976 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1977 types form a lattice under type intersection.
1978 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1979 ** ASSOC now ignores NIL elements in an alist.
1980 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1981 and (1+ MOST-POSITIVE-FIXNUM) answers.
1982 ** The addition of a method with invalid qualifiers to a generic
1983 function does not cause an error to be signalled immediately;
1984 a warning is signalled, and the error is generated only on
1985 calling the generic function.
1986 * changed .fasl file version number, in part to add type codes for
1987 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1988 obscure ANSI requirements
1990 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1991 * fixed bug 148: failure to inline-expand a local function left
1992 garbage, confusing the compiler.
1993 * fixed bugs 3cd: structure slot readers perform type check if the
1994 slot can have an invalid value (i.e. it is either not initialized
1995 or can be written with a less specific slot writer).
1996 * bug fix: the compiler now traps array references to elements off
1997 the end of an array; previously, the bounds checking in some
1998 circumstances could go off-by-one.
1999 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2001 * type declarations for array element types now obey the description
2002 on the CLHS page "Declaration TYPE", as per discussions on
2003 sbcl-help around 2003-05-08. This means that a declaration
2004 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2005 declaration, all references to BAR will be asserted or assumed
2006 (with THE, so dependent on compiler policy) to involve objects of
2007 type FOO. Note that no such declaration is implied in
2008 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2009 * declared types of functions from the "Conditions"
2010 chapter. (reported by Paul Dietz)
2011 * bug fix: CERROR accepts a function as its first argument.
2012 * bug fix: NTH an NTHCDR accept a bignum as index
2013 arguments. (reported by Adam Warner)
2014 * optimization: character compare routines now optimize comparing
2015 against a constant character. (reported by Gilbert Baumann)
2016 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2017 argument now works properly on the MIPS platform.
2018 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2019 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2020 declared to have a functional type, does not cause an error
2022 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2023 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2024 * fixed bug: if last continuation of a deleted block has a
2025 destination, this destination should be deleted too. (reported by
2027 * fixed a bug in the bootstrap process: the host compiler's values
2028 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2029 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2030 test case from Patrik Nordebo)
2031 * improved the ability of the disassembler on the PPC platform to
2032 provide helpful disassembly notes.
2033 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2034 the class in more cases than previously.
2035 * bug fix: FILE-POSITION now understands :START and :END for
2036 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2037 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2038 8) data. (thanks to David Lichteblau)
2039 * bug fix: it is possible to add a method to a generic function
2040 without lambda list.
2041 * bug fix: reader failed to signal END-OF-FILE inside an
2042 object representation. (reported by Nikodemus Siivola)
2043 * fixed some bugs revealed by Paul Dietz' test suite:
2044 ** LAST and [N]BUTLAST should accept a bignum.
2045 ** condition slot accessors are methods.
2046 ** (VECTOR NIL) is a subtype of STRING.
2048 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2049 * minor incompatible change: some nonsensical specialized lambda
2050 lists (used in DEFMETHOD) which were previously ignored now signal
2052 * minor incompatible change: the system is now aware of the types of
2053 variables in the COMMON-LISP package, and will signal errors for
2054 most violations of these type constraints (where previously they
2055 were silently accepted).
2056 * minor incompatible change: COMPILE-FILE now uses the freedom
2057 afforded (ANSI 3.2.2.3) to use derived function types for
2058 functions defined in the same file. This also permits the system
2059 to warn on static type mismatches and function
2060 redefinition. (Currently it does not work with high DEBUG level.)
2061 * minor incompatible change: VALUES declaration is disabled.
2062 * When issuing notes, the compiler now signals a condition of type
2063 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2064 restart for use in user handlers. It is expected that the
2065 COMPILER-NOTE condition will eventually become a condition
2066 supertype to a hierarchy of note types, which will then be
2067 handleable in a similar fashion. However, at the moment, no such
2068 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2069 but it's an implementation detail, not a classification for the
2071 * Changes in type checking closed the following bugs:
2072 ** type checking of unused values (192b, 194d, 203);
2073 ** template selection based on unsafe type assertions (192c, 236);
2074 ** type checking in branches (194bc).
2075 * A short form of VALUES type specifier has ANSI meaning (it has
2076 increased the number of situations when SBCL cannot perform type
2078 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2079 keywords or constants is permissible.
2080 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2081 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2082 classes. (thanks to Antonio Martinez)
2083 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2084 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2085 argument to a pretty-print stream. Instead, it leaves any such
2086 operations to DESCRIBE-OBJECT methods.
2087 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2089 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2090 lambda lists. (thanks to David Lichteblau)
2091 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2092 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2093 (thanks to Gerd Moellmann)
2094 * fixed bug 246: increased compilation speed of long
2095 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2097 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2099 * DEFCONSTANT now throws a condition of type
2100 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2101 constant to a non-EQL value; CONTINUE and ABORT restarts
2102 respectively change and preserve the value.
2103 * fixed bug 63: The code walker, part of the implementation of CLOS,
2104 is now better at handling symbol macros.
2105 * bug fix: There is no longer an internal implementation type named
2106 CL:LENGTH. (reported by Raymond Toy)
2107 * bug fix: In macro-like defining macros/special operators the
2108 implicit block does not enclose the lambda list.
2109 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2110 enclosed in parentheses) are not suitable as type specifiers, and
2111 their use properly signals an error now.
2112 * bug fix: An argument count mismatch for a type specifier in code
2113 being compiled no longer causes an unhandled error at compile
2114 time, but signals a compile-time warning.
2115 * fixed simple vector readable printing
2116 * bug fix: DESCRIBE takes more care over whether the class
2117 precedence list slot of a class is bound before accessing it.
2118 (reported by Markus Krummenacker)
2119 * bug fix: FORMATTER can successfully compile pretty-printer format
2120 strings which use variants of the ~* directive inside.
2121 * bug fix: SEARCH now applies its TEST predicate to the elements of
2122 the arguments in the correct order. (thanks to Wolfhard Buss)
2123 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2124 in type inference. (thanks to Robert E. Brown)
2125 * bug fix: Reading in symbols with an explicit package name of ""
2126 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2127 * fixed some bugs revealed by Paul Dietz' test suite:
2128 ** NIL is now allowed as a structure slot name.
2129 ** Arbitrary numbers, not just REALs, are allowed in certain
2130 circumstances in LOOP for-as-arithmetic clauses.
2131 ** Multiple class redefinitions before slot access no longer
2132 causes a type error.
2133 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2134 association between the name and a class.
2135 ** Generic functions with non-standard method-combination and over
2136 five methods all of which return constants no longer return NIL
2137 after the first few invocations. (thanks to Gerd Moellmann)
2138 ** CALL-NEXT-METHOD with no arguments now passes the original
2139 values of the arguments, even in the presence of assignment.
2140 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2142 ** DISASSEMBLE works with closures and funcallable instances.
2143 ** ADD-METHOD now returns the generic function, not the new method.
2144 ** FIND-METHOD signals an error if the lengths of the specializers
2145 is incompatible with the generic function, even if the ERRORP
2147 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2148 which its argument is a member.
2149 ** DEFCLASS only redefines the class named by its class-name
2150 argument if that name is the proper name of the class;
2151 otherwise, it creates a new class.
2152 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2153 of the UNBOUND-SLOT condition to the name of the slot.
2154 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2155 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2156 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2157 SLOT-MAKUNBOUND in the specified fashion.
2159 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2160 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2161 cross-compilation host. As a consequence, we can now bootstrap our
2162 way up to SBCL starting with a bare gcc toolchain and human-readable
2163 source code (first the source to CLISP, then the source to SBCL).
2164 * A contributed module containing a partial implementation of the
2165 simple-streams interface has been included. (thanks to Rudi
2167 * A contributed module implementing the RFC1321 Message Digest
2168 Algorithm, known as MD5, has been included.
2169 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2170 no longer has any effect, as the code controlled by this feature
2171 has been deleted. (As far as we know, no-one has ever built using
2172 this feature, and its semantics were confused in any case).
2173 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2174 work on conditions (as required by the ANSI specification),
2175 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2176 expected behaviour on conditions. Users should note, however,
2177 that such behaviour is not required by the ANSI specification,
2178 and so use of this behaviour may render their code unportable.
2179 * fixed some bugs revealed by Paul Dietz' test suite:
2180 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2182 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2183 ** SXHASH on condition objects no longer returns NIL.
2184 ** :ALLOCATION :CLASS slots are better treated; their values are
2185 updated on class redefinition, and initforms inherited from
2186 superclasses are applied.
2187 ** REMOVE-METHOD returns its generic function argument even when
2188 no method was removed.
2189 ** SHARED-INITIALIZE now initializes the values of the requested
2190 slots, including those with :ALLOCATION :CLASS.
2191 ** ALLOCATE-INSTANCE now works on structure classes defined via
2192 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2194 ** SLOT-EXISTS-P now works on conditions, as well as structures
2196 ** MAKE-LOAD-FORM now has the required methods on
2197 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2198 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2199 keyword, and now implements the SLOT-NAMES argument.
2200 ** methods with &OPTIONAL arguments no longer allow too many
2201 arguments to be passed in the call without error.
2202 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2203 option is consistent with the required arguments of the generic
2204 function lambda list.
2205 * bug fix: REQUIRE accepts a string designator. (Thanks to
2207 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2208 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2209 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2210 * bug fix: The compiler now checks for duplicated variables in macro
2212 * bug fix: SETQ on globals returns the correct value.
2213 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2214 not a COMPILER-ERROR (followed by some other strange error on
2215 choosing the CONTINUE restart).
2216 * bug fix: make.sh and friends are now more consistent in the way that
2217 they look for GNU "make".
2219 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2220 * experimental native threads support (on x86 Linux >=2.4 only).
