1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
3 * incompatible change: SB-EXT package no longer contains the
4 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
5 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
7 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
9 * bug fix: loading of default sysinit file works. (thanks to Leonid
11 * bug fix: better detection of circularities in the file-compiler.
12 (reported by Marco Monteiro)
14 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
15 * feature: implemented the READER-METHOD-CLASS and
16 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
18 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
20 * optimization: faster LOGCOUNT implementation on x86 and x86-64
21 (thanks to Lutz Euler)
22 * optimization: hashing of general arrays and vectors has been
23 improved. (reported by Any Fingerhut)
24 * enhancement: SB-INTROSPECT is now able to find definitions of
25 profiled functions. (thanks to Troels Henriksen)
26 * enhancement: compiler-macro expansion applies now to FUNCALL forms
27 as well, allowing compiler-macros for SETF-functions to expand.
28 * enhancement: step-instrumentation no longer wraps known
29 single-value functions with multiple-value context, allowing
30 better type inference.
31 * fixed bug #337: use of MAKE-METHOD in method combination now works
32 even in the presence of user-defined method classes. (reported by
33 Bruno Haible and Pascal Costanza)
34 * fixed bug #339(c): if there are applicable methods not part of any
35 long-form method-combination group, call INVALID-METHOD-ERROR.
36 (reported by Bruno Haible)
37 * fixed bug #361: the :FUNCTION initarg in the protocol for
38 initialization of methods can now be used to override
39 internally-produced optimized functions. (reported by Bruno
41 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
42 system-provided lambda expression no longer cause warnings about
43 unbound #:|pv-table| symbols.
44 * bug fix: improved the handling of type declarations and the
45 detection of violations for keyword arguments with non-constant
47 * bug fix: potentially erronous calls to PATHNAME and
48 MERGE-PATHNAMES were being flushed in some cases.
49 (reported by Richard Kreuter)
50 * bug fix: compiled calls to TYPEP were mishandling obsolete
51 instances. (reported by James Bielman and Attila Lendvai)
52 * bug fix: format strings with ~> without matching ~< no longer
53 trigger an AVER, but signal an understandable error instead.
54 (reported by Antonio Martinez)
55 * bug fix: specifying an output-file without a pathname-type for
56 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
57 of input-file instead of "fasl". (reported by Robert Dodier)
58 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
59 (reported by James Y Knight).
60 * bug fix: compiler-macros-function did not consider the environment
61 argument for shadowing by local functions.
62 * bug fix: compiler-macros expansion was inhibited by local INLINE
64 * bug fix: inline expansions of known functions were subject to
65 step-instrumentation in high debug policies, leading to problems
67 * bug fix: compiler failed to differentiate between different CONS
69 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
70 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
71 * bug fix: #368: incorrect use of expressed vs. upgraded array
73 * bug fix: #306a: more precise unions of array types.
74 * thread-safety improvements:
75 ** CONDITION-WAIT could return early on Linux, if the thread was
76 interrupted and subsequently continued with SIGCONT.
77 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
79 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
80 * added support for the ucs-2 external format. (contributed by Ivan
82 * minor incompatible change: pretty printing of objects of type
83 (cons symbol) is, in the default pprint-dispatch-table, now
84 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
86 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
87 called later in a class's lifetime, possibly as late as when the
88 first instance of the class is created. Previously,
89 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
90 class became finalizeable.
91 * fixed bug: FILE-POSITION sometimes returned inconsistent results
92 for multibyte external-format streams. (thanks to "vbzoli")
93 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
94 with :ALLOCATION :CLASS inherited from superclasses of the
96 * fixed bug: anonymous classes can now be created using the :NAME
97 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
98 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
99 * fixed bug: core-files saved with :EXECUTABLE T can again be
100 executed when SBCL_HOME isn't set. (reported by James Knight)
101 * fixed bug: toplevel LOCALLY forms with declarations could
102 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
103 * fixed bug: printing from several different threads using different
104 values of *print-case* could cause invalid output, due to
105 some internal special variables of the printer not being bound
106 thread-locally (reported by Max Mikhanosha)
107 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
108 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
109 (reported by Pascal Costanza)
110 * fixed bug: FINALIZE-INHERITANCE is called from
111 REINITIALIZE-INSTANCE on classes when the class has previously
112 been finalized, as required by AMOP.
113 * minor code generation optimizations:
114 ** better register allocation in CLOS dispatching functions
115 ** overflow detection when coercing signed bytes to fixnums on x86-64
116 ** is now implemented with one IMUL instruction instead of three shifts
117 ** more efficient bit-vector access on x86 and x86-64
118 ** more efficient access to raw structure slots on x86 and x86-64
119 * fixed some bugs revealed by Paul Dietz' test suite:
120 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
123 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
124 * feature: thread support on Solaris/x86, and experimental thread support
126 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
128 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
129 default method for SLOT-UNBOUND.
130 * minor incompatible change: prevent the user from specializing the
131 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
132 somewhat counter to the intent of the protocol, I (CSR) think, and
133 additionally it just doesn't work in SBCL as currently
134 implemented, thanks to optimizations (that are always valid for
135 the other three SLOT-VALUEish functions, but not for the setter).
136 * minor incompatible change: the :SB-LDB feature is now enabled by
137 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
138 the low-level debugger.
139 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
140 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
141 down to an SB-SYS:FD-STREAM, allowing direct communication in
142 more cases. Temporary files and pipes are still used as a fallback
144 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
145 on the non-reentrant C functions localtime(3) and gmtime(3).
146 * bug fix: global optimization policy was not visible in LOCALLY and
148 * bug fix: class objects can be used as specializers in methods.
149 (reported by Pascal Costanza)
150 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
151 with a type error. (reported by blitz_ on #lisp)
152 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
153 failed with a type error. (reported by Pascal Bourguignon)
154 * bug fix: merging pathnames against defaults with :DIRECTORY
155 starting with '(:RELATIVE :BACK) should preserve the :BACK.
156 (reported by James Y Knight)
157 * bug fix: saving large (>2GB) cores on x86-64 now works
158 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
159 constant index and a (simple-array (signed-byte 32)) array
160 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
161 error (patch by Robert J. Macomber)
162 * feature: TIME now displays the amount of run-time spent in GC
163 * bug fix: The debugger now does a better job of respecting
164 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
165 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
166 workaround for bug 403.)
167 * bug fix: floating point exception handling now works on FreeBSD
168 (thanks to NIIMI Satoshi)
169 * fixed some bugs revealed by Paul Dietz' test suite:
170 ** MISC.641: LET-conversion were not supposed to work in late
172 * improvements to the Win32/x86 port:
173 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
174 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
176 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
177 * new feature: source path information is generated for macro-expansion
178 errors for use in IDE's like Slime (thanks to Helmut Eller)
179 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
180 * bug fix: compiler does not loop forever on an invalid type in
182 * improvement: compilation of most CLOS applications is significantly
184 * optimization: added a limited bytecode compiler for simple toplevel
185 forms, speeding up compilation and FASL loading
187 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
188 * minor incompatible change: in sbcl-0.9.11 (but not earlier
189 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
190 into subclasses of the finalized class. Now user calls to
191 FINALIZE-INHERITANCE finalize just the one class, and calls by the
192 system return before any subclasses are finalized.
193 * minor incompatible change: The reader no longer ignores errors
194 regarding non-existent packages in #+ and #- feature tests.
195 * new feature: command line options --no-sysinit, --no-userinit to
196 inhibit loading the corresponding init files
197 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
198 for sending data through UDP sockets (thanks to François-René Rideau)
199 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
200 error is signalled instead (thanks to François-René Rideau)
201 * minor incompatible change: Error signalling behaviour of lexical
202 operations violating package locks has changed slightly. Refer to
203 documentation on package locks for details.
204 * enhancement: EVAL can process IF-expressions without resorting to the
206 * bug fix: Some operations involving SIN, COS, or TAN weren't being
207 constant-folded properly due to missing out-of-line functions.
208 (reported by Mika Pihlajamäki)
209 * bug fix: LISTEN sometimes returned T even in cases where no data was
210 immediately available from the stream
211 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
212 were reversed. (reported by Levente Mészáros)
213 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
214 when given a vector displaced to an adjustable array. (reported
216 * bug fix: derivation of float boundaries from numbers outside the
217 appropriate float range (reported by John Wiseman)
218 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
220 * bug fix: precomputing cache entries for generic functions with
221 some subclasses of specializers as yet invalid does not attempt to
222 fill a cache line with a negative offset. (reported by Levente
224 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
225 allow condition class objects as arguments to DOCUMENTATION and
226 (SETF DOCUMENTATION); only find and set documentation for
227 structure names for the STRUCTURE doc-type. (suggested by Gary
229 * improvements to the Win32/x86 port:
230 ** (user-homedir-pathname) and default initialization file
231 locations now know about the user's "Documents and Settings"
232 directory (thanks to Yaroslav Kavenchuk)
233 ** run-program is implemented (thanks to Mike Thomas)
234 ** sockets support (thanks to Timothy Ritchey)
235 ** better backtrace support (thanks to Alastair Bridgewater)
236 ** sb-grovel supported
237 ** asdf-install and sb-posix work somewhat
238 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
239 * improvements to the Solaris/x86 port:
240 ** works on Solaris 11/Solaris Express
241 ** floating-point exception handling support
242 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
243 * fixed some bugs revealed by Paul Dietz' test suite:
244 ** REMOVE-METHOD returns its generic function argument even when
245 the method is not one of the generic functions' methods.
246 ** objects declared as MEMBER types can be admissible arguments to
248 ** printing characters should simply be printed by the FORMAT ~:C
250 ** compiler failure when compiling functions with hairy constant
251 defaults for optional parameters.
252 ** compiler produces wrong code when MAYBE-INLINE-expanding a
253 function, which is already optimized.
255 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
256 * new platform: experimental support for SBCL x86/Darwin, including
257 MacOS X 10.4.5 on Intel.
258 * new feature: Unicode character names are now known to the system
259 (through CHAR-NAME and NAME-CHAR).
260 * new feature: the filesystem location of SBCL's core file is
261 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
262 * minor incompatible change: the contrib modules SB-POSIX and
263 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
264 this change is to make it easier to distribute
265 location-independent binaries.
266 * bug fix: as implied by AMOP, standardized classes no longer have
267 slots named by external symbols of public packages. (reported by
269 * optimization: calling functions via constant symbols -- as in
270 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
271 via the function object as in (FUNCALL #'FOO).
272 * optimization: on x86 and x86-64, the calling convention for the
273 case of dealing with an unknown number of values has been altered
274 to be friendlier to the prediction heuristics implemented,
275 particularly with reference to CALL and RET pairing. (thanks to
276 Alastair Bridgewater)
277 * optimization: on x86, the code for access to thread-local symbol
278 values and binding/unbinding of thread-local symbols is smaller.
279 (thanks to Alastair Bridgewater)
280 * enhancement: CONSTANTP is now able to determine constantness of
281 more complex forms, including calls to constant-foldable standardized
282 functions and some special forms beyond QUOTE.
283 * fixed bug: occasional GC crashes on Solaris/x86
284 * optimization: x86-64 supports stack allocation of results of simple
285 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
286 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
287 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
288 (thanks to James Knight)
289 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
290 by the standard (thanks to Stephen Compall)
292 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
293 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
294 be used for bundling the runtime and the core file into one
295 executable binary. This feature is not currently supported on all SBCL
296 platforms. (thanks to James Bielman and NIIMI Satoshi)
297 * new feature: a generational or ephemeral garbage collector is now
298 the default on the PowerPC platform (both Linux and Darwin). The
299 old Cheney (stop and copy) collector is a build-time option.
300 (thanks to Cyrus Harmon, after Raymond Toy)
301 * minor incompatible change: the method by which SBCL finds its
302 contributed modules has changed; it no longer relies on symbolic
303 links from an $SBCL_HOME/systems directory, but searches directly
304 in the subdirectories of $SBCL_HOME.
305 * enhancement: the dynamic heap size on the Linux/PPC platform is
306 markedly larger, even using the older Cheney garbage collector.
307 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
308 certain complicated string types. (reported by Gary King)
309 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
310 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
311 * fixed bug: DOCUMENTATION on structures no longer signals an
312 error if no structure type of the right name exists. (reported by
314 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
315 applicable non-standard primary or :AROUND method on
316 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
317 no longer cause unbound variable errors. (reported by Kevin Reid)
318 * fixed bug: metacircle resolution in cases where methods have slots
319 added before the slots from STANDARD-METHOD. (reported by Jean
321 * fixed bug: the Power PC assembly code for calling into lisp
322 sometimes computed its return address wrongly (depending on the
323 15th bit of the address, fixed at link time). This used to
324 manifest itself in a segmentation violation while building PCL.
325 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
326 many others over the years)
327 * fixed bug: ROOM no longer reports silly numbers for stack usage if
328 the stack pointer (treated as a signed integer) is negative.
329 (thanks to Peter van Eynde)
331 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
332 * new platform: experimental support for the Windows operating
333 system has been added. (thanks to Alastair Bridgewater)
334 * fixed several bugs in and robustified the PPC FFI (including
335 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
336 * bug fix: allow non-simple string symbol names (reported by Paul
338 * bug fix: interrupt handling on NetBSD (thanks to Richard M
340 * bug fix: saving a core corrupted callbacks on x86/x86-64
341 * bug fix: closed a loophole in metacircularity detection and
342 grounding in the PCL implementation of CLOS.
343 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
344 (reported by tomppa on #lisp)
345 * bug fix: clear the direction flag on Lisp -> C transitions, as
346 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
347 (reported by Andrej Grozin and Hendrik Maryns)
348 * optimization: major improvements to GC efficiency on GENCGC platforms
349 * optimization: faster implementation of EQUAL
350 * optimization: emit more efficient opcodes for some common
351 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
352 * optimization: several other minor code-generation improvements on x86-64
353 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
355 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
356 * minor incompatible change: (SETF CLASS-NAME) and (SETF
357 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
358 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
359 by ANSI as a generic function, and for consistency (SETF
360 GENERIC-FUNCTION-NAME) is treated likewise.
361 * fixed bug #233.b: make constraint propagation notice when a variable
362 value is changed after it is referenced but before it is used
363 * fixed bug #296: no more arbitrary behaviour from filenames with
364 odd characters as --load arguments. (reported by Adam Warner)
365 * fixed bug #390: :CHARACTER-SET pathname components now work as
366 expected. (reported by Tim Daly Jr)
367 * fixed bug #391: complicated :TYPE intersections in slot
368 definitions no longer cause an error in PCL internals.
369 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
370 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
371 returning the number of octets which would be written to the
372 file-stream. (thanks to Robert J. Macomber)
373 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
374 arguments to the internal stream functions. (thanks to David
376 * bug fix: stack exhaustion is now handled on the x86/SunOS
377 platform. (thanks to Daisuke Homma)
378 * bug fix: more accurate ROOM results on GENCGC platforms
379 * optimization: improved type inference for arithmetic-for
380 index variables in LOOP
381 * optimization: faster floating-point SQRT on x86-64
382 * fixed some bugs revealed by Paul Dietz' test suite:
383 ** DOCUMENTATION returns NIL instead of "" for method combinations
384 that don't have a docstring
386 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
387 * minor incompatible change: (SETF CLASS-NAME) and (SETF
388 GENERIC-FUNCTION-NAME) are no longer generic functions, and
389 therefore are not customizeable by user code (as seems to be at
390 least permitted and maybe required by AMOP). As a consolation,
391 however, the SBCL implementation of these functions now calls
392 REINITIALIZE-INSTANCE as specified by AMOP.
393 * bug fix: slot-definition documentation is propagated as per ANSI
394 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
395 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
396 argument, as do eslotd initialization methods. (from Pascal
397 Costanza's "Closer" project)
398 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
399 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
401 * bug fix: it is now possible to have more than one subclass of
402 STANDARD-GENERIC-FUNCTION without causing stack overflow.
403 (reported by Bruno Haible, Pascal Costanza and others)
404 * bug fix: the dependent update protocol now works for generic
405 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
407 * bug fix: condition-class instances corresponding to
408 DEFINE-CONDITION forms are now created eagerly. (reported by
409 Kalle Olavi Niemitalo on comp.lang.lisp)
410 * bug fix: floating point printing is more accurate in some
411 circumstances. (thanks to Simon Alexander)
412 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
413 merged with *DEFAULT-PATHNAME-DEFAULTS*.
414 * bug fix: callbacks on OS X now preserve stack-alignment.
415 (thanks to Cyrus Harmon)
416 * enhancement: the x86-64 disassembler is much better at
417 disassembling SSE instructions. (thanks to Lutz Euler)
418 * enhancement: CHECK-TYPE on a variable now allows type-inference
419 (thanks to Vincent Arkesteijn)
420 * optimization: improved performance of EUC-JP external format.
421 (thanks to NIIMI Satoshi)
422 * optimization: performance improvements to IO on file streams of
423 :ELEMENT-TYPE CHARACTER
424 * optimization: much faster memory allocation on x86-64
425 * optimization: faster garbage collections (latency and throughput)
427 * optimization: faster bignum subtraction and division on x86 and x86-64
428 (thanks to Lutz Euler)
429 * optimization: more accurate type-inference for loops that use a
430 floating point index variable or a negative step.
432 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
433 * bug fix: add a workaround to SBCL looping infinitely at startup on
434 Linux kernels with apparently buggy implementations of personality().
435 (thanks to Svein Ove Aas)
436 * bug fix: Unicode symbols are correctly printed in LDB backtraces
437 (thanks to David Lichteblau)
438 * bug fix: local bindings shadow global symbol macros for
439 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
440 * optimization: non-open coded uses of numeric comparison operators
441 (e.g. >) no longer cons when called with more than one parameter
442 on platforms supporting dynamic-extent allocation.
443 * enhancement: saving cores with foreign code loaded is now supported
444 on MIPS/Linux in addition to the previously supported platforms.
445 * bug fix: FIND-RESTART now tests for activity, not applicability when given
446 a restart object as identifier. (reported by Helmut Eller for CMUCL)
447 * bug fix: division by zero in sb-sprof when no samples were collected
448 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
449 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
450 (thanks to Alexander Kjeldaas)
451 * new feature: ignoring the return values of destructive functions
452 like NREVERSE signals a compile-time style-warning.
453 (thanks to Kevin Reid)
454 * new feature: the alignment of alien structure fields can be
455 explicitly specified. (thanks to Cyrus Harmon)
456 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
457 and dump core on SIGQUIT
459 ** incompatible change: threads do not inherit values of specials
460 from their parents (see manual)
461 ** bug fix: threads stacks belonging to dead threads are freed by the
462 next exiting thread, no need to gc to collect thread stacks anymore
463 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
464 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
465 ** bug fix: dynamic variable and thread start related gc lossage
466 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
468 * fixed some bugs revealed by Paul Dietz' test suite:
469 ** SUBTYPEP is slightly more accurate on heinously complicated
470 CONS types where some of the members have uncertain (in the
471 NIL, NIL sense) type relationships to each other.
473 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
474 no longer purified unless :PURIFY T is explicitly specified.