2221 This is not compiled in by default: you need to add :SB-THREAD to
2222 the target features. See the "Beyond ANSI" chapter of the manual
2224 * fix for longstanding nonANSIism: The old distinction between
2225 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2226 The return value from CL:FIND-CLASS is now a CLOS class, and
2227 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2228 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2229 * An interface to the MetaObject Protocol, as described in Kiczales,
2230 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2231 MIT Press, 1991, is available from the SB-MOP package.
2232 * incompatible change: the SB-PCL package should now be considered
2233 a private implementation detail, and no longer a semi-private MOP
2235 * minor incompatible change: due to rearrangement for threads, the
2236 control stack and binding stack are now allocated at arbitrary
2237 addresses instead of being hardcoded per-port. Users affected by
2238 this probably have to be doing advanced things with shared
2239 libraries, and will know who they are.
2240 * minor incompatible change: Previously, all --eval forms used were
2241 processed with READ before any of them were processed with EVAL.
2242 Now each --eval form is processed with both READ and EVAL before
2243 the next --eval form is processed. (Thus package operations like
2244 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2245 work as the user might reasonably expect.)
2246 * minor incompatible change: *STANDARD-INPUT* is now only an
2247 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2249 * minor incompatible change: Y-OR-N-P is now character-oriented,
2250 not line oriented. Also, YES-OR-NO-P now works without errors.
2251 (thanks to Antonio Martinez)
2252 * sb-aclrepl module improvements: an integrated inspector, added
2253 repl features, and a bug fix to :trace command.
2254 * Known functions, which cannot be open coded by the backend, are
2255 considered to be able to check types of their arguments. (fixing
2256 a bug report by Nathan J. Froyd)
2257 * fixed a bug in computing method discriminating functions: It is
2258 now possible to define methods specialized on classes which have
2259 forward-referenced superclasses. (thanks to Gerd Moellmann)
2260 * fixed evaluation order in optional entries (reported by Gilbert
2262 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2263 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2264 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2265 required-by-AMOP NAME argument, as well as CLASS and
2266 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2267 * fixed bug 20: DEFMETHOD can define methods using names that are
2268 not the proper names of classes to designate class specializers.
2269 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2270 Unix file descriptors, instead of blowing up. (thanks to Antonio
2272 * Garbage collection refactoring: user-visible change is that a
2273 call to the GC function during WITHOUT-GCING will not do garbage
2274 collection until the end of the WITHOUT-GCING. If you were doing
2275 this you were probably losing anyway.
2276 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2277 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2278 TYPEP the latter but not the former.
2279 * The compiler issues a full WARNING for calls to undefined functions
2280 with names from the CL package.
2281 * MAP-INTO for a vector destination is open coded. (reported by
2282 Brian Downing on c.l.l)
2283 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2284 documentation string.
2285 * fixed some bugs revealed by Paul Dietz' test suite:
2286 ** COPY-ALIST now signals an error if its argument is a dotted
2288 ** Condition slots are now accessed more correctly in the presence
2289 of multiple initargs for a given slot.
2290 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2291 exclude restarts of the same name associated with a different
2293 ** DEFCLASS of forward-referenced classes with another
2294 forward-referenced class in the superclasses list no longer
2296 ** Condition slots are now initialized once each, not multiple
2297 times. (thanks to Gerd Moellmann)
2298 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2300 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2301 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2302 arguments contain duplicated elements.
2303 ** RESTART-CASE understands local macros.
2304 ** RESTART-CASE associates exactly its own restarts with a condition.
2305 ** ENDP in safe mode checks its argument to be of type LIST.
2306 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2307 in question is unbound.
2308 ** Optimized MAKE-INSTANCE functions no longer cause internal
2309 assertion failures in the presence of duplicate initargs.
2310 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2311 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2312 propagated even in the ctor optimized implementation of
2314 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2316 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2317 * a better implementation of SXHASH on (simple) bit vectors,
2318 measured both in execution speed and in distribution of results
2319 over the positive fixnums, has been installed. Likewise, a better
2320 implementation of EQUAL for simple bit vectors is now available.
2321 * fixed CEILING optimization for a divisor of form 2^k.
2322 * fixed bug 240 (emitting extra style warnings "using the lexical
2323 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2324 by Antonio Martinez)
2325 * fixed SXHASH, giving different results for NIL depending on type
2326 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2327 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2328 variable. (found by Rolf Wester)
2329 * SBCL does not ignore type declarations for special
2330 variables. (reported by rif on c.l.l 2003-03-05)
2331 * some bug fixes in contrib/sb-aclrepl/
2332 * fixed some bugs revealed by Paul Dietz' test suite:
2333 ** a bug in the CONS type specifier, whereby the CAR and CDR
2334 types got intertwined, has been fixed;
2335 ** the type system is now able to reason about the interaction
2336 between INTEGER and RATIO types more completely;
2337 ** APPEND, [N]REVERSE and NRECONC check that those their
2338 arguments, which must be proper lists, are really so;
2339 ** An array specialized to be unable to hold elements has been
2340 implemented, as required -- yes, really -- by ANSI;
2341 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2342 on malformed property lists;
2344 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2345 * incompatible packaging change: in line with Unix convention,
2346 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2347 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2348 * REQUIRE and PROVIDE are now optionally capable of doing something
2349 useful. See the documentation string for REQUIRE.
2350 * infrastructure for a managed SBCL contrib system: contributed
2351 modules in this release include:
2352 ** the ASDF system definition facility;
2353 ** an interface to the BSD Sockets API;
2354 ** an ACL-like convenience interface to the repl;
2355 (thanks to Kevin Rosenberg)
2356 ** an implementation of ROTATE-BYTE, with efficient implementation
2358 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2359 gives a closer approximation to the right answer than previously.
2360 (thanks to Raymond Toy)
2361 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2362 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2363 optional environment arguments, as required by ANSI.
2364 * fixed bugs in other functions taking environment objects, allowing
2365 calls with an explicit NIL environment argument to be compiled
2367 * fixed bug 228: primary return values from
2368 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2369 COMPILE or FUNCTION.
2370 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2371 :TYPE will no longer signal errors on innocuous objects.
2372 * fixed bug 231b: SETQ is better at respecting type declarations in
2373 the lexical environment.
2374 * fixed a bug in DEFCLASS: classes named by symbols with no or
2375 unprintable packages can now be defined.
2376 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2377 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2378 * fixed compiler failure related to checking types of functions.
2379 (reported by Robert E. Brown)
2380 * the compiler is now much more consistent in its error-checking
2381 treatment of bounding index arguments to sequence functions: in
2382 (SAFETY 3) code, errors will be signalled in almost all cases if
2383 invalid sequence bounding indices are passed to functions defined
2384 by ANSI to operate on sequences.
2385 * fixed a bug in the build procedure: documentation of SBCL-specific
2386 packages is now preserved and available in the final Lisp image.
2387 * lifted FDEFINITION lookup out of loops in the implementation of
2388 many list operations. (thanks to Robert E. Brown)
2389 * fixed a bug in the reader: the #n# reader macro now works for
2390 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2391 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2392 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2393 E. Brown and Tony Martinez respectively)
2394 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2395 SLOT-BOUNDP: in particular, it is now able to optimize them much
2396 better, and is now not vulnerable to having packages renamed.
2397 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2398 always called when a slot is not present in an instance. (thanks
2400 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2401 longer signal an error on unreading a character following EOF on
2402 the previous constituent stream. (thanks to Tony Martinez)
2403 * fixed some bugs revealed by Paul Dietz' test suite:
2404 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2405 not just nonnegative fixnums;
2406 ** the logical bit-array operators such as BIT-AND now accept an
2407 explicit NIL for their "opt-arg" argument (to indicate a
2408 freshly-consed result bit-array);
2409 ** ELT now signals an error on an invalid sequence index in safe
2411 ** the type system is now cleverer about negations of numeric
2412 types, and consequently understands the BIGNUM and RATIO types
2414 ** the type system is now cleverer about the interaction between
2415 INTEGER and RATIO types: while bugs still remain, many more
2416 cases are accurately computed;
2417 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2418 if it is in the last clause;
2419 ** CONSTANTLY now correctly returns a side-effect-free function in
2421 ** DECLARE is no longer treated as a special-operator; in
2422 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2423 * incremented fasl file version number due to the change in the
2424 DEFSTRUCT-SLOT-DESCRIPTION structure.
2426 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2427 * minor incompatible change: code processed by the "interpreter" or
2428 EVAL now has a compilation optimization policy of (DEBUG 2)
2429 (changed from (DEBUG 1)) to improve debuggability of interactive
2430 development, and to allow the use of the debug RETURN command in
2432 * an experimental implementation of the RETURN command for the
2433 debugger has been included. (thanks to Frederik Kuivinen)
2434 * fixed bug 62: constraints were not propagated into a loop.
2435 * fixed bug in embedded calls of SORT (reported and investigated by
2437 * fixed some bugs revealed by Paul Dietz' test suite:
2438 ** printing and reading of arrays with some dimensions having
2439 length 0 (thanks to Gerd Moellmann);
2440 ** BOA constructor with &AUX argument without a default value does
2441 not cause a type error;
2442 ** CONSTANTP now returns true for all self-evaluating objects.
2444 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2445 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2446 accessors that are related by inheritance, as specified in the
2447 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2448 Valtteri Vuorikoski)
2449 * The compiler is now able to inline functions that were defined in
2450 a complex lexical environment (e.g. inside a MACROLET).
2451 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2453 * Support for the upcoming FreeBSD-5.0 release has been included.
2454 (thanks to Dag-Erling Smorgrav)
2455 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2456 effect when it is not in a toplevel context.
2457 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2458 stand a better chance of being correct. (thanks to Gerd
2460 * fixed bug in COERCE, which now signals an error on coercing a
2461 rational to a bounded real type which excludes the expected
2463 * The compiler is now able to derive types more accurately from the
2464 COERCE and COMPILE functions.
2465 * fixed bug 223: functional binding is considered to be constant
2466 only for symbols in the CL package.
2467 * fixed bug 231: SETQ did not check the type of a variable being set
2468 (reported by Robert E. Brown)
2469 * A new optimization for MAKE-INSTANCE has been included, fixing
2470 various bugs (including relating to :ALLOCATION :CLASS slots and
2471 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2473 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2474 ** As required by ANSI, LOOP now disallows anonymous collection
2475 clauses such as COLLECT I in conjunction with aggregate boolean
2476 clauses such as THEREIS (= I 1);
2477 ** LOOP now signals an error when any variable is reused in the
2478 same loop (including the potentially useful construct analogous
2479 to WITH A = 1 WITH A = (1+ A);
2480 ** IT is only a special loop symbol within the first clause of a
2481 conditional loop clause;
2482 ** LOOP with a typed iteration variable over a hashtable now
2483 signals a type error iff it should.
2484 * fixed some other bugs revealed by Paul Dietz' test suite:
2485 ** FILE-STREAM now names the class previously known as FD-STREAM;
2486 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2487 argument) no longer signals an error;
2488 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2489 of the slot symbol, rather than using the current package
2490 ((:CONC-NAME "") continues to intern the slot's name in the
2492 * incremented fasl file version number, because of the incompatible
2493 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2494 of the new implementation of DEFINE-COMPILER-MACRO.
2496 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2497 * Support for building SBCL for MIPS platforms running in
2498 little-endian mode has now been checked in, and basic
2499 functionality on said platforms verified.
2500 * minor incompatible change: PCL now records the pathname of a file
2501 in which methods and the like are defined, rather than its
2503 * minor incompatible change: TRUENAME now considers the truename of
2504 a file naming a directory to be the pathname with :DIRECTORY
2505 component indicating that directory.
2506 * minor incompatible change: a NAMED clause in the extended form of
2507 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2508 reason for the previous behaviour is unclear.
2509 * more systematization and improvement of CLOS and MOP conformance
2510 in PCL (thanks to Gerd Moellman and Pierre Mai):
2511 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2513 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2514 primary methods with no specializers;
2515 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2517 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2518 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2519 FORWARD-REFERENCED-CLASSes; error reporting on
2520 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2522 ** SXHASH on CLOS instances now uses a slot internal to the
2523 instance to return different numbers on distinct instances,
2524 while preserving the same return value through invocations of
2526 ** DEFMETHOD signals errors when methods with longer incongruent
2527 lambda lists are added to generic functions;
2528 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2529 CLASS, as specified in AMOP;
2530 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2531 on the order returned by the primary method for classes of
2532 class STANDARD-CLASS;
2533 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2534 * fixed some bugs shown by Paul Dietz' test suite:
2535 ** DOLIST puts its body in TAGBODY;
2536 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2538 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2539 value producing form;
2540 ** if more variables are given to PROGV than values, extra
2541 variables are bound and made to have no value;
2542 ** NSUBSTITUTE on list arguments gets the right answer with
2544 ** ELT signals an error of type TYPE-ERROR when the index argument
2545 is not a valid sequence index;
2546 ** LOOP signals (at macroexpansion time) an error of type
2547 PROGRAM-ERROR when duplicate variable names are found;
2548 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2549 ** FUNCALL of special-operators now cause an error of type
2551 ** PSETQ now works as required in the presence of side-effecting
2552 symbol-macro places;
2553 ** NCONC accepts any object as its last argument;
2554 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2556 ** loop-for-as-package does not require a package to be explicitely
2558 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2560 * fixed bug 166: compiler preserves "there is a way to go"
2561 invariant when deleting code.
2562 * fixed bug 172: macro lambda lists with required arguments after
2563 &REST arguments now cause an error to be signalled. (thanks to
2565 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2566 bug. (thanks to Matthew Danish)
2567 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2569 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2570 arguments are lexically rebound. (thanks to Gerd Moellmann and
2572 * fixed bug 194: error messages are now more informative when there
2573 is no primary method applicable in a call to a generic
2574 function. (thanks to Gerd Moellmann)
2575 * fixed bug in command line argument checking (thanks to Julian
2577 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2578 (thanks to Matthew Danish)
2579 * incremented fasl file version number, because of the
2580 SXHASH-related changes in the layout of CLOS data structures
2582 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2583 * minor incompatible change: The runtime (the Unix executable named
2584 "sbcl") is now much pickier about the .core files it will load.
2585 Essentially it now requires .core files to descend from the same
2586 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2587 as the runtime does. (The intent is to prevent the crashes which
2588 can occur, and which can even be reported as mysterious failures,
2589 when people patch the sources or change the build parameters
2590 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2591 sbcl and .core files.)
2592 * fixed bug: VALUES-LIST is no longer optimized away.
2593 * fixed bug 142: The FFI conversion of C string values to Lisp
2594 string values no longer conses excessively. (thanks to Nathan
2595 Froyd porting Raymond Toy's fix to CMU CL)
2596 * began to systematize and improve MOP conformance in PCL (thanks to
2597 Nathan Froyd, Gerd Moellman and Pierre Mai):
2598 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2600 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2601 ** FINALIZE-INHERITANCE is now called on class finalization;
2602 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2603 argument precedence order.
2604 * fixed bug 202: The compiler no longer fails on functions whose
2605 derived types contradict their declared type.
2606 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2607 so it can be non-toplevel.
2608 * The fasl file version number has changed (because of the new
2609 implementation of DEFMACRO).
2610 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2611 safe code, that any length requirement by their type-specifier
2612 argument is valid. The exceptions to this are described in bug
2614 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2615 functions now check the types of their inputs as required by ANSI.
2616 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2617 introduced symbol is DECLAREd to be SPECIAL.
2618 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2619 * fixed bug: :COUNT argument to sequence functions may be negative
2620 * fixed bug: body of DO-SYMBOLS may contain declarations
2621 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2622 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2623 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2624 symbol macro only once
2625 * fixed printing of call frame when argument list is unavailable
2626 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2627 * compiler no longer signals WARNING on unknown keyword
2630 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2631 * A beta-quality port to the mips architecture running Linux,
2632 based on the old CMUCL backend, has been made. It has been tested
2633 on a big-endian kernel, and works sufficiently well to be able to
2634 rebuild itself; it has not been tested in little-endian mode.
2635 * fixed an inconsistency between gencgc.c and purify.c which made
2636 dumping/loading .core files unreliable
2637 * fixed bug 120a: The compiler now deals correctly with IFs where
2638 the consequent is the same as the alternative, instead of
2639 misderiving the return type. (thanks to Alexey Dejneka)
2640 * fixed bug 113: Logical pathnames are now dumpable (the logical
2641 host is resolved at load-time, throwing an error if it is not
2643 * fixed bug 174: FORMAT's error message is slightly clearer when a
2644 non-printing character is used in a format directive.
2645 * fixed several bugs in compiler checking of type declarations, i.e.
2646 violations of the Python "declarations are assertions" principle
2647 (thanks to Alexey Dejneka)
2648 * fixed several bugs in PCL's error checking (thanks to Gerd
2650 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2652 * fixed bug in compilation of functions as first class values
2653 (thanks to Antonio Martinez-Shotton)
2654 * The compiler's handling TYPE-ERRORs which it can prove will
2655 inevitably happen at runtime has been cleaned up and corrected
2656 in several ways. (thanks to Alexey Dejneka)
2657 * improved argument type checking for various basic arithmetic
2658 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2659 lavished on them in the past that they can be compiled in many
2660 ways in different special cases
2661 * fixed bug 181: compiler checks validity of user supplied type
2663 * cleaned up code flushing in optimization: Function calls which
2664 should signal errors for safety purposes (e.g. which ANSI says
2665 should signal errors when their arguments are of incorrect type)
2666 are no longer optimized away.
2667 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2668 * incremented fasl file version number, because changes in the
2669 implementation of sequence functions like COERCE caused
2670 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2671 in old inline expansions) to become undefined. (Actually these
2672 changes were later undone, so we might very well be binary
2673 compatible with 0.7.7 after all, but leaving the version number
2674 incremented seemed like the simplest and most conservative
2677 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2678 * An alpha-quality port to the parisc architecture running Linux,
2679 based on the old CMUCL backend, has been made. This, even more so
2680 than the other backends, should be considered still a work in
2681 progress; known problems include that the Linux kernel in 64-bit
2682 mode does not propagate the correct sigcontext structure to
2683 userspace, and consequently SBCL on a parisc64 kernel will not
2685 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2686 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2687 suppressed.) Also now that the compiler is looking at declarations
2688 in the environment, it checks optimization declarations as well,
2689 and suppresses inlining when (> DEBUG SPEED).
2690 * More fixes have been made to treatment of floating point exception
2691 treatment and other Unix signals. In particular, floating point
2692 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2693 * The detection and handling of control stack exhaustion (infinite
2694 or very deeply nested recursion) has changed. Stack exhaustion
2695 detection is now done by write-protecting pages at the OS level
2696 and applies at all optimization settings; when found, a
2697 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2698 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2699 be caught using IGNORE-ERRORS.
2700 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2701 correctly when transforming recursive LABELS functions to LETs.
2702 (thanks to Alexey Dejneka)
2703 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2704 that are names of constants or global variables.
2705 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2706 alien routines with docstrings.
2707 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2708 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2710 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2711 (thanks to Eric Marsden)
2712 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2713 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2714 * Bug fix: The SPARC backend can now compile functions involving
2715 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2716 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2717 to a functions expecting a PATHNAME-DESIGNATOR.
2718 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2719 lambda lists. (thanks to Alexey Dejneka)
2720 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2721 functions so that EXPORTing the name of the class doesn't cause
2722 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2723 to Antonio Martinez for reporting this)
2724 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2725 constant-folded without failing an assertion. (thanks to Einar
2726 Floystad Dorum for reporting this)
2727 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2728 types are now dealt with more correctly by the compiler.