475 ** Non-purified cores are significantly smaller than before
477 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
478 * new feature: timers based on Zach Beane's excellent timer package
479 * added support for the following external formats: koi8-u,
480 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
481 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
482 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
483 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
484 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
485 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
486 * incompatible change: a threaded SBCL will no longer revert to
487 non-threaded mode on non-NPTL systems, but refuse to start entirely.
488 * bug fix: interrupts are disabled until startup is complete; no
489 more sigsegvs when receiving a signal to soon
490 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
492 * bug fix: add a workaround for the memory randomization features in
493 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
494 workaround will only be in effect on systems with the proc filesystem
496 * bug fix: printing objects of type HASH-TABLE signals a
497 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
499 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
500 have been running for over 50 days. (reported by Gilbert Baumann)
501 * bug fix: the logic for getting names of functions gets less
502 confused when confronded with alternate-metaclass
503 funcallable-instances. (reported by Cyrus Harmon)
504 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
505 consistent, even on internal alternate-metaclass objects.
506 * bug fix: flush closure information collected by physical
507 environment analysis prepass before the main pass. (bug reported
509 * bug fix: compiler pack phase does not modify a hashtable, which is
510 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
511 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
512 may be expanded. (reported by Alexander <malishev> on c.l.l)
513 * bug fix: no more style-warnings when compiling code using LAST
514 under high-SPEED low-SPACE settings. (reported by David Wallin
516 * fixed some incompatibilities between SBCL's MOP and the MOP
518 ** the METAOBJECT class is now implemented;
519 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
520 STANDARD-OBJECT, as required;
521 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
522 now compatible (as required by VALIDATE-SUPERCLASS); there
523 remains a consistency requirement over the presence or absence
524 of the FUNCTION class in the superclass list of finalized
525 classes; see the manual for more details;
526 ** the :AROUND method for COMPUTE-SLOTS on
527 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
528 requested slot ordering.
530 ** bug fix: parent thread now can be gc'ed even with a live
532 ** bug fix: binding a special with PROGV to no value is not
533 the same as that symbol not having been bound (thanks to
535 ** bug fix: binding specials is thread safe (thanks to
537 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
538 and SB-SPROF do not die with 'no handler for signal XX in
539 interrupt_handle_now(..)' anymore
540 ** bug fix: WITH-TIMEOUT works with multiple running threads
541 * fixed some bugs revealed by Paul Dietz' test suite:
542 ** ENSURE-GENERIC-FUNCTION should take a method class object for
543 the :method-class keyword argument.
545 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
546 * new port: the Solaris operating system on x86 processors is now
547 mostly supported, though some rough edges in the environment
548 remain. (thanks to Daisuke Homma)
549 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
550 space for its heap. (thanks to Thiemo Seufer)
551 * optimization: SBCL on MIPS platforms now supports dynamic-extent
552 closures. (thanks to Thiemo Seufer)
553 * minor incompatible change: eof selects abort in the debugger.
554 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
555 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
557 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
558 deprecated, and will go away in a future revision of SBCL.
559 * minor incompatible change: GC-ON and GC-OFF are no longer
560 implemented with a counter, it does not matter how many times gc
561 is switched on or off
562 * bug fix: discriminating functions for generic function classes
563 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
564 make invalid assumptions about method precedence order. (reported
566 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
567 platforms. (thanks to Thiemo Seufer)
568 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
569 * bug fix: correct pathname printing with printer escaping is on.
570 (thanks to Kevin Reid)
571 * bug fix: complex VOP definitions in "user-space" no longer trigger
572 package locks. (reported by Zach Beane)
573 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
574 for funcallable-instance functions, and is no different from
576 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
577 characters in the current readtable. (reported by Nicholas Neuss)
578 * bug fix: SUBTYPEP on various CONS types returns more a more
579 accurate acknowledgment of its certainty.
580 * optimizations: REMOVE-DUPLICATES now runs in linear time on
581 lists in some cases. This partially fixes bug 384.
582 * flush all standard streams before prompting in the REPL and the
584 * bug fix: signal handling and triggering gc do not conflict
585 directly anymore, in particular a high frequency sb-sprof does
586 not prevent gc from running
587 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
588 approximation for timezone and DST information between the
589 universal time epoch and the smallest negative 32-bit time_t.
590 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
591 year 1899 so long as the encoded time is non-negative.
592 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
593 the correct answer for non-constant rotate amounts when performing
594 an inline 32-bit rotation.
596 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
597 there is only one thread in the session
598 ** bug fix: memory leak for streams created in one thread and
599 written to in another
600 ** bug fix: lockup when compiled with gcc4
601 ** bug fix: race that allows the gc to be triggered when gc is
603 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
604 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
606 ** bug fix: finished threads are gc'ed properly
607 * fixed some bugs revealed by Paul Dietz' test suite:
608 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
609 has arguments with a different set of applicable methods from
610 the orignal arguments.
611 ** The type error thrown by MAP now has a correct expected-type
613 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
614 name a compiled function.
615 ** Three MISC tests where a large bignum was improperly coerced to
616 a float that couldn't represent that bignum during type
617 derivation were fixed.
618 ** SUBTYPEP can now handle types involving the negation of a
619 list-form FUNCTION type.
620 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
621 as (COMPLEX (AND RATIO (NOT FIXNUM))).
622 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
624 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
625 * New feature: Experimental support for bivalent streams: streams
626 opened with :element-type :default now allow character and binary
627 (unsigned-byte 8) I/O
628 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
629 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
631 * bug fix: no more highly sporadic "couldn't check whether ~S is
632 readable" when reading a stream and an interrupt hits in the middle
633 of a select system call
634 * compiler better recognizes complex arrays (reported by Eduardo
636 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
637 for variables on SBCL built with linkage-tables. (reported by Luis
639 * various error reporting improvements.
640 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
641 (Thanks to Thiemo Seufer)
642 * bug fix: nested reader invokations work correctly
643 * bug fix: it is possible to have simultaneous references to foreign
644 code and foreign data with the same name.
646 ** added x86-64 support
647 ** incompatible change: the threading api now works with thread
648 objects instead of thread ids
649 ** bug fix: threads are protected from signals and interruption when
650 starting up or going down
651 ** bug fix: a race where an exiting thread could lose its stack to gc
652 ** bug fix: don't halt on infinite error in threads if possible
653 ** fixed numerous gc deadlocks introduced in the pthread merge
654 ** bug fix: fixed thread safety issues in read and print
655 ** bug fix: debugger doesn't hang on session lock if interrupted at
656 an inappropriate moment
657 ** bug fix: run-program is now thread safe(r)
658 ** bug fix: inner with-recursive-lock no longer releases the mutex
659 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
660 * fixed some bugs revealed by Paul Dietz' test suite:
661 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
662 EXPECTED-TYPE slots filled.
663 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
665 ** MAKE-CONDITION accepts classes as type-designators.
666 ** COMPILE may never return NIL.
667 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
668 range before calling Unix time functions
670 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
671 * numerous signal handling fixes to increase stability
672 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
673 * minor incompatible change: we now correctly canonize default
674 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
675 as per the MOP, rather than the historical (INITARG INITFUNCTION
676 INITFORM). (reported by Bruno Haible)
677 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
678 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
679 for more information.
680 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
681 pathname is a directory pathname.
682 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
683 * dynamic space size on PPC has been increased to 768Mb. (thanks to
685 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
686 :METACLASS argument in addition to a class name. (reported by
687 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
688 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
689 (reported by Bruno Haible; patch from Gerd Moellmann)
690 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
692 * bug fix: more cleanups to the floating point exception handling on
693 x86-64 (thanks to James Knight)
694 * bug fix: the compiler does not try to constant fold calls of
695 COERCE and BIT-* functions when they return freshly constructed
696 objects. (reported by Nikodemus Siivola and Paul Dietz)
697 * optimization: improved the MIPS versions of generic (in the
698 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
699 * optimization: direct conversion of (unsigned-byte 32) to floats on
700 the PowerPC platform.
701 * optimization: structure instances with raw slots now use less
702 memory, and probably show better memory locality. (thanks to
704 * optimization: DYNAMIC-EXTENT declarations for lists and closures
705 are treated as requests for stack allocation on the x86-64,
706 Alpha-32, PPC and SPARC platforms.
707 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
708 the wrong file descriptor; implementation of SOCKET-OPEN-P.
709 (thanks to Tony Martinez)
711 ** gcing a dead thread can no longer lead to lockups
712 ** threads block signals until they are set up properly
713 ** errno is no longer shared by threads
714 ** interrupt-thread restores the eflags register on x86
715 ** fixed some lockups due to gc/thread interaction
716 * fixed some bugs revealed by Paul Dietz' test suite:
717 ** invalid dotted lists no longer raise a read error when
719 ** signal an error if a symbol that names a declaration is used
720 as the name of a type, or vice versa
721 ** allow using the (declare (typespec var*)) abbreviation for
722 (declare (type typespec var*)) with all type specifiers
723 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
724 ** always return NIL from PPRINT-POP when OBJECT is NIL
725 ** don't signal errors when pretty-printing malformed LABELS,
726 FLET or MACROLET forms
727 ** declarations in a DOLIST body are also in scope for the
729 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
731 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
734 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
735 * fixed cross-compiler leakages that prevented building a 32-bit
736 target with a 64-bit host compiler.
737 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
738 opened with :IF-EXISTS :OVERWRITE.
739 * fixed bug 281: error for an invalid qualifier in a short-form method
740 combination method is not signalled until the faulty method is called.
741 * bug fix: iteration variable type inferrer failed to deal with open
742 intervals. (reported by Alan Shields)
743 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
744 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
745 or more runtime options were provided to the sbcl binary.
746 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
747 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
748 * optimization: there's now a fast-path for fixnum arguments in the
749 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
750 * optimization: the code generated on x86-64 is more compact thanks
751 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
752 * bug fix: floating-point exception handling now partly works on
753 x86-64. (Thanks to James Knight)
754 * improvement to the MIPS backend from Thiemo Seufer: C-style
755 64-bit long long arguments and return values to alien functions
757 * fixed some bugs revealed by Paul Dietz' test suite:
758 ** the type-error signalled from WARN has a filled-in DATUM slot.
759 ** the type-error required when a stream is not associated with
760 a file has the stream as its datum.
761 ** type-errors on single-floats on x86-64 no longer have
762 :INVALID-OBJECT as the datum
763 ** the type-errors signalled for invalid function names now have
764 a correct expected type
765 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
766 for structures defined with a :TYPE. Documentation strings for
767 typed structures are no longer immediately discarded
768 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
769 broadcast streams and synonym streams. FILE-LENGTH now also works
770 on broadcast streams.
772 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
773 * incompatible change: the --noprogrammer option, deprecated since
774 version 0.7.5, has been removed. Please use the equivalent
775 --disable-debugger option instead.
776 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
778 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
779 inoperational for a while now) has been completely removed.
780 * null lexical environments are now printed as #<NULL-LEXENV>,
781 significantly reducing the amount of clutter in typical backtraces.
782 * documentation on weak pointers, finalization, and after GC hooks
783 has been added to the manual.
784 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
785 as other specialized array types, is much faster. SUBSEQ and
786 COPY-SEQ on such arrays have also been sped up.
787 * optimization: EQL is now more efficient when at least other argument
788 is known to be of type (OR FIXNUM (NOT NUMBER)).
789 * fixed bug: the runtime is now less vulnerable to changes in the
790 size of the SBCL object on OS X, and virtual memory is reserved for
791 all spaces that need to be at a fixed address.
792 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
793 * fixed bug: finalizers and after GC hooks that cause consing are now
795 * fixed bug: compiler error messages and summaries are now printed to
796 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
797 * fixed inference of the upper bound of an iteration variable.
798 (reported by Rajat Datta).
799 * fixed bug 211e: calling local functions with duplicated constant
800 keyword argument no longer causes a bogus style warning about an
802 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
803 ftype declaration to be disregarded. (reported by Dave Roberts)
804 * fixed bug 373: caused by erronous compilation of references to alien
805 variables in the runtime on ppc/darwin.
806 * fixed bug 376: CONJUGATE type deriver.
807 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
808 (reported by Baughn on #lisp)
809 * a cross-compiler bug on non-x86 platforms has been identified and
810 fixed. (thanks to Bruno Haible)
811 * improvements to the MIPS runtime code for increased stability.
812 (thanks to Thiemo Seufer)
813 * increased the maximimum compact environment size to allow
814 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
815 * partial workaround for bug 135: don't name the function we're
816 calling for hairy cases of EVAL, so as not to accumulate one environment
817 entry per call to EVAL. (reported by Kevin Reid)
818 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
819 * optimization: type testing for non-vector arrays should be faster.
820 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
822 * bug fix: setting 31st element of a bit vector to zero did not work
824 * bug fix: redefining a class definition which failed due to a
825 previous accessor / function clash now works (but see BUGS entry
826 #380 for more problems in this area). (thanks to Zach Beane)
827 * the long form of DEFINE-METHOD-COMBINATION disables method group
828 checking when given a single method group with pattern *. (thanks
830 * on x86 compiler supports stack allocation of results of simple
831 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
832 * support for building and running on Mac OS X 10.4 "Tiger" has been added
833 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
834 * fixed some bugs related to Unicode integration:
835 ** the restarts for recovering from input and output encoding
836 errors only appear when there is in fact such an error to
838 * fixed some bugs revealed by Paul Dietz' test suite:
839 ** MISC.549 and similar: late transformation of unsafe type
840 assertions into derived types caused unexpected code
842 ** SCALE-FLOAT type deriver is less wrong.
843 ** type derivers for EXP, LOG and similar functions compute result
844 types for complex arguments better.
845 ** (MISC.563) CONJUGATE type deriver works for very restricted
847 ** out-of-line type testers for character strings are available.
848 ** EQUAL compiler transform understands specialness of objects
850 ** accessing double-floats stored on the stack now works on x86-64.
851 ** debugger internals could sometimes create invalid lispobjs,
852 resulting in GC crashes.
853 ** MISC.548: type check weakening can convert required type into
855 ** initialization forms for bindings are not in scope of free special
858 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
859 * incompatible change: thread support for non-NPTL systems has
860 been removed - locking is buggy and unreliable. A threaded
861 SBCL build will now warn at startup and refuse to create
862 new threads, unless futex support is detected in the kernel
863 * incompatible change: the top level REPL now has only an ABORT
864 restart associated with it, not TOPLEVEL and ABORT as it used to.
865 TOP and TOPLEVEL are now available as debugger commands for
866 returning to the top level.
867 * incompatible change: forms evaluated in the REPL now use the
868 global optimization policy.
869 * incompatible change: user- and system-initialization files are
870 no longer processed with LOAD, but by READ and EVAL; hence the
871 global optimization policy, startup package, readtable, etc,
873 * The .fasl file format number has been incremented because of
874 various incompatible changes.
875 * internal entry point details and argument counts no longer appear
876 in backtraces unless explicitly requested by setting
877 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
878 * built-in and standard functions no longer have names like "top
879 level local call to FOO".
880 * fixed bug 32: functions defined in non-null lexical environments
881 now have more legible printed representation
882 * fixed bug 33: functions defined in non-null lexical environemnts
883 are now more amenable to inspection by INSPECT.
884 * workaround for bug 354: XEPs no longer appear in backtraces unless
885 explicitly requested.
886 * fixed bug: receiving the signal which results from attempting to
887 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
888 SIGBUS on other *BSDs) on architectures where the C stack is also
889 the Lisp stack (x86 and x86-64) from foreign code no longer leads
890 to debugger confusion or wild instability. (reported by Cheuksan
892 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
893 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
894 (reported by Lutz Euler)
895 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
896 compiled code. (reported by Paul Dietz)
897 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
898 specialized parameter to a method to something that is not TYPEP
899 the specializer is now possible.
900 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
901 face of package deletion.
902 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
903 pathname. (reported by David Tolpin on c.l.l)
904 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
905 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
906 STRING-STREAM and FILE-STREAM, respectively.
907 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
908 than previously; in addition, its test suite is now run on build.
909 (thanks to Luke Gorrie)
910 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
911 * optimization: added a immediate representation for single-floats
913 * fixed some bugs revealed by Paul Dietz' test suite:
914 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
915 ** attempting to create a package with a colliding nickname causes
916 correctable errors to be signalled.
917 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
918 ** several x86-64 backend bugs related to sign-extension of immediate
921 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
922 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
923 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
925 * fixed loading of multiply forward-referenced layouts.
926 (thanks to Cheuksan Wang)
927 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
928 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
929 (see COMPILE-FILE documentation for details of :PRINT option)
930 * fixed bugs 19 and 317: fixed-format floating point printing is
931 more accurate. This also fixes a bug reported by Adam Warner
932 related to the ~@F format directive.
933 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
935 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
936 dispatch table if the from-char is a dispatch macro character.
937 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
938 by the wordsize no longer ignore the last word. (reported by Lutz
940 * fixed bug in type checking of dynamic-extent variables. (reported
942 * optimization: sequence traversal functions use their freedom to
943 coerce function designators to functions.
944 * optimization: code with many calls to CLOS methods specialized on
945 CLOS classes has had redundant type checks removed.
946 * optimization: added declarations to speed up operations that access
947 the internal character database (for example STRING-UPCASE)
948 * optimization: comparison operations between floats and sufficiently small
949 fixnums no longer create extra rationals
950 * fixed some bugs related to Unicode integration:
951 ** portions of multibyte characters at the end of buffers for
952 character-based file input are correctly transferred to the
953 start of the buffer at the next read.
954 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
955 passing it through to OPEN.
956 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
957 argument given to internal calls to OPEN.
958 ** the built-in comment readers (introduced by character sequences
959 ";" and "#|") are more forgiving to encoding errors; they will
960 STYLE-WARN and then attempt to resync the stream at a character
961 boundary. (thanks to Teemu Kalvas)
962 * fixed some bugs in the x86-64 port:
963 ** Negative short int return values from c-calls are sign-extended
965 ** The stack is aligned to 16-bytes for c-calls, as required by
966 the ABI. (thanks to Cheuksan Wang)
967 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
968 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
969 for enums. (thanks to Vincent Arkesteijn)
970 ** Multiple small optimizations and bugfixes for floating point
972 * fixed some bugs revealed by Paul Dietz' test suite:
973 ** Space, Tab, Linefeed, Return and Page have the invalid
974 secondary constituent character trait.
975 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
977 ** WITH-INPUT-FROM-STRING should only update the index place on
979 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
981 ** Bit-array operations (BIT-AND and similar) worked incorrectly
982 with one-dimensional arrays with fill pointers.
983 ** TYPE-OF failed on a complex with an integer realpart and a
985 ** compiler failure during type inference for the code of form
986 (IF (EQL X (THE ...)) ...) (MISC.535).
988 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
989 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
990 platform. The port supports 61-bit fixnums, large memory spaces
991 and reloading shared object files.