2729 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2730 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2731 behaviour analogous to RENAME-FILE. This puts its behaviour more
2732 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2734 * The fasl file version number has changed again. (because of the
2735 bug fix involving the names of PCL MAKE-INSTANCE functions)
2737 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2738 * bug fix: Floating point exceptions are treated much more
2739 consistently on the x86/Linux and PPC/Linux platforms.
2740 * Array initialization with :INITIAL-ELEMENT is now much faster for
2741 cases when the compiler cannot open code the array creation, but
2742 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2743 array accesses have also seen a speed increase.
2744 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2745 specified. (This was at the root of some bad interactions between
2746 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2748 * bug fix: Internal error arguments for undefined functions are now
2749 computed correctly on the PPC/Linux platform.
2750 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2751 Raymond Toy's patch for CMU CL)
2752 * Support for the Solaris 9 operating environment has been included
2753 (thanks to Daniel Merritt)
2754 * A very ugly but hopefully complete draft of the missing FFI chapter
2755 of the manual has been created by reformatting the corresponding
2756 CMU CL manual chapter into (currently very ugly and incoherent)
2757 DocBook and bringing it up to date for SBCL behavior. Thus, the
2758 manual is now essentially complete, at least by my extreme
2759 once-and-only-once standards, whereby it's acceptable to refer to
2760 the doc strings of SB-EXT functions as the primary documentation.
2761 * The fasl file version number has changed again, due to cleanup of
2762 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2763 is no longer a static symbol.)
2765 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2766 * SBCL now builds with OpenMCL (version 0.12) as the
2767 cross-compilation host; also, more progress has been made toward
2768 bootstrapping under CLISP.
2769 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2771 * bug 158 fixed: The compiler can now deal with integer loop
2772 increments different from 1; fixing this turned out also to fix
2774 * bug 169 fixed: no more bogus warnings about using lexical bindings
2775 despite the presence of perfectly good SPECIAL declarations (thanks
2776 to David Lichteblau)
2777 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2778 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2779 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2781 * bug 180 fixed: Method combination specifications no longer ignore
2782 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2783 * bug fix: Structure type predicate functions now check their argument
2784 count as they should.
2785 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2786 correctly. (thanks to Pierre Mai)
2787 * minor incompatible change: The --noprogrammer option is deprecated
2788 in favor of the new --disable-debugger option, which is very similar.
2789 (The major difference is that it takes effect at a slightly different
2790 time at startup, causing handling of errors in --sysinit and
2791 --userinit files will be affected differently.) The
2792 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2793 been added to allow this functionality to be controlled from ordinary
2794 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2795 maintainers, who might want to run non-interactive scripts to
2796 build SBCL cores which will later be used interactively.)
2797 * minor incompatible change: The LOAD function no longer, when given
2798 a wild pathname to load, loads all files matching that pathname.
2799 Instead, an error of type FILE-ERROR is signalled.
2801 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2802 * bug 147 fixed: The compiler preserves its block link/count
2803 invariants more correctly now so that it doesn't crash. (thanks
2805 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2807 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2808 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2809 no-value-for-this-character as NIL (as specified by ANSI).
2810 * HOST-NAMESTRING on physical pathnames now returns a string that is
2811 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2812 (thanks to Christophe Rhodes)
2813 * The Alpha port handles icache flushing more correctly. (thanks to
2815 * More progress has been made toward bootstrapping under CLISP. (thanks
2816 to Christophe Rhodes)
2817 * The fasl file format has changed again, because dynamic loading
2818 on OpenBSD (which has non-ELF object files) motivated some cleanups
2819 in the way that foreign symbols are transformed and passed around.
2820 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2821 is no longer treated as whitespace by the reader, but instead as
2822 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2823 |A
\7fB|, instead of A as it used to.
2825 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2826 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2827 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2828 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2829 this as of 0.7.1.45, but was left out of the previous news section
2830 (thanks to Dan Barlow)
2831 * SBCL now runs on the Solaris operating system on SPARC architectures
2832 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2833 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2834 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2835 * SPARC backend cleanups, allowing builds of cores optimized for V8
2836 and V9 SPARCS, and also emission of code targeted to a particular
2837 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2839 * SBCL is closer to bootstrapping under CLISP, thanks to various
2840 fixes by Christophe Rhodes.
2841 * The fasl file format has changed again, to allow the compiler's
2842 INFO database to support symbol macros.
2843 * The user manual (in doc/) is formatted into HTML more nicely.
2844 (thanks to coreythomas)
2845 * The system is smarter about SUBTYPEP relationships, especially
2846 those involving NOT types (including types such as ATOM which are
2847 represented internally using NOT types). Thus SUBTYPEP is less
2848 likely to return (VALUES NIL NIL) in general, and in particular
2849 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2851 * The fasl file format has changed again, because the internal
2852 representation of types now includes a new slot to support the new
2853 SUBTYPEP-of-NOT-types logic.
2854 * (not a change in the main branch of SBCL, but a related prototype
2855 which can hopefully be merged into the main branch of SBCL in the
2856 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2857 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2858 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2859 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2860 dispatch macro character combinations. (thanks to Alexey Dejneka)
2861 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2862 that superficially look like logical namestrings correctly.
2863 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2865 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2866 default supplied-p) for &optional and &key arguments. (thanks to
2869 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2870 * incompatible change: The compiler is now less aggressive about
2871 tail call optimization, doing it only when (> SPACE DEBUG) or
2872 (> SPEED DEBUG). (This is an incompatible change because there are
2873 programs which relied on the old CMU-CL-style behavior to optimize
2874 away their unbounded recursion which will now die of stack overflow.)
2875 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2876 for non-GENCGC systems has been increased to 20M (since that
2877 seems much closer to the likely performance optimum for modern
2878 systems than the old 4M value was)
2879 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2880 and *DEBUG-PRINT-LEVEL*
2881 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2882 of CMU CL's support for SPARC, and various endianness and other
2883 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2884 * new syntactic sugar for the Unix command line: --load foo.bar is now
2885 an alternate notation for --eval '(load "foo.bar")'.
2887 ** The system now detects stack overflow and handles it gracefully,
2888 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2889 optimization settings. (This is a good thing in general, and
2890 its introduction in this version should be particularly timely
2891 for anyone whose code fails because of suppression of tail
2893 ** The system now hunts for the C variable "environ" in a more
2894 devious way, to avoid segfaults when the C library version differs
2895 between compile time and run time. (thanks to Christophe Rhodes)
2896 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2897 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2898 ** The compiler no longer issues bogus style warnings for undefined
2899 classes in the same source file as the DEFCLASSes which defined
2900 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2902 ** fixes in CONDITION class precedence list for undefined function
2903 errors (thanks to Alexei Dejneka)
2904 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2905 correctly. (thanks to Dan Barlow)
2906 ** portability fixes aiming at bootstrapping under CLISP (thanks
2907 to Dave McDonald and Christophe Rhodes)
2908 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2909 ** fixes in type translation and and type inference (thanks to
2911 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2912 ** various fixes in the new ports (thanks to Dan Barlow)
2913 * several changes related to debugging:
2914 ** suppression of tail recursion, as noted above
2915 ** stack overflow detection, as noted above
2916 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2917 is now the default. (For some time encapsulation has been more
2918 reliable than the breakpoint-based :ENCAPSULATE NIL
2919 implementation, at least on X86 systems; and I just noticed that
2920 encapsulation also seems closer to the spirit of the ANSI
2923 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2925 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2926 up properly again. (There was a packaging bug in 0.7.0 which
2927 left their definitions in SB-SYS::LOAD-FOREIGN and
2928 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2929 vital for most things which interface to C-level interfaces,
2930 like extensions working with sockets or databases or
2931 Perl-compatible regexes or whatever, and the need to fix
2932 this bug is the main reason that 0.7.1 was released so
2934 ** DEFGENERIC is now choosier about the methods it redefines, so that
2935 reLOADing a previously-LOADed file containing DEFGENERICs does
2936 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2937 cycle now works as it should. (thanks to Alexey Dejneka)
2938 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2939 fixed by Christophe Rhodes. (He actually submitted this patch
2940 months ago, and I delayed until after 0.7.0.)
2941 ** Bug 111 (internal compiler confusion about runtime checks on
2942 FUNCTION types) was fixed by Alexey Dejneka.
2943 * Some internal cleanups (getting rid of variables which aren't
2944 needed now that the byte interpreter is gone) caused the fasl
2945 file format number to change again.
2947 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2948 * major incompatible change: The default fasl file extension, i.e. the
2949 default extension for files produced by COMPILE-FILE, has changed
2950 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2952 ** There are many changes in the implementation of the compiler.
2953 SBCL is now essentially a compiler-only implementation of ANSI
2954 Common Lisp. EVAL still "interprets" a few special cases, but
2955 almost all the interesting cases are handled by creating
2956 a LAMBDA expression, calling COMPILE on it, then calling
2957 FUNCALL on the result.
2958 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2959 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2960 Since the code is newer, there might still be some new bugs
2961 (though not as many as before Martin Atzmueller's fixes:-). But
2962 the new code is substantially simpler and clearer, and hopefully
2963 any remaining bugs will be simpler, less fundamental, and more
2964 fixable then the bugs in the old code.
2965 ** The revised compiler is still a little unsteady on its feet.
2967 *** The debugging information it produces (particularly the names
2968 of FUNCTION objects) is sometimes much less useful than what
2969 the old compiler produced.
2970 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2971 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2972 within a MACROLET) has been temporarily weakened.
2973 ** There are new compiler optimizations for various functions:
2974 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2975 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2976 *** the math functions TRUNCATE, FLOOR, and CEILING
2977 *** the function-of-all-trades COERCE
2978 Mostly these should be transparent, but there's one
2979 potentially-annoying problem (bug 117): when the compiler
2980 inline-expands a function and does type analysis on the result,
2981 it can create control paths which have type mismatches, and
2982 when it can't prove that those control paths aren't taken,
2983 it will issue WARNINGs about the type mismatches. This is
2984 a particular problem in practice for the new sequence functions.