992 * enhancement: saving cores with foreign code loaded is now
993 supported on ppc/Darwin in addition to the previously supported
995 * enhancement: the statistical profiler now walks deeper into the
996 call stack for more meaningful call-graphs and accrued time
997 reports (x86/x86-64 only). It also now reports time spent in
999 * enhancement: it is now possible to trace most individual methods
1000 of a generic function in addition to tracing the generic function
1002 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1004 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1005 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1006 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1007 produces an error. (thanks to Vincent Arkesteijn)
1008 * bug fix: NAMESTRING on pathnames with :WILD components in their
1009 directories works correctly. (thanks to Artem V. Andreev)
1010 * fixed bug 125: compiler preserves identity of closures. (reported
1012 * bug fixed: functions with &REST arguments sometimes failed with
1013 "Undefined function" when compiled with (DEBUG 3). (reported by
1015 * bug fix: overflow during compiling of setting element of a bit
1016 vector with constant index and value. (reported by Timmy Douglas)
1017 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1018 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1019 * on x86 compiler partially supports stack allocation of dynamic-extent
1021 * GO and RETURN-FROM do not check the extent of their exit points
1022 when compiled with SAFETY 0.
1023 * fixed some bugs related to Unicode integration:
1024 ** encoding and decoding errors are now much more robustly
1025 handled; it should now be possible to recover even from invalid
1026 input or output to the terminal. (thanks to Teemu Kalvas)
1027 ** provided a first cut at implementing STRING-TO-OCTETS and
1028 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1029 ** altered the SB-MD5 contributed module slightly, changing the
1030 interface just enough for it to be supportable for builds where
1031 lisp characters are not eight bits.
1032 * fixed some bugs revealed by Paul Dietz' test suite:
1033 ** the FORMATTER-generated functions for ~V[ conditionals require
1034 the correct number of arguments.
1035 ** READ-FROM-STRING returns the mandated second value when applied
1036 to displaced strings.
1037 ** the #\Rubout and #\Backspace characters are treated as invalid
1038 constituent characters by the tokenizer.
1040 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1041 * new feature: reloading changed shared object files with
1042 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1043 * new feature: references to foreign variables and functions
1044 can now be compiled and loaded before the corresponding shared
1045 object file is loaded, as long as the foreign definitions are
1046 available at runtime.
1047 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1048 supported. (thanks to Dan Debertin)
1049 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1050 just keywords. (thanks to Vincent Arkesteijn)
1051 * fixed bug #331: structure-class instances corresponding to
1052 DEFSTRUCT forms are now created eagerly.
1053 * fixed bug #345: backtraces from calls to undefined functions work
1054 on x86 as well. Related bug #61 is now also partially fixed on x86
1055 (backtraces from throws to unknown catch tags.)
1056 * bug fix: lambda-list parsing is now stricter vrt. order and number
1057 of lambda-list keywords.
1058 * bug fix: as specified by AMOP, an error is signalled if a
1059 class-option appears multiple times in a DEFCLASS form. (reported
1061 * bug fix: RANDOM can be compiled when the compiler derives the type
1062 of its numeric argument as a disjoint set of small integers.
1063 (reported by Paul Dietz)
1064 * bug fix: starting a core saved with shared objects loaded when
1065 those objects are not available no longer causes threaded SBCL to
1066 hang. (reported by Sean Ross)
1067 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1069 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1070 argument, which must always be NIL. (reported by Kalle Niemitalo)
1071 * bug fix: printing 1.0d+23 no longer results in an error.
1072 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1073 * bug fix: structure slot setters preserve evaluation order. (thanks
1075 * fixed some bugs related to Unicode integration:
1076 ** RUN-PROGRAM can allow its child to take input from a Lisp
1077 stream. (reported by Stefan Scholl)
1078 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1079 locales. (reported by Ken Causey)
1080 * fixed some bugs revealed by Paul Dietz' test suite:
1081 ** INCF, DECF and REMF evaluate their place form as specified in
1083 ** FORMATTER expands ~{ ~} iteration directives with V or #
1084 parameters correctly.
1085 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1086 consequent uses no arguments correctly.
1087 ** the system has a partial understanding of the (COMPLEX RATIO)
1089 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1092 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1093 * new feature: a build-time option (controlled by the :SB-UNICODE
1094 keyword feature, enabled by default) for building the system with
1095 support for the entire 21-bit character space defined by the
1097 * new feature: the system now has rudimentary external-format
1098 support; the primary user-visible change at this time is that
1099 characters with the high bit set (such as lower-case-e-acute) will
1100 print correctly to a terminal in a UTF-8 environment.
1101 * minor incompatible change: BASE-CHAR no longer names a class;
1102 however, CHARACTER continues to do so, as required by ANSI.
1103 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1104 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1105 has been moved to the SB-EXT package (temporarily re-exported from
1107 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1108 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1110 * fixed bug #351: better error-handlind and reporting for malformed
1112 * fixed bug #350: bignum-printing is now more memory-efficient,
1113 allowing printing of very large bignums, eg. (expt 2 10000000).
1114 (reported by Bruno Haible)
1115 * fixed bug #302: better primitive-type selection for intersection
1117 * fixed bug #308: non-graphic characters now all have names, as
1118 required. (reported by Bruno Haible)
1119 * bug fix: redefining a class with different superclasses now correctly
1120 removes it from the direct-subclasses of its previous superclasses.
1121 (reported by David Morse)
1122 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1123 new value now works. (reported by Bruno Haible)
1124 * bug fix: correct canonicalization of multiple non-standard slot
1125 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1126 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1127 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1128 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1129 now exists, an signals an error.
1130 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1131 during macroexpansion for non-iteration variables as well. (reported
1132 by Bruno Haible for CMUCL)
1133 * bug fix: Cyclic structures and unprintable objects in compiler
1134 messages no longer cause errors. (reported by Bruno Haible)
1135 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1136 and READ-FROM-STRING all now return a primary value of NIL if
1137 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1138 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1139 (reported by Bruno Haible for CMUCL)
1140 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1141 do not designate a condition. (reported by Bruno Haible for
1143 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1144 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1145 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1146 permitting resolution in favour of any of the conflicting symbols.
1147 (reported by Bruno Haible for CMUCL)
1148 * bug fix: EQUAL compiler optimizations is less aggressive on
1149 strings which can potentially compare true despite having distinct
1150 specialized array element types.
1151 * bug fix: unit enumerations can be defined without dividing by
1152 zero. (thanks to Vincent Arkesteijn)
1153 * FORMAT compile-time argument count checking has been enhanced.
1154 (report from Bruno Haible for CMUCL)
1155 * a partial workaround for the bug 262: the compiler does not try to
1156 inline-expand a local function doing RETURN-FROM from a deleted
1157 BLOCK. (thanks to Peter Denno for the bug report and to David
1158 Wragg for the simple test case)
1159 * fixed some bugs revealed by Paul Dietz' test suite:
1160 ** RENAME-PACKAGE allows all package designators as new package
1162 ** constraint propagation and lambda variable substitution are
1163 more cautious in dealing with partially deleted code.
1164 ** compiler sometimes lost reoptimization passes.
1165 ** CERROR, given a condition as condition designator, passes its
1166 remaining arguments to the continue format control without
1168 ** Case-altering FORMAT directives work correctly on non-ASCII
1170 ** The REAL type specifier handles bounds outside the floating
1171 point ranges without signalling FLOATING-POINT-OVERFLOW.
1172 ** Functions with IR1-transformations can create intercomponent
1173 references to global functions.
1174 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1176 ** FORMAT ~R treats a nil value for its first parameter correctly.
1178 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1179 * enhancement: saving cores with foreign code loaded is now
1180 supported on x86/NetBSD and sparc/Linux in addition to the previously
1181 supported platforms.
1182 * bug fix: on some platforms repeated installations caused multiple
1183 copies of HTML documentation to be installed -- should not happen
1184 any more. (reported by Stefan Scholl)
1185 * bug fix: parsing self-recursive alien record types multiple times
1186 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1187 original patch by Helmut Eller for CMUCL)
1188 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1189 (thanks to Richard Kreuter)
1190 * bug fix: defining classes whose accessors are methods on existing
1191 generic functions in other (locked) packages no longer signals
1192 bogus package lock violations. (reported by François-René Rideau)
1193 * bug fix: special variables as DEFMETHOD parameters no longer have
1194 associated bogus type declarations. (reported by David Wragg and
1196 * bug fix: read-write consistency on streams of element-type
1197 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1198 * bug fix: redefiniton of the only method of a generic function with
1199 no DEFGENERIC no longer emits a full WARNING. In addition,
1200 redefinition of generic functions with no DEFGENERIC to an
1201 incompatible lambda list now signals an error. (thanks to Zach
1203 * bug fix: DEFGENERIC now works even when there's a function of the
1204 same name in an enclosing lexical environment. (thanks to Zach
1206 * fixed compiler failure, caused by instrumenting code during
1207 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1208 * optimization: added loop analysis and improved register allocation
1209 * fixed some bugs revealed by Paul Dietz' test suite:
1210 ** POSITION on displaced vectors with non-zero displacement
1211 returns the right answer.
1212 ** (SIMPLE-STRING) is a valid type specifier for sequence
1214 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1216 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1217 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1219 ** The justification version of the FORMAT ~< directive treats
1220 non-zero minpad parameter correctly.
1222 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1223 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1224 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1225 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1226 the supported interface.
1227 * new feature: Single-stepping of code compiled with DEBUG 2 or
1228 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1229 * new feature: saving cores with foreign code loaded is now
1230 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1231 Timothy Moore's work for CMUCL)
1232 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1233 parameters to * instead of NIL if no initform is supplied.
1234 (reported by Johan Bockgaard)
1235 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1236 correctly when FUNCALL appears as the car of the form. Note:
1237 despite this FUNCALL forms are not currently subject to
1238 compiler-macro expansion. (port of Raymond Toy's fix for the
1239 same from CMUCL, reported by Johan Bockgaard)
1240 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1241 (thanks for Teemu Kalvas)
1242 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1243 case instead of terminating the iteration (thanks for Julian
1244 Squires, Sean Champ and Raymond Toy)
1245 * bug fix: incorrect expansion of defgeneric that caused a style
1246 warning. (thanks for Zach Beane)
1247 * bug fix: slot accessor effective method computation works properly
1248 for classes with multiple non-standard applicable methods on
1249 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1250 * on x86 compiler supports stack allocation of results of LIST and
1251 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1252 CMUCL implementation by Gerd Moellmann)
1253 * fixed some bugs revealed by Paul Dietz' test suite:
1254 ** FORMAT strings with both the ~<~:;~> form of the justification
1255 directive and pretty-printing directives cause an error.
1257 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1258 * incompatible change: the internal functions
1259 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1260 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1261 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1262 instead of the old functions.
1263 * new feature: on platforms where "dladdr" is available foreign
1264 function names now appear in backtraces. (based on Helmut Eller's
1266 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1267 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1269 * documentation: ASDF manual is now distributed with SBCL.
1270 * bug fix: SBCL can now load its contributed modules with REQUIRE
1271 even if the system-provided entries have been removed from
1272 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1274 * bug fix: backtraces involving undefined functions or assembly
1275 routines are more informative. (thanks to Brian Downing)
1276 * bug fix: mutually referent alien structures now work correctly.
1277 (reported by Rick Taube)
1278 * bug fix: structures defined by WITH-ALIEN can be referred to
1279 within other definitions in the same WITH-ALIEN.
1280 * bug fix: division operators (MOD, TRUNCATE and the like) with
1281 constant zero divisors and integer dividends no longer generate
1283 * bug fix: provide default methods for INPUT-STREAM-P and
1284 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1285 * bug fix: improve the detection and resolution of MOP metacycles
1286 (where effective method computation depends on the generic function
1287 having its effective method computed). (reported by Bruno Haible)
1288 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1289 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1290 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1291 * bug fix: pathnames with relative directory components can be
1292 represented relative to default pathnames.
1293 * optimization: in taking the GCD of bignums, reduce the two bignums
1294 to approximately the same size (using Euclid's algorithm) before
1295 applying the more sophisticated binary GCD. (thanks to Juho
1297 * optimization: COUNT on bitvectors now operates word-at-a-time.
1298 * optimization: ASH with a positive, but not necessarily constant,
1299 (leftwards) shift, when in a modular context, is compiled to a
1301 * fixed some bugs revealed by Paul Dietz' test suite:
1302 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1303 if the corresponding argument is NIL.
1304 ** FORMAT directives accepting a minpad parameter treat negative
1306 ** literal commas as character parameters to FORMAT directives are
1308 ** literal spaces directly after ~<Newline> directives within a
1309 format-logical-block (~:< ~@:>) do not induce :FILL-style
1310 conditional newlines.
1311 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1312 as specified: it no longer includes conditional newlines).
1313 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1315 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1316 :PER-LINE-PREFIX argument does not evaluate to a string.
1318 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1319 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1320 the manual for details; for now, package locks can be disabled by
1321 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1322 if no major problems are found then it is likely that they will be
1323 compiled in unconditionally.
1324 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1325 unsupported operators on all platforms. To load a shared library
1326 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1327 object file, link it into a shared library outside of SBCL and
1328 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1330 * bug workaround: ROOM T and the GC were not working together
1331 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1332 reporting facility aren't preserved. That reporting has been
1333 disabled, so now until and unless someone figures out how to make it
1334 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1335 * minor incompatible change: as threatened around sbcl-0.8.0, the
1336 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1337 The symbols are also exported from SB-PCL for backwards
1338 compatibility, but more so than before SB-PCL should be treated as
1339 an implementation-internal package.
1340 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1342 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1343 It is known as of this release not to work on the Alpha, however.
1344 * fixed bug #167: errors signalled due to illegal syntax in method
1345 bodies are now more legible.
1346 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1347 designators and can hence be used with TYPEP.
1348 * fixed bug #333: CHECK-TYPE now ensures that the type error
1349 signalled, if any, has the right object to be accessed by
1350 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1351 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1352 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1354 * fixed bug #334: programmatic addition of slots using specialized
1355 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1356 and :ALLOCATION :CLASS slots.
1357 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1358 float exponents. (rereported by Peter Seibel)
1359 * fixed a bug: #\Space (and other whitespace characters) are no
1360 longer considered to be macro characters in standard syntax by
1361 GET-MACRO-CHARACTER.
1362 * fixed bug: initialization of condition class metaobjects no longer
1363 causes an instance of the condition to be created. (reported by
1365 * fixed bug: it is now possible to have slots such that
1366 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1367 description is neither :INSTANCE nor :CLASS.
1368 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1369 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1370 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1371 non-standard SLOT-DEFINITION-ALLOCATION values.
1372 * fixed bugs: various race conditions handling exiting threads.
1373 CL-PPCRE's thread torture test now passes.
1374 * fixed arguably-a-bug: GC time proportional to number of threads in
1375 system even when most of them are idle
1376 * optimization: improved performance of BIT and SBIT on bit-vectors.
1377 * .fasl file incompatibility: The fasl file version number has
1378 been incremented because of changes associated with package locks.
1380 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1381 * minor incompatible change: the system no longer provides
1382 optimization hints (conditions which are TYPEP
1383 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1384 mode; these hints are still emitted when higher SPEED optimization
1386 * new contrib module: a sampling profiler (profiling by statistical
1387 sampling, rather than by instrumenting functions) is available as
1388 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1389 * the behaviour of the standard function ED is now customizeable by
1390 third parties through a hook variable: see ED's documentation
1391 string for information on the protocol.
1392 * the compiler no longer emits efficiency notes for (FUNCALL X)
1393 when the type of X is uncertain under default optimization
1395 * fixed bug 276: mutating a binding of a specialized parameter to a
1396 method to something that is not TYPEP the specializer is now
1398 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1399 exhibit the required behaviour on the x86 platform. (thanks to
1400 Peter van Eynde, Eric Marsden and Bruno Haible)
1401 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1402 even for difficult arguments. (reported by Peter Graves)
1403 * fixed bug 141a: the backquote printer now descends quoted
1405 * fixed another bug in backquote printing: no more destructive
1406 modification of the form's list structure. (reported by Brian
1408 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1409 move between its address being taken and the call to
1410 interrupt_thread, fixing a crashing race condition.
1411 * the SB-POSIX contrib implementation has been adjusted so that it
1412 no longer exhibits ridiculously poor performance when constructing
1413 instances corresponding to C structs.
1415 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1416 * minor incompatible change: the sb-grovel contrib now treats C
1417 structures as alien (in the sense of SB-ALIEN) objects rather than
1418 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1419 has implications for memory management of client code
1420 (sb-grovel-returned objects must now be manually managed) and for
1421 type safety (alien objects now have full types).
1422 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1423 used to control emission of compiler diagnostics, rather than the
1424 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1425 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1426 quality should be considered deprecated.
1427 * install.sh now installs the user manual as well
1428 * (not quite a new documentable feature, but worth considering in
1429 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1430 beginnings of a semantically meaningful condition hierarchy is
1431 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1433 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1434 designator as the defaults argument.
1435 * fixed bug: Displaced arrays whose displaced-to array has become
1436 too small now cause ARRAY-DIMENSION to signal an error, providing
1437 for safer bounds-checking. (reported by Bruno Haible)
1438 * fixed bug: DEFCLASS slot definitions with identical :READER and
1439 :WRITER names now signal a reasonable error. (reported by Thomas
1441 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1443 * fixed bug: Invalid slot specification errors now print correctly.
1444 (thanks to Zach Beane)
1445 * fixed bug 320: Shared to local slot value transfers in class
1446 redefinitions now happen corrently from superclasses as
1447 well. (reported by Bruno Haible)
1448 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1450 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1451 improper lists correctly. (reported by Bruno Haible)
1452 * fixed bug 313: source-transform for <fun-name> was erroneously
1453 applied to a call of a value of a variable with name <fun-name>.
1454 (reported by Antonio Menezes Leitao)
1455 * fixed bug 307: The obsolete instance protocol ensures that
1456 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1457 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1458 remove dead unknown-values globs from the middle of the stack.
1459 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1460 easier to resume long computations after using *BREAK-ON-SIGNALS*
1461 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1462 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1463 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1464 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1465 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1466 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1468 * fixed a (fairly theoretical) bug in string printing: if
1469 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1470 does not have array-element-type equal to the most general string
1472 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1473 function name. (thanks to Nikodemus Siivola)
1474 * fixed bug: compile-time format string checker failed on
1475 non-closed ~{. (reported by Thomas F Burdick)
1476 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1477 don't warn on legal loop code involving more than one aggregate
1478 boolean. (thanks to Nikodemus Siivola)
1479 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1480 in keyword position to MAKE-PACKAGE was not being reported as
1482 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1483 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1485 * fixed bugs 280 and 312: the checking for multiple definitions in a
1486 file is less likely to become confused by uses of inline
1488 * fixed bug: the #S reader macro performs the keyword coercion
1489 specified for slot names. (reported by Kalle Niemitalo)
1490 * fixed bug: lambda lists may contain symbols whose names start with
1491 & but are not lambda-list-keywords; their occurrence triggers a
1493 * fixed bug 321: define-method-combination argument lambda lists do
1494 not cause invalid code to be generated when &optional or &aux
1495 variables have default values. (reported by Bruno Haible)
1496 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1497 have CLOS classes; this appears also to have fixed reported
1498 instability in stack exhaustion detection.
1499 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1500 MUFFLE-WARNING when no associated restart is present is now
1502 * optimization: rearranged the expansion of various defining macros
1503 so that each expands into only one top-level form in a
1504 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1506 * optimization: used a previously little-used slot in symbols to
1507 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1509 * fixed some bugs revealed by Paul Dietz' test suite:
1510 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1512 ** Fixed an optimization bug related to inheritance of initforms
1513 from local to shared slots.
1514 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1515 ** CAST optimizer forgot to flush argument derived type.
1516 ** print/read consistency on floats is now orders of magnitude
1517 more likely. (thanks also to Bruno Haible for a similar report
1519 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1520 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1521 for objects of type REAL. Make it so.
1522 ** SXHASH is specified (infelicitously) to respect similarity,
1523 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1524 it so. (thanks to Markus Ziegler)
1525 ** on the Alpha, the compiler succeeds in compiling functions
1526 returning a known number of arguments greater than 63.
1527 ** fixed handling of invalid NIL arguments in keyword position
1529 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1531 ** provided a readably-printable representation for RANDOM-STATE
1533 ** ensured that pathnames, where they have a namestring, always
1534 print using #P"..." syntax.
1536 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1537 * Support for the forthcoming 2.0 version of the NetBSD kernel
1538 running on x86 hardware has been added. (thanks to Perry
1539 E. Metzger most immediately, and others for their past work)
1540 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1541 current release version; SBCL's OpenBSD support had been broken
1542 since about the time of OpenBSD's switch to ELF binary format.)
1543 * [placeholder for DX summary]
1544 ** user code with &REST lists declared dynamic-extent, under high
1545 speed or space and low safety and debug optimization policy.
1546 * The manual has been converted to Texinfo format and the debugger
1547 chapter from the cmucl manual has been added.
1548 * A facility has been added to extract documentation strings from
1549 sbcl and store them as Texinfo-formatted snippets for inclusion in
1550 the manual (via Texinfo's @include directive)
1551 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1552 the test case to Dave Roberts)
1553 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1554 pointers. (reported by Sean Ross)
1555 * bug fix: PROFILE output is printed nicely even for large numerical
1556 values. (thanks to Zach Beane)
1557 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1558 greater than 32 handle EOF correctly.
1559 * bug fix: on X86 an immediate argument of the IMUL instruction is
1560 correctly printed in disassembly. (thanks to Lutz Euler)
1561 * bug fix: class slots in redefined classes preserve their old
1562 values. (thanks to Bruno Haible and Nikodemus Siivola)
1563 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1564 succeeds. (reported by Marco Baringer)
1565 * fixed some bugs revealed by Paul Dietz' test suite:
1566 ** READ-SEQUENCE now works on ECHO-STREAMs.
1567 ** RATIONALIZE works more according to its specification. (thanks
1570 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1571 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1572 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1573 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1574 should matter to you only if you rebind the printer control
1575 variables and then find you want different bindings in the
1576 debugger than in the ordinary execution of your program.)
1577 * The runtime build system has been tweaked to support building
1578 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1579 assembler and linker. (thanks to Nikodemus Siivola)
1580 * Unbound, undefined, undeclared variables now trigger full
1581 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1582 more often programmer error than deliberate exploitation of undefined
1584 * optimization: the hash algorithm for strings has changed to one
1585 that is less vulnerable to spurious collisions. (thanks to Juho
1587 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1588 needless bounds checking. (thanks to Juho Snellman)
1589 * optimization: implemented multiplication as a modular
1590 (UNSIGNED-BYTE 32) operation on the PPC backend.
1591 * fixed some bugs revealed by Paul Dietz' test suite:
1592 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1594 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1595 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1596 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1598 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1600 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1602 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1603 their output stream on EOF from read.
1604 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1605 have been read to end-of-file.
1606 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1608 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1609 description of determination of which consecutive characters
1611 ** Printing the "Space" character with escaping on now yields "#\\ ",
1612 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1613 ** Reading floating-point numbers with *READ-BASE* set to a number
1614 less than 10 works correctly.
1615 ** Reading floating-point numbers with *READ-BASE* set to a number
1616 more than 10 works correctly.
1617 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1618 the readtable currently in effect.
1620 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1621 * minor incompatible change: parsing of namestrings on a physical
1622 (Unix) host has changed; numbers after the final #\. in a
1623 namestring are no longer interpreted as a version field. This is
1624 intented to be largely invisible to the user, except that the
1625 meaning of the namestring "*.*.*" has changed: it now refers to a
1626 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1627 should usually be replaced by
1628 (make-pathname :name :wild :type :wild :version :wild)
1629 with the added benefit that this is more likely to be portable.
1630 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1631 option to OPEN now signals an error if the file being opened
1632 exists; this may have an impact on existing code.
1633 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1634 signals received, on the PowerPC platforms (both Linux and
1635 Darwin). (thanks to Pierre Mai for pointing out the location of
1637 * several fixes on OS X: The system now builds and runs cleanly on
1638 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1639 (thanks to Brian Mastenbrook)
1640 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1641 second-resolution: integer multiples of 1/3600 between -24 and 24.
1642 (thanks to Vincent Arkesteijn)
1643 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1644 type when called with 1 argument; PEEK-CHAR checked type of
1645 PEEK-TYPE only after having read first character from a
1646 stream. (reported by Peter Graves)
1647 * bug fix: the garbage collector now has much better locality
1648 behaviour, and in particular no longer treats all memory as being
1649 exhausted when many small objects point to each other in a deeply
1651 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1652 recognized as being TYPEP their class.
1653 * bug fix: the PUSHNEW documentation string has been corrected.
1654 (thanks to Vincent Arkesteijn)
1655 * bug fix: defaulting of the value for the last of an atypically
1656 large number of multiple values being bound was not being
1657 performed correctly on the Alpha or PPC platforms
1658 * optimization: implemented multiplication as a modular
1659 (UNSIGNED-BYTE 32) operation on the x86 backend.
1660 * optimization: SEARCH on simple-base-strings can now be open-coded.
1661 (see also contrib/compiler-extras.lisp for inspiration for
1662 teaching the compiler about the Boyer-Moore algorithm).
1663 * value, returned by MAX (and MIN) called with several EQUALP, but
1664 not EQL, arguments now does not depend on compiler settings.
1665 * fixed some bugs revealed by Paul Dietz' test suite:
1666 ** in stack analysis liveness information is propagated from
1667 non-local entry points.
1668 ** pathwise CAST removing failed when the CAST node did not start
1670 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1671 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1673 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1674 host is already defined.
1675 ** RENAME-FILE works on streams instead of signalling an internal
1677 ** PEEK-CHAR uses the current readtable when determining whether
1678 or not a character is whitespace.
1679 ** MERGE-PATHNAMES handles the case when the pathname does not
1680 specify a name while the default-pathname specifies a version.
1681 ** Pathnames now stand a better chance of respecting print/read
1683 ** Attempting to use standardized file system operators with a
1684 pathname with invalid :DIRECTORY components signals a
1686 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1688 ** DIRECTORY on logical pathnames is more correct.
1689 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1690 signal a TYPE-ERROR if their argument is not a stream.
1691 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1692 designator argument does not designate a stream.
1693 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1694 examining the synonym.
1695 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1697 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1698 element-type larger than ([UN]SIGNED-BYTE 32).
1700 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1701 * When built with the :SB-FUTEX feature, threaded builds now take
1702 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1703 for faster/more reliable mutex and condition variable support.
1704 * Incompatible change (but one you probably shouldn't have been using
1705 anyway): the interface and code for arbitrating between multiple
1706 threads in the same user session has been redesigned.
1707 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1708 error when called without an explicit environment argument.
1709 (thanks to Nikodemus Siivola)
1710 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1711 confused about their position. (thanks to Adam Warner and Gerd
1713 * bug fix: Pretty printing backquoted forms with unquotations in the
1714 argument list position of various code constructs such as LAMBDA
1715 now works correctly. (reported by Paul Dietz)
1716 * bug fix: Pretty printing unquotations no longer loses all
1717 stream position information.
1718 * optimization: performance of string output streams is now less
1719 poor for multiple small sequence writes.
1720 * optimization: performance of CSUBTYPEP in the presence of complex
1721 expressions involving CONS and NOT many times has been improved.
1722 (reported by Paul Dietz)
1723 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1725 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1726 streams for the REPL, for the debugger, and so on. (thanks to
1728 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1729 * fixed some bugs revealed by Paul Dietz' test suite:
1730 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1731 is ignored if it is too small, rather than propagating through
1732 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1733 ** extremely complex negations of CONS types were not being
1734 sufficiently canonicalized, leading to inconsistencies in
1736 ** VALUES tranformer lost derived type.
1738 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1739 * fixed a bootstrapping bug: the build process no longer assumes
1740 that the various BOOLE-related constants have the same value in
1741 host and target lisps. (noted by Paul Dietz' test suite on an
1742 SBCL binary built from CLISP)
1743 * The system can now be dynamically linked on the MIPS platform,
1744 which enables dynamic loading of foreign code from Lisp. (thanks
1745 to Ralf Baechle for discussions on the MIPS ABI)
1746 * The system now records debugging information for its own source
1747 files in a filesystem-position-independent manner, relative to
1748 the "SYS" logical host.
1749 * fixed a compiler bug: MV-LET convertion did not check references
1750 to the "max args" entry point. (reported by Brian Downing)
1751 * tweaked disassembly notes to be less confident about proclaiming
1752 some instruction as an LRA. (thanks to Brian Downing)
1753 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1754 now each have their own history, command character, and other
1755 characteristics. (thanks to David Lichteblau)
1756 * fixed some bugs revealed by Paul Dietz' test suite:
1757 ** compiler failure in compiling LOGAND expressions including a
1759 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1760 shift greater than 32.
1761 ** FLUSH-DEST did not mark blocks for type check regeneration.
1762 ** HANDLER-CASE failed to accept declarations in handler clauses
1763 in some circumstances.
1765 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1766 * New code in contrib/sb-introspect (still probably not entirely
1767 stable yet) provides some support for smart Lisp development
1768 environments like SLIME.
1769 * The conditions signalled for errors occurring when loading .fasl
1770 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1771 in a way which should help ASDF recover gracefully.
1772 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1773 stuff has been cleaned up. If you code contrib/ stuff, this might
1774 affect you, and you can look at contrib/README, contrib/STANDARDS,
1775 and/or the 0.8.4.27 diff to check.
1776 * In full calls the compiler now does not generate checks for declared
1777 argument types for all arguments.
1778 * various threading fixes
1779 ** and some experimental patches which didn't make it into
1780 the main tree for this release, but which are shipped in
1781 contrib/experimental-thread.patch as a possible fix for some
1782 failures (deadlock, spinning...) in GC-intensive multithreaded
1784 * fixed PPC build problem (source code incompatibility of different
1785 library versions): added offsetof() hackery which attempts to divine
1786 where glibc maintainers put uc_mcontext today
1787 * fixed bug 282: compiler does not trust type assertions while passing
1788 arguments to a full call.
1789 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1790 &OPTIONAL VALUES type parameter.
1791 * fix bug 214: algorithm for noting rejected templates is now more
1792 similar to that of template seletion. (also reported by rydis on
1794 * fixed bug 141b: printing backquoted information readably and prettily
1795 inserts a space where necessary.
1796 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1797 (CHARACTER 10) are now reported as errors, rather than propagated
1798 as unknown types. (reported by piso on #lisp)
1799 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1800 with values NIL and :ERROR. (thanks to Milan Zamazal)
1801 * fixed bug 191c: CLOS now does proper keyword argument checking as
1802 described in CLHS 7.6.5 and 7.6.5.1.
1803 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1804 longer raise an error; further, using a list as a for-as-arithmetic
1805 counter now raises a meaningful error.
1806 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1807 now understood by sequence creation functions such as MAKE-SEQUENCE
1809 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1810 a STRING-INPUT-STREAM.
1811 * compiler enhancement: SIGNUM is now better able to derive the type
1813 * type declarations inside WITH-SLOTS are checked. (reported by
1815 * fixed some bugs revealed by Paul Dietz' test suite:
1816 ** incorrect optimization of TRUNCATE for a positive first
1817 argument and negative second.
1818 ** compiler failure in let-convertion during flushing dead code.
1819 ** compiler failure while deriving type of TRUNCATE on an
1820 interval, containing 0.
1821 ** ASH of a negative bignum by a negative bignum count now returns
1823 ** intersection of CONS types now canonicalizes properly, fixing
1824 inconsistencies in SUBTYPEP.
1826 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1827 * incompatible change: The --disable-debugger command line
1828 option now clobbers the debugger at a more fundamental
1829 level, by redefining #'INVOKE-DEBUGGER instead of by
1830 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1831 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1832 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1833 enter the debugger REPL and then suffer recursive errors
1834 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1835 while under the new system, BREAK in --disable-debugger mode
1836 terminates the system just as an unhandled error would.
1837 * fixed compiler performance when processing loops with a step >1;
1838 * bug fix: DOCUMENTATION now retrieves generic function
1839 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1840 support has been systematized, and now supports the methods
1841 specified by ANSI, along with a default method and a method for
1842 slot documentation. (reported by Nathan Froyd)
1843 * bug fix: effective methods associated with a generic function are
1844 no longer cached over a change of that generic function's method
1845 combination. (reported by Andreas Fuchs)
1846 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1847 names a non-existent directory. (thanks to Andreas Fuchs)
1848 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1849 the PPC platform, lead to stack corruption; this has been fixed.
1850 (reported by Rainer Joswig)
1851 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1852 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1853 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1854 arithmetic in general, is now much more reliable on the PPC
1856 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1857 platform now returns the right answer.
1858 * optimization: restored some effective method precomputation in
1859 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1860 precomputation is now tunable.
1861 * optimization: compiler-internal data structure use has been
1862 reviewed, and changes have been made that should improve the
1863 performance of the compiler by about 20%.
1864 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1865 simple-base-strings and simple-bit-vectors is improved.
1866 * optimization: the optimization of 32-bit logical and arithmetic
1867 functions introduced in version 0.8.3 on the x86 has been
1868 implemented on the mips, ppc and sparc platforms; an
1869 implementation of the same facility, but for 64-bit arithmetic,
1870 has been added for the alpha.
1871 * microoptimization: the compiler is better able to make use of the
1872 x86 LEA instruction for multiplication by constants.
1873 * bug fix: in some situations compiler did not report usage of
1874 generic arithmetic in (SPEED 3) policy.
1875 * bug 145b fix: compiler used wrong type specifier while converting
1876 MEMBER-types to numeric.
1877 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1879 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1880 index argument now works properly on the Alpha platform.
1881 * bug fix: floating point exception treatment on the Alpha platform
1883 * bug fix: FILE-POSITION works much better on string input and
1884 output streams. (thanks to Nikodemus Siivola)
1885 * bug fix: many threading/garbage collection symptoms sorted.
1886 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1887 might be pseudo-atomic.
1888 * internal change: Stopping for GC is now done with signals not
1889 ptrace. GC is now done in whichever thread wanted it, instead of
1891 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1893 * bug fix: result form in DO is not contained in the implicit
1895 * incompatible change: ICR structure is changed; the value part of
1896 CONTINUATION is now called LVAR; corresponding functions are
1897 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1898 * added type deriver for ISQRT (thanks to Robert E. Brown).
1899 * bug fix: better support for loading from the command line when an
1900 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1902 * fixed some bugs revealed by Paul Dietz' test suite:
1903 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1904 ** ROUND and FROUND now give the right answer when given very
1905 small float arguments.
1906 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1908 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1909 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1910 ** LCM should return a non-negative integer.
1911 ** PARSE-INTEGER returned the index of a terminator instead of the
1912 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1913 ** PARSE-INTEGER returned an incorrect index being applied to a
1915 ** LCM with two arguments of 0 returns 0 rather than signalling
1917 ** unsigned addition of a 32-bit constant with the high bit set no
1918 longer causes an internal compiler error on the x86.
1919 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1920 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1921 with negative last argument.
1922 ** byte specifiers with zero size and position no longer cause
1923 an error during type derivation.
1924 ** bignum multiplication on the Alpha platform now returns the
1926 * porting: The system now builds on SuSE AMD64, although it still
1927 generates a 32-bit binary.
1928 * .fasl file incompatibility: The fasl file version number has
1929 been incremented (because of the changes to internal compiler
1930 data structures referred to above).
1932 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1933 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1934 more accurately, on the Darwin kernel running on PowerPC hardware.
1935 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1936 * Compiler code deletion notes now signal a condition of type
1937 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1938 an associated MUFFLE-WARNING restart.
1939 * The compiler now performs limited argument count validation of
1940 constant format strings in FORMAT, and where appropriate in ERROR,
1941 CERROR and WARN. (thanks to Gerd Moellmann)
1942 * New ASDF-INSTALL contrib can be used for automatic download and
1943 installation of third-party Lisp code from CCLAN or other sites
1945 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1946 which forces another thread to execute a function supplied by the
1948 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1949 now accept and act upon their :ELEMENT-TYPE keyword argument.
1950 (reported by Martin Atzmueller, Edi Weitz)
1951 * bug fix: FILE-POSITION now accepts position designators up to
1952 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1953 is the greater. (thanks to Patrik Nordebo)
1954 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1955 Kalle Olavi Niemitalo)
1956 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1957 printable. (reported by Eric Marsden)
1958 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1959 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1960 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1961 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1962 (thanks to Stig Sandoe)
1963 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1964 UNREAD-CHAR now correctly includes the unread character in the
1965 target sequence. (thanks to Gerd Moellmann)
1966 * bug fix in threaded builds: the system can now be suspended and
1967 resumed by shell job control with minimal disruption.
1968 * bug fixes in times and timezones >2038AD
1969 * better handling of "where is GNU make?" problem in build scripts
1970 (thanks to Nikodemus Siivola)
1971 * new optimization: inside a named function any reference to a
1972 function with the same name is considered to be a self-reference;
1973 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1974 optimization quality.
1975 * new optimization on x86: logical functions and + now have
1976 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1977 used when the result is truncated to 32 bits.
1978 * VALUES declaration is partially enabled.
1979 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1980 * fixed some bugs revealed by Paul Dietz' test suite:
1981 ** The system now obeys the constraint imposed by
1982 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1983 types form a lattice under type intersection.
1984 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1985 ** ASSOC now ignores NIL elements in an alist.
1986 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1987 and (1+ MOST-POSITIVE-FIXNUM) answers.
1988 ** The addition of a method with invalid qualifiers to a generic
1989 function does not cause an error to be signalled immediately;
1990 a warning is signalled, and the error is generated only on
1991 calling the generic function.
1992 * changed .fasl file version number, in part to add type codes for
1993 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1994 obscure ANSI requirements
1996 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1997 * fixed bug 148: failure to inline-expand a local function left
1998 garbage, confusing the compiler.
1999 * fixed bugs 3cd: structure slot readers perform type check if the
2000 slot can have an invalid value (i.e. it is either not initialized
2001 or can be written with a less specific slot writer).
2002 * bug fix: the compiler now traps array references to elements off
2003 the end of an array; previously, the bounds checking in some
2004 circumstances could go off-by-one.
2005 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2007 * type declarations for array element types now obey the description
2008 on the CLHS page "Declaration TYPE", as per discussions on
2009 sbcl-help around 2003-05-08. This means that a declaration
2010 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2011 declaration, all references to BAR will be asserted or assumed
2012 (with THE, so dependent on compiler policy) to involve objects of
2013 type FOO. Note that no such declaration is implied in
2014 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2015 * declared types of functions from the "Conditions"
2016 chapter. (reported by Paul Dietz)
2017 * bug fix: CERROR accepts a function as its first argument.