2985 It's not clear how this should be fixed, and for now, a
2986 workaround is given in the entry for 117 in the BUGS file.
2987 ** (Because of the interaction between the two previous items --
2988 occasional inlining problems and new inline expansions -- some
2989 of the new sequence function optimizations won't really kick in
2990 completely until debugging information, and then inlining, are
2991 straightened out in some future version.)
2992 * minor incompatible changes:
2993 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2994 DIRECTORY no longer implicitly promotes NIL slots of its
2995 pathname argument to :WILD. In particular, when you ask for the
2996 contents of a directory (which you used to be able to do without
2997 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2998 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2999 ** changes in behavior that ANSI explicitly defines to be
3000 implementation dependent:
3001 *** The new compiler-only implementation still conforms with ANSI,
3002 but acts a little different than before. Besides the obvious
3003 changes in performance tradeoffs (that the cost per form passed
3004 to EVAL has gone up, and the cost per form executed by EVAL
3005 has gone down), the behavior of the system changes a little
3006 because there are no longer any interpreted function objects.
3007 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3008 e.g. doing COMPILE on the output of interactive DEFUN is
3010 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3012 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3013 has changed from (:CL) to NIL.
3014 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3015 e.g. #\Newline and #\Tab, don't have names specified in the
3016 ANSI Common Lisp standard, is now based on their ASCII symbolic
3017 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3018 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3019 are no longer used for output.
3020 ** changes in internal implementation constants:
3021 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3022 4 million. (If your application spends a lot of time GCing and
3023 you have a lot of RAM, you might want to experiment with
3024 increasing it even more.)
3025 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3026 However, almost all old code should still continue to work without
3027 immediate update, as SB-C-CALL is now a (deprecated) nickname
3029 ** Old operator names in the style DEF-FOO are now deprecated in
3030 favor of new corresponding names DEFINE-FOO, for consistency with
3031 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3032 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3033 internal symbols, but a few supported extensions like
3034 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3035 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3036 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3037 etc. as you get deeper into recursive calls to the debugger
3038 command loop, instead of the old "5]", "5]]", "5]]]"
3039 sequence. (I was motivated to do this when squabbles between
3040 ILISP and SBCL left me very deeply nested in the debugger. In the
3041 short term, this change will probably provoke more ILISP/SBCL
3042 squabbles, but hopefully it will be an improvement in the long run.)
3043 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3044 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3045 function, since declaiming return types involving aliens is
3046 (1) annoyingly messy to do by hand and (2) vital to efficient
3047 compilation of code which calls such functions.
3048 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3049 longer reexported by the SB-EXT package. They're solely useful
3050 for alien code, so it seems more logical that you should get
3051 them from the SB-ALIEN package, not in SB-EXT.
3052 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3053 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3054 features. Instead, the code that they used to control is always
3055 built into the system.
3056 * many other bug fixes
3057 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3058 advantage of the new EVAL-WHEN stuff and to clean them up in
3059 general, and they are now more ANSI-compliant in a number of
3060 ways. Martin Atzmueller is responsible for a lot of this.
3061 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3063 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3064 *** correct ERROR type for various file operations
3065 *** some fixes for Lisp streams
3066 *** DEFMETHOD syntax checking
3067 *** changing old weird representation of debug information as
3068 strings (which, among their other deficiencies, don't transform
3069 correctly when you rename packages, and don't change their
3070 print representation when you change things like *PACKAGE*
3071 and *PRINT-LENGTH*) to symbols and lists of symbols
3072 He also made several improvements and fixed several bugs in DESCRIBE.
3073 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3075 *** misbehavior of WRITE-STRING/WRITE-LINE
3076 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3077 and several other LOOP problems as well
3078 *** DIRECTORY when similar filenames are present
3079 *** DEFGENERIC with :METHOD options
3080 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3081 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3082 *** argument ordering in FIND with :TEST option
3083 *** mishandled package designator argument in APROPOS-LIST
3084 *** various problems in the backquote readmacro
3085 *** a bug in APROPOS
3086 *** probably some others that I'm not describing very well here,
3087 since the CVS log documents them by reference to sbcl-devel
3088 messages, and the SourceForge archives aren't working well.:-(
3089 ** Dan Barlow improved the Alpha port (and is making progress on the
3090 PPC port, for those of you who think different).
3091 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3092 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3093 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3094 continues to work on the SPARC port (for those of you in a position
3095 to look down upon our little PC-compatible boxes from a great height).
3096 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3097 and uses that copy, rather than the current dynamic value, when
3098 it's trying to decide whether to truncate output. Thus e.g.
3099 (let ((*print-lines* 50))
3100 (pprint-logical-block (stream nil)
3102 (let ((*print-lines* 8))
3103 (print (aref possiblybigthings i) stream)))))
3104 should now truncate the logical block only at 50 lines, instead of
3105 often truncating it at 8 lines, as it did before.
3106 * The doc/cmucl/ directory, containing old CMU CL documentation
3107 from the time of the fork, is no longer part of the base system.
3108 SourceForge has shut down its anonymous FTP service, and with it
3109 my original plan for distributing the old CMU CL documentation
3110 there. For now, if you need these files you can download an old
3111 SBCL source release and extract them from it.
3112 * The fasl file version number changed again, for dozens of reasons,
3113 some of which are apparent above.
3115 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3116 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3117 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3118 patch, so that DISASSEMBLE output is much nicer.
3119 * The code in the SB-PROFILE package now seems reasonably stable.
3120 I still haven't decided what the final interface should look like
3121 (I'd like PROFILE to interact cleanly with TRACE, since both
3122 facilities use function encapsulation) but if you have a need
3123 for profiling now, you can probably use it successfully with
3124 the current CMU-CL-style interface.
3125 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3126 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3127 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3128 link now returns the dangling link itself, and for similar
3129 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3130 link itself. (In these cases the old code signalled an error and
3131 looped endlessly, respectively.) Thus, DIRECTORY now works even
3132 in the presence of dangling and cyclic symbolic links.
3133 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3134 is now a supported extension again, since the consensus on
3135 sbcl-devel was that it can be useful for ordinary development
3136 work, not just for debugging SBCL itself.
3137 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3138 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3139 that currently-defined functions might be redefined later with
3140 different return types.
3141 * Hash tables can be printed readably, as inspired by CMU CL code
3142 of Eric Marsden and SBCL code of Martin Atzmueller.
3143 * better error handling in CLOS method combination, thanks to
3144 Martin Atzmueller porting Pierre Mai's CMU CL patches
3145 * more overflow fixes for >16Mbyte I/O buffers
3146 * A bug in READ has been fixed, so that now a single Ctrl-D
3147 character suffices to cause end-of-file on character streams.
3148 In particular, now you only need one Ctrl-D at the command
3149 line (not two) to exit SBCL.
3150 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3152 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3153 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3154 does the right thing.
3155 * The implementation of some type tests, especially for CONDITION
3156 types, is now tidier and maybe faster, due to CMU CL code
3157 originally by Douglas Crosher, ported by Martin Atzmueller.
3158 * Some math functions have been fixed, and there are new
3159 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3160 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3161 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3162 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3163 likely to be built into sbcl-0.7.0.)
3164 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3165 CL, ported to SBCL by Martin Atzmueller)
3166 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3167 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3168 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3169 * There's a new slam.sh hack to shorten the edit/compile/debug
3170 cycle for low-level changes to SBCL itself, and a new
3171 :SB-AFTER-XC-CORE target feature to control the generation of
3172 the after-xc.core file needed by slam.sh.
3173 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3174 COMPILE-FILE is no longer supported, so that now every function
3175 gets an entry point, so that block compilation looks a little
3176 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3177 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3178 returns the number of bytes consed since the system started,
3179 rather than the number consed since the first time the function
3180 was called. (The new definition parallels ANSI functions like
3181 CL:GET-INTERNAL-RUN-TIME.)
3182 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3183 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3184 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3185 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3186 * Fasl file version numbers are now independent of the target CPU,
3187 since historically most system changes which required version
3188 number changes have affected all CPUs equally. Similarly,
3189 the byte fasl file version is now equal to the ordinary
3192 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3193 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3194 is no longer recognized. I apologize for this, because it was
3195 listed in SB-EXT as a supported extension, but I found that
3196 its existing behavior was poorly specified, as well as incorrectly
3197 specified, and it looked like too much of a mess to straighten it
3198 out. I have enough on my hands trying to get ANSI stuff to work..
3199 * many patches ported from CMU CL by Martin Atzmueller, with
3200 half a dozen bug fixes in pretty-printing and the debugger, and
3201 half a dozen others elsewhere
3202 * fixed bug 13: Floating point infinities are now supported again.
3203 They might still be a little bit flaky, but thanks to bug reports
3204 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3205 as flaky as they were.
3206 * The --noprogrammer command line option is now supported. (Its
3207 behavior is slightly different in detail from what the old man
3208 page claimed it would do, but it's still appropriate under the
3209 same circumstances that the man page talks about.)
3210 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3211 are now supported, and enabled by default. Thus, the compiler can
3212 handle many floating point and complex operations much less
3213 inefficiently. (Thus e.g. you can implement a complex FFT
3215 * The compiler now detects type mismatches between DECLAIM FTYPE
3216 and DEFUN better, and implements CHECK-TYPE more correctly, and
3217 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3218 thanks to patches from Martin Atzmueller.
3219 * various fixes to make the cross-compiler more portable to
3220 ANSI-conforming-but-different cross-compilation hosts (notably
3221 Lispworks for Windows, following bug reports from Arthur Lemmens)
3222 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3223 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3224 fixed thanks to Pierre Mai's CMU CL patches.
3225 * fixes to keep the system from overflowing internal counters when
3226 it tries to use i/o buffers larger than 16M bytes
3227 * fixed bug 45a: Various internal functions required to support
3228 complex special functions have been merged from CMU CL sources.