2018 * bug fix: NTH an NTHCDR accept a bignum as index
2019 arguments. (reported by Adam Warner)
2020 * optimization: character compare routines now optimize comparing
2021 against a constant character. (reported by Gilbert Baumann)
2022 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2023 argument now works properly on the MIPS platform.
2024 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2025 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2026 declared to have a functional type, does not cause an error
2028 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2029 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2030 * fixed bug: if last continuation of a deleted block has a
2031 destination, this destination should be deleted too. (reported by
2033 * fixed a bug in the bootstrap process: the host compiler's values
2034 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2035 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2036 test case from Patrik Nordebo)
2037 * improved the ability of the disassembler on the PPC platform to
2038 provide helpful disassembly notes.
2039 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2040 the class in more cases than previously.
2041 * bug fix: FILE-POSITION now understands :START and :END for
2042 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2043 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2044 8) data. (thanks to David Lichteblau)
2045 * bug fix: it is possible to add a method to a generic function
2046 without lambda list.
2047 * bug fix: reader failed to signal END-OF-FILE inside an
2048 object representation. (reported by Nikodemus Siivola)
2049 * fixed some bugs revealed by Paul Dietz' test suite:
2050 ** LAST and [N]BUTLAST should accept a bignum.
2051 ** condition slot accessors are methods.
2052 ** (VECTOR NIL) is a subtype of STRING.
2054 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2055 * minor incompatible change: some nonsensical specialized lambda
2056 lists (used in DEFMETHOD) which were previously ignored now signal
2058 * minor incompatible change: the system is now aware of the types of
2059 variables in the COMMON-LISP package, and will signal errors for
2060 most violations of these type constraints (where previously they
2061 were silently accepted).
2062 * minor incompatible change: COMPILE-FILE now uses the freedom
2063 afforded (ANSI 3.2.2.3) to use derived function types for
2064 functions defined in the same file. This also permits the system
2065 to warn on static type mismatches and function
2066 redefinition. (Currently it does not work with high DEBUG level.)
2067 * minor incompatible change: VALUES declaration is disabled.
2068 * When issuing notes, the compiler now signals a condition of type
2069 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2070 restart for use in user handlers. It is expected that the
2071 COMPILER-NOTE condition will eventually become a condition
2072 supertype to a hierarchy of note types, which will then be
2073 handleable in a similar fashion. However, at the moment, no such
2074 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2075 but it's an implementation detail, not a classification for the
2077 * Changes in type checking closed the following bugs:
2078 ** type checking of unused values (192b, 194d, 203);
2079 ** template selection based on unsafe type assertions (192c, 236);
2080 ** type checking in branches (194bc).
2081 * A short form of VALUES type specifier has ANSI meaning (it has
2082 increased the number of situations when SBCL cannot perform type
2084 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2085 keywords or constants is permissible.
2086 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2087 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2088 classes. (thanks to Antonio Martinez)
2089 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2090 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2091 argument to a pretty-print stream. Instead, it leaves any such
2092 operations to DESCRIBE-OBJECT methods.
2093 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2095 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2096 lambda lists. (thanks to David Lichteblau)
2097 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2098 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2099 (thanks to Gerd Moellmann)
2100 * fixed bug 246: increased compilation speed of long
2101 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2103 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2105 * DEFCONSTANT now throws a condition of type
2106 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2107 constant to a non-EQL value; CONTINUE and ABORT restarts
2108 respectively change and preserve the value.
2109 * fixed bug 63: The code walker, part of the implementation of CLOS,
2110 is now better at handling symbol macros.
2111 * bug fix: There is no longer an internal implementation type named
2112 CL:LENGTH. (reported by Raymond Toy)
2113 * bug fix: In macro-like defining macros/special operators the
2114 implicit block does not enclose the lambda list.
2115 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2116 enclosed in parentheses) are not suitable as type specifiers, and
2117 their use properly signals an error now.
2118 * bug fix: An argument count mismatch for a type specifier in code
2119 being compiled no longer causes an unhandled error at compile
2120 time, but signals a compile-time warning.
2121 * fixed simple vector readable printing
2122 * bug fix: DESCRIBE takes more care over whether the class
2123 precedence list slot of a class is bound before accessing it.
2124 (reported by Markus Krummenacker)
2125 * bug fix: FORMATTER can successfully compile pretty-printer format
2126 strings which use variants of the ~* directive inside.
2127 * bug fix: SEARCH now applies its TEST predicate to the elements of
2128 the arguments in the correct order. (thanks to Wolfhard Buss)
2129 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2130 in type inference. (thanks to Robert E. Brown)
2131 * bug fix: Reading in symbols with an explicit package name of ""
2132 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2133 * fixed some bugs revealed by Paul Dietz' test suite:
2134 ** NIL is now allowed as a structure slot name.
2135 ** Arbitrary numbers, not just REALs, are allowed in certain
2136 circumstances in LOOP for-as-arithmetic clauses.
2137 ** Multiple class redefinitions before slot access no longer
2138 causes a type error.
2139 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2140 association between the name and a class.
2141 ** Generic functions with non-standard method-combination and over
2142 five methods all of which return constants no longer return NIL
2143 after the first few invocations. (thanks to Gerd Moellmann)
2144 ** CALL-NEXT-METHOD with no arguments now passes the original
2145 values of the arguments, even in the presence of assignment.
2146 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2148 ** DISASSEMBLE works with closures and funcallable instances.
2149 ** ADD-METHOD now returns the generic function, not the new method.
2150 ** FIND-METHOD signals an error if the lengths of the specializers
2151 is incompatible with the generic function, even if the ERRORP
2153 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2154 which its argument is a member.
2155 ** DEFCLASS only redefines the class named by its class-name
2156 argument if that name is the proper name of the class;
2157 otherwise, it creates a new class.
2158 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2159 of the UNBOUND-SLOT condition to the name of the slot.
2160 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2161 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2162 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2163 SLOT-MAKUNBOUND in the specified fashion.
2165 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2166 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2167 cross-compilation host. As a consequence, we can now bootstrap our
2168 way up to SBCL starting with a bare gcc toolchain and human-readable
2169 source code (first the source to CLISP, then the source to SBCL).
2170 * A contributed module containing a partial implementation of the
2171 simple-streams interface has been included. (thanks to Rudi
2173 * A contributed module implementing the RFC1321 Message Digest
2174 Algorithm, known as MD5, has been included.
2175 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2176 no longer has any effect, as the code controlled by this feature
2177 has been deleted. (As far as we know, no-one has ever built using
2178 this feature, and its semantics were confused in any case).
2179 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2180 work on conditions (as required by the ANSI specification),
2181 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2182 expected behaviour on conditions. Users should note, however,
2183 that such behaviour is not required by the ANSI specification,
2184 and so use of this behaviour may render their code unportable.
2185 * fixed some bugs revealed by Paul Dietz' test suite:
2186 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2188 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2189 ** SXHASH on condition objects no longer returns NIL.
2190 ** :ALLOCATION :CLASS slots are better treated; their values are
2191 updated on class redefinition, and initforms inherited from
2192 superclasses are applied.
2193 ** REMOVE-METHOD returns its generic function argument even when
2194 no method was removed.
2195 ** SHARED-INITIALIZE now initializes the values of the requested
2196 slots, including those with :ALLOCATION :CLASS.
2197 ** ALLOCATE-INSTANCE now works on structure classes defined via
2198 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2200 ** SLOT-EXISTS-P now works on conditions, as well as structures
2202 ** MAKE-LOAD-FORM now has the required methods on
2203 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2204 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2205 keyword, and now implements the SLOT-NAMES argument.
2206 ** methods with &OPTIONAL arguments no longer allow too many
2207 arguments to be passed in the call without error.
2208 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2209 option is consistent with the required arguments of the generic
2210 function lambda list.
2211 * bug fix: REQUIRE accepts a string designator. (Thanks to
2213 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2214 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2215 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2216 * bug fix: The compiler now checks for duplicated variables in macro
2218 * bug fix: SETQ on globals returns the correct value.
2219 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2220 not a COMPILER-ERROR (followed by some other strange error on
2221 choosing the CONTINUE restart).
2222 * bug fix: make.sh and friends are now more consistent in the way that
2223 they look for GNU "make".
2225 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2226 * experimental native threads support (on x86 Linux >=2.4 only).
2227 This is not compiled in by default: you need to add :SB-THREAD to
2228 the target features. See the "Beyond ANSI" chapter of the manual
2230 * fix for longstanding nonANSIism: The old distinction between
2231 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2232 The return value from CL:FIND-CLASS is now a CLOS class, and
2233 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2234 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2235 * An interface to the MetaObject Protocol, as described in Kiczales,
2236 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2237 MIT Press, 1991, is available from the SB-MOP package.
2238 * incompatible change: the SB-PCL package should now be considered
2239 a private implementation detail, and no longer a semi-private MOP
2241 * minor incompatible change: due to rearrangement for threads, the
2242 control stack and binding stack are now allocated at arbitrary
2243 addresses instead of being hardcoded per-port. Users affected by
2244 this probably have to be doing advanced things with shared
2245 libraries, and will know who they are.
2246 * minor incompatible change: Previously, all --eval forms used were
2247 processed with READ before any of them were processed with EVAL.
2248 Now each --eval form is processed with both READ and EVAL before
2249 the next --eval form is processed. (Thus package operations like
2250 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2251 work as the user might reasonably expect.)
2252 * minor incompatible change: *STANDARD-INPUT* is now only an
2253 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2255 * minor incompatible change: Y-OR-N-P is now character-oriented,
2256 not line oriented. Also, YES-OR-NO-P now works without errors.
2257 (thanks to Antonio Martinez)
2258 * sb-aclrepl module improvements: an integrated inspector, added
2259 repl features, and a bug fix to :trace command.
2260 * Known functions, which cannot be open coded by the backend, are
2261 considered to be able to check types of their arguments. (fixing
2262 a bug report by Nathan J. Froyd)
2263 * fixed a bug in computing method discriminating functions: It is
2264 now possible to define methods specialized on classes which have
2265 forward-referenced superclasses. (thanks to Gerd Moellmann)
2266 * fixed evaluation order in optional entries (reported by Gilbert
2268 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2269 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2270 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2271 required-by-AMOP NAME argument, as well as CLASS and
2272 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2273 * fixed bug 20: DEFMETHOD can define methods using names that are
2274 not the proper names of classes to designate class specializers.
2275 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2276 Unix file descriptors, instead of blowing up. (thanks to Antonio
2278 * Garbage collection refactoring: user-visible change is that a
2279 call to the GC function during WITHOUT-GCING will not do garbage
2280 collection until the end of the WITHOUT-GCING. If you were doing
2281 this you were probably losing anyway.
2282 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2283 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2284 TYPEP the latter but not the former.
2285 * The compiler issues a full WARNING for calls to undefined functions
2286 with names from the CL package.
2287 * MAP-INTO for a vector destination is open coded. (reported by
2288 Brian Downing on c.l.l)
2289 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2290 documentation string.
2291 * fixed some bugs revealed by Paul Dietz' test suite:
2292 ** COPY-ALIST now signals an error if its argument is a dotted
2294 ** Condition slots are now accessed more correctly in the presence
2295 of multiple initargs for a given slot.
2296 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2297 exclude restarts of the same name associated with a different
2299 ** DEFCLASS of forward-referenced classes with another
2300 forward-referenced class in the superclasses list no longer
2302 ** Condition slots are now initialized once each, not multiple
2303 times. (thanks to Gerd Moellmann)
2304 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2306 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2307 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2308 arguments contain duplicated elements.
2309 ** RESTART-CASE understands local macros.
2310 ** RESTART-CASE associates exactly its own restarts with a condition.
2311 ** ENDP in safe mode checks its argument to be of type LIST.
2312 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2313 in question is unbound.
2314 ** Optimized MAKE-INSTANCE functions no longer cause internal
2315 assertion failures in the presence of duplicate initargs.
2316 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2317 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2318 propagated even in the ctor optimized implementation of
2320 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2322 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2323 * a better implementation of SXHASH on (simple) bit vectors,
2324 measured both in execution speed and in distribution of results
2325 over the positive fixnums, has been installed. Likewise, a better
2326 implementation of EQUAL for simple bit vectors is now available.
2327 * fixed CEILING optimization for a divisor of form 2^k.
2328 * fixed bug 240 (emitting extra style warnings "using the lexical
2329 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2330 by Antonio Martinez)
2331 * fixed SXHASH, giving different results for NIL depending on type
2332 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2333 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2334 variable. (found by Rolf Wester)
2335 * SBCL does not ignore type declarations for special
2336 variables. (reported by rif on c.l.l 2003-03-05)
2337 * some bug fixes in contrib/sb-aclrepl/
2338 * fixed some bugs revealed by Paul Dietz' test suite:
2339 ** a bug in the CONS type specifier, whereby the CAR and CDR
2340 types got intertwined, has been fixed;
2341 ** the type system is now able to reason about the interaction
2342 between INTEGER and RATIO types more completely;
2343 ** APPEND, [N]REVERSE and NRECONC check that those their
2344 arguments, which must be proper lists, are really so;
2345 ** An array specialized to be unable to hold elements has been
2346 implemented, as required -- yes, really -- by ANSI;
2347 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2348 on malformed property lists;
2350 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2351 * incompatible packaging change: in line with Unix convention,
2352 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2353 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2354 * REQUIRE and PROVIDE are now optionally capable of doing something
2355 useful. See the documentation string for REQUIRE.
2356 * infrastructure for a managed SBCL contrib system: contributed
2357 modules in this release include:
2358 ** the ASDF system definition facility;
2359 ** an interface to the BSD Sockets API;
2360 ** an ACL-like convenience interface to the repl;
2361 (thanks to Kevin Rosenberg)
2362 ** an implementation of ROTATE-BYTE, with efficient implementation
2364 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2365 gives a closer approximation to the right answer than previously.
2366 (thanks to Raymond Toy)
2367 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2368 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2369 optional environment arguments, as required by ANSI.
2370 * fixed bugs in other functions taking environment objects, allowing
2371 calls with an explicit NIL environment argument to be compiled
2373 * fixed bug 228: primary return values from
2374 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2375 COMPILE or FUNCTION.
2376 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2377 :TYPE will no longer signal errors on innocuous objects.
2378 * fixed bug 231b: SETQ is better at respecting type declarations in
2379 the lexical environment.
2380 * fixed a bug in DEFCLASS: classes named by symbols with no or
2381 unprintable packages can now be defined.
2382 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2383 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2384 * fixed compiler failure related to checking types of functions.
2385 (reported by Robert E. Brown)
2386 * the compiler is now much more consistent in its error-checking
2387 treatment of bounding index arguments to sequence functions: in
2388 (SAFETY 3) code, errors will be signalled in almost all cases if
2389 invalid sequence bounding indices are passed to functions defined
2390 by ANSI to operate on sequences.
2391 * fixed a bug in the build procedure: documentation of SBCL-specific
2392 packages is now preserved and available in the final Lisp image.
2393 * lifted FDEFINITION lookup out of loops in the implementation of
2394 many list operations. (thanks to Robert E. Brown)
2395 * fixed a bug in the reader: the #n# reader macro now works for
2396 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2397 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2398 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2399 E. Brown and Tony Martinez respectively)
2400 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2401 SLOT-BOUNDP: in particular, it is now able to optimize them much
2402 better, and is now not vulnerable to having packages renamed.
2403 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2404 always called when a slot is not present in an instance. (thanks
2406 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2407 longer signal an error on unreading a character following EOF on
2408 the previous constituent stream. (thanks to Tony Martinez)
2409 * fixed some bugs revealed by Paul Dietz' test suite:
2410 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2411 not just nonnegative fixnums;
2412 ** the logical bit-array operators such as BIT-AND now accept an
2413 explicit NIL for their "opt-arg" argument (to indicate a
2414 freshly-consed result bit-array);
2415 ** ELT now signals an error on an invalid sequence index in safe
2417 ** the type system is now cleverer about negations of numeric
2418 types, and consequently understands the BIGNUM and RATIO types
2420 ** the type system is now cleverer about the interaction between
2421 INTEGER and RATIO types: while bugs still remain, many more
2422 cases are accurately computed;
2423 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2424 if it is in the last clause;
2425 ** CONSTANTLY now correctly returns a side-effect-free function in
2427 ** DECLARE is no longer treated as a special-operator; in
2428 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2429 * incremented fasl file version number due to the change in the
2430 DEFSTRUCT-SLOT-DESCRIPTION structure.
2432 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2433 * minor incompatible change: code processed by the "interpreter" or
2434 EVAL now has a compilation optimization policy of (DEBUG 2)
2435 (changed from (DEBUG 1)) to improve debuggability of interactive
2436 development, and to allow the use of the debug RETURN command in
2438 * an experimental implementation of the RETURN command for the
2439 debugger has been included. (thanks to Frederik Kuivinen)
2440 * fixed bug 62: constraints were not propagated into a loop.
2441 * fixed bug in embedded calls of SORT (reported and investigated by
2443 * fixed some bugs revealed by Paul Dietz' test suite:
2444 ** printing and reading of arrays with some dimensions having
2445 length 0 (thanks to Gerd Moellmann);
2446 ** BOA constructor with &AUX argument without a default value does
2447 not cause a type error;
2448 ** CONSTANTP now returns true for all self-evaluating objects.
2450 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2451 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2452 accessors that are related by inheritance, as specified in the
2453 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2454 Valtteri Vuorikoski)
2455 * The compiler is now able to inline functions that were defined in
2456 a complex lexical environment (e.g. inside a MACROLET).
2457 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2459 * Support for the upcoming FreeBSD-5.0 release has been included.
2460 (thanks to Dag-Erling Smorgrav)
2461 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2462 effect when it is not in a toplevel context.
2463 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2464 stand a better chance of being correct. (thanks to Gerd
2466 * fixed bug in COERCE, which now signals an error on coercing a
2467 rational to a bounded real type which excludes the expected
2469 * The compiler is now able to derive types more accurately from the
2470 COERCE and COMPILE functions.
2471 * fixed bug 223: functional binding is considered to be constant
2472 only for symbols in the CL package.
2473 * fixed bug 231: SETQ did not check the type of a variable being set
2474 (reported by Robert E. Brown)
2475 * A new optimization for MAKE-INSTANCE has been included, fixing
2476 various bugs (including relating to :ALLOCATION :CLASS slots and
2477 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2479 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2480 ** As required by ANSI, LOOP now disallows anonymous collection
2481 clauses such as COLLECT I in conjunction with aggregate boolean
2482 clauses such as THEREIS (= I 1);
2483 ** LOOP now signals an error when any variable is reused in the
2484 same loop (including the potentially useful construct analogous
2485 to WITH A = 1 WITH A = (1+ A);
2486 ** IT is only a special loop symbol within the first clause of a
2487 conditional loop clause;
2488 ** LOOP with a typed iteration variable over a hashtable now
2489 signals a type error iff it should.