3229 (When I was first setting up SBCL, I misunderstood a compile-time
3230 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3231 * improved support for type intersection and union, fixing bug 12
3232 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3233 more obscure bugs as well
3234 * some steps toward byte-compiling non-performance-critical
3235 parts of the system, courtesy of patches from Martin Atzmueller
3236 * Christophe Rhodes has made some debian packages of sbcl at
3237 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3238 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3239 stable, but are nonetheless usable. When he's ready, I'd be happy
3240 to add them to the SourceForge "File Releases" section. (And if
3241 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3242 * new fasl file format version number (because of changes in
3243 internal representation of (OR ..) types to accommodate the new
3244 support for (AND ..) types, among other things)
3246 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3247 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3249 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3251 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3252 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3253 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3254 * DESCRIBE now works on structure objects again.
3255 * Most function call argument type mismatches are now handled as
3256 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3257 know whether the function will be redefined before the call is
3258 executed. (The compiler could flag local calls with full WARNINGs,
3259 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3260 it doesn't keep track of enough information to know whether calls
3261 are local in this sense.)
3262 * Compiler output is now more verbose, with messages truncated
3263 later than before. (There should be some supported way for users
3264 to override the default verbosity, but I haven't decided how to
3265 provide it yet, so this behavior is still controlled by the internal
3266 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3267 src/compiler/ir1util.lisp.)
3268 * Fasl file format version numbers have increased again, because
3269 support for the Gray streams extension changes the layout of the
3270 system's STREAM objects.
3271 * The Gray subclassable streams extension now works, thanks to a
3272 patch from Martin Atzmueller.
3273 * The full LOAD-FOREIGN extension (not just the primitive
3274 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3275 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3276 but like most other programs, it defaults to copying the Unix
3277 environment from the original process instead of starting the
3278 new process in an empty environment.
3279 * Extensions which manipulate the Unix environment now support
3280 an :ENVIRONMENT keyword option which doesn't smash case or
3281 do other bad things. The CMU-CL-style :ENV option is retained
3282 for porting convenience.
3283 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3284 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3286 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3288 * A patch from Martin Atzmueller seems to have solved the SIGINT
3289 problem, and as far as we know, signal-handling now works cleanly.
3290 (If you find any new bugs, please report them!)
3291 * The system no longer defaults Lisp source file names to types
3292 ".l", ".cl", or ".lsp", but only to ".lisp".
3293 * The compiler no longer uses special default file extensions for
3294 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3295 seems to expect a single default extension for all compiled code,
3296 and there's no compelling reason to try to stretch the standard
3297 to allow two different extensions.) Instead, byte-compiled files
3298 default to the same extension as native-compiled files.
3299 * Fasl file format version numbers have increased again, because
3300 a rearrangement of internal implementation packages made some
3301 dumped symbols in old fasl files unreadable in new cores.
3302 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3303 many fewer weird special cases.
3304 * Bug #17 (differing COMPILE-FILE behavior between logical and
3305 physical pathnames) has been fixed, and some related misbehavior too,
3306 thanks to a patch from Martin Atzmueller.
3307 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3308 by Tim Moore, ported to SBCL by Martin Atzmueller.
3309 * Martin Atzmueller fixed several filesystem-related problems,
3310 including bug #36, in part by porting CMU CL patches, which were
3311 written in part by Paul Werkowski.
3312 * More compiler warnings in src/runtime/ are gone, thanks to
3313 more patches from Martin Atzmueller.
3314 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3317 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3319 * DESCRIBE now works on CONDITION objects.
3320 * The debugger now handles errors which arise when trying to print
3321 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3323 * The build system now uses an additional file, customize-target-features.lisp,
3324 to allow local modifications to the target *FEATURES* list. (The point of
3325 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3326 debugging features enabled, without having to worry about propagating it
3327 into everyone's system when I do a "cvs update".) When no
3328 customize-target-features.lisp file exists, the target *FEATURES* list
3329 should be constructed the same way as before.
3330 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3331 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3332 warn about behavior which is undefined under the ANSI spec. Note: This
3333 is specified by ANSI, but it's not very popular with programmers.
3334 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3335 macro in the SBCL sources for an example of a workaround which you
3336 might use to make portable ANSI-standard code which does what you want.
3337 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3338 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3339 when it's used as a non-toplevel form. (This is required in order
3340 to implement the DEFCONSTANT-EQX macro.)
3341 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3342 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3343 funny things, due to bugs in the implementation of EVAL-WHEN
3344 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3345 non-toplevel DEFCONSTANTs are uncommon.)
3346 * The core file version number and fasl file version number have been
3347 incremented, because the old noncompliant DEFCONSTANT behavior involved
3348 calling functions which no longer exist, and because I also took the
3349 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3350 * fixed bug 1 (error handling before read-eval-print loop starts), and
3351 redid debugger restarts and related debugger commands somewhat while
3353 ** The QUIT debugger command is gone, since it did something
3354 rather different than the SB-EXT:QUIT command, and since it never
3355 worked properly outside the main toplevel read/eval/print loop.
3356 Invoking the new TOPLEVEL restart provides the same functionality.
3357 ** The GO debugger command is also gone, since you can just invoke
3358 the CONTINUE restart directly instead.
3359 ** The TOP debugger command is also gone, since it's redundant with the
3360 FRAME 0 command, and since it interfered with abbreviations for the
3362 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3364 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3365 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3366 some porting work by Martin Atzmueller)
3367 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3368 with a boatload of patches which he ported to SBCL. Now that those
3370 ** The system tries to make sure that its low-priority messages
3371 are prefixed by semicolons, to help people who like to use
3372 syntax highlighting in their ILISP buffer. (This patch
3373 was originally due to Raymond Toy.)
3374 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3375 patches originally written by Raymond Toy.
3376 ** The compiler understands coercion between single-value and
3377 multiple-VALUES type expressions better, getting rid of some very
3378 weird behavior, thanks to patches originally by Robert MacLachlan
3379 and Douglas Crosher.
3380 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3381 lambda lists, thanks to a patch originally by Pierre Mai.
3382 ** The system no longer bogusly warns about "abbreviated type
3384 ** The compiler gets less confused by inlining and RETURN-FROM,
3385 thanks to some patches originally by Tim Moore.
3386 ** The system no longer hangs when dumping circular lists to fasl
3387 files, thanks to a patch originally from Douglas Crosher.
3388 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3389 undefined function error.
3390 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3391 in structure slot initforms) for now, documented workaround instead:-|
3392 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3393 * fixed bug 5: added stubs for various Gray stream functions called
3394 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3395 installed, at least appropriate type errors are generated
3396 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3397 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3398 consistently in DEFMETHOD forms.
3399 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3400 it doesn't seem to affect SBCL after all
3401 * The C runtime system now builds with better optimization and many
3402 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3404 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3406 * The system is now under CVS at SourceForge (instead of the
3407 CVS repository on my home machine).
3408 * The new signal handling code has been tweaked to treat register
3409 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3410 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3411 downstream, e.g. in debug-int.lisp, has implicit dependencies
3412 on the unsignedness of integer representation of machine words,
3413 and that caused the system to bomb out with infinite regress
3414 when trying to recover from type errors involving signed values,
3415 e.g. (BUTLAST '(1 2 3) -1).)
3416 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3417 (This was one of the bugs Peter Van Eynde reported back in July.)
3418 * The system now uses code inspired by Colin Walters' O(N)
3419 implementation of MAP (from the cmucl-imp@cons.org mailing
3420 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3421 inline the MAP operation, and there is more than one
3422 sequence argument to the MAP call (so that it can't just
3423 do ETYPECASE once and for all based on the type of the
3424 single sequence argument). (The old non-inline implementation
3425 of the general M-argument sequence-of-length-N case required
3426 O(M*N*N) time when any of the sequence arguments were LISTs.)
3427 * The QUIT :UNIX-CODE keyword argument has been renamed to
3428 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3429 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3431 * Raymond Wiker's patches to port dynamic loading from Linux to
3432 FreeBSD have been added.
3433 * The BUGS file is now more nearly up to date, thanks in large part
3434 to Martin Atzmueller's review of it.
3435 * The debugger now flushes standard output streams before it begins
3436 its output ("debugger invoked" and so forth).
3437 * The core version number and fasl file version number have both
3438 been incremented, because of incompatible changes in the layout
3440 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3441 used to not be called for a saved Lisp image.)
3442 * Martin Atzmueller's version of a patch to fix a compiler crash,
3443 as posted on sbcl-devel 13 September 2000, has been installed.
3444 * Instead of installing Martin Atzmueller's patch for the
3445 compiler transform for SUBSEQ, I deleted the compiler transform,
3446 and transforms for some similar consing operations.
3447 * A bug in signal handling which kept TRACE from working on OpenBSD
3449 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3450 away by constant folding
3451 * The system now defines its address space constants in one place
3452 (in the Lisp sources), and propagates them automatically elsewhere
3453 (through GENESIS and the sbcl.h file). Therefore, patching the
3454 address map is less unnecessarily tedious and error-prone. The
3455 Lisp names of address space constants have also been systematized.
3456 * CVS tags like dollar-Header-dollar have been removed from
3457 the sources, because they have never saved me trouble and
3458 they've been source of trouble working with patches and other
3459 diff-related operations.
3460 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3461 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3463 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3465 * The system has been ported to OpenBSD.
3466 * The system now compiles with a simple "sh make.sh" on the systems
3467 that it's supported on. I.e., now you no longer need to tweak
3468 text in the target-features.lisp-expr and symlinks in src/runtime/
3469 by hand, the make.sh takes care of it for you.
3470 * The system is no longer so grossly inefficient when compiling code
3471 involving vectors implemented as general (not simple) vectors (VECTOR T),
3472 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3473 which dares to use the various sequence functions on non-simple
3474 vectors, takes less of a performance hit.