2490 * fixed some other bugs revealed by Paul Dietz' test suite:
2491 ** FILE-STREAM now names the class previously known as FD-STREAM;
2492 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2493 argument) no longer signals an error;
2494 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2495 of the slot symbol, rather than using the current package
2496 ((:CONC-NAME "") continues to intern the slot's name in the
2498 * incremented fasl file version number, because of the incompatible
2499 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2500 of the new implementation of DEFINE-COMPILER-MACRO.
2502 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2503 * Support for building SBCL for MIPS platforms running in
2504 little-endian mode has now been checked in, and basic
2505 functionality on said platforms verified.
2506 * minor incompatible change: PCL now records the pathname of a file
2507 in which methods and the like are defined, rather than its
2509 * minor incompatible change: TRUENAME now considers the truename of
2510 a file naming a directory to be the pathname with :DIRECTORY
2511 component indicating that directory.
2512 * minor incompatible change: a NAMED clause in the extended form of
2513 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2514 reason for the previous behaviour is unclear.
2515 * more systematization and improvement of CLOS and MOP conformance
2516 in PCL (thanks to Gerd Moellman and Pierre Mai):
2517 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2519 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2520 primary methods with no specializers;
2521 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2523 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2524 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2525 FORWARD-REFERENCED-CLASSes; error reporting on
2526 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2528 ** SXHASH on CLOS instances now uses a slot internal to the
2529 instance to return different numbers on distinct instances,
2530 while preserving the same return value through invocations of
2532 ** DEFMETHOD signals errors when methods with longer incongruent
2533 lambda lists are added to generic functions;
2534 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2535 CLASS, as specified in AMOP;
2536 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2537 on the order returned by the primary method for classes of
2538 class STANDARD-CLASS;
2539 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2540 * fixed some bugs shown by Paul Dietz' test suite:
2541 ** DOLIST puts its body in TAGBODY;
2542 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2544 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2545 value producing form;
2546 ** if more variables are given to PROGV than values, extra
2547 variables are bound and made to have no value;
2548 ** NSUBSTITUTE on list arguments gets the right answer with
2550 ** ELT signals an error of type TYPE-ERROR when the index argument
2551 is not a valid sequence index;
2552 ** LOOP signals (at macroexpansion time) an error of type
2553 PROGRAM-ERROR when duplicate variable names are found;
2554 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2555 ** FUNCALL of special-operators now cause an error of type
2557 ** PSETQ now works as required in the presence of side-effecting
2558 symbol-macro places;
2559 ** NCONC accepts any object as its last argument;
2560 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2562 ** loop-for-as-package does not require a package to be explicitely
2564 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2566 * fixed bug 166: compiler preserves "there is a way to go"
2567 invariant when deleting code.
2568 * fixed bug 172: macro lambda lists with required arguments after
2569 &REST arguments now cause an error to be signalled. (thanks to
2571 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2572 bug. (thanks to Matthew Danish)
2573 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2575 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2576 arguments are lexically rebound. (thanks to Gerd Moellmann and
2578 * fixed bug 194: error messages are now more informative when there
2579 is no primary method applicable in a call to a generic
2580 function. (thanks to Gerd Moellmann)
2581 * fixed bug in command line argument checking (thanks to Julian
2583 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2584 (thanks to Matthew Danish)
2585 * incremented fasl file version number, because of the
2586 SXHASH-related changes in the layout of CLOS data structures
2588 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2589 * minor incompatible change: The runtime (the Unix executable named
2590 "sbcl") is now much pickier about the .core files it will load.
2591 Essentially it now requires .core files to descend from the same
2592 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2593 as the runtime does. (The intent is to prevent the crashes which
2594 can occur, and which can even be reported as mysterious failures,
2595 when people patch the sources or change the build parameters
2596 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2597 sbcl and .core files.)
2598 * fixed bug: VALUES-LIST is no longer optimized away.
2599 * fixed bug 142: The FFI conversion of C string values to Lisp
2600 string values no longer conses excessively. (thanks to Nathan
2601 Froyd porting Raymond Toy's fix to CMU CL)
2602 * began to systematize and improve MOP conformance in PCL (thanks to
2603 Nathan Froyd, Gerd Moellman and Pierre Mai):
2604 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2606 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2607 ** FINALIZE-INHERITANCE is now called on class finalization;
2608 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2609 argument precedence order.
2610 * fixed bug 202: The compiler no longer fails on functions whose
2611 derived types contradict their declared type.
2612 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2613 so it can be non-toplevel.
2614 * The fasl file version number has changed (because of the new
2615 implementation of DEFMACRO).
2616 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2617 safe code, that any length requirement by their type-specifier
2618 argument is valid. The exceptions to this are described in bug
2620 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2621 functions now check the types of their inputs as required by ANSI.
2622 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2623 introduced symbol is DECLAREd to be SPECIAL.
2624 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2625 * fixed bug: :COUNT argument to sequence functions may be negative
2626 * fixed bug: body of DO-SYMBOLS may contain declarations
2627 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2628 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2629 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2630 symbol macro only once
2631 * fixed printing of call frame when argument list is unavailable
2632 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2633 * compiler no longer signals WARNING on unknown keyword
2636 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2637 * A beta-quality port to the mips architecture running Linux,
2638 based on the old CMUCL backend, has been made. It has been tested
2639 on a big-endian kernel, and works sufficiently well to be able to
2640 rebuild itself; it has not been tested in little-endian mode.
2641 * fixed an inconsistency between gencgc.c and purify.c which made
2642 dumping/loading .core files unreliable
2643 * fixed bug 120a: The compiler now deals correctly with IFs where
2644 the consequent is the same as the alternative, instead of
2645 misderiving the return type. (thanks to Alexey Dejneka)
2646 * fixed bug 113: Logical pathnames are now dumpable (the logical
2647 host is resolved at load-time, throwing an error if it is not
2649 * fixed bug 174: FORMAT's error message is slightly clearer when a
2650 non-printing character is used in a format directive.
2651 * fixed several bugs in compiler checking of type declarations, i.e.
2652 violations of the Python "declarations are assertions" principle
2653 (thanks to Alexey Dejneka)
2654 * fixed several bugs in PCL's error checking (thanks to Gerd
2656 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2658 * fixed bug in compilation of functions as first class values
2659 (thanks to Antonio Martinez-Shotton)
2660 * The compiler's handling TYPE-ERRORs which it can prove will
2661 inevitably happen at runtime has been cleaned up and corrected
2662 in several ways. (thanks to Alexey Dejneka)
2663 * improved argument type checking for various basic arithmetic
2664 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2665 lavished on them in the past that they can be compiled in many
2666 ways in different special cases
2667 * fixed bug 181: compiler checks validity of user supplied type
2669 * cleaned up code flushing in optimization: Function calls which
2670 should signal errors for safety purposes (e.g. which ANSI says
2671 should signal errors when their arguments are of incorrect type)
2672 are no longer optimized away.
2673 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2674 * incremented fasl file version number, because changes in the
2675 implementation of sequence functions like COERCE caused
2676 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2677 in old inline expansions) to become undefined. (Actually these
2678 changes were later undone, so we might very well be binary
2679 compatible with 0.7.7 after all, but leaving the version number
2680 incremented seemed like the simplest and most conservative
2683 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2684 * An alpha-quality port to the parisc architecture running Linux,
2685 based on the old CMUCL backend, has been made. This, even more so
2686 than the other backends, should be considered still a work in
2687 progress; known problems include that the Linux kernel in 64-bit
2688 mode does not propagate the correct sigcontext structure to
2689 userspace, and consequently SBCL on a parisc64 kernel will not
2691 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2692 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2693 suppressed.) Also now that the compiler is looking at declarations
2694 in the environment, it checks optimization declarations as well,
2695 and suppresses inlining when (> DEBUG SPEED).
2696 * More fixes have been made to treatment of floating point exception
2697 treatment and other Unix signals. In particular, floating point
2698 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2699 * The detection and handling of control stack exhaustion (infinite
2700 or very deeply nested recursion) has changed. Stack exhaustion
2701 detection is now done by write-protecting pages at the OS level
2702 and applies at all optimization settings; when found, a
2703 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2704 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2705 be caught using IGNORE-ERRORS.
2706 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2707 correctly when transforming recursive LABELS functions to LETs.
2708 (thanks to Alexey Dejneka)
2709 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2710 that are names of constants or global variables.
2711 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2712 alien routines with docstrings.
2713 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2714 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2716 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2717 (thanks to Eric Marsden)
2718 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2719 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2720 * Bug fix: The SPARC backend can now compile functions involving
2721 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2722 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2723 to a functions expecting a PATHNAME-DESIGNATOR.
2724 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2725 lambda lists. (thanks to Alexey Dejneka)
2726 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2727 functions so that EXPORTing the name of the class doesn't cause
2728 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2729 to Antonio Martinez for reporting this)
2730 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2731 constant-folded without failing an assertion. (thanks to Einar
2732 Floystad Dorum for reporting this)
2733 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2734 types are now dealt with more correctly by the compiler.
2735 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2736 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2737 behaviour analogous to RENAME-FILE. This puts its behaviour more
2738 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2740 * The fasl file version number has changed again. (because of the
2741 bug fix involving the names of PCL MAKE-INSTANCE functions)
2743 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2744 * bug fix: Floating point exceptions are treated much more
2745 consistently on the x86/Linux and PPC/Linux platforms.
2746 * Array initialization with :INITIAL-ELEMENT is now much faster for
2747 cases when the compiler cannot open code the array creation, but
2748 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2749 array accesses have also seen a speed increase.
2750 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2751 specified. (This was at the root of some bad interactions between
2752 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2754 * bug fix: Internal error arguments for undefined functions are now
2755 computed correctly on the PPC/Linux platform.
2756 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2757 Raymond Toy's patch for CMU CL)
2758 * Support for the Solaris 9 operating environment has been included
2759 (thanks to Daniel Merritt)
2760 * A very ugly but hopefully complete draft of the missing FFI chapter
2761 of the manual has been created by reformatting the corresponding
2762 CMU CL manual chapter into (currently very ugly and incoherent)
2763 DocBook and bringing it up to date for SBCL behavior. Thus, the
2764 manual is now essentially complete, at least by my extreme
2765 once-and-only-once standards, whereby it's acceptable to refer to
2766 the doc strings of SB-EXT functions as the primary documentation.
2767 * The fasl file version number has changed again, due to cleanup of
2768 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2769 is no longer a static symbol.)
2771 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2772 * SBCL now builds with OpenMCL (version 0.12) as the
2773 cross-compilation host; also, more progress has been made toward
2774 bootstrapping under CLISP.
2775 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2777 * bug 158 fixed: The compiler can now deal with integer loop
2778 increments different from 1; fixing this turned out also to fix
2780 * bug 169 fixed: no more bogus warnings about using lexical bindings
2781 despite the presence of perfectly good SPECIAL declarations (thanks
2782 to David Lichteblau)
2783 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2784 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2785 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2787 * bug 180 fixed: Method combination specifications no longer ignore
2788 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2789 * bug fix: Structure type predicate functions now check their argument
2790 count as they should.
2791 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2792 correctly. (thanks to Pierre Mai)
2793 * minor incompatible change: The --noprogrammer option is deprecated
2794 in favor of the new --disable-debugger option, which is very similar.
2795 (The major difference is that it takes effect at a slightly different
2796 time at startup, causing handling of errors in --sysinit and
2797 --userinit files will be affected differently.) The
2798 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2799 been added to allow this functionality to be controlled from ordinary
2800 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2801 maintainers, who might want to run non-interactive scripts to
2802 build SBCL cores which will later be used interactively.)
2803 * minor incompatible change: The LOAD function no longer, when given
2804 a wild pathname to load, loads all files matching that pathname.
2805 Instead, an error of type FILE-ERROR is signalled.
2807 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2808 * bug 147 fixed: The compiler preserves its block link/count
2809 invariants more correctly now so that it doesn't crash. (thanks
2811 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2813 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2814 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2815 no-value-for-this-character as NIL (as specified by ANSI).
2816 * HOST-NAMESTRING on physical pathnames now returns a string that is
2817 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2818 (thanks to Christophe Rhodes)
2819 * The Alpha port handles icache flushing more correctly. (thanks to
2821 * More progress has been made toward bootstrapping under CLISP. (thanks
2822 to Christophe Rhodes)
2823 * The fasl file format has changed again, because dynamic loading
2824 on OpenBSD (which has non-ELF object files) motivated some cleanups
2825 in the way that foreign symbols are transformed and passed around.
2826 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2827 is no longer treated as whitespace by the reader, but instead as
2828 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2829 |A
\7fB|, instead of A as it used to.
2831 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2832 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2833 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2834 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2835 this as of 0.7.1.45, but was left out of the previous news section
2836 (thanks to Dan Barlow)
2837 * SBCL now runs on the Solaris operating system on SPARC architectures
2838 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2839 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2840 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2841 * SPARC backend cleanups, allowing builds of cores optimized for V8
2842 and V9 SPARCS, and also emission of code targeted to a particular
2843 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2845 * SBCL is closer to bootstrapping under CLISP, thanks to various
2846 fixes by Christophe Rhodes.
2847 * The fasl file format has changed again, to allow the compiler's
2848 INFO database to support symbol macros.
2849 * The user manual (in doc/) is formatted into HTML more nicely.
2850 (thanks to coreythomas)
2851 * The system is smarter about SUBTYPEP relationships, especially
2852 those involving NOT types (including types such as ATOM which are
2853 represented internally using NOT types). Thus SUBTYPEP is less
2854 likely to return (VALUES NIL NIL) in general, and in particular
2855 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2857 * The fasl file format has changed again, because the internal
2858 representation of types now includes a new slot to support the new
2859 SUBTYPEP-of-NOT-types logic.
2860 * (not a change in the main branch of SBCL, but a related prototype
2861 which can hopefully be merged into the main branch of SBCL in the
2862 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2863 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2864 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2865 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2866 dispatch macro character combinations. (thanks to Alexey Dejneka)
2867 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2868 that superficially look like logical namestrings correctly.
2869 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2871 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2872 default supplied-p) for &optional and &key arguments. (thanks to
2875 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2876 * incompatible change: The compiler is now less aggressive about
2877 tail call optimization, doing it only when (> SPACE DEBUG) or
2878 (> SPEED DEBUG). (This is an incompatible change because there are
2879 programs which relied on the old CMU-CL-style behavior to optimize
2880 away their unbounded recursion which will now die of stack overflow.)
2881 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2882 for non-GENCGC systems has been increased to 20M (since that
2883 seems much closer to the likely performance optimum for modern
2884 systems than the old 4M value was)
2885 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2886 and *DEBUG-PRINT-LEVEL*
2887 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2888 of CMU CL's support for SPARC, and various endianness and other
2889 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2890 * new syntactic sugar for the Unix command line: --load foo.bar is now
2891 an alternate notation for --eval '(load "foo.bar")'.
2893 ** The system now detects stack overflow and handles it gracefully,
2894 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2895 optimization settings. (This is a good thing in general, and
2896 its introduction in this version should be particularly timely
2897 for anyone whose code fails because of suppression of tail
2899 ** The system now hunts for the C variable "environ" in a more
2900 devious way, to avoid segfaults when the C library version differs
2901 between compile time and run time. (thanks to Christophe Rhodes)
2902 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2903 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2904 ** The compiler no longer issues bogus style warnings for undefined
2905 classes in the same source file as the DEFCLASSes which defined
2906 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2908 ** fixes in CONDITION class precedence list for undefined function
2909 errors (thanks to Alexei Dejneka)
2910 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2911 correctly. (thanks to Dan Barlow)
2912 ** portability fixes aiming at bootstrapping under CLISP (thanks
2913 to Dave McDonald and Christophe Rhodes)
2914 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2915 ** fixes in type translation and and type inference (thanks to
2917 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2918 ** various fixes in the new ports (thanks to Dan Barlow)
2919 * several changes related to debugging:
2920 ** suppression of tail recursion, as noted above
2921 ** stack overflow detection, as noted above
2922 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2923 is now the default. (For some time encapsulation has been more
2924 reliable than the breakpoint-based :ENCAPSULATE NIL
2925 implementation, at least on X86 systems; and I just noticed that
2926 encapsulation also seems closer to the spirit of the ANSI
2929 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2931 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2932 up properly again. (There was a packaging bug in 0.7.0 which
2933 left their definitions in SB-SYS::LOAD-FOREIGN and
2934 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2935 vital for most things which interface to C-level interfaces,
2936 like extensions working with sockets or databases or
2937 Perl-compatible regexes or whatever, and the need to fix
2938 this bug is the main reason that 0.7.1 was released so
2940 ** DEFGENERIC is now choosier about the methods it redefines, so that
2941 reLOADing a previously-LOADed file containing DEFGENERICs does
2942 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2943 cycle now works as it should. (thanks to Alexey Dejneka)
2944 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2945 fixed by Christophe Rhodes. (He actually submitted this patch
2946 months ago, and I delayed until after 0.7.0.)
2947 ** Bug 111 (internal compiler confusion about runtime checks on
2948 FUNCTION types) was fixed by Alexey Dejneka.
2949 * Some internal cleanups (getting rid of variables which aren't
2950 needed now that the byte interpreter is gone) caused the fasl
2951 file format number to change again.
2953 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2954 * major incompatible change: The default fasl file extension, i.e. the
2955 default extension for files produced by COMPILE-FILE, has changed
2956 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2958 ** There are many changes in the implementation of the compiler.
2959 SBCL is now essentially a compiler-only implementation of ANSI
2960 Common Lisp. EVAL still "interprets" a few special cases, but
2961 almost all the interesting cases are handled by creating
2962 a LAMBDA expression, calling COMPILE on it, then calling
2963 FUNCALL on the result.
2964 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2965 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2966 Since the code is newer, there might still be some new bugs
2967 (though not as many as before Martin Atzmueller's fixes:-). But
2968 the new code is substantially simpler and clearer, and hopefully
2969 any remaining bugs will be simpler, less fundamental, and more
2970 fixable then the bugs in the old code.
2971 ** The revised compiler is still a little unsteady on its feet.
2973 *** The debugging information it produces (particularly the names
2974 of FUNCTION objects) is sometimes much less useful than what
2975 the old compiler produced.
2976 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2977 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2978 within a MACROLET) has been temporarily weakened.
2979 ** There are new compiler optimizations for various functions:
2980 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2981 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2982 *** the math functions TRUNCATE, FLOOR, and CEILING
2983 *** the function-of-all-trades COERCE
2984 Mostly these should be transparent, but there's one
2985 potentially-annoying problem (bug 117): when the compiler
2986 inline-expands a function and does type analysis on the result,
2987 it can create control paths which have type mismatches, and
2988 when it can't prove that those control paths aren't taken,
2989 it will issue WARNINGs about the type mismatches. This is
2990 a particular problem in practice for the new sequence functions.
2991 It's not clear how this should be fixed, and for now, a
2992 workaround is given in the entry for 117 in the BUGS file.
2993 ** (Because of the interaction between the two previous items --
2994 occasional inlining problems and new inline expansions -- some
2995 of the new sequence function optimizations won't really kick in
2996 completely until debugging information, and then inlining, are
2997 straightened out in some future version.)
2998 * minor incompatible changes:
2999 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3000 DIRECTORY no longer implicitly promotes NIL slots of its
3001 pathname argument to :WILD. In particular, when you ask for the
3002 contents of a directory (which you used to be able to do without
3003 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3004 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3005 ** changes in behavior that ANSI explicitly defines to be
3006 implementation dependent:
3007 *** The new compiler-only implementation still conforms with ANSI,
3008 but acts a little different than before. Besides the obvious
3009 changes in performance tradeoffs (that the cost per form passed
3010 to EVAL has gone up, and the cost per form executed by EVAL
3011 has gone down), the behavior of the system changes a little
3012 because there are no longer any interpreted function objects.
3013 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3014 e.g. doing COMPILE on the output of interactive DEFUN is
3016 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3018 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3019 has changed from (:CL) to NIL.
3020 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3021 e.g. #\Newline and #\Tab, don't have names specified in the
3022 ANSI Common Lisp standard, is now based on their ASCII symbolic
3023 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3024 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3025 are no longer used for output.
3026 ** changes in internal implementation constants:
3027 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3028 4 million. (If your application spends a lot of time GCing and
3029 you have a lot of RAM, you might want to experiment with
3030 increasing it even more.)
3031 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3032 However, almost all old code should still continue to work without
3033 immediate update, as SB-C-CALL is now a (deprecated) nickname
3035 ** Old operator names in the style DEF-FOO are now deprecated in
3036 favor of new corresponding names DEFINE-FOO, for consistency with
3037 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3038 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3039 internal symbols, but a few supported extensions like
3040 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3041 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3042 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3043 etc. as you get deeper into recursive calls to the debugger
3044 command loop, instead of the old "5]", "5]]", "5]]]"
3045 sequence. (I was motivated to do this when squabbles between
3046 ILISP and SBCL left me very deeply nested in the debugger. In the
3047 short term, this change will probably provoke more ILISP/SBCL
3048 squabbles, but hopefully it will be an improvement in the long run.)
3049 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3050 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3051 function, since declaiming return types involving aliens is
3052 (1) annoyingly messy to do by hand and (2) vital to efficient
3053 compilation of code which calls such functions.
3054 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3055 longer reexported by the SB-EXT package. They're solely useful
3056 for alien code, so it seems more logical that you should get
3057 them from the SB-ALIEN package, not in SB-EXT.
3058 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3059 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3060 features. Instead, the code that they used to control is always
3061 built into the system.
3062 * many other bug fixes
3063 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3064 advantage of the new EVAL-WHEN stuff and to clean them up in
3065 general, and they are now more ANSI-compliant in a number of
3066 ways. Martin Atzmueller is responsible for a lot of this.
3067 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3069 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3070 *** correct ERROR type for various file operations
3071 *** some fixes for Lisp streams
3072 *** DEFMETHOD syntax checking
3073 *** changing old weird representation of debug information as
3074 strings (which, among their other deficiencies, don't transform
3075 correctly when you rename packages, and don't change their
3076 print representation when you change things like *PACKAGE*
3077 and *PRINT-LENGTH*) to symbols and lists of symbols
3078 He also made several improvements and fixed several bugs in DESCRIBE.
3079 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3081 *** misbehavior of WRITE-STRING/WRITE-LINE
3082 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3083 and several other LOOP problems as well
3084 *** DIRECTORY when similar filenames are present
3085 *** DEFGENERIC with :METHOD options
3086 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3087 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3088 *** argument ordering in FIND with :TEST option
3089 *** mishandled package designator argument in APROPOS-LIST
3090 *** various problems in the backquote readmacro
3091 *** a bug in APROPOS
3092 *** probably some others that I'm not describing very well here,
3093 since the CVS log documents them by reference to sbcl-devel
3094 messages, and the SourceForge archives aren't working well.:-(
3095 ** Dan Barlow improved the Alpha port (and is making progress on the
3096 PPC port, for those of you who think different).
3097 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3098 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3099 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3100 continues to work on the SPARC port (for those of you in a position
3101 to look down upon our little PC-compatible boxes from a great height).
3102 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3103 and uses that copy, rather than the current dynamic value, when
3104 it's trying to decide whether to truncate output. Thus e.g.
3105 (let ((*print-lines* 50))
3106 (pprint-logical-block (stream nil)
3108 (let ((*print-lines* 8))
3109 (print (aref possiblybigthings i) stream)))))
3110 should now truncate the logical block only at 50 lines, instead of
3111 often truncating it at 8 lines, as it did before.
3112 * The doc/cmucl/ directory, containing old CMU CL documentation
3113 from the time of the fork, is no longer part of the base system.
3114 SourceForge has shut down its anonymous FTP service, and with it
3115 my original plan for distributing the old CMU CL documentation
3116 there. For now, if you need these files you can download an old
3117 SBCL source release and extract them from it.
3118 * The fasl file version number changed again, for dozens of reasons,
3119 some of which are apparent above.
3121 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3122 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3123 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3124 patch, so that DISASSEMBLE output is much nicer.
3125 * The code in the SB-PROFILE package now seems reasonably stable.
3126 I still haven't decided what the final interface should look like
3127 (I'd like PROFILE to interact cleanly with TRACE, since both
3128 facilities use function encapsulation) but if you have a need
3129 for profiling now, you can probably use it successfully with
3130 the current CMU-CL-style interface.
3131 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3132 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3133 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3134 link now returns the dangling link itself, and for similar
3135 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3136 link itself. (In these cases the old code signalled an error and
3137 looped endlessly, respectively.) Thus, DIRECTORY now works even
3138 in the presence of dangling and cyclic symbolic links.
3139 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3140 is now a supported extension again, since the consensus on
3141 sbcl-devel was that it can be useful for ordinary development
3142 work, not just for debugging SBCL itself.
3143 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3144 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3145 that currently-defined functions might be redefined later with
3146 different return types.
3147 * Hash tables can be printed readably, as inspired by CMU CL code
3148 of Eric Marsden and SBCL code of Martin Atzmueller.
3149 * better error handling in CLOS method combination, thanks to
3150 Martin Atzmueller porting Pierre Mai's CMU CL patches
3151 * more overflow fixes for >16Mbyte I/O buffers
3152 * A bug in READ has been fixed, so that now a single Ctrl-D
3153 character suffices to cause end-of-file on character streams.
3154 In particular, now you only need one Ctrl-D at the command
3155 line (not two) to exit SBCL.
3156 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3158 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3159 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3160 does the right thing.
3161 * The implementation of some type tests, especially for CONDITION
3162 types, is now tidier and maybe faster, due to CMU CL code
3163 originally by Douglas Crosher, ported by Martin Atzmueller.
3164 * Some math functions have been fixed, and there are new
3165 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3166 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3167 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3168 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3169 likely to be built into sbcl-0.7.0.)
3170 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3171 CL, ported to SBCL by Martin Atzmueller)
3172 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3173 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3174 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3175 * There's a new slam.sh hack to shorten the edit/compile/debug
3176 cycle for low-level changes to SBCL itself, and a new
3177 :SB-AFTER-XC-CORE target feature to control the generation of
3178 the after-xc.core file needed by slam.sh.
3179 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3180 COMPILE-FILE is no longer supported, so that now every function
3181 gets an entry point, so that block compilation looks a little
3182 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3183 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3184 returns the number of bytes consed since the system started,
3185 rather than the number consed since the first time the function
3186 was called. (The new definition parallels ANSI functions like
3187 CL:GET-INTERNAL-RUN-TIME.)
3188 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3189 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3190 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3191 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3192 * Fasl file version numbers are now independent of the target CPU,
3193 since historically most system changes which required version
3194 number changes have affected all CPUs equally. Similarly,
3195 the byte fasl file version is now equal to the ordinary
3198 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3199 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3200 is no longer recognized. I apologize for this, because it was
3201 listed in SB-EXT as a supported extension, but I found that
3202 its existing behavior was poorly specified, as well as incorrectly
3203 specified, and it looked like too much of a mess to straighten it
3204 out. I have enough on my hands trying to get ANSI stuff to work..
3205 * many patches ported from CMU CL by Martin Atzmueller, with
3206 half a dozen bug fixes in pretty-printing and the debugger, and
3207 half a dozen others elsewhere
3208 * fixed bug 13: Floating point infinities are now supported again.
3209 They might still be a little bit flaky, but thanks to bug reports
3210 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3211 as flaky as they were.
3212 * The --noprogrammer command line option is now supported. (Its
3213 behavior is slightly different in detail from what the old man
3214 page claimed it would do, but it's still appropriate under the
3215 same circumstances that the man page talks about.)
3216 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3217 are now supported, and enabled by default. Thus, the compiler can
3218 handle many floating point and complex operations much less
3219 inefficiently. (Thus e.g. you can implement a complex FFT
3221 * The compiler now detects type mismatches between DECLAIM FTYPE
3222 and DEFUN better, and implements CHECK-TYPE more correctly, and
3223 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3224 thanks to patches from Martin Atzmueller.
3225 * various fixes to make the cross-compiler more portable to
3226 ANSI-conforming-but-different cross-compilation hosts (notably
3227 Lispworks for Windows, following bug reports from Arthur Lemmens)
3228 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3229 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3230 fixed thanks to Pierre Mai's CMU CL patches.
3231 * fixes to keep the system from overflowing internal counters when
3232 it tries to use i/o buffers larger than 16M bytes
3233 * fixed bug 45a: Various internal functions required to support
3234 complex special functions have been merged from CMU CL sources.
3235 (When I was first setting up SBCL, I misunderstood a compile-time
3236 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3237 * improved support for type intersection and union, fixing bug 12
3238 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3239 more obscure bugs as well
3240 * some steps toward byte-compiling non-performance-critical
3241 parts of the system, courtesy of patches from Martin Atzmueller
3242 * Christophe Rhodes has made some debian packages of sbcl at
3243 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3244 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3245 stable, but are nonetheless usable. When he's ready, I'd be happy
3246 to add them to the SourceForge "File Releases" section. (And if
3247 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3248 * new fasl file format version number (because of changes in
3249 internal representation of (OR ..) types to accommodate the new
3250 support for (AND ..) types, among other things)
3252 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3253 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3255 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3257 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3258 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3259 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3260 * DESCRIBE now works on structure objects again.
3261 * Most function call argument type mismatches are now handled as
3262 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3263 know whether the function will be redefined before the call is
3264 executed. (The compiler could flag local calls with full WARNINGs,
3265 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3266 it doesn't keep track of enough information to know whether calls
3267 are local in this sense.)
3268 * Compiler output is now more verbose, with messages truncated
3269 later than before. (There should be some supported way for users
3270 to override the default verbosity, but I haven't decided how to
3271 provide it yet, so this behavior is still controlled by the internal
3272 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3273 src/compiler/ir1util.lisp.)
3274 * Fasl file format version numbers have increased again, because
3275 support for the Gray streams extension changes the layout of the
3276 system's STREAM objects.
3277 * The Gray subclassable streams extension now works, thanks to a
3278 patch from Martin Atzmueller.
3279 * The full LOAD-FOREIGN extension (not just the primitive
3280 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3281 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3282 but like most other programs, it defaults to copying the Unix
3283 environment from the original process instead of starting the
3284 new process in an empty environment.
3285 * Extensions which manipulate the Unix environment now support
3286 an :ENVIRONMENT keyword option which doesn't smash case or
3287 do other bad things. The CMU-CL-style :ENV option is retained
3288 for porting convenience.
3289 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3290 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3292 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3294 * A patch from Martin Atzmueller seems to have solved the SIGINT
3295 problem, and as far as we know, signal-handling now works cleanly.
3296 (If you find any new bugs, please report them!)
3297 * The system no longer defaults Lisp source file names to types
3298 ".l", ".cl", or ".lsp", but only to ".lisp".
3299 * The compiler no longer uses special default file extensions for
3300 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3301 seems to expect a single default extension for all compiled code,
3302 and there's no compelling reason to try to stretch the standard
3303 to allow two different extensions.) Instead, byte-compiled files
3304 default to the same extension as native-compiled files.
3305 * Fasl file format version numbers have increased again, because
3306 a rearrangement of internal implementation packages made some
3307 dumped symbols in old fasl files unreadable in new cores.
3308 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3309 many fewer weird special cases.
3310 * Bug #17 (differing COMPILE-FILE behavior between logical and
3311 physical pathnames) has been fixed, and some related misbehavior too,
3312 thanks to a patch from Martin Atzmueller.
3313 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3314 by Tim Moore, ported to SBCL by Martin Atzmueller.
3315 * Martin Atzmueller fixed several filesystem-related problems,
3316 including bug #36, in part by porting CMU CL patches, which were
3317 written in part by Paul Werkowski.
3318 * More compiler warnings in src/runtime/ are gone, thanks to
3319 more patches from Martin Atzmueller.
3320 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3323 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3325 * DESCRIBE now works on CONDITION objects.
3326 * The debugger now handles errors which arise when trying to print
3327 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3329 * The build system now uses an additional file, customize-target-features.lisp,
3330 to allow local modifications to the target *FEATURES* list. (The point of
3331 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3332 debugging features enabled, without having to worry about propagating it
3333 into everyone's system when I do a "cvs update".) When no
3334 customize-target-features.lisp file exists, the target *FEATURES* list
3335 should be constructed the same way as before.
3336 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3337 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3338 warn about behavior which is undefined under the ANSI spec. Note: This
3339 is specified by ANSI, but it's not very popular with programmers.
3340 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3341 macro in the SBCL sources for an example of a workaround which you
3342 might use to make portable ANSI-standard code which does what you want.
3343 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3344 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3345 when it's used as a non-toplevel form. (This is required in order
3346 to implement the DEFCONSTANT-EQX macro.)
3347 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3348 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3349 funny things, due to bugs in the implementation of EVAL-WHEN
3350 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3351 non-toplevel DEFCONSTANTs are uncommon.)
3352 * The core file version number and fasl file version number have been
3353 incremented, because the old noncompliant DEFCONSTANT behavior involved
3354 calling functions which no longer exist, and because I also took the
3355 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3356 * fixed bug 1 (error handling before read-eval-print loop starts), and
3357 redid debugger restarts and related debugger commands somewhat while
3359 ** The QUIT debugger command is gone, since it did something
3360 rather different than the SB-EXT:QUIT command, and since it never
3361 worked properly outside the main toplevel read/eval/print loop.
3362 Invoking the new TOPLEVEL restart provides the same functionality.
3363 ** The GO debugger command is also gone, since you can just invoke
3364 the CONTINUE restart directly instead.
3365 ** The TOP debugger command is also gone, since it's redundant with the
3366 FRAME 0 command, and since it interfered with abbreviations for the
3368 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3370 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3371 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3372 some porting work by Martin Atzmueller)
3373 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3374 with a boatload of patches which he ported to SBCL. Now that those
3376 ** The system tries to make sure that its low-priority messages
3377 are prefixed by semicolons, to help people who like to use
3378 syntax highlighting in their ILISP buffer. (This patch
3379 was originally due to Raymond Toy.)
3380 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3381 patches originally written by Raymond Toy.
3382 ** The compiler understands coercion between single-value and
3383 multiple-VALUES type expressions better, getting rid of some very
3384 weird behavior, thanks to patches originally by Robert MacLachlan
3385 and Douglas Crosher.
3386 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3387 lambda lists, thanks to a patch originally by Pierre Mai.
3388 ** The system no longer bogusly warns about "abbreviated type
3390 ** The compiler gets less confused by inlining and RETURN-FROM,
3391 thanks to some patches originally by Tim Moore.
3392 ** The system no longer hangs when dumping circular lists to fasl
3393 files, thanks to a patch originally from Douglas Crosher.
3394 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3395 undefined function error.
3396 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3397 in structure slot initforms) for now, documented workaround instead:-|
3398 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3399 * fixed bug 5: added stubs for various Gray stream functions called
3400 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3401 installed, at least appropriate type errors are generated
3402 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3403 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3404 consistently in DEFMETHOD forms.
3405 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3406 it doesn't seem to affect SBCL after all
3407 * The C runtime system now builds with better optimization and many
3408 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3410 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3412 * The system is now under CVS at SourceForge (instead of the
3413 CVS repository on my home machine).
3414 * The new signal handling code has been tweaked to treat register
3415 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3416 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3417 downstream, e.g. in debug-int.lisp, has implicit dependencies
3418 on the unsignedness of integer representation of machine words,
3419 and that caused the system to bomb out with infinite regress
3420 when trying to recover from type errors involving signed values,
3421 e.g. (BUTLAST '(1 2 3) -1).)
3422 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3423 (This was one of the bugs Peter Van Eynde reported back in July.)
3424 * The system now uses code inspired by Colin Walters' O(N)
3425 implementation of MAP (from the cmucl-imp@cons.org mailing
3426 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3427 inline the MAP operation, and there is more than one
3428 sequence argument to the MAP call (so that it can't just
3429 do ETYPECASE once and for all based on the type of the
3430 single sequence argument). (The old non-inline implementation
3431 of the general M-argument sequence-of-length-N case required
3432 O(M*N*N) time when any of the sequence arguments were LISTs.)
3433 * The QUIT :UNIX-CODE keyword argument has been renamed to
3434 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3435 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3437 * Raymond Wiker's patches to port dynamic loading from Linux to
3438 FreeBSD have been added.
3439 * The BUGS file is now more nearly up to date, thanks in large part
3440 to Martin Atzmueller's review of it.
3441 * The debugger now flushes standard output streams before it begins
3442 its output ("debugger invoked" and so forth).
3443 * The core version number and fasl file version number have both
3444 been incremented, because of incompatible changes in the layout
3446 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3447 used to not be called for a saved Lisp image.)
3448 * Martin Atzmueller's version of a patch to fix a compiler crash,
3449 as posted on sbcl-devel 13 September 2000, has been installed.
3450 * Instead of installing Martin Atzmueller's patch for the
3451 compiler transform for SUBSEQ, I deleted the compiler transform,
3452 and transforms for some similar consing operations.
3453 * A bug in signal handling which kept TRACE from working on OpenBSD
3455 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3456 away by constant folding
3457 * The system now defines its address space constants in one place
3458 (in the Lisp sources), and propagates them automatically elsewhere
3459 (through GENESIS and the sbcl.h file). Therefore, patching the
3460 address map is less unnecessarily tedious and error-prone. The
3461 Lisp names of address space constants have also been systematized.
3462 * CVS tags like dollar-Header-dollar have been removed from
3463 the sources, because they have never saved me trouble and
3464 they've been source of trouble working with patches and other
3465 diff-related operations.
3466 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3467 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3469 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3471 * The system has been ported to OpenBSD.
3472 * The system now compiles with a simple "sh make.sh" on the systems
3473 that it's supported on. I.e., now you no longer need to tweak
3474 text in the target-features.lisp-expr and symlinks in src/runtime/
3475 by hand, the make.sh takes care of it for you.
3476 * The system is no longer so grossly inefficient when compiling code
3477 involving vectors implemented as general (not simple) vectors (VECTOR T),
3478 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3479 which dares to use the various sequence functions on non-simple
3480 vectors, takes less of a performance hit.
3481 * There is now a primitive type predicate VECTOR-T-P
3482 to test for the (VECTOR T) type, so that e.g.
3483 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3484 can now be compiled with some semblance of efficiency. (The old code
3485 turned the type declaration into a full call to %TYPEP at runtime!)
3486 * AREF on (VECTOR T) is still not fast, since it's still compiled
3487 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3488 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3489 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3490 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3491 without making full calls to SUBTYPEP at runtime.
3492 (Some analogous efficiency issues for non-simple vectors specialized to
3493 element types other than T, or for non-simple multidimensional arrays,
3494 have not been addressed. They could almost certainly be handled the
3495 same way if anyone is motivated to do so.)
3496 * The changes in array handling break binary compatibility, so
3497 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3498 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3499 instead of (VALUES T T).
3500 * By following the instructions that Dan Barlow posted to sbcl-devel
3501 on 2 July 2000, I was able to enable primitive dynamic object
3502 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3503 functionality is not implemented (since it calls ld to resolve
3504 library references automatically, requiring RUN-PROGRAM for its
3505 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3506 try to resolve library references) is now supported.
3507 * The system now flushes the standard output streams when it terminates,
3508 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3509 them at several other probably-convenient times, e.g. in each pass of
3510 the toplevel read-eval-print loop, and after evaluating a form given
3511 as an "--eval" command-line option. (These changes were motivated by a
3512 discussion of stream flushing issues on cmucl-imp in August 2000.)
3513 * The source transform for TYPEP of array types no longer assumes
3514 that an array whose element type is a not-yet-defined type
3515 is implemented as an array of T, but instead punts, so that the
3516 type will be interpreted at runtime.
3517 * There is now some support for cross-compiling in make.sh: each of
3518 the phases of make.sh has its own script. (This should be transparent
3519 to people doing ordinary, non-cross-compile builds.)
3520 * Since my laptop doesn't have hundreds of megabytes of memory like
3521 my desktop machine, I became more motivated to do some items on
3522 my to-do list in order to reduce the size of the system a little:
3523 ** Arrange for various needed-only-at-cold-init things to be
3524 uninterned after cold init. To support this, those things have
3525 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3526 symbols with such names are now uninterned after cold init).
3527 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3528 fasl files for cold load.
3529 ** Remove the old compiler structure pooling code (which used to
3530 be conditional on the target feature :SB-ALLOC) completely.
3531 ** Redo the representation of some data in cold init to be more compact.
3532 (I also looked into supporting byte compiled code at bootstrap time,
3533 which would probably reduce the size of the system a lot, but that
3534 looked too complicated, so I punted for now.)
3535 * The maximum signal nesting depth in the src/runtime/ support code has
3536 been reduced from 4096 to 256. (I don't know any reason for the very
3537 large old value. If the new smaller value turns out to break something,
3538 I'll probably just bump it back up.)
3539 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3541 * Many, many bugs reported by Peter Van Eynde have been added to
3542 the BUGS list; some have even been fixed.
3543 * While enabling dynamic object file loading, I tried to make the
3544 code easier to understand, renaming various functions and variables
3545 with less ambiguous names, and changing some function calling
3546 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3547 * While trying to figure out how to do the OpenBSD port, I tried to
3548 clean up some of the code in src/runtime/. In particular, I dropped
3549 support for non-POSIX signal handling, added various comments,
3550 tweaked the code to reduce the number of compilation warnings, and
3551 renamed some files to increase consistency.
3552 * To support the new automatic configuration functionality in make.sh,
3553 the source file target-features.lisp-expr has been replaced with the
3554 source file base-target-features.lisp-expr and the machine-generated
3555 file local-target-features.lisp-expr.
3556 * fixed a stupid quoting error in make.sh so that using CMU CL
3557 "lisp -batch" as cross-compilation host works again
3559 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3561 * DESCRIBE no longer tries to call itself recursively to describe
3562 bound/fbound values, so that it no longer fails on symbols which are
3563 bound to themselves (like keywords, T, and NIL).
3564 * DESCRIBE now works on generic functions.
3565 * The printer now prints less-screwed-up representations of closures
3566 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3567 * A private symbol is used instead of the :EMPTY keyword previously
3568 used to mark empty slots in hash tables. Thus
3569 (DEFVAR *HT* (MAKE-HASH-TABLE))
3570 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3571 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3572 now does what ANSI says that it should. (You can still get
3573 similar noncompliant behavior if bang on the hash table
3574 implementation with all the symbols you get back from
3575 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3576 This breaks binary compatibility, since tests for equality to
3577 :EMPTY are wired into things like the macroexpansion of
3578 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3580 * There's now a minimal placeholder implementation for CL:STEP,
3581 as required by ANSI.
3582 * An obscure bug in the interaction of the normal compiler, the byte
3583 compiler, inlining, and structure predicates has been patched
3584 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3585 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3587 * Missing ordinary arguments in a macro call are now detected even
3588 when the macro lambda list contains &KEY or &REST.
3589 * The debugger no longer complains about encountering the top of the
3590 stack when you type "FRAME 0" to explicitly instruct it to go to
3591 the top of the stack. And it now prints the frame you request even
3592 if it's the current frame (instead of saying "You are here.").
3593 * As specified by ANSI, the system now always prints keywords
3594 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3596 * The default initial SIZE of HASH-TABLEs is now smaller.
3597 * Type information from CLOS class dispatch is now propagated
3598 into DEFMETHOD bodies, so that e.g.
3599 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3601 is now basically equivalent to
3602 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3603 (DECLARE (TYPE SINGLE-FLOAT X))
3605 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3606 operation, without having to do run-time type dispatch.
3607 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3608 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3609 * Since I don't seem to be making big file reorganizations very often
3610 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3611 is overflowing my ability to conveniently back them up), I've finally
3612 checked the system into CVS. (The CVS repository is on my home system,
3613 not at SourceForge -- putting it on SourceForge might come later.)
3614 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3615 high-level GC-NOTIFY-FOO functions send their output. (There's
3616 still very little control of where low-level verbose GC functions
3617 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3618 less than it used to -- the GC-NOTIFY-FOO functions are now under
3619 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3620 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3621 in only one place in the source code, and propagates it automatically
3622 everywhere that it's needed. Thus e.g. when I bump the version from
3623 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3624 * The C source files now include boilerplate legalese and documentation
3625 at the head of each file (just as the Lisp source files already did).
3626 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3627 to his page will be replaced with a link to his new CLiki service.
3629 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3631 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3632 * The build process now looks for GNU make under the default name "gmake",
3633 instead of "make" as it used to. If GNU make is not available as "gmake"
3634 on your system, you can change this default behavior by setting the
3635 GNUMAKE environment variable.
3636 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3637 can build without error under CMU CL.
3639 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3641 * There is now a partial SBCL user manual (with some new text and some
3642 text cribbed from the CMU CL manual).
3643 * The beginnings of a profiler have been added (starting with the
3644 CMU CL profiler and simplifying and cleaning up). Eventually the
3645 main interface should be through the TRACE macro, but for now,
3646 it's still accessed through vaguely CMU-CL-style functions and macros
3647 exported from the package SB-PROFILE.
3648 * Some problems left over from porting CMU CL to the new
3649 cross-compilation bootstrap process have been cleaned up:
3650 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3651 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3652 part when getting rid of the compiler *BACKEND* stuff.)
3653 ** The value of *NULL-TYPE* was screwed up, because it was
3654 being initialized before the type system knew the final
3655 definition of the 'NULL type. This screwed up several key
3656 optimizations in the compiler, causing inefficiency in all sorts
3657 of places. (I found it because I wanted to understand why
3658 GET-INTERNAL-RUN-TIME was consing.)
3659 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3660 PROCLAIM FTYPE information. Unfortunately this broke binary
3661 compatibility again, since now the forms output by DEFGENERIC
3662 to refer to functions which didn't exist in 0.6.3.
3663 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3664 can use the new (as of 0.6.3) transform for SOME into MAP into
3666 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3667 values to (INTEGER 0 1000000), so that the time code will no longer
3668 occasionally get blown up by Linux returning 1000000 microseconds
3669 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3670 its output conform to the ANSI spec. (Alas, this makes its output
3671 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3672 * A full call to MAP NIL with a single sequence argument no longer conses.
3673 * fixes to problems pointed out by Martin Atzmueller:
3674 * The manual page no longer talks about multiprocessing as though
3675 it were currently supported.
3676 * The ILISP support patches have been removed from the distribution,
3677 because as of version 5.10.1, ILISP now supports SBCL without us
3678 having to maintain patches.
3679 * added a modified version of Raymond Toy's recent CMU CL patch for
3680 EQUALP comparison of HASH-TABLE
3682 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3684 * The system still can't cross-compile itself with
3685 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3686 entails), but at least it can compile more of itself that way
3687 than it used to be able to, and various buglets which were uncovered
3688 by trying to cross-compile itself that way have now been fixed.
3689 * This release breaks binary compatibility again. This time
3690 at least I've incremented the FASL file format version to 2, so that the
3691 problem can be detected reliably instead of just causing weird errors.
3692 * various new style warnings:
3693 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3694 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3695 ** using the lexical binding of a variable named in the *FOO* style
3696 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3697 as specified by ANSI.
3698 * *RANDOM-STATE* is no longer automatically initialized from
3699 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3700 default behavior of the system is to repeat its behavior every time
3701 it's run. If you'd like to change this behavior, you can always
3702 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3703 old convention there was no comparably easy way to get the system to
3704 repeat its behavior every time it was run.
3705 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3706 FTYPE declarations has been removed, in favor of their ANSI
3707 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3708 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3709 the types of their sequence arguments can be determined at compile time.
3710 This is done through a new open code expansion for MAP which eliminates
3711 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3712 argument types can be determined at compile time.
3713 * The optimizer now transforms COERCE into an identity operation when it
3714 can prove that the coerced object is already of the correct type. (This
3715 can be a win for machine generated code, including the output of other
3716 optimization transforms, such as the MAP transform above.)
3717 * Credit information has been moved from source file headers into CREDITS.
3718 * Source file headers have been made more standard.
3719 * The CASE macro now compiles without complaining even when it has
3722 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3724 * (Note that the way that the PCL macroexpansions were rewritten
3725 to accommodate the change in DEFGENERIC below breaks binary
3726 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3727 not run under sbcl-0.6.2. Once we get out of alpha releases,
3728 i.e. hit release 1.0.0, we'll probably try to maintain binary
3729 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3730 and sbcl-1.4.4. Until then, however, it might be fairly common
3731 for maintenance releases to break binary compatibility.)
3732 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3734 * The reporting of unhandled signals has been changed to print some
3735 explanatory text as well as the report form. (Previously only
3736 the report form was printed.)
3737 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3738 it defines, so that the compiler no longer issues undefined function
3739 warnings for compiled-but-not-yet-loaded generic functions.
3740 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3741 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3742 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3743 section of the ANSI spec are supported.
3744 * The "" nickname for the "KEYWORD" package has been removed.
3745 The reader still handles symbol tokens which begin with a package marker
3746 as keywords, but it doesn't expose its mechanism for doing so in the
3747 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3748 * The system now issues STYLE-WARNINGs for contradictory TYPE
3749 proclamations. (Warnings for contradictory FTYPE proclamations would
3750 be nice too, but those can't be done usefully unless the type system
3751 is made smarter about FUNCTION types.)
3752 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3753 systematized, so that "*target-*.lisp is supposed to exist only on the
3754 target and imply that there's a related file which exists on the
3755 host, and *host-*.lisp is supposed to exist only on the host and imply
3756 that there's a related file which exists on the target. This involves a
3757 lot of renaming. Hopefully the acute confusion caused by the renaming
3758 will be justified by the reduction in chronic confusion..
3759 ** runtime-type.lisp -> early-target-type.lisp
3760 ** target-type.lisp -> late-target-type.lisp
3761 ** early-host-format.lisp -> early-format.lisp
3762 ** late-host-format.lisp -> late-format.lisp
3763 ** host-error.lisp -> misc-error.lisp
3764 ** early-error.lisp -> early-target-error.lisp
3765 ** late-error.lisp -> late-target-error.lisp
3766 ** host-defboot.lisp -> early-defboot.lisp
3767 ** code/misc.lisp -> code/target-misc.lisp
3768 ** code/host-misc.lisp -> code/misc.lisp
3769 ** code/numbers.lisp -> code/target-numbers.lisp
3770 ** code/early-numbers.lisp -> numbers.lisp
3771 ** early-host-type.lisp -> early-type.lisp
3772 ** late-host-type.lisp -> late-type.lisp
3773 ** host-typep.lisp -> typep.lisp
3774 ** load.lisp -> target-load.lisp
3775 ** host-load.lisp -> load.lisp
3776 ** host-disassem.lisp -> disassem.lisp
3777 ** host-insts.lisp -> insts.lisp
3778 ** byte-comp.lisp -> target-byte-comp.lisp
3779 ** host-byte-comp.lisp -> byte-comp.lisp
3780 ** host-signal.lisp -> signal.lisp
3781 ** host-defstruct.lisp -> defstruct.lisp
3782 ** late-target-type.lisp -> deftypes-for-target.lisp
3783 Furthermore, several other previously target-only files foo.lisp (e.g.
3784 hash-table.lisp and random.lisp) have been split into a target-and-host
3785 foo.lisp file and a target-only target-foo.lisp file, with their key type
3786 definitions in the target-and-host part, so that the cross-compiler will
3787 know more about target types.
3788 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3789 gone away. In their place are various *BACKEND-FOO* variables
3790 corresponding to the slots of the old structure.
3791 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3792 into the target SBCL has been fixed.
3793 * The chill.lisp system for loading cold code into a running SBCL
3795 * Support for the CMU CL "scavenger hook" extension has been removed.
3796 (It was undocumented and unused in the CMU CL sources that SBCL was
3797 derived from, and stale in sbcl-0.6.1.)
3798 * Various errors in the cross-compiler type system were detected
3799 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3800 (enabling various consistency checks). Many of them were fixed,
3801 but some hard problems remain, so the compiler is back to
3802 running without *TYPE-SYSTEM-INITIALIZED* for now.
3803 * As part of the cross-compiler type system cleanup, I implemented
3804 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3805 * I have started adding UNCROSS calls throughout the type system
3806 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3807 on cross-compiler input files will be able to go away, and various
3809 * CONSTANTP now returns true for quoted forms (as explicitly required
3812 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3814 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3815 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3816 (MAKE-LIST -1). In the longer term, it ought to become true
3817 that declarations are assertions even at SAFETY 1. For now, it's not
3818 quite true even at SAFETY 3, but it's at least more nearly true..
3819 (Note that this change seems to increases the size of the system by
3820 O(5%) and to decrease the speed of the compiler by 20% or more.)
3821 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3822 for the problem of printing dozens of lines of distracting information
3823 about low-level system machinery as part of the top stack frame
3824 on entry to the debugger when an undefined function was called.
3825 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3826 is not reset to COMMON-LISP-USER.
3827 * Compilation of stuff related to dyncount.lisp has been made conditional
3828 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3829 smaller. The various dyncount-related symbols have been moved into
3830 a new "SB-DYNCOUNT" package.
3831 * tty-inspect.lisp has been renamed to inspect.lisp.
3832 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3833 feature has gone away. (When we eventually port to other flavors of
3834 libc and/or Unix, we'll try to make the differences between flavors
3835 invisible at the user level.)
3836 * Various other *FEATURES* tags, and/or their associated conditionals,
3837 have been removed if obsolescent, or given better documentation, or
3838 sometimes given more-mnemonic names.
3840 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3842 * tidied up "make.sh" script
3843 * tidied up system directory structure
3844 * better "clean.sh" behavior
3845 * added doc/FOR-CMUCL-DEVELOPERS
3846 * many many small tweaks to output format, e.g. removing possibly-confusing
3847 trailing #\. character in DESCRIBE-INSTANCE
3848 * (EQUALP #\A 'A) no longer signals an error.
3849 * new hashing code, including EQUALP hashing
3850 * tidied up Lisp initialization and toplevel
3851 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3852 * command line argument processing
3853 * added POSIX-GETENV function to deal with Unix-ish environment variables
3854 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3855 terminating SBCL on EOF
3856 * non-verbose GC by default
3857 * There is no more "sbcl" shell script; the sbcl file is now the C
3858 runtime executable (just like CMU CL).
3859 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3861 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3862 symbol and package information as Lisp native symbol and package objects
3863 instead of strings naming symbols and strings naming packages. This way,
3864 whenever packages are renamed (as in warm init), debug information is
3865 transformed along with everything else.
3866 * tweaked the optimization policy declarations which control the building
3867 of SBCL itself. Now, among other things, the system no longer saves
3868 source location debugging information. (This helps two problems at once
3869 by reducing SBCL size and by keeping SBCL from trying to look for its
3870 sources -- which may not exist -- when reporting errors.)
3871 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3872 debugging and testing purposes
3873 * cleaned up printing, making the printer call PRINT-OBJECT for
3874 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3875 methods, giving nearly-ANSI behavior
3876 * converted almost all special variables to use *FOO* naming convention
3877 * deleted PARSE-TIME functionality, since it can be done portably
3878 * moved some files out of cold init into warm init
3879 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3881 * regularized formatting of source files
3882 * added an install.sh script
3883 * fixed ridiculous memory usage of cross-compiler by making
3884 compiler/alloc.lisp not try to do pooling unless it can hook
3885 itself into the GC of the cross-compilation host. Now the system
3886 builds nicely on my old laptop.
3887 * added :SB-ALLOC in target-features.lisp-expr
3888 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3889 was not implemented)
3890 * re-did condition handling and note reporting in the compiler. Notes
3891 are no longer handled by signalling conditions. Style warnings
3892 and warnings are handled more correctly and reported in such a way
3893 that it's easy to find one or the other in your output (so that you
3894 can e.g. figure out which of many problems caused COMPILE-FILE to
3896 * changed the severity of several compiler warnings from full WARNING
3897 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3898 compiler note reporting so that it doesn't use the condition system
3899 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3900 COMPILE-FILE command)
3901 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3902 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3903 no longer supported, since their implementation was deeply intertwingled
3904 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3905 * removed bogus "support" for compiler macros named (SETF FOO), and
3906 removed the compiler macro for SETF INFO (but only after making a fool
3907 of myself on the cmucl-imp mailing list by posting a bogus patch for
3908 DEFINE-COMPILER-MACRO..)
3909 * Compiled files containing forms which have side effects on the Lisp
3910 reader (such as DEFPACKAGE forms) are now handled more correctly.
3911 (Compiler queuing of top level lambdas has been suppressed by setting
3912 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3913 * deleted various currently-unused source files, e.g. gengc.lisp. They
3914 may be added back at some point e.g. when porting to other architectures,
3915 but until they are it's distracting to distribute them and to try to
3917 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3918 there were so many of them they're just distractions, and UNCROSS is
3919 known to be able to handle the current sources
3920 * moved PROFILE functionality into TRACE, so that it will be clear
3921 how the wrapping and unwrapping of functions when you profile them
3922 interacts with the wrapping and unwrapping of functions when you
3923 trace them. (Actually, the functionality isn't there yet, but at least
3924 the interface specification is there. Hopefully, the functionality will
3925 arrive with some future maintenance release.)
3926 * removed host-oops.lisp
3927 * changed signature of QUIT function to allow UNIX-CODE argument
3928 * fixed READ-SEQUENCE bug
3929 * tweaked verbose GC output so that it looks more like the progress
3930 output that ANSI specifies for functions like LOAD
3931 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3932 * added <http://sbcl.sourceforge.com> to the banner information printed by