3475 * There is now a primitive type predicate VECTOR-T-P
3476 to test for the (VECTOR T) type, so that e.g.
3477 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3478 can now be compiled with some semblance of efficiency. (The old code
3479 turned the type declaration into a full call to %TYPEP at runtime!)
3480 * AREF on (VECTOR T) is still not fast, since it's still compiled
3481 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3482 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3483 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3484 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3485 without making full calls to SUBTYPEP at runtime.
3486 (Some analogous efficiency issues for non-simple vectors specialized to
3487 element types other than T, or for non-simple multidimensional arrays,
3488 have not been addressed. They could almost certainly be handled the
3489 same way if anyone is motivated to do so.)
3490 * The changes in array handling break binary compatibility, so
3491 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3492 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3493 instead of (VALUES T T).
3494 * By following the instructions that Dan Barlow posted to sbcl-devel
3495 on 2 July 2000, I was able to enable primitive dynamic object
3496 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3497 functionality is not implemented (since it calls ld to resolve
3498 library references automatically, requiring RUN-PROGRAM for its
3499 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3500 try to resolve library references) is now supported.
3501 * The system now flushes the standard output streams when it terminates,
3502 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3503 them at several other probably-convenient times, e.g. in each pass of
3504 the toplevel read-eval-print loop, and after evaluating a form given
3505 as an "--eval" command-line option. (These changes were motivated by a
3506 discussion of stream flushing issues on cmucl-imp in August 2000.)
3507 * The source transform for TYPEP of array types no longer assumes
3508 that an array whose element type is a not-yet-defined type
3509 is implemented as an array of T, but instead punts, so that the
3510 type will be interpreted at runtime.
3511 * There is now some support for cross-compiling in make.sh: each of
3512 the phases of make.sh has its own script. (This should be transparent
3513 to people doing ordinary, non-cross-compile builds.)
3514 * Since my laptop doesn't have hundreds of megabytes of memory like
3515 my desktop machine, I became more motivated to do some items on
3516 my to-do list in order to reduce the size of the system a little:
3517 ** Arrange for various needed-only-at-cold-init things to be
3518 uninterned after cold init. To support this, those things have
3519 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3520 symbols with such names are now uninterned after cold init).
3521 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3522 fasl files for cold load.
3523 ** Remove the old compiler structure pooling code (which used to
3524 be conditional on the target feature :SB-ALLOC) completely.
3525 ** Redo the representation of some data in cold init to be more compact.
3526 (I also looked into supporting byte compiled code at bootstrap time,
3527 which would probably reduce the size of the system a lot, but that
3528 looked too complicated, so I punted for now.)
3529 * The maximum signal nesting depth in the src/runtime/ support code has
3530 been reduced from 4096 to 256. (I don't know any reason for the very
3531 large old value. If the new smaller value turns out to break something,
3532 I'll probably just bump it back up.)
3533 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3535 * Many, many bugs reported by Peter Van Eynde have been added to
3536 the BUGS list; some have even been fixed.
3537 * While enabling dynamic object file loading, I tried to make the
3538 code easier to understand, renaming various functions and variables
3539 with less ambiguous names, and changing some function calling
3540 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3541 * While trying to figure out how to do the OpenBSD port, I tried to
3542 clean up some of the code in src/runtime/. In particular, I dropped
3543 support for non-POSIX signal handling, added various comments,
3544 tweaked the code to reduce the number of compilation warnings, and
3545 renamed some files to increase consistency.
3546 * To support the new automatic configuration functionality in make.sh,
3547 the source file target-features.lisp-expr has been replaced with the
3548 source file base-target-features.lisp-expr and the machine-generated
3549 file local-target-features.lisp-expr.
3550 * fixed a stupid quoting error in make.sh so that using CMU CL
3551 "lisp -batch" as cross-compilation host works again
3553 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3555 * DESCRIBE no longer tries to call itself recursively to describe
3556 bound/fbound values, so that it no longer fails on symbols which are
3557 bound to themselves (like keywords, T, and NIL).
3558 * DESCRIBE now works on generic functions.
3559 * The printer now prints less-screwed-up representations of closures
3560 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3561 * A private symbol is used instead of the :EMPTY keyword previously
3562 used to mark empty slots in hash tables. Thus
3563 (DEFVAR *HT* (MAKE-HASH-TABLE))
3564 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3565 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3566 now does what ANSI says that it should. (You can still get
3567 similar noncompliant behavior if bang on the hash table
3568 implementation with all the symbols you get back from
3569 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3570 This breaks binary compatibility, since tests for equality to
3571 :EMPTY are wired into things like the macroexpansion of
3572 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3574 * There's now a minimal placeholder implementation for CL:STEP,
3575 as required by ANSI.
3576 * An obscure bug in the interaction of the normal compiler, the byte
3577 compiler, inlining, and structure predicates has been patched
3578 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3579 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3581 * Missing ordinary arguments in a macro call are now detected even
3582 when the macro lambda list contains &KEY or &REST.
3583 * The debugger no longer complains about encountering the top of the
3584 stack when you type "FRAME 0" to explicitly instruct it to go to
3585 the top of the stack. And it now prints the frame you request even
3586 if it's the current frame (instead of saying "You are here.").
3587 * As specified by ANSI, the system now always prints keywords
3588 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3590 * The default initial SIZE of HASH-TABLEs is now smaller.
3591 * Type information from CLOS class dispatch is now propagated
3592 into DEFMETHOD bodies, so that e.g.
3593 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3595 is now basically equivalent to
3596 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3597 (DECLARE (TYPE SINGLE-FLOAT X))
3599 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3600 operation, without having to do run-time type dispatch.
3601 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3602 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3603 * Since I don't seem to be making big file reorganizations very often
3604 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3605 is overflowing my ability to conveniently back them up), I've finally
3606 checked the system into CVS. (The CVS repository is on my home system,
3607 not at SourceForge -- putting it on SourceForge might come later.)
3608 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3609 high-level GC-NOTIFY-FOO functions send their output. (There's
3610 still very little control of where low-level verbose GC functions
3611 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3612 less than it used to -- the GC-NOTIFY-FOO functions are now under
3613 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3614 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3615 in only one place in the source code, and propagates it automatically
3616 everywhere that it's needed. Thus e.g. when I bump the version from
3617 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3618 * The C source files now include boilerplate legalese and documentation
3619 at the head of each file (just as the Lisp source files already did).
3620 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3621 to his page will be replaced with a link to his new CLiki service.
3623 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3625 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3626 * The build process now looks for GNU make under the default name "gmake",
3627 instead of "make" as it used to. If GNU make is not available as "gmake"
3628 on your system, you can change this default behavior by setting the
3629 GNUMAKE environment variable.
3630 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3631 can build without error under CMU CL.
3633 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3635 * There is now a partial SBCL user manual (with some new text and some
3636 text cribbed from the CMU CL manual).
3637 * The beginnings of a profiler have been added (starting with the
3638 CMU CL profiler and simplifying and cleaning up). Eventually the
3639 main interface should be through the TRACE macro, but for now,
3640 it's still accessed through vaguely CMU-CL-style functions and macros
3641 exported from the package SB-PROFILE.
3642 * Some problems left over from porting CMU CL to the new
3643 cross-compilation bootstrap process have been cleaned up:
3644 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3645 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3646 part when getting rid of the compiler *BACKEND* stuff.)
3647 ** The value of *NULL-TYPE* was screwed up, because it was
3648 being initialized before the type system knew the final
3649 definition of the 'NULL type. This screwed up several key
3650 optimizations in the compiler, causing inefficiency in all sorts
3651 of places. (I found it because I wanted to understand why
3652 GET-INTERNAL-RUN-TIME was consing.)
3653 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3654 PROCLAIM FTYPE information. Unfortunately this broke binary
3655 compatibility again, since now the forms output by DEFGENERIC
3656 to refer to functions which didn't exist in 0.6.3.
3657 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3658 can use the new (as of 0.6.3) transform for SOME into MAP into
3660 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3661 values to (INTEGER 0 1000000), so that the time code will no longer
3662 occasionally get blown up by Linux returning 1000000 microseconds
3663 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3664 its output conform to the ANSI spec. (Alas, this makes its output
3665 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3666 * A full call to MAP NIL with a single sequence argument no longer conses.
3667 * fixes to problems pointed out by Martin Atzmueller:
3668 * The manual page no longer talks about multiprocessing as though
3669 it were currently supported.
3670 * The ILISP support patches have been removed from the distribution,
3671 because as of version 5.10.1, ILISP now supports SBCL without us
3672 having to maintain patches.
3673 * added a modified version of Raymond Toy's recent CMU CL patch for
3674 EQUALP comparison of HASH-TABLE
3676 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3678 * The system still can't cross-compile itself with
3679 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3680 entails), but at least it can compile more of itself that way
3681 than it used to be able to, and various buglets which were uncovered
3682 by trying to cross-compile itself that way have now been fixed.
3683 * This release breaks binary compatibility again. This time
3684 at least I've incremented the FASL file format version to 2, so that the
3685 problem can be detected reliably instead of just causing weird errors.
3686 * various new style warnings:
3687 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3688 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3689 ** using the lexical binding of a variable named in the *FOO* style
3690 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3691 as specified by ANSI.
3692 * *RANDOM-STATE* is no longer automatically initialized from
3693 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3694 default behavior of the system is to repeat its behavior every time
3695 it's run. If you'd like to change this behavior, you can always
3696 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3697 old convention there was no comparably easy way to get the system to
3698 repeat its behavior every time it was run.
3699 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3700 FTYPE declarations has been removed, in favor of their ANSI
3701 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3702 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3703 the types of their sequence arguments can be determined at compile time.
3704 This is done through a new open code expansion for MAP which eliminates
3705 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3706 argument types can be determined at compile time.
3707 * The optimizer now transforms COERCE into an identity operation when it
3708 can prove that the coerced object is already of the correct type. (This
3709 can be a win for machine generated code, including the output of other
3710 optimization transforms, such as the MAP transform above.)
3711 * Credit information has been moved from source file headers into CREDITS.
3712 * Source file headers have been made more standard.
3713 * The CASE macro now compiles without complaining even when it has
3716 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3718 * (Note that the way that the PCL macroexpansions were rewritten
3719 to accommodate the change in DEFGENERIC below breaks binary
3720 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3721 not run under sbcl-0.6.2. Once we get out of alpha releases,
3722 i.e. hit release 1.0.0, we'll probably try to maintain binary
3723 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3724 and sbcl-1.4.4. Until then, however, it might be fairly common
3725 for maintenance releases to break binary compatibility.)
3726 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3728 * The reporting of unhandled signals has been changed to print some
3729 explanatory text as well as the report form. (Previously only
3730 the report form was printed.)
3731 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3732 it defines, so that the compiler no longer issues undefined function
3733 warnings for compiled-but-not-yet-loaded generic functions.
3734 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3735 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3736 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3737 section of the ANSI spec are supported.
3738 * The "" nickname for the "KEYWORD" package has been removed.
3739 The reader still handles symbol tokens which begin with a package marker
3740 as keywords, but it doesn't expose its mechanism for doing so in the
3741 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3742 * The system now issues STYLE-WARNINGs for contradictory TYPE
3743 proclamations. (Warnings for contradictory FTYPE proclamations would
3744 be nice too, but those can't be done usefully unless the type system
3745 is made smarter about FUNCTION types.)
3746 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3747 systematized, so that "*target-*.lisp is supposed to exist only on the
3748 target and imply that there's a related file which exists on the
3749 host, and *host-*.lisp is supposed to exist only on the host and imply
3750 that there's a related file which exists on the target. This involves a
3751 lot of renaming. Hopefully the acute confusion caused by the renaming
3752 will be justified by the reduction in chronic confusion..
3753 ** runtime-type.lisp -> early-target-type.lisp
3754 ** target-type.lisp -> late-target-type.lisp
3755 ** early-host-format.lisp -> early-format.lisp
3756 ** late-host-format.lisp -> late-format.lisp
3757 ** host-error.lisp -> misc-error.lisp
3758 ** early-error.lisp -> early-target-error.lisp
3759 ** late-error.lisp -> late-target-error.lisp
3760 ** host-defboot.lisp -> early-defboot.lisp
3761 ** code/misc.lisp -> code/target-misc.lisp
3762 ** code/host-misc.lisp -> code/misc.lisp
3763 ** code/numbers.lisp -> code/target-numbers.lisp
3764 ** code/early-numbers.lisp -> numbers.lisp
3765 ** early-host-type.lisp -> early-type.lisp
3766 ** late-host-type.lisp -> late-type.lisp
3767 ** host-typep.lisp -> typep.lisp
3768 ** load.lisp -> target-load.lisp
3769 ** host-load.lisp -> load.lisp
3770 ** host-disassem.lisp -> disassem.lisp
3771 ** host-insts.lisp -> insts.lisp
3772 ** byte-comp.lisp -> target-byte-comp.lisp
3773 ** host-byte-comp.lisp -> byte-comp.lisp
3774 ** host-signal.lisp -> signal.lisp
3775 ** host-defstruct.lisp -> defstruct.lisp
3776 ** late-target-type.lisp -> deftypes-for-target.lisp
3777 Furthermore, several other previously target-only files foo.lisp (e.g.
3778 hash-table.lisp and random.lisp) have been split into a target-and-host
3779 foo.lisp file and a target-only target-foo.lisp file, with their key type
3780 definitions in the target-and-host part, so that the cross-compiler will
3781 know more about target types.
3782 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3783 gone away. In their place are various *BACKEND-FOO* variables
3784 corresponding to the slots of the old structure.
3785 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3786 into the target SBCL has been fixed.
3787 * The chill.lisp system for loading cold code into a running SBCL
3789 * Support for the CMU CL "scavenger hook" extension has been removed.
3790 (It was undocumented and unused in the CMU CL sources that SBCL was
3791 derived from, and stale in sbcl-0.6.1.)
3792 * Various errors in the cross-compiler type system were detected
3793 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3794 (enabling various consistency checks). Many of them were fixed,
3795 but some hard problems remain, so the compiler is back to
3796 running without *TYPE-SYSTEM-INITIALIZED* for now.
3797 * As part of the cross-compiler type system cleanup, I implemented
3798 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3799 * I have started adding UNCROSS calls throughout the type system
3800 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3801 on cross-compiler input files will be able to go away, and various
3803 * CONSTANTP now returns true for quoted forms (as explicitly required
3806 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3808 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3809 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3810 (MAKE-LIST -1). In the longer term, it ought to become true
3811 that declarations are assertions even at SAFETY 1. For now, it's not
3812 quite true even at SAFETY 3, but it's at least more nearly true..
3813 (Note that this change seems to increases the size of the system by
3814 O(5%) and to decrease the speed of the compiler by 20% or more.)
3815 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3816 for the problem of printing dozens of lines of distracting information
3817 about low-level system machinery as part of the top stack frame
3818 on entry to the debugger when an undefined function was called.
3819 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3820 is not reset to COMMON-LISP-USER.
3821 * Compilation of stuff related to dyncount.lisp has been made conditional
3822 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3823 smaller. The various dyncount-related symbols have been moved into
3824 a new "SB-DYNCOUNT" package.
3825 * tty-inspect.lisp has been renamed to inspect.lisp.
3826 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3827 feature has gone away. (When we eventually port to other flavors of
3828 libc and/or Unix, we'll try to make the differences between flavors
3829 invisible at the user level.)
3830 * Various other *FEATURES* tags, and/or their associated conditionals,
3831 have been removed if obsolescent, or given better documentation, or
3832 sometimes given more-mnemonic names.
3834 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3836 * tidied up "make.sh" script
3837 * tidied up system directory structure
3838 * better "clean.sh" behavior
3839 * added doc/FOR-CMUCL-DEVELOPERS
3840 * many many small tweaks to output format, e.g. removing possibly-confusing
3841 trailing #\. character in DESCRIBE-INSTANCE
3842 * (EQUALP #\A 'A) no longer signals an error.
3843 * new hashing code, including EQUALP hashing
3844 * tidied up Lisp initialization and toplevel
3845 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3846 * command line argument processing
3847 * added POSIX-GETENV function to deal with Unix-ish environment variables
3848 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3849 terminating SBCL on EOF
3850 * non-verbose GC by default
3851 * There is no more "sbcl" shell script; the sbcl file is now the C
3852 runtime executable (just like CMU CL).
3853 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3855 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3856 symbol and package information as Lisp native symbol and package objects
3857 instead of strings naming symbols and strings naming packages. This way,
3858 whenever packages are renamed (as in warm init), debug information is
3859 transformed along with everything else.
3860 * tweaked the optimization policy declarations which control the building
3861 of SBCL itself. Now, among other things, the system no longer saves
3862 source location debugging information. (This helps two problems at once
3863 by reducing SBCL size and by keeping SBCL from trying to look for its
3864 sources -- which may not exist -- when reporting errors.)
3865 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3866 debugging and testing purposes
3867 * cleaned up printing, making the printer call PRINT-OBJECT for
3868 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3869 methods, giving nearly-ANSI behavior
3870 * converted almost all special variables to use *FOO* naming convention
3871 * deleted PARSE-TIME functionality, since it can be done portably
3872 * moved some files out of cold init into warm init
3873 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3875 * regularized formatting of source files
3876 * added an install.sh script
3877 * fixed ridiculous memory usage of cross-compiler by making
3878 compiler/alloc.lisp not try to do pooling unless it can hook
3879 itself into the GC of the cross-compilation host. Now the system
3880 builds nicely on my old laptop.
3881 * added :SB-ALLOC in target-features.lisp-expr
3882 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3883 was not implemented)
3884 * re-did condition handling and note reporting in the compiler. Notes
3885 are no longer handled by signalling conditions. Style warnings
3886 and warnings are handled more correctly and reported in such a way
3887 that it's easy to find one or the other in your output (so that you
3888 can e.g. figure out which of many problems caused COMPILE-FILE to
3890 * changed the severity of several compiler warnings from full WARNING
3891 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3892 compiler note reporting so that it doesn't use the condition system
3893 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3894 COMPILE-FILE command)
3895 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3896 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3897 no longer supported, since their implementation was deeply intertwingled
3898 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3899 * removed bogus "support" for compiler macros named (SETF FOO), and
3900 removed the compiler macro for SETF INFO (but only after making a fool
3901 of myself on the cmucl-imp mailing list by posting a bogus patch for
3902 DEFINE-COMPILER-MACRO..)
3903 * Compiled files containing forms which have side effects on the Lisp
3904 reader (such as DEFPACKAGE forms) are now handled more correctly.
3905 (Compiler queuing of top level lambdas has been suppressed by setting
3906 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3907 * deleted various currently-unused source files, e.g. gengc.lisp. They
3908 may be added back at some point e.g. when porting to other architectures,
3909 but until they are it's distracting to distribute them and to try to
3911 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3912 there were so many of them they're just distractions, and UNCROSS is
3913 known to be able to handle the current sources
3914 * moved PROFILE functionality into TRACE, so that it will be clear
3915 how the wrapping and unwrapping of functions when you profile them
3916 interacts with the wrapping and unwrapping of functions when you
3917 trace them. (Actually, the functionality isn't there yet, but at least
3918 the interface specification is there. Hopefully, the functionality will
3919 arrive with some future maintenance release.)
3920 * removed host-oops.lisp
3921 * changed signature of QUIT function to allow UNIX-CODE argument
3922 * fixed READ-SEQUENCE bug
3923 * tweaked verbose GC output so that it looks more like the progress
3924 output that ANSI specifies for functions like LOAD
3925 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3926 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob