1 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
3 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
4 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now
5 part of the supported interface.
6 * new feature: saving cores with foreign code loaded is now
7 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based
8 on Timothy Moore's work for CMUCL)
9 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
10 (thanks for Teemu Kalvas)
11 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
12 case instead of terminating the iteration (thanks for Julian Squires,
13 Sean Champ and Raymond Toy)
14 * bug fix: incorrect expansion of defgeneric that caused
15 a style warning. (thanks for Zach Beane)
17 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
18 * incompatible change: the internal functions
19 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
20 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
21 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
22 instead of the old functions.
23 * new feature: on platforms where "dladdr" is available foreign
24 function names now appear in backtraces. (based on Helmut Eller's
26 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
27 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
29 * documentation: ASDF manual is now distributed with SBCL.
30 * bug fix: SBCL can now load its contributed modules with REQUIRE
31 even if the system-provided entries have been removed from
32 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
34 * bug fix: backtraces involving undefined functions or assembly
35 routines are more informative. (thanks to Brian Downing)
36 * bug fix: mutually referent alien structures now work correctly.
37 (reported by Rick Taube)
38 * bug fix: structures defined by WITH-ALIEN can be referred to
39 within other definitions in the same WITH-ALIEN.
40 * bug fix: division operators (MOD, TRUNCATE and the like) with
41 constant zero divisors and integer dividends no longer generate
43 * bug fix: provide default methods for INPUT-STREAM-P and
44 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
45 * bug fix: improve the detection and resolution of MOP metacycles
46 (where effective method computation depends on the generic function
47 having its effective method computed). (reported by Bruno Haible)
48 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
49 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
50 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
51 * bug fix: pathnames with relative directory components can be
52 represented relative to default pathnames.
53 * optimization: in taking the GCD of bignums, reduce the two bignums
54 to approximately the same size (using Euclid's algorithm) before
55 applying the more sophisticated binary GCD. (thanks to Juho
57 * optimization: COUNT on bitvectors now operates word-at-a-time.
58 * optimization: ASH with a positive, but not necessarily constant,
59 (leftwards) shift, when in a modular context, is compiled to a
61 * fixed some bugs revealed by Paul Dietz' test suite:
62 ** FORMAT variable parameters ("~V<char>") are defaulted properly
63 if the corresponding argument is NIL.
64 ** FORMAT directives accepting a minpad parameter treat negative
66 ** literal commas as character parameters to FORMAT directives are
68 ** literal spaces directly after ~<Newline> directives within a
69 format-logical-block (~:< ~@:>) do not induce :FILL-style
71 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
72 as specified: it no longer includes conditional newlines).
73 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
75 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
76 :PER-LINE-PREFIX argument does not evaluate to a string.
78 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
79 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
80 the manual for details; for now, package locks can be disabled by
81 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
82 if no major problems are found then it is likely that they will be
83 compiled in unconditionally.
84 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
85 unsupported operators on all platforms. To load a shared library
86 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
87 object file, link it into a shared library outside of SBCL and
88 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
90 * bug workaround: ROOM T and the GC were not working together
91 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
92 reporting facility aren't preserved. That reporting has been
93 disabled, so now until and unless someone figures out how to make it
94 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
95 * minor incompatible change: as threatened around sbcl-0.8.0, the
96 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
97 The symbols are also exported from SB-PCL for backwards
98 compatibility, but more so than before SB-PCL should be treated as
99 an implementation-internal package.
100 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
102 * the SB-SPROF contrib now works on (most) non-x86 architectures.
103 It is known as of this release not to work on the Alpha, however.
104 * fixed bug #167: errors signalled due to illegal syntax in method
105 bodies are now more legible.
106 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
107 designators and can hence be used with TYPEP.
108 * fixed bug #333: CHECK-TYPE now ensures that the type error
109 signalled, if any, has the right object to be accessed by
110 TYPE-ERROR-DATUM. (reported by Tony Martinez)
111 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
112 5.1.2.3 for multiple-value place subforms. (reported by Kalle
114 * fixed bug #334: programmatic addition of slots using specialized
115 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
116 and :ALLOCATION :CLASS slots.
117 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
118 float exponents. (rereported by Peter Seibel)
119 * fixed a bug: #\Space (and other whitespace characters) are no
120 longer considered to be macro characters in standard syntax by
122 * fixed bug: initialization of condition class metaobjects no longer
123 causes an instance of the condition to be created. (reported by
125 * fixed bug: it is now possible to have slots such that
126 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
127 description is neither :INSTANCE nor :CLASS.
128 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
129 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
130 SLOT-VALUE-USING-CLASS), and no longer causes errors with
131 non-standard SLOT-DEFINITION-ALLOCATION values.
132 * fixed bugs: various race conditions handling exiting threads.
133 CL-PPCRE's thread torture test now passes.
134 * fixed arguably-a-bug: GC time proportional to number of threads in
135 system even when most of them are idle
136 * optimization: improved performance of BIT and SBIT on bit-vectors.
137 * .fasl file incompatibility: The fasl file version number has
138 been incremented because of changes associated with package locks.
140 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
141 * minor incompatible change: the system no longer provides
142 optimization hints (conditions which are TYPEP
143 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
144 mode; these hints are still emitted when higher SPEED optimization
146 * new contrib module: a sampling profiler (profiling by statistical
147 sampling, rather than by instrumenting functions) is available as
148 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
149 * the behaviour of the standard function ED is now customizeable by
150 third parties through a hook variable: see ED's documentation
151 string for information on the protocol.
152 * the compiler no longer emits efficiency notes for (FUNCALL X)
153 when the type of X is uncertain under default optimization
155 * fixed bug 276: mutating a binding of a specialized parameter to a
156 method to something that is not TYPEP the specializer is now
158 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
159 exhibit the required behaviour on the x86 platform. (thanks to
160 Peter van Eynde, Eric Marsden and Bruno Haible)
161 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
162 even for difficult arguments. (reported by Peter Graves)
163 * fixed bug 141a: the backquote printer now descends quoted
165 * fixed another bug in backquote printing: no more destructive
166 modification of the form's list structure. (reported by Brian
168 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
169 move between its address being taken and the call to
170 interrupt_thread, fixing a crashing race condition.
171 * the SB-POSIX contrib implementation has been adjusted so that it
172 no longer exhibits ridiculously poor performance when constructing
173 instances corresponding to C structs.
175 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
176 * minor incompatible change: the sb-grovel contrib now treats C
177 structures as alien (in the sense of SB-ALIEN) objects rather than
178 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
179 has implications for memory management of client code
180 (sb-grovel-returned objects must now be manually managed) and for
181 type safety (alien objects now have full types).
182 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
183 used to control emission of compiler diagnostics, rather than the
184 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
185 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
186 quality should be considered deprecated.
187 * install.sh now installs the user manual as well
188 * (not quite a new documentable feature, but worth considering in
189 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
190 beginnings of a semantically meaningful condition hierarchy is
191 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
193 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
194 designator as the defaults argument.
195 * fixed bug: Displaced arrays whose displaced-to array has become
196 too small now cause ARRAY-DIMENSION to signal an error, providing
197 for safer bounds-checking. (reported by Bruno Haible)
198 * fixed bug: DEFCLASS slot definitions with identical :READER and
199 :WRITER names now signal a reasonable error. (reported by Thomas
201 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
203 * fixed bug: Invalid slot specification errors now print correctly.
204 (thanks to Zach Beane)
205 * fixed bug 320: Shared to local slot value transfers in class
206 redefinitions now happen corrently from superclasses as
207 well. (reported by Bruno Haible)
208 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
210 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
211 improper lists correctly. (reported by Bruno Haible)
212 * fixed bug 313: source-transform for <fun-name> was erroneously
213 applied to a call of a value of a variable with name <fun-name>.
214 (reported by Antonio Menezes Leitao)
215 * fixed bug 307: The obsolete instance protocol ensures that
216 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
217 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
218 remove dead unknown-values globs from the middle of the stack.
219 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
220 easier to resume long computations after using *BREAK-ON-SIGNALS*
221 to diagnose and fix failures. (thanks to Nikodemus Siivola)
222 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
223 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
224 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
225 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
226 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
228 * fixed a (fairly theoretical) bug in string printing: if
229 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
230 does not have array-element-type equal to the most general string
232 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
233 function name. (thanks to Nikodemus Siivola)
234 * fixed bug: compile-time format string checker failed on
235 non-closed ~{. (reported by Thomas F Burdick)
236 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
237 don't warn on legal loop code involving more than one aggregate
238 boolean. (thanks to Nikodemus Siivola)
239 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
240 in keyword position to MAKE-PACKAGE was not being reported as
242 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
243 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
245 * fixed bugs 280 and 312: the checking for multiple definitions in a
246 file is less likely to become confused by uses of inline
248 * fixed bug: the #S reader macro performs the keyword coercion
249 specified for slot names. (reported by Kalle Niemitalo)
250 * fixed bug: lambda lists may contain symbols whose names start with
251 & but are not lambda-list-keywords; their occurrence triggers a
253 * fixed bug 321: define-method-combination argument lambda lists do
254 not cause invalid code to be generated when &optional or &aux
255 variables have default values. (reported by Bruno Haible)
256 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
257 have CLOS classes; this appears also to have fixed reported
258 instability in stack exhaustion detection.
259 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
260 MUFFLE-WARNING when no associated restart is present is now
262 * optimization: rearranged the expansion of various defining macros
263 so that each expands into only one top-level form in a
264 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
266 * optimization: used a previously little-used slot in symbols to
267 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
269 * fixed some bugs revealed by Paul Dietz' test suite:
270 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
272 ** Fixed an optimization bug related to inheritance of initforms
273 from local to shared slots.
274 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
275 ** CAST optimizer forgot to flush argument derived type.
276 ** print/read consistency on floats is now orders of magnitude
277 more likely. (thanks also to Bruno Haible for a similar report
279 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
280 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
281 for objects of type REAL. Make it so.
282 ** SXHASH is specified (infelicitously) to respect similarity,
283 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
284 it so. (thanks to Markus Ziegler)
285 ** on the Alpha, the compiler succeeds in compiling functions
286 returning a known number of arguments greater than 63.
287 ** fixed handling of invalid NIL arguments in keyword position
289 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
291 ** provided a readably-printable representation for RANDOM-STATE
293 ** ensured that pathnames, where they have a namestring, always
294 print using #P"..." syntax.
296 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
297 * Support for the forthcoming 2.0 version of the NetBSD kernel
298 running on x86 hardware has been added. (thanks to Perry
299 E. Metzger most immediately, and others for their past work)
300 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
301 current release version; SBCL's OpenBSD support had been broken
302 since about the time of OpenBSD's switch to ELF binary format.)
303 * [placeholder for DX summary]
304 ** user code with &REST lists declared dynamic-extent, under high
305 speed or space and low safety and debug optimization policy.
306 * The manual has been converted to Texinfo format and the debugger
307 chapter from the cmucl manual has been added.
308 * A facility has been added to extract documentation strings from
309 sbcl and store them as Texinfo-formatted snippets for inclusion in
310 the manual (via Texinfo's @include directive)
311 * bug fix: compiler emitted division in optimized DEREF. (thanks for
312 the test case to Dave Roberts)
313 * bug fix: multidimensional simple arrays loaded from FASLs had fill
314 pointers. (reported by Sean Ross)
315 * bug fix: PROFILE output is printed nicely even for large numerical
316 values. (thanks to Zach Beane)
317 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
318 greater than 32 handle EOF correctly.
319 * bug fix: on X86 an immediate argument of the IMUL instruction is
320 correctly printed in disassembly. (thanks to Lutz Euler)
321 * bug fix: class slots in redefined classes preserve their old
322 values. (thanks to Bruno Haible and Nikodemus Siivola)
323 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
324 succeeds. (reported by Marco Baringer)
325 * fixed some bugs revealed by Paul Dietz' test suite:
326 ** READ-SEQUENCE now works on ECHO-STREAMs.
327 ** RATIONALIZE works more according to its specification. (thanks
330 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
331 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
332 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
333 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
334 should matter to you only if you rebind the printer control
335 variables and then find you want different bindings in the
336 debugger than in the ordinary execution of your program.)
337 * The runtime build system has been tweaked to support building
338 (on SPARC/SunOS) using a C compiler which invokes Sun's own
339 assembler and linker. (thanks to Nikodemus Siivola)
340 * Unbound, undefined, undeclared variables now trigger full
341 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
342 more often programmer error than deliberate exploitation of undefined
344 * optimization: the hash algorithm for strings has changed to one
345 that is less vulnerable to spurious collisions. (thanks to Juho
347 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
348 needless bounds checking. (thanks to Juho Snellman)
349 * optimization: implemented multiplication as a modular
350 (UNSIGNED-BYTE 32) operation on the PPC backend.
351 * fixed some bugs revealed by Paul Dietz' test suite:
352 ** ADJUST-ARRAY now copies the datum in a zero rank array if
354 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
355 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
356 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
358 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
360 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
362 ** ECHO-STREAMs no longer attempt to echo the end of file value to
363 their output stream on EOF from read.
364 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
365 have been read to end-of-file.
366 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
368 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
369 description of determination of which consecutive characters
371 ** Printing the "Space" character with escaping on now yields "#\\ ",
372 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
373 ** Reading floating-point numbers with *READ-BASE* set to a number
374 less than 10 works correctly.
375 ** Reading floating-point numbers with *READ-BASE* set to a number
376 more than 10 works correctly.
377 ** Printing with *PRINT-READABLY* targets the standard readtable, not
378 the readtable currently in effect.
380 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
381 * minor incompatible change: parsing of namestrings on a physical
382 (Unix) host has changed; numbers after the final #\. in a
383 namestring are no longer interpreted as a version field. This is
384 intented to be largely invisible to the user, except that the
385 meaning of the namestring "*.*.*" has changed: it now refers to a
386 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
387 should usually be replaced by
388 (make-pathname :name :wild :type :wild :version :wild)
389 with the added benefit that this is more likely to be portable.
390 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
391 option to OPEN now signals an error if the file being opened
392 exists; this may have an impact on existing code.
393 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
394 signals received, on the PowerPC platforms (both Linux and
395 Darwin). (thanks to Pierre Mai for pointing out the location of
397 * several fixes on OS X: The system now builds and runs cleanly on
398 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
399 (thanks to Brian Mastenbrook)
400 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
401 second-resolution: integer multiples of 1/3600 between -24 and 24.
402 (thanks to Vincent Arkesteijn)
403 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
404 type when called with 1 argument; PEEK-CHAR checked type of
405 PEEK-TYPE only after having read first character from a
406 stream. (reported by Peter Graves)
407 * bug fix: the garbage collector now has much better locality
408 behaviour, and in particular no longer treats all memory as being
409 exhausted when many small objects point to each other in a deeply
411 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
412 recognized as being TYPEP their class.
413 * bug fix: the PUSHNEW documentation string has been corrected.
414 (thanks to Vincent Arkesteijn)
415 * bug fix: defaulting of the value for the last of an atypically
416 large number of multiple values being bound was not being
417 performed correctly on the Alpha or PPC platforms
418 * optimization: implemented multiplication as a modular
419 (UNSIGNED-BYTE 32) operation on the x86 backend.
420 * optimization: SEARCH on simple-base-strings can now be open-coded.
421 (see also contrib/compiler-extras.lisp for inspiration for
422 teaching the compiler about the Boyer-Moore algorithm).
423 * value, returned by MAX (and MIN) called with several EQUALP, but
424 not EQL, arguments now does not depend on compiler settings.
425 * fixed some bugs revealed by Paul Dietz' test suite:
426 ** in stack analysis liveness information is propagated from
427 non-local entry points.
428 ** pathwise CAST removing failed when the CAST node did not start
430 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
431 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
433 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
434 host is already defined.
435 ** RENAME-FILE works on streams instead of signalling an internal
437 ** PEEK-CHAR uses the current readtable when determining whether
438 or not a character is whitespace.
439 ** MERGE-PATHNAMES handles the case when the pathname does not
440 specify a name while the default-pathname specifies a version.
441 ** Pathnames now stand a better chance of respecting print/read
443 ** Attempting to use standardized file system operators with a
444 pathname with invalid :DIRECTORY components signals a
446 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
448 ** DIRECTORY on logical pathnames is more correct.
449 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
450 signal a TYPE-ERROR if their argument is not a stream.
451 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
452 designator argument does not designate a stream.
453 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
454 examining the synonym.
455 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
457 ** OPEN and WITH-OPEN-STREAM allow opening streams with
458 element-type larger than ([UN]SIGNED-BYTE 32).
460 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
461 * When built with the :SB-FUTEX feature, threaded builds now take
462 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
463 for faster/more reliable mutex and condition variable support.
464 * Incompatible change (but one you probably shouldn't have been using
465 anyway): the interface and code for arbitrating between multiple
466 threads in the same user session has been redesigned.
467 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
468 error when called without an explicit environment argument.
469 (thanks to Nikodemus Siivola)
470 * bug fix: buffered :DIRECTION :IO streams are less likely to become
471 confused about their position. (thanks to Adam Warner and Gerd
473 * bug fix: Pretty printing backquoted forms with unquotations in the
474 argument list position of various code constructs such as LAMBDA
475 now works correctly. (reported by Paul Dietz)
476 * bug fix: Pretty printing unquotations no longer loses all
477 stream position information.
478 * optimization: performance of string output streams is now less
479 poor for multiple small sequence writes.
480 * optimization: performance of CSUBTYPEP in the presence of complex
481 expressions involving CONS and NOT many times has been improved.
482 (reported by Paul Dietz)
483 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
485 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
486 streams for the REPL, for the debugger, and so on. (thanks to
488 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
489 * fixed some bugs revealed by Paul Dietz' test suite:
490 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
491 is ignored if it is too small, rather than propagating through
492 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
493 ** extremely complex negations of CONS types were not being
494 sufficiently canonicalized, leading to inconsistencies in
496 ** VALUES tranformer lost derived type.
498 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
499 * fixed a bootstrapping bug: the build process no longer assumes
500 that the various BOOLE-related constants have the same value in
501 host and target lisps. (noted by Paul Dietz' test suite on an
502 SBCL binary built from CLISP)
503 * The system can now be dynamically linked on the MIPS platform,
504 which enables dynamic loading of foreign code from Lisp. (thanks
505 to Ralf Baechle for discussions on the MIPS ABI)
506 * The system now records debugging information for its own source
507 files in a filesystem-position-independent manner, relative to
508 the "SYS" logical host.
509 * fixed a compiler bug: MV-LET convertion did not check references
510 to the "max args" entry point. (reported by Brian Downing)
511 * tweaked disassembly notes to be less confident about proclaiming
512 some instruction as an LRA. (thanks to Brian Downing)
513 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
514 now each have their own history, command character, and other
515 characteristics. (thanks to David Lichteblau)
516 * fixed some bugs revealed by Paul Dietz' test suite:
517 ** compiler failure in compiling LOGAND expressions including a
519 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
520 shift greater than 32.
521 ** FLUSH-DEST did not mark blocks for type check regeneration.
522 ** HANDLER-CASE failed to accept declarations in handler clauses
523 in some circumstances.
525 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
526 * New code in contrib/sb-introspect (still probably not entirely
527 stable yet) provides some support for smart Lisp development
528 environments like SLIME.
529 * The conditions signalled for errors occurring when loading .fasl
530 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
531 in a way which should help ASDF recover gracefully.
532 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
533 stuff has been cleaned up. If you code contrib/ stuff, this might
534 affect you, and you can look at contrib/README, contrib/STANDARDS,
535 and/or the 0.8.4.27 diff to check.
536 * In full calls the compiler now does not generate checks for declared
537 argument types for all arguments.
538 * various threading fixes
539 ** and some experimental patches which didn't make it into
540 the main tree for this release, but which are shipped in
541 contrib/experimental-thread.patch as a possible fix for some
542 failures (deadlock, spinning...) in GC-intensive multithreaded
544 * fixed PPC build problem (source code incompatibility of different
545 library versions): added offsetof() hackery which attempts to divine
546 where glibc maintainers put uc_mcontext today
547 * fixed bug 282: compiler does not trust type assertions while passing
548 arguments to a full call.
549 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
550 &OPTIONAL VALUES type parameter.
551 * fix bug 214: algorithm for noting rejected templates is now more
552 similar to that of template seletion. (also reported by rydis on
554 * fixed bug 141b: printing backquoted information readably and prettily
555 inserts a space where necessary.
556 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
557 (CHARACTER 10) are now reported as errors, rather than propagated
558 as unknown types. (reported by piso on #lisp)
559 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
560 with values NIL and :ERROR. (thanks to Milan Zamazal)
561 * fixed bug 191c: CLOS now does proper keyword argument checking as
562 described in CLHS 7.6.5 and 7.6.5.1.
563 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
564 longer raise an error; further, using a list as a for-as-arithmetic
565 counter now raises a meaningful error.
566 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
567 now understood by sequence creation functions such as MAKE-SEQUENCE
569 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
570 a STRING-INPUT-STREAM.
571 * compiler enhancement: SIGNUM is now better able to derive the type
573 * type declarations inside WITH-SLOTS are checked. (reported by
575 * fixed some bugs revealed by Paul Dietz' test suite:
576 ** incorrect optimization of TRUNCATE for a positive first
577 argument and negative second.
578 ** compiler failure in let-convertion during flushing dead code.
579 ** compiler failure while deriving type of TRUNCATE on an
580 interval, containing 0.
581 ** ASH of a negative bignum by a negative bignum count now returns
583 ** intersection of CONS types now canonicalizes properly, fixing
584 inconsistencies in SUBTYPEP.
586 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
587 * incompatible change: The --disable-debugger command line
588 option now clobbers the debugger at a more fundamental
589 level, by redefining #'INVOKE-DEBUGGER instead of by
590 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
591 is specified by ANSI to ignore *DEBUGGER-HOOK* and
592 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
593 enter the debugger REPL and then suffer recursive errors
594 because *DEBUG-IO* is also messed up in --disable-debugger mode;
595 while under the new system, BREAK in --disable-debugger mode
596 terminates the system just as an unhandled error would.
597 * fixed compiler performance when processing loops with a step >1;
598 * bug fix: DOCUMENTATION now retrieves generic function
599 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
600 support has been systematized, and now supports the methods
601 specified by ANSI, along with a default method and a method for
602 slot documentation. (reported by Nathan Froyd)
603 * bug fix: effective methods associated with a generic function are
604 no longer cached over a change of that generic function's method
605 combination. (reported by Andreas Fuchs)
606 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
607 names a non-existent directory. (thanks to Andreas Fuchs)
608 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
609 the PPC platform, lead to stack corruption; this has been fixed.
610 (reported by Rainer Joswig)
611 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
612 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
613 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
614 arithmetic in general, is now much more reliable on the PPC
616 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
617 platform now returns the right answer.
618 * optimization: restored some effective method precomputation in
619 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
620 precomputation is now tunable.
621 * optimization: compiler-internal data structure use has been
622 reviewed, and changes have been made that should improve the
623 performance of the compiler by about 20%.
624 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
625 simple-base-strings and simple-bit-vectors is improved.
626 * optimization: the optimization of 32-bit logical and arithmetic
627 functions introduced in version 0.8.3 on the x86 has been
628 implemented on the mips, ppc and sparc platforms; an
629 implementation of the same facility, but for 64-bit arithmetic,
630 has been added for the alpha.
631 * microoptimization: the compiler is better able to make use of the
632 x86 LEA instruction for multiplication by constants.
633 * bug fix: in some situations compiler did not report usage of
634 generic arithmetic in (SPEED 3) policy.
635 * bug 145b fix: compiler used wrong type specifier while converting
636 MEMBER-types to numeric.
637 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
639 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
640 index argument now works properly on the Alpha platform.
641 * bug fix: floating point exception treatment on the Alpha platform
643 * bug fix: FILE-POSITION works much better on string input and
644 output streams. (thanks to Nikodemus Siivola)
645 * bug fix: many threading/garbage collection symptoms sorted.
646 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
647 might be pseudo-atomic.
648 * internal change: Stopping for GC is now done with signals not
649 ptrace. GC is now done in whichever thread wanted it, instead of
651 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
653 * bug fix: result form in DO is not contained in the implicit
655 * incompatible change: ICR structure is changed; the value part of
656 CONTINUATION is now called LVAR; corresponding functions are
657 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
658 * added type deriver for ISQRT (thanks to Robert E. Brown).
659 * bug fix: better support for loading from the command line when an
660 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
662 * fixed some bugs revealed by Paul Dietz' test suite:
663 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
664 ** ROUND and FROUND now give the right answer when given very
665 small float arguments.
666 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
668 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
669 ** (GCD 0 <negative-integer>) returned <negative-integer>.
670 ** LCM should return a non-negative integer.
671 ** PARSE-INTEGER returned the index of a terminator instead of the
672 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
673 ** PARSE-INTEGER returned an incorrect index being applied to a
675 ** LCM with two arguments of 0 returns 0 rather than signalling
677 ** unsigned addition of a 32-bit constant with the high bit set no
678 longer causes an internal compiler error on the x86.
679 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
680 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
681 with negative last argument.
682 ** byte specifiers with zero size and position no longer cause
683 an error during type derivation.
684 ** bignum multiplication on the Alpha platform now returns the
686 * porting: The system now builds on SuSE AMD64, although it still
687 generates a 32-bit binary.
688 * .fasl file incompatibility: The fasl file version number has
689 been incremented (because of the changes to internal compiler
690 data structures referred to above).
692 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
693 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
694 more accurately, on the Darwin kernel running on PowerPC hardware.
695 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
696 * Compiler code deletion notes now signal a condition of type
697 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
698 an associated MUFFLE-WARNING restart.
699 * The compiler now performs limited argument count validation of
700 constant format strings in FORMAT, and where appropriate in ERROR,
701 CERROR and WARN. (thanks to Gerd Moellmann)
702 * New ASDF-INSTALL contrib can be used for automatic download and
703 installation of third-party Lisp code from CCLAN or other sites
705 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
706 which forces another thread to execute a function supplied by the
708 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
709 now accept and act upon their :ELEMENT-TYPE keyword argument.
710 (reported by Martin Atzmueller, Edi Weitz)
711 * bug fix: FILE-POSITION now accepts position designators up to
712 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
713 is the greater. (thanks to Patrik Nordebo)
714 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
715 Kalle Olavi Niemitalo)
716 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
717 printable. (reported by Eric Marsden)
718 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
719 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
720 * x86 bug fix in control stack exhaustion checking: now shows backtrace
721 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
722 (thanks to Stig Sandoe)
723 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
724 UNREAD-CHAR now correctly includes the unread character in the
725 target sequence. (thanks to Gerd Moellmann)
726 * bug fix in threaded builds: the system can now be suspended and
727 resumed by shell job control with minimal disruption.
728 * bug fixes in times and timezones >2038AD
729 * better handling of "where is GNU make?" problem in build scripts
730 (thanks to Nikodemus Siivola)
731 * new optimization: inside a named function any reference to a
732 function with the same name is considered to be a self-reference;
733 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
734 optimization quality.
735 * new optimization on x86: logical functions and + now have
736 optimized (UNSIGNED-BYTE 32) versions, which are automatically
737 used when the result is truncated to 32 bits.
738 * VALUES declaration is partially enabled.
739 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
740 * fixed some bugs revealed by Paul Dietz' test suite:
741 ** The system now obeys the constraint imposed by
742 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
743 types form a lattice under type intersection.
744 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
745 ** ASSOC now ignores NIL elements in an alist.
746 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
747 and (1+ MOST-POSITIVE-FIXNUM) answers.
748 ** The addition of a method with invalid qualifiers to a generic
749 function does not cause an error to be signalled immediately;
750 a warning is signalled, and the error is generated only on
751 calling the generic function.
752 * changed .fasl file version number, in part to add type codes for
753 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
754 obscure ANSI requirements
756 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
757 * fixed bug 148: failure to inline-expand a local function left
758 garbage, confusing the compiler.
759 * fixed bugs 3cd: structure slot readers perform type check if the
760 slot can have an invalid value (i.e. it is either not initialized
761 or can be written with a less specific slot writer).
762 * bug fix: the compiler now traps array references to elements off
763 the end of an array; previously, the bounds checking in some
764 circumstances could go off-by-one.
765 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
767 * type declarations for array element types now obey the description
768 on the CLHS page "Declaration TYPE", as per discussions on
769 sbcl-help around 2003-05-08. This means that a declaration
770 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
771 declaration, all references to BAR will be asserted or assumed
772 (with THE, so dependent on compiler policy) to involve objects of
773 type FOO. Note that no such declaration is implied in
774 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
775 * declared types of functions from the "Conditions"
776 chapter. (reported by Paul Dietz)
777 * bug fix: CERROR accepts a function as its first argument.
778 * bug fix: NTH an NTHCDR accept a bignum as index
779 arguments. (reported by Adam Warner)
780 * optimization: character compare routines now optimize comparing
781 against a constant character. (reported by Gilbert Baumann)
782 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
783 argument now works properly on the MIPS platform.
784 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
785 * repeated evaluation of the same DEFSTRUCT, a slot of which is
786 declared to have a functional type, does not cause an error
788 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
789 many :DEFAULT-INITARGS. (reported by Istvan Marko)
790 * fixed bug: if last continuation of a deleted block has a
791 destination, this destination should be deleted too. (reported by
793 * fixed a bug in the bootstrap process: the host compiler's values
794 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
795 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
796 test case from Patrik Nordebo)
797 * improved the ability of the disassembler on the PPC platform to
798 provide helpful disassembly notes.
799 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
800 the class in more cases than previously.
801 * bug fix: FILE-POSITION now understands :START and :END for
802 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
803 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
804 8) data. (thanks to David Lichteblau)
805 * bug fix: it is possible to add a method to a generic function
807 * bug fix: reader failed to signal END-OF-FILE inside an
808 object representation. (reported by Nikodemus Siivola)
809 * fixed some bugs revealed by Paul Dietz' test suite:
810 ** LAST and [N]BUTLAST should accept a bignum.
811 ** condition slot accessors are methods.
812 ** (VECTOR NIL) is a subtype of STRING.
814 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
815 * minor incompatible change: some nonsensical specialized lambda
816 lists (used in DEFMETHOD) which were previously ignored now signal
818 * minor incompatible change: the system is now aware of the types of
819 variables in the COMMON-LISP package, and will signal errors for
820 most violations of these type constraints (where previously they
821 were silently accepted).
822 * minor incompatible change: COMPILE-FILE now uses the freedom
823 afforded (ANSI 3.2.2.3) to use derived function types for
824 functions defined in the same file. This also permits the system
825 to warn on static type mismatches and function
826 redefinition. (Currently it does not work with high DEBUG level.)
827 * minor incompatible change: VALUES declaration is disabled.
828 * When issuing notes, the compiler now signals a condition of type
829 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
830 restart for use in user handlers. It is expected that the
831 COMPILER-NOTE condition will eventually become a condition
832 supertype to a hierarchy of note types, which will then be
833 handleable in a similar fashion. However, at the moment, no such
834 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
835 but it's an implementation detail, not a classification for the
837 * Changes in type checking closed the following bugs:
838 ** type checking of unused values (192b, 194d, 203);
839 ** template selection based on unsafe type assertions (192c, 236);
840 ** type checking in branches (194bc).
841 * A short form of VALUES type specifier has ANSI meaning (it has
842 increased the number of situations when SBCL cannot perform type
844 * fixed bug in DEFSTRUCT: once again, naming structure slots with
845 keywords or constants is permissible.
846 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
847 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
848 classes. (thanks to Antonio Martinez)
849 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
850 outputs FRESH-LINE or TERPRI, and no longer converts its stream
851 argument to a pretty-print stream. Instead, it leaves any such
852 operations to DESCRIBE-OBJECT methods.
853 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
855 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
856 lambda lists. (thanks to David Lichteblau)
857 * bug fix: Defining a generic function with a :METHOD-CLASS being a
858 subclass of STANDARD-METHOD no longer causes stack exhaustion.
859 (thanks to Gerd Moellmann)
860 * fixed bug 246: increased compilation speed of long
861 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
863 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
865 * DEFCONSTANT now throws a condition of type
866 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
867 constant to a non-EQL value; CONTINUE and ABORT restarts
868 respectively change and preserve the value.
869 * fixed bug 63: The code walker, part of the implementation of CLOS,
870 is now better at handling symbol macros.
871 * bug fix: There is no longer an internal implementation type named
872 CL:LENGTH. (reported by Raymond Toy)
873 * bug fix: In macro-like defining macros/special operators the
874 implicit block does not enclose the lambda list.
875 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
876 enclosed in parentheses) are not suitable as type specifiers, and
877 their use properly signals an error now.
878 * bug fix: An argument count mismatch for a type specifier in code
879 being compiled no longer causes an unhandled error at compile
880 time, but signals a compile-time warning.
881 * fixed simple vector readable printing
882 * bug fix: DESCRIBE takes more care over whether the class
883 precedence list slot of a class is bound before accessing it.
884 (reported by Markus Krummenacker)
885 * bug fix: FORMATTER can successfully compile pretty-printer format
886 strings which use variants of the ~* directive inside.
887 * bug fix: SEARCH now applies its TEST predicate to the elements of
888 the arguments in the correct order. (thanks to Wolfhard Buss)
889 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
890 in type inference. (thanks to Robert E. Brown)
891 * bug fix: Reading in symbols with an explicit package name of ""
892 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
893 * fixed some bugs revealed by Paul Dietz' test suite:
894 ** NIL is now allowed as a structure slot name.
895 ** Arbitrary numbers, not just REALs, are allowed in certain
896 circumstances in LOOP for-as-arithmetic clauses.
897 ** Multiple class redefinitions before slot access no longer
899 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
900 association between the name and a class.
901 ** Generic functions with non-standard method-combination and over
902 five methods all of which return constants no longer return NIL
903 after the first few invocations. (thanks to Gerd Moellmann)
904 ** CALL-NEXT-METHOD with no arguments now passes the original
905 values of the arguments, even in the presence of assignment.
906 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
908 ** DISASSEMBLE works with closures and funcallable instances.
909 ** ADD-METHOD now returns the generic function, not the new method.
910 ** FIND-METHOD signals an error if the lengths of the specializers
911 is incompatible with the generic function, even if the ERRORP
913 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
914 which its argument is a member.
915 ** DEFCLASS only redefines the class named by its class-name
916 argument if that name is the proper name of the class;
917 otherwise, it creates a new class.
918 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
919 of the UNBOUND-SLOT condition to the name of the slot.
920 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
921 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
922 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
923 SLOT-MAKUNBOUND in the specified fashion.
925 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
926 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
927 cross-compilation host. As a consequence, we can now bootstrap our
928 way up to SBCL starting with a bare gcc toolchain and human-readable
929 source code (first the source to CLISP, then the source to SBCL).
930 * A contributed module containing a partial implementation of the
931 simple-streams interface has been included. (thanks to Rudi
933 * A contributed module implementing the RFC1321 Message Digest
934 Algorithm, known as MD5, has been included.
935 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
936 no longer has any effect, as the code controlled by this feature
937 has been deleted. (As far as we know, no-one has ever built using
938 this feature, and its semantics were confused in any case).
939 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
940 work on conditions (as required by the ANSI specification),
941 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
942 expected behaviour on conditions. Users should note, however,
943 that such behaviour is not required by the ANSI specification,
944 and so use of this behaviour may render their code unportable.
945 * fixed some bugs revealed by Paul Dietz' test suite:
946 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
948 ** &ENVIRONMENT parameter in macro lambda list is bound first.
949 ** SXHASH on condition objects no longer returns NIL.
950 ** :ALLOCATION :CLASS slots are better treated; their values are
951 updated on class redefinition, and initforms inherited from
952 superclasses are applied.
953 ** REMOVE-METHOD returns its generic function argument even when
954 no method was removed.
955 ** SHARED-INITIALIZE now initializes the values of the requested
956 slots, including those with :ALLOCATION :CLASS.
957 ** ALLOCATE-INSTANCE now works on structure classes defined via
958 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
960 ** SLOT-EXISTS-P now works on conditions, as well as structures
962 ** MAKE-LOAD-FORM now has the required methods on
963 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
964 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
965 keyword, and now implements the SLOT-NAMES argument.
966 ** methods with &OPTIONAL arguments no longer allow too many
967 arguments to be passed in the call without error.
968 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
969 option is consistent with the required arguments of the generic
970 function lambda list.
971 * bug fix: REQUIRE accepts a string designator. (Thanks to
973 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
974 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
975 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
976 * bug fix: The compiler now checks for duplicated variables in macro
978 * bug fix: SETQ on globals returns the correct value.
979 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
980 not a COMPILER-ERROR (followed by some other strange error on
981 choosing the CONTINUE restart).
982 * bug fix: make.sh and friends are now more consistent in the way that
983 they look for GNU "make".
985 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
986 * experimental native threads support (on x86 Linux >=2.4 only).
987 This is not compiled in by default: you need to add :SB-THREAD to
988 the target features. See the "Beyond ANSI" chapter of the manual
990 * fix for longstanding nonANSIism: The old distinction between
991 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
992 The return value from CL:FIND-CLASS is now a CLOS class, and
993 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
994 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
995 * An interface to the MetaObject Protocol, as described in Kiczales,
996 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
997 MIT Press, 1991, is available from the SB-MOP package.
998 * incompatible change: the SB-PCL package should now be considered
999 a private implementation detail, and no longer a semi-private MOP
1001 * minor incompatible change: due to rearrangement for threads, the
1002 control stack and binding stack are now allocated at arbitrary
1003 addresses instead of being hardcoded per-port. Users affected by
1004 this probably have to be doing advanced things with shared
1005 libraries, and will know who they are.
1006 * minor incompatible change: Previously, all --eval forms used were
1007 processed with READ before any of them were processed with EVAL.
1008 Now each --eval form is processed with both READ and EVAL before
1009 the next --eval form is processed. (Thus package operations like
1010 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1011 work as the user might reasonably expect.)
1012 * minor incompatible change: *STANDARD-INPUT* is now only an
1013 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1015 * minor incompatible change: Y-OR-N-P is now character-oriented,
1016 not line oriented. Also, YES-OR-NO-P now works without errors.
1017 (thanks to Antonio Martinez)
1018 * sb-aclrepl module improvements: an integrated inspector, added
1019 repl features, and a bug fix to :trace command.
1020 * Known functions, which cannot be open coded by the backend, are
1021 considered to be able to check types of their arguments. (fixing
1022 a bug report by Nathan J. Froyd)
1023 * fixed a bug in computing method discriminating functions: It is
1024 now possible to define methods specialized on classes which have
1025 forward-referenced superclasses. (thanks to Gerd Moellmann)
1026 * fixed evaluation order in optional entries (reported by Gilbert
1028 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1029 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1030 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1031 required-by-AMOP NAME argument, as well as CLASS and
1032 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1033 * fixed bug 20: DEFMETHOD can define methods using names that are
1034 not the proper names of classes to designate class specializers.
1035 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1036 Unix file descriptors, instead of blowing up. (thanks to Antonio
1038 * Garbage collection refactoring: user-visible change is that a
1039 call to the GC function during WITHOUT-GCING will not do garbage
1040 collection until the end of the WITHOUT-GCING. If you were doing
1041 this you were probably losing anyway.
1042 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1043 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1044 TYPEP the latter but not the former.
1045 * The compiler issues a full WARNING for calls to undefined functions
1046 with names from the CL package.
1047 * MAP-INTO for a vector destination is open coded. (reported by
1048 Brian Downing on c.l.l)
1049 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1050 documentation string.
1051 * fixed some bugs revealed by Paul Dietz' test suite:
1052 ** COPY-ALIST now signals an error if its argument is a dotted
1054 ** Condition slots are now accessed more correctly in the presence
1055 of multiple initargs for a given slot.
1056 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1057 exclude restarts of the same name associated with a different
1059 ** DEFCLASS of forward-referenced classes with another
1060 forward-referenced class in the superclasses list no longer
1062 ** Condition slots are now initialized once each, not multiple
1063 times. (thanks to Gerd Moellmann)
1064 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1066 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1067 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1068 arguments contain duplicated elements.
1069 ** RESTART-CASE understands local macros.
1070 ** RESTART-CASE associates exactly its own restarts with a condition.
1071 ** ENDP in safe mode checks its argument to be of type LIST.
1072 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1073 in question is unbound.
1074 ** Optimized MAKE-INSTANCE functions no longer cause internal
1075 assertion failures in the presence of duplicate initargs.
1076 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1077 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1078 propagated even in the ctor optimized implementation of
1080 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1082 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1083 * a better implementation of SXHASH on (simple) bit vectors,
1084 measured both in execution speed and in distribution of results
1085 over the positive fixnums, has been installed. Likewise, a better
1086 implementation of EQUAL for simple bit vectors is now available.
1087 * fixed CEILING optimization for a divisor of form 2^k.
1088 * fixed bug 240 (emitting extra style warnings "using the lexical
1089 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1090 by Antonio Martinez)
1091 * fixed SXHASH, giving different results for NIL depending on type
1092 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1093 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1094 variable. (found by Rolf Wester)
1095 * SBCL does not ignore type declarations for special
1096 variables. (reported by rif on c.l.l 2003-03-05)
1097 * some bug fixes in contrib/sb-aclrepl/
1098 * fixed some bugs revealed by Paul Dietz' test suite:
1099 ** a bug in the CONS type specifier, whereby the CAR and CDR
1100 types got intertwined, has been fixed;
1101 ** the type system is now able to reason about the interaction
1102 between INTEGER and RATIO types more completely;
1103 ** APPEND, [N]REVERSE and NRECONC check that those their
1104 arguments, which must be proper lists, are really so;
1105 ** An array specialized to be unable to hold elements has been
1106 implemented, as required -- yes, really -- by ANSI;
1107 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1108 on malformed property lists;
1110 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1111 * incompatible packaging change: in line with Unix convention,
1112 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1113 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1114 * REQUIRE and PROVIDE are now optionally capable of doing something
1115 useful. See the documentation string for REQUIRE.
1116 * infrastructure for a managed SBCL contrib system: contributed
1117 modules in this release include:
1118 ** the ASDF system definition facility;
1119 ** an interface to the BSD Sockets API;
1120 ** an ACL-like convenience interface to the repl;
1121 (thanks to Kevin Rosenberg)
1122 ** an implementation of ROTATE-BYTE, with efficient implementation
1124 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1125 gives a closer approximation to the right answer than previously.
1126 (thanks to Raymond Toy)
1127 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1128 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1129 optional environment arguments, as required by ANSI.
1130 * fixed bugs in other functions taking environment objects, allowing
1131 calls with an explicit NIL environment argument to be compiled
1133 * fixed bug 228: primary return values from
1134 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1135 COMPILE or FUNCTION.
1136 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1137 :TYPE will no longer signal errors on innocuous objects.
1138 * fixed bug 231b: SETQ is better at respecting type declarations in
1139 the lexical environment.
1140 * fixed a bug in DEFCLASS: classes named by symbols with no or
1141 unprintable packages can now be defined.
1142 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1143 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1144 * fixed compiler failure related to checking types of functions.
1145 (reported by Robert E. Brown)
1146 * the compiler is now much more consistent in its error-checking
1147 treatment of bounding index arguments to sequence functions: in
1148 (SAFETY 3) code, errors will be signalled in almost all cases if
1149 invalid sequence bounding indices are passed to functions defined
1150 by ANSI to operate on sequences.
1151 * fixed a bug in the build procedure: documentation of SBCL-specific
1152 packages is now preserved and available in the final Lisp image.
1153 * lifted FDEFINITION lookup out of loops in the implementation of
1154 many list operations. (thanks to Robert E. Brown)
1155 * fixed a bug in the reader: the #n# reader macro now works for
1156 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1157 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1158 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1159 E. Brown and Tony Martinez respectively)
1160 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1161 SLOT-BOUNDP: in particular, it is now able to optimize them much
1162 better, and is now not vulnerable to having packages renamed.
1163 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1164 always called when a slot is not present in an instance. (thanks
1166 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1167 longer signal an error on unreading a character following EOF on
1168 the previous constituent stream. (thanks to Tony Martinez)
1169 * fixed some bugs revealed by Paul Dietz' test suite:
1170 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1171 not just nonnegative fixnums;
1172 ** the logical bit-array operators such as BIT-AND now accept an
1173 explicit NIL for their "opt-arg" argument (to indicate a
1174 freshly-consed result bit-array);
1175 ** ELT now signals an error on an invalid sequence index in safe
1177 ** the type system is now cleverer about negations of numeric
1178 types, and consequently understands the BIGNUM and RATIO types
1180 ** the type system is now cleverer about the interaction between
1181 INTEGER and RATIO types: while bugs still remain, many more
1182 cases are accurately computed;
1183 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1184 if it is in the last clause;
1185 ** CONSTANTLY now correctly returns a side-effect-free function in
1187 ** DECLARE is no longer treated as a special-operator; in
1188 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1189 * incremented fasl file version number due to the change in the
1190 DEFSTRUCT-SLOT-DESCRIPTION structure.
1192 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1193 * minor incompatible change: code processed by the "interpreter" or
1194 EVAL now has a compilation optimization policy of (DEBUG 2)
1195 (changed from (DEBUG 1)) to improve debuggability of interactive
1196 development, and to allow the use of the debug RETURN command in
1198 * an experimental implementation of the RETURN command for the
1199 debugger has been included. (thanks to Frederik Kuivinen)
1200 * fixed bug 62: constraints were not propagated into a loop.
1201 * fixed bug in embedded calls of SORT (reported and investigated by
1203 * fixed some bugs revealed by Paul Dietz' test suite:
1204 ** printing and reading of arrays with some dimensions having
1205 length 0 (thanks to Gerd Moellmann);
1206 ** BOA constructor with &AUX argument without a default value does
1207 not cause a type error;
1208 ** CONSTANTP now returns true for all self-evaluating objects.
1210 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1211 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1212 accessors that are related by inheritance, as specified in the
1213 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1214 Valtteri Vuorikoski)
1215 * The compiler is now able to inline functions that were defined in
1216 a complex lexical environment (e.g. inside a MACROLET).
1217 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1219 * Support for the upcoming FreeBSD-5.0 release has been included.
1220 (thanks to Dag-Erling Smorgrav)
1221 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1222 effect when it is not in a toplevel context.
1223 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1224 stand a better chance of being correct. (thanks to Gerd
1226 * fixed bug in COERCE, which now signals an error on coercing a
1227 rational to a bounded real type which excludes the expected
1229 * The compiler is now able to derive types more accurately from the
1230 COERCE and COMPILE functions.
1231 * fixed bug 223: functional binding is considered to be constant
1232 only for symbols in the CL package.
1233 * fixed bug 231: SETQ did not check the type of a variable being set
1234 (reported by Robert E. Brown)
1235 * A new optimization for MAKE-INSTANCE has been included, fixing
1236 various bugs (including relating to :ALLOCATION :CLASS slots and
1237 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1239 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1240 ** As required by ANSI, LOOP now disallows anonymous collection
1241 clauses such as COLLECT I in conjunction with aggregate boolean
1242 clauses such as THEREIS (= I 1);
1243 ** LOOP now signals an error when any variable is reused in the
1244 same loop (including the potentially useful construct analogous
1245 to WITH A = 1 WITH A = (1+ A);
1246 ** IT is only a special loop symbol within the first clause of a
1247 conditional loop clause;
1248 ** LOOP with a typed iteration variable over a hashtable now
1249 signals a type error iff it should.
1250 * fixed some other bugs revealed by Paul Dietz' test suite:
1251 ** FILE-STREAM now names the class previously known as FD-STREAM;
1252 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1253 argument) no longer signals an error;
1254 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1255 of the slot symbol, rather than using the current package
1256 ((:CONC-NAME "") continues to intern the slot's name in the
1258 * incremented fasl file version number, because of the incompatible
1259 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1260 of the new implementation of DEFINE-COMPILER-MACRO.
1262 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1263 * Support for building SBCL for MIPS platforms running in
1264 little-endian mode has now been checked in, and basic
1265 functionality on said platforms verified.
1266 * minor incompatible change: PCL now records the pathname of a file
1267 in which methods and the like are defined, rather than its
1269 * minor incompatible change: TRUENAME now considers the truename of
1270 a file naming a directory to be the pathname with :DIRECTORY
1271 component indicating that directory.
1272 * minor incompatible change: a NAMED clause in the extended form of
1273 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1274 reason for the previous behaviour is unclear.
1275 * more systematization and improvement of CLOS and MOP conformance
1276 in PCL (thanks to Gerd Moellman and Pierre Mai):
1277 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1279 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1280 primary methods with no specializers;
1281 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1283 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1284 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1285 FORWARD-REFERENCED-CLASSes; error reporting on
1286 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1288 ** SXHASH on CLOS instances now uses a slot internal to the
1289 instance to return different numbers on distinct instances,
1290 while preserving the same return value through invocations of
1292 ** DEFMETHOD signals errors when methods with longer incongruent
1293 lambda lists are added to generic functions;
1294 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1295 CLASS, as specified in AMOP;
1296 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1297 on the order returned by the primary method for classes of
1298 class STANDARD-CLASS;
1299 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1300 * fixed some bugs shown by Paul Dietz' test suite:
1301 ** DOLIST puts its body in TAGBODY;
1302 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1304 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1305 value producing form;
1306 ** if more variables are given to PROGV than values, extra
1307 variables are bound and made to have no value;
1308 ** NSUBSTITUTE on list arguments gets the right answer with
1310 ** ELT signals an error of type TYPE-ERROR when the index argument
1311 is not a valid sequence index;
1312 ** LOOP signals (at macroexpansion time) an error of type
1313 PROGRAM-ERROR when duplicate variable names are found;
1314 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1315 ** FUNCALL of special-operators now cause an error of type
1317 ** PSETQ now works as required in the presence of side-effecting
1318 symbol-macro places;
1319 ** NCONC accepts any object as its last argument;
1320 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1322 ** loop-for-as-package does not require a package to be explicitely
1324 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1326 * fixed bug 166: compiler preserves "there is a way to go"
1327 invariant when deleting code.
1328 * fixed bug 172: macro lambda lists with required arguments after
1329 &REST arguments now cause an error to be signalled. (thanks to
1331 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1332 bug. (thanks to Matthew Danish)
1333 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1335 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1336 arguments are lexically rebound. (thanks to Gerd Moellmann and
1338 * fixed bug 194: error messages are now more informative when there
1339 is no primary method applicable in a call to a generic
1340 function. (thanks to Gerd Moellmann)
1341 * fixed bug in command line argument checking (thanks to Julian
1343 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1344 (thanks to Matthew Danish)
1345 * incremented fasl file version number, because of the
1346 SXHASH-related changes in the layout of CLOS data structures
1348 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1349 * minor incompatible change: The runtime (the Unix executable named
1350 "sbcl") is now much pickier about the .core files it will load.
1351 Essentially it now requires .core files to descend from the same
1352 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1353 as the runtime does. (The intent is to prevent the crashes which
1354 can occur, and which can even be reported as mysterious failures,
1355 when people patch the sources or change the build parameters
1356 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1357 sbcl and .core files.)
1358 * fixed bug: VALUES-LIST is no longer optimized away.
1359 * fixed bug 142: The FFI conversion of C string values to Lisp
1360 string values no longer conses excessively. (thanks to Nathan
1361 Froyd porting Raymond Toy's fix to CMU CL)
1362 * began to systematize and improve MOP conformance in PCL (thanks to
1363 Nathan Froyd, Gerd Moellman and Pierre Mai):
1364 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1366 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1367 ** FINALIZE-INHERITANCE is now called on class finalization;
1368 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1369 argument precedence order.
1370 * fixed bug 202: The compiler no longer fails on functions whose
1371 derived types contradict their declared type.
1372 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1373 so it can be non-toplevel.
1374 * The fasl file version number has changed (because of the new
1375 implementation of DEFMACRO).
1376 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1377 safe code, that any length requirement by their type-specifier
1378 argument is valid. The exceptions to this are described in bug
1380 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1381 functions now check the types of their inputs as required by ANSI.
1382 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1383 introduced symbol is DECLAREd to be SPECIAL.
1384 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1385 * fixed bug: :COUNT argument to sequence functions may be negative
1386 * fixed bug: body of DO-SYMBOLS may contain declarations
1387 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1388 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1389 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1390 symbol macro only once
1391 * fixed printing of call frame when argument list is unavailable
1392 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1393 * compiler no longer signals WARNING on unknown keyword
1396 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1397 * A beta-quality port to the mips architecture running Linux,
1398 based on the old CMUCL backend, has been made. It has been tested
1399 on a big-endian kernel, and works sufficiently well to be able to
1400 rebuild itself; it has not been tested in little-endian mode.
1401 * fixed an inconsistency between gencgc.c and purify.c which made
1402 dumping/loading .core files unreliable
1403 * fixed bug 120a: The compiler now deals correctly with IFs where
1404 the consequent is the same as the alternative, instead of
1405 misderiving the return type. (thanks to Alexey Dejneka)
1406 * fixed bug 113: Logical pathnames are now dumpable (the logical
1407 host is resolved at load-time, throwing an error if it is not
1409 * fixed bug 174: FORMAT's error message is slightly clearer when a
1410 non-printing character is used in a format directive.
1411 * fixed several bugs in compiler checking of type declarations, i.e.
1412 violations of the Python "declarations are assertions" principle
1413 (thanks to Alexey Dejneka)
1414 * fixed several bugs in PCL's error checking (thanks to Gerd
1416 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1418 * fixed bug in compilation of functions as first class values
1419 (thanks to Antonio Martinez-Shotton)
1420 * The compiler's handling TYPE-ERRORs which it can prove will
1421 inevitably happen at runtime has been cleaned up and corrected
1422 in several ways. (thanks to Alexey Dejneka)
1423 * improved argument type checking for various basic arithmetic
1424 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1425 lavished on them in the past that they can be compiled in many
1426 ways in different special cases
1427 * fixed bug 181: compiler checks validity of user supplied type
1429 * cleaned up code flushing in optimization: Function calls which
1430 should signal errors for safety purposes (e.g. which ANSI says
1431 should signal errors when their arguments are of incorrect type)
1432 are no longer optimized away.
1433 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1434 * incremented fasl file version number, because changes in the
1435 implementation of sequence functions like COERCE caused
1436 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1437 in old inline expansions) to become undefined. (Actually these
1438 changes were later undone, so we might very well be binary
1439 compatible with 0.7.7 after all, but leaving the version number
1440 incremented seemed like the simplest and most conservative
1443 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1444 * An alpha-quality port to the parisc architecture running Linux,
1445 based on the old CMUCL backend, has been made. This, even more so
1446 than the other backends, should be considered still a work in
1447 progress; known problems include that the Linux kernel in 64-bit
1448 mode does not propagate the correct sigcontext structure to
1449 userspace, and consequently SBCL on a parisc64 kernel will not
1451 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1452 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1453 suppressed.) Also now that the compiler is looking at declarations
1454 in the environment, it checks optimization declarations as well,
1455 and suppresses inlining when (> DEBUG SPEED).
1456 * More fixes have been made to treatment of floating point exception
1457 treatment and other Unix signals. In particular, floating point
1458 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1459 * The detection and handling of control stack exhaustion (infinite
1460 or very deeply nested recursion) has changed. Stack exhaustion
1461 detection is now done by write-protecting pages at the OS level
1462 and applies at all optimization settings; when found, a
1463 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1464 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1465 be caught using IGNORE-ERRORS.
1466 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1467 correctly when transforming recursive LABELS functions to LETs.
1468 (thanks to Alexey Dejneka)
1469 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1470 that are names of constants or global variables.
1471 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1472 alien routines with docstrings.
1473 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1474 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1476 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1477 (thanks to Eric Marsden)
1478 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1479 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1480 * Bug fix: The SPARC backend can now compile functions involving
1481 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1482 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1483 to a functions expecting a PATHNAME-DESIGNATOR.
1484 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1485 lambda lists. (thanks to Alexey Dejneka)
1486 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1487 functions so that EXPORTing the name of the class doesn't cause
1488 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1489 to Antonio Martinez for reporting this)
1490 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1491 constant-folded without failing an assertion. (thanks to Einar
1492 Floystad Dorum for reporting this)
1493 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1494 types are now dealt with more correctly by the compiler.
1495 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1496 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1497 behaviour analogous to RENAME-FILE. This puts its behaviour more
1498 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1500 * The fasl file version number has changed again. (because of the
1501 bug fix involving the names of PCL MAKE-INSTANCE functions)
1503 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1504 * bug fix: Floating point exceptions are treated much more
1505 consistently on the x86/Linux and PPC/Linux platforms.
1506 * Array initialization with :INITIAL-ELEMENT is now much faster for
1507 cases when the compiler cannot open code the array creation, but
1508 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1509 array accesses have also seen a speed increase.
1510 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1511 specified. (This was at the root of some bad interactions between
1512 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1514 * bug fix: Internal error arguments for undefined functions are now
1515 computed correctly on the PPC/Linux platform.
1516 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1517 Raymond Toy's patch for CMU CL)
1518 * Support for the Solaris 9 operating environment has been included
1519 (thanks to Daniel Merritt)
1520 * A very ugly but hopefully complete draft of the missing FFI chapter
1521 of the manual has been created by reformatting the corresponding
1522 CMU CL manual chapter into (currently very ugly and incoherent)
1523 DocBook and bringing it up to date for SBCL behavior. Thus, the
1524 manual is now essentially complete, at least by my extreme
1525 once-and-only-once standards, whereby it's acceptable to refer to
1526 the doc strings of SB-EXT functions as the primary documentation.
1527 * The fasl file version number has changed again, due to cleanup of
1528 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1529 is no longer a static symbol.)
1531 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1532 * SBCL now builds with OpenMCL (version 0.12) as the
1533 cross-compilation host; also, more progress has been made toward
1534 bootstrapping under CLISP.
1535 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1537 * bug 158 fixed: The compiler can now deal with integer loop
1538 increments different from 1; fixing this turned out also to fix
1540 * bug 169 fixed: no more bogus warnings about using lexical bindings
1541 despite the presence of perfectly good SPECIAL declarations (thanks
1542 to David Lichteblau)
1543 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1544 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1545 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1547 * bug 180 fixed: Method combination specifications no longer ignore
1548 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1549 * bug fix: Structure type predicate functions now check their argument
1550 count as they should.
1551 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1552 correctly. (thanks to Pierre Mai)
1553 * minor incompatible change: The --noprogrammer option is deprecated
1554 in favor of the new --disable-debugger option, which is very similar.
1555 (The major difference is that it takes effect at a slightly different
1556 time at startup, causing handling of errors in --sysinit and
1557 --userinit files will be affected differently.) The
1558 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1559 been added to allow this functionality to be controlled from ordinary
1560 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1561 maintainers, who might want to run non-interactive scripts to
1562 build SBCL cores which will later be used interactively.)
1563 * minor incompatible change: The LOAD function no longer, when given
1564 a wild pathname to load, loads all files matching that pathname.
1565 Instead, an error of type FILE-ERROR is signalled.
1567 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1568 * bug 147 fixed: The compiler preserves its block link/count
1569 invariants more correctly now so that it doesn't crash. (thanks
1571 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1573 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1574 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1575 no-value-for-this-character as NIL (as specified by ANSI).
1576 * HOST-NAMESTRING on physical pathnames now returns a string that is
1577 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1578 (thanks to Christophe Rhodes)
1579 * The Alpha port handles icache flushing more correctly. (thanks to
1581 * More progress has been made toward bootstrapping under CLISP. (thanks
1582 to Christophe Rhodes)
1583 * The fasl file format has changed again, because dynamic loading
1584 on OpenBSD (which has non-ELF object files) motivated some cleanups
1585 in the way that foreign symbols are transformed and passed around.
1586 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1587 is no longer treated as whitespace by the reader, but instead as
1588 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1589 |A
\7fB|, instead of A as it used to.
1591 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1592 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1593 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1594 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1595 this as of 0.7.1.45, but was left out of the previous news section
1596 (thanks to Dan Barlow)
1597 * SBCL now runs on the Solaris operating system on SPARC architectures
1598 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1599 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1600 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1601 * SPARC backend cleanups, allowing builds of cores optimized for V8
1602 and V9 SPARCS, and also emission of code targeted to a particular
1603 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1605 * SBCL is closer to bootstrapping under CLISP, thanks to various
1606 fixes by Christophe Rhodes.
1607 * The fasl file format has changed again, to allow the compiler's
1608 INFO database to support symbol macros.
1609 * The user manual (in doc/) is formatted into HTML more nicely.
1610 (thanks to coreythomas)
1611 * The system is smarter about SUBTYPEP relationships, especially
1612 those involving NOT types (including types such as ATOM which are
1613 represented internally using NOT types). Thus SUBTYPEP is less
1614 likely to return (VALUES NIL NIL) in general, and in particular
1615 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1617 * The fasl file format has changed again, because the internal
1618 representation of types now includes a new slot to support the new
1619 SUBTYPEP-of-NOT-types logic.
1620 * (not a change in the main branch of SBCL, but a related prototype
1621 which can hopefully be merged into the main branch of SBCL in the
1622 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1623 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1624 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1625 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1626 dispatch macro character combinations. (thanks to Alexey Dejneka)
1627 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1628 that superficially look like logical namestrings correctly.
1629 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1631 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1632 default supplied-p) for &optional and &key arguments. (thanks to
1635 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1636 * incompatible change: The compiler is now less aggressive about
1637 tail call optimization, doing it only when (> SPACE DEBUG) or
1638 (> SPEED DEBUG). (This is an incompatible change because there are
1639 programs which relied on the old CMU-CL-style behavior to optimize
1640 away their unbounded recursion which will now die of stack overflow.)
1641 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1642 for non-GENCGC systems has been increased to 20M (since that
1643 seems much closer to the likely performance optimum for modern
1644 systems than the old 4M value was)
1645 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1646 and *DEBUG-PRINT-LEVEL*
1647 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1648 of CMU CL's support for SPARC, and various endianness and other
1649 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1650 * new syntactic sugar for the Unix command line: --load foo.bar is now
1651 an alternate notation for --eval '(load "foo.bar")'.
1653 ** The system now detects stack overflow and handles it gracefully,
1654 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1655 optimization settings. (This is a good thing in general, and
1656 its introduction in this version should be particularly timely
1657 for anyone whose code fails because of suppression of tail
1659 ** The system now hunts for the C variable "environ" in a more
1660 devious way, to avoid segfaults when the C library version differs
1661 between compile time and run time. (thanks to Christophe Rhodes)
1662 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1663 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1664 ** The compiler no longer issues bogus style warnings for undefined
1665 classes in the same source file as the DEFCLASSes which defined
1666 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1668 ** fixes in CONDITION class precedence list for undefined function
1669 errors (thanks to Alexei Dejneka)
1670 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1671 correctly. (thanks to Dan Barlow)
1672 ** portability fixes aiming at bootstrapping under CLISP (thanks
1673 to Dave McDonald and Christophe Rhodes)
1674 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1675 ** fixes in type translation and and type inference (thanks to
1677 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1678 ** various fixes in the new ports (thanks to Dan Barlow)
1679 * several changes related to debugging:
1680 ** suppression of tail recursion, as noted above
1681 ** stack overflow detection, as noted above
1682 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1683 is now the default. (For some time encapsulation has been more
1684 reliable than the breakpoint-based :ENCAPSULATE NIL
1685 implementation, at least on X86 systems; and I just noticed that
1686 encapsulation also seems closer to the spirit of the ANSI
1689 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1691 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1692 up properly again. (There was a packaging bug in 0.7.0 which
1693 left their definitions in SB-SYS::LOAD-FOREIGN and
1694 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1695 vital for most things which interface to C-level interfaces,
1696 like extensions working with sockets or databases or
1697 Perl-compatible regexes or whatever, and the need to fix
1698 this bug is the main reason that 0.7.1 was released so
1700 ** DEFGENERIC is now choosier about the methods it redefines, so that
1701 reLOADing a previously-LOADed file containing DEFGENERICs does
1702 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1703 cycle now works as it should. (thanks to Alexey Dejneka)
1704 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1705 fixed by Christophe Rhodes. (He actually submitted this patch
1706 months ago, and I delayed until after 0.7.0.)
1707 ** Bug 111 (internal compiler confusion about runtime checks on
1708 FUNCTION types) was fixed by Alexey Dejneka.
1709 * Some internal cleanups (getting rid of variables which aren't
1710 needed now that the byte interpreter is gone) caused the fasl
1711 file format number to change again.
1713 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1714 * major incompatible change: The default fasl file extension, i.e. the
1715 default extension for files produced by COMPILE-FILE, has changed
1716 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1718 ** There are many changes in the implementation of the compiler.
1719 SBCL is now essentially a compiler-only implementation of ANSI
1720 Common Lisp. EVAL still "interprets" a few special cases, but
1721 almost all the interesting cases are handled by creating
1722 a LAMBDA expression, calling COMPILE on it, then calling
1723 FUNCALL on the result.
1724 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1725 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1726 Since the code is newer, there might still be some new bugs
1727 (though not as many as before Martin Atzmueller's fixes:-). But
1728 the new code is substantially simpler and clearer, and hopefully
1729 any remaining bugs will be simpler, less fundamental, and more
1730 fixable then the bugs in the old code.
1731 ** The revised compiler is still a little unsteady on its feet.
1733 *** The debugging information it produces (particularly the names
1734 of FUNCTION objects) is sometimes much less useful than what
1735 the old compiler produced.
1736 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1737 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1738 within a MACROLET) has been temporarily weakened.
1739 ** There are new compiler optimizations for various functions:
1740 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1741 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1742 *** the math functions TRUNCATE, FLOOR, and CEILING
1743 *** the function-of-all-trades COERCE
1744 Mostly these should be transparent, but there's one
1745 potentially-annoying problem (bug 117): when the compiler
1746 inline-expands a function and does type analysis on the result,
1747 it can create control paths which have type mismatches, and
1748 when it can't prove that those control paths aren't taken,
1749 it will issue WARNINGs about the type mismatches. This is
1750 a particular problem in practice for the new sequence functions.
1751 It's not clear how this should be fixed, and for now, a
1752 workaround is given in the entry for 117 in the BUGS file.
1753 ** (Because of the interaction between the two previous items --
1754 occasional inlining problems and new inline expansions -- some
1755 of the new sequence function optimizations won't really kick in
1756 completely until debugging information, and then inlining, are
1757 straightened out in some future version.)
1758 * minor incompatible changes:
1759 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1760 DIRECTORY no longer implicitly promotes NIL slots of its
1761 pathname argument to :WILD. In particular, when you ask for the
1762 contents of a directory (which you used to be able to do without
1763 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1764 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1765 ** changes in behavior that ANSI explicitly defines to be
1766 implementation dependent:
1767 *** The new compiler-only implementation still conforms with ANSI,
1768 but acts a little different than before. Besides the obvious
1769 changes in performance tradeoffs (that the cost per form passed
1770 to EVAL has gone up, and the cost per form executed by EVAL
1771 has gone down), the behavior of the system changes a little
1772 because there are no longer any interpreted function objects.
1773 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1774 e.g. doing COMPILE on the output of interactive DEFUN is
1776 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1778 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1779 has changed from (:CL) to NIL.
1780 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1781 e.g. #\Newline and #\Tab, don't have names specified in the
1782 ANSI Common Lisp standard, is now based on their ASCII symbolic
1783 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1784 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1785 are no longer used for output.
1786 ** changes in internal implementation constants:
1787 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1788 4 million. (If your application spends a lot of time GCing and
1789 you have a lot of RAM, you might want to experiment with
1790 increasing it even more.)
1791 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1792 However, almost all old code should still continue to work without
1793 immediate update, as SB-C-CALL is now a (deprecated) nickname
1795 ** Old operator names in the style DEF-FOO are now deprecated in
1796 favor of new corresponding names DEFINE-FOO, for consistency with
1797 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1798 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1799 internal symbols, but a few supported extensions like
1800 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1801 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1802 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1803 etc. as you get deeper into recursive calls to the debugger
1804 command loop, instead of the old "5]", "5]]", "5]]]"
1805 sequence. (I was motivated to do this when squabbles between
1806 ILISP and SBCL left me very deeply nested in the debugger. In the
1807 short term, this change will probably provoke more ILISP/SBCL
1808 squabbles, but hopefully it will be an improvement in the long run.)
1809 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1810 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1811 function, since declaiming return types involving aliens is
1812 (1) annoyingly messy to do by hand and (2) vital to efficient
1813 compilation of code which calls such functions.
1814 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1815 longer reexported by the SB-EXT package. They're solely useful
1816 for alien code, so it seems more logical that you should get
1817 them from the SB-ALIEN package, not in SB-EXT.
1818 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1819 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1820 features. Instead, the code that they used to control is always
1821 built into the system.
1822 * many other bug fixes
1823 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1824 advantage of the new EVAL-WHEN stuff and to clean them up in
1825 general, and they are now more ANSI-compliant in a number of
1826 ways. Martin Atzmueller is responsible for a lot of this.
1827 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1829 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1830 *** correct ERROR type for various file operations
1831 *** some fixes for Lisp streams
1832 *** DEFMETHOD syntax checking
1833 *** changing old weird representation of debug information as
1834 strings (which, among their other deficiencies, don't transform
1835 correctly when you rename packages, and don't change their
1836 print representation when you change things like *PACKAGE*
1837 and *PRINT-LENGTH*) to symbols and lists of symbols
1838 He also made several improvements and fixed several bugs in DESCRIBE.
1839 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1841 *** misbehavior of WRITE-STRING/WRITE-LINE
1842 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1843 and several other LOOP problems as well
1844 *** DIRECTORY when similar filenames are present
1845 *** DEFGENERIC with :METHOD options
1846 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1847 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1848 *** argument ordering in FIND with :TEST option
1849 *** mishandled package designator argument in APROPOS-LIST
1850 *** various problems in the backquote readmacro
1851 *** a bug in APROPOS
1852 *** probably some others that I'm not describing very well here,
1853 since the CVS log documents them by reference to sbcl-devel
1854 messages, and the SourceForge archives aren't working well.:-(
1855 ** Dan Barlow improved the Alpha port (and is making progress on the
1856 PPC port, for those of you who think different).
1857 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
1858 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
1859 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
1860 continues to work on the SPARC port (for those of you in a position
1861 to look down upon our little PC-compatible boxes from a great height).
1862 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
1863 and uses that copy, rather than the current dynamic value, when
1864 it's trying to decide whether to truncate output. Thus e.g.
1865 (let ((*print-lines* 50))
1866 (pprint-logical-block (stream nil)
1868 (let ((*print-lines* 8))
1869 (print (aref possiblybigthings i) stream)))))
1870 should now truncate the logical block only at 50 lines, instead of
1871 often truncating it at 8 lines, as it did before.
1872 * The doc/cmucl/ directory, containing old CMU CL documentation
1873 from the time of the fork, is no longer part of the base system.
1874 SourceForge has shut down its anonymous FTP service, and with it
1875 my original plan for distributing the old CMU CL documentation
1876 there. For now, if you need these files you can download an old
1877 SBCL source release and extract them from it.
1878 * The fasl file version number changed again, for dozens of reasons,
1879 some of which are apparent above.
1881 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
1882 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
1883 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
1884 patch, so that DISASSEMBLE output is much nicer.
1885 * The code in the SB-PROFILE package now seems reasonably stable.
1886 I still haven't decided what the final interface should look like
1887 (I'd like PROFILE to interact cleanly with TRACE, since both
1888 facilities use function encapsulation) but if you have a need
1889 for profiling now, you can probably use it successfully with
1890 the current CMU-CL-style interface.
1891 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
1892 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
1893 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
1894 link now returns the dangling link itself, and for similar
1895 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
1896 link itself. (In these cases the old code signalled an error and
1897 looped endlessly, respectively.) Thus, DIRECTORY now works even
1898 in the presence of dangling and cyclic symbolic links.
1899 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
1900 is now a supported extension again, since the consensus on
1901 sbcl-devel was that it can be useful for ordinary development
1902 work, not just for debugging SBCL itself.
1903 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
1904 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
1905 that currently-defined functions might be redefined later with
1906 different return types.
1907 * Hash tables can be printed readably, as inspired by CMU CL code
1908 of Eric Marsden and SBCL code of Martin Atzmueller.
1909 * better error handling in CLOS method combination, thanks to
1910 Martin Atzmueller porting Pierre Mai's CMU CL patches
1911 * more overflow fixes for >16Mbyte I/O buffers
1912 * A bug in READ has been fixed, so that now a single Ctrl-D
1913 character suffices to cause end-of-file on character streams.
1914 In particular, now you only need one Ctrl-D at the command
1915 line (not two) to exit SBCL.
1916 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
1918 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
1919 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
1920 does the right thing.
1921 * The implementation of some type tests, especially for CONDITION
1922 types, is now tidier and maybe faster, due to CMU CL code
1923 originally by Douglas Crosher, ported by Martin Atzmueller.
1924 * Some math functions have been fixed, and there are new
1925 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
1926 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
1927 * (There are also some new optimizers in contrib/*-extras.lisp. Those
1928 aren't built into sbcl-0.6.13, but are a sneak preview of what's
1929 likely to be built into sbcl-0.7.0.)
1930 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
1931 CL, ported to SBCL by Martin Atzmueller)
1932 * DESCRIBE now gives more information in some cases. (Pierre Mai's
1933 patch to CMU CL, ported to SBCL by Martin Atzmueller)
1934 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
1935 * There's a new slam.sh hack to shorten the edit/compile/debug
1936 cycle for low-level changes to SBCL itself, and a new
1937 :SB-AFTER-XC-CORE target feature to control the generation of
1938 the after-xc.core file needed by slam.sh.
1939 * minor incompatible change: The ENTRY-POINTS &KEY argument to
1940 COMPILE-FILE is no longer supported, so that now every function
1941 gets an entry point, so that block compilation looks a little
1942 more like the plain vanilla ANSI section 3.2.2.3 scheme.
1943 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
1944 returns the number of bytes consed since the system started,
1945 rather than the number consed since the first time the function
1946 was called. (The new definition parallels ANSI functions like
1947 CL:GET-INTERNAL-RUN-TIME.)
1948 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
1949 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
1950 supported. Now DIRECTORY always does the abstract Common-Lisp-y
1951 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
1952 * Fasl file version numbers are now independent of the target CPU,
1953 since historically most system changes which required version
1954 number changes have affected all CPUs equally. Similarly,
1955 the byte fasl file version is now equal to the ordinary
1958 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
1959 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
1960 is no longer recognized. I apologize for this, because it was
1961 listed in SB-EXT as a supported extension, but I found that
1962 its existing behavior was poorly specified, as well as incorrectly
1963 specified, and it looked like too much of a mess to straighten it
1964 out. I have enough on my hands trying to get ANSI stuff to work..
1965 * many patches ported from CMU CL by Martin Atzmueller, with
1966 half a dozen bug fixes in pretty-printing and the debugger, and
1967 half a dozen others elsewhere
1968 * fixed bug 13: Floating point infinities are now supported again.
1969 They might still be a little bit flaky, but thanks to bug reports
1970 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
1971 as flaky as they were.
1972 * The --noprogrammer command line option is now supported. (Its
1973 behavior is slightly different in detail from what the old man
1974 page claimed it would do, but it's still appropriate under the
1975 same circumstances that the man page talks about.)
1976 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
1977 are now supported, and enabled by default. Thus, the compiler can
1978 handle many floating point and complex operations much less
1979 inefficiently. (Thus e.g. you can implement a complex FFT
1981 * The compiler now detects type mismatches between DECLAIM FTYPE
1982 and DEFUN better, and implements CHECK-TYPE more correctly, and
1983 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
1984 thanks to patches from Martin Atzmueller.
1985 * various fixes to make the cross-compiler more portable to
1986 ANSI-conforming-but-different cross-compilation hosts (notably
1987 Lispworks for Windows, following bug reports from Arthur Lemmens)
1988 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
1989 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
1990 fixed thanks to Pierre Mai's CMU CL patches.
1991 * fixes to keep the system from overflowing internal counters when
1992 it tries to use i/o buffers larger than 16M bytes
1993 * fixed bug 45a: Various internal functions required to support
1994 complex special functions have been merged from CMU CL sources.
1995 (When I was first setting up SBCL, I misunderstood a compile-time
1996 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
1997 * improved support for type intersection and union, fixing bug 12
1998 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
1999 more obscure bugs as well
2000 * some steps toward byte-compiling non-performance-critical
2001 parts of the system, courtesy of patches from Martin Atzmueller
2002 * Christophe Rhodes has made some debian packages of sbcl at
2003 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2004 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2005 stable, but are nonetheless usable. When he's ready, I'd be happy
2006 to add them to the SourceForge "File Releases" section. (And if
2007 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2008 * new fasl file format version number (because of changes in
2009 internal representation of (OR ..) types to accommodate the new
2010 support for (AND ..) types, among other things)
2012 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2013 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2015 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2017 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2018 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2019 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2020 * DESCRIBE now works on structure objects again.
2021 * Most function call argument type mismatches are now handled as
2022 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2023 know whether the function will be redefined before the call is
2024 executed. (The compiler could flag local calls with full WARNINGs,
2025 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2026 it doesn't keep track of enough information to know whether calls
2027 are local in this sense.)
2028 * Compiler output is now more verbose, with messages truncated
2029 later than before. (There should be some supported way for users
2030 to override the default verbosity, but I haven't decided how to
2031 provide it yet, so this behavior is still controlled by the internal
2032 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2033 src/compiler/ir1util.lisp.)
2034 * Fasl file format version numbers have increased again, because
2035 support for the Gray streams extension changes the layout of the
2036 system's STREAM objects.
2037 * The Gray subclassable streams extension now works, thanks to a
2038 patch from Martin Atzmueller.
2039 * The full LOAD-FOREIGN extension (not just the primitive
2040 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2041 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2042 but like most other programs, it defaults to copying the Unix
2043 environment from the original process instead of starting the
2044 new process in an empty environment.
2045 * Extensions which manipulate the Unix environment now support
2046 an :ENVIRONMENT keyword option which doesn't smash case or
2047 do other bad things. The CMU-CL-style :ENV option is retained
2048 for porting convenience.
2049 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2050 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2052 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2054 * A patch from Martin Atzmueller seems to have solved the SIGINT
2055 problem, and as far as we know, signal-handling now works cleanly.
2056 (If you find any new bugs, please report them!)
2057 * The system no longer defaults Lisp source file names to types
2058 ".l", ".cl", or ".lsp", but only to ".lisp".
2059 * The compiler no longer uses special default file extensions for
2060 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2061 seems to expect a single default extension for all compiled code,
2062 and there's no compelling reason to try to stretch the standard
2063 to allow two different extensions.) Instead, byte-compiled files
2064 default to the same extension as native-compiled files.
2065 * Fasl file format version numbers have increased again, because
2066 a rearrangement of internal implementation packages made some
2067 dumped symbols in old fasl files unreadable in new cores.
2068 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2069 many fewer weird special cases.
2070 * Bug #17 (differing COMPILE-FILE behavior between logical and
2071 physical pathnames) has been fixed, and some related misbehavior too,
2072 thanks to a patch from Martin Atzmueller.
2073 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2074 by Tim Moore, ported to SBCL by Martin Atzmueller.
2075 * Martin Atzmueller fixed several filesystem-related problems,
2076 including bug #36, in part by porting CMU CL patches, which were
2077 written in part by Paul Werkowski.
2078 * More compiler warnings in src/runtime/ are gone, thanks to
2079 more patches from Martin Atzmueller.
2080 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2083 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2085 * DESCRIBE now works on CONDITION objects.
2086 * The debugger now handles errors which arise when trying to print
2087 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2089 * The build system now uses an additional file, customize-target-features.lisp,
2090 to allow local modifications to the target *FEATURES* list. (The point of
2091 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2092 debugging features enabled, without having to worry about propagating it
2093 into everyone's system when I do a "cvs update".) When no
2094 customize-target-features.lisp file exists, the target *FEATURES* list
2095 should be constructed the same way as before.
2096 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2097 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2098 warn about behavior which is undefined under the ANSI spec. Note: This
2099 is specified by ANSI, but it's not very popular with programmers.
2100 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2101 macro in the SBCL sources for an example of a workaround which you
2102 might use to make portable ANSI-standard code which does what you want.
2103 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2104 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2105 when it's used as a non-toplevel form. (This is required in order
2106 to implement the DEFCONSTANT-EQX macro.)
2107 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2108 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2109 funny things, due to bugs in the implementation of EVAL-WHEN
2110 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2111 non-toplevel DEFCONSTANTs are uncommon.)
2112 * The core file version number and fasl file version number have been
2113 incremented, because the old noncompliant DEFCONSTANT behavior involved
2114 calling functions which no longer exist, and because I also took the
2115 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2116 * fixed bug 1 (error handling before read-eval-print loop starts), and
2117 redid debugger restarts and related debugger commands somewhat while
2119 ** The QUIT debugger command is gone, since it did something
2120 rather different than the SB-EXT:QUIT command, and since it never
2121 worked properly outside the main toplevel read/eval/print loop.
2122 Invoking the new TOPLEVEL restart provides the same functionality.
2123 ** The GO debugger command is also gone, since you can just invoke
2124 the CONTINUE restart directly instead.
2125 ** The TOP debugger command is also gone, since it's redundant with the
2126 FRAME 0 command, and since it interfered with abbreviations for the
2128 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2130 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2131 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2132 some porting work by Martin Atzmueller)
2133 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2134 with a boatload of patches which he ported to SBCL. Now that those
2136 ** The system tries to make sure that its low-priority messages
2137 are prefixed by semicolons, to help people who like to use
2138 syntax highlighting in their ILISP buffer. (This patch
2139 was originally due to Raymond Toy.)
2140 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2141 patches originally written by Raymond Toy.
2142 ** The compiler understands coercion between single-value and
2143 multiple-VALUES type expressions better, getting rid of some very
2144 weird behavior, thanks to patches originally by Robert MacLachlan
2145 and Douglas Crosher.
2146 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2147 lambda lists, thanks to a patch originally by Pierre Mai.
2148 ** The system no longer bogusly warns about "abbreviated type
2150 ** The compiler gets less confused by inlining and RETURN-FROM,
2151 thanks to some patches originally by Tim Moore.
2152 ** The system no longer hangs when dumping circular lists to fasl
2153 files, thanks to a patch originally from Douglas Crosher.
2154 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2155 undefined function error.
2156 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2157 in structure slot initforms) for now, documented workaround instead:-|
2158 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2159 * fixed bug 5: added stubs for various Gray stream functions called
2160 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2161 installed, at least appropriate type errors are generated
2162 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2163 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2164 consistently in DEFMETHOD forms.
2165 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2166 it doesn't seem to affect SBCL after all
2167 * The C runtime system now builds with better optimization and many
2168 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2170 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2172 * The system is now under CVS at SourceForge (instead of the
2173 CVS repository on my home machine).
2174 * The new signal handling code has been tweaked to treat register
2175 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2176 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2177 downstream, e.g. in debug-int.lisp, has implicit dependencies
2178 on the unsignedness of integer representation of machine words,
2179 and that caused the system to bomb out with infinite regress
2180 when trying to recover from type errors involving signed values,
2181 e.g. (BUTLAST '(1 2 3) -1).)
2182 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2183 (This was one of the bugs Peter Van Eynde reported back in July.)
2184 * The system now uses code inspired by Colin Walters' O(N)
2185 implementation of MAP (from the cmucl-imp@cons.org mailing
2186 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2187 inline the MAP operation, and there is more than one
2188 sequence argument to the MAP call (so that it can't just
2189 do ETYPECASE once and for all based on the type of the
2190 single sequence argument). (The old non-inline implementation
2191 of the general M-argument sequence-of-length-N case required
2192 O(M*N*N) time when any of the sequence arguments were LISTs.)
2193 * The QUIT :UNIX-CODE keyword argument has been renamed to
2194 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2195 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2197 * Raymond Wiker's patches to port dynamic loading from Linux to
2198 FreeBSD have been added.
2199 * The BUGS file is now more nearly up to date, thanks in large part
2200 to Martin Atzmueller's review of it.
2201 * The debugger now flushes standard output streams before it begins
2202 its output ("debugger invoked" and so forth).
2203 * The core version number and fasl file version number have both
2204 been incremented, because of incompatible changes in the layout
2206 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2207 used to not be called for a saved Lisp image.)
2208 * Martin Atzmueller's version of a patch to fix a compiler crash,
2209 as posted on sbcl-devel 13 September 2000, has been installed.
2210 * Instead of installing Martin Atzmueller's patch for the
2211 compiler transform for SUBSEQ, I deleted the compiler transform,
2212 and transforms for some similar consing operations.
2213 * A bug in signal handling which kept TRACE from working on OpenBSD
2215 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2216 away by constant folding
2217 * The system now defines its address space constants in one place
2218 (in the Lisp sources), and propagates them automatically elsewhere
2219 (through GENESIS and the sbcl.h file). Therefore, patching the
2220 address map is less unnecessarily tedious and error-prone. The
2221 Lisp names of address space constants have also been systematized.
2222 * CVS tags like dollar-Header-dollar have been removed from
2223 the sources, because they have never saved me trouble and
2224 they've been source of trouble working with patches and other
2225 diff-related operations.
2226 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2227 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2229 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2231 * The system has been ported to OpenBSD.
2232 * The system now compiles with a simple "sh make.sh" on the systems
2233 that it's supported on. I.e., now you no longer need to tweak
2234 text in the target-features.lisp-expr and symlinks in src/runtime/
2235 by hand, the make.sh takes care of it for you.
2236 * The system is no longer so grossly inefficient when compiling code
2237 involving vectors implemented as general (not simple) vectors (VECTOR T),
2238 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2239 which dares to use the various sequence functions on non-simple
2240 vectors, takes less of a performance hit.
2241 * There is now a primitive type predicate VECTOR-T-P
2242 to test for the (VECTOR T) type, so that e.g.
2243 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2244 can now be compiled with some semblance of efficiency. (The old code
2245 turned the type declaration into a full call to %TYPEP at runtime!)
2246 * AREF on (VECTOR T) is still not fast, since it's still compiled
2247 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2248 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2249 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2250 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2251 without making full calls to SUBTYPEP at runtime.
2252 (Some analogous efficiency issues for non-simple vectors specialized to
2253 element types other than T, or for non-simple multidimensional arrays,
2254 have not been addressed. They could almost certainly be handled the
2255 same way if anyone is motivated to do so.)
2256 * The changes in array handling break binary compatibility, so
2257 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2258 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2259 instead of (VALUES T T).
2260 * By following the instructions that Dan Barlow posted to sbcl-devel
2261 on 2 July 2000, I was able to enable primitive dynamic object
2262 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2263 functionality is not implemented (since it calls ld to resolve
2264 library references automatically, requiring RUN-PROGRAM for its
2265 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2266 try to resolve library references) is now supported.
2267 * The system now flushes the standard output streams when it terminates,
2268 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2269 them at several other probably-convenient times, e.g. in each pass of
2270 the toplevel read-eval-print loop, and after evaluating a form given
2271 as an "--eval" command-line option. (These changes were motivated by a
2272 discussion of stream flushing issues on cmucl-imp in August 2000.)
2273 * The source transform for TYPEP of array types no longer assumes
2274 that an array whose element type is a not-yet-defined type
2275 is implemented as an array of T, but instead punts, so that the
2276 type will be interpreted at runtime.
2277 * There is now some support for cross-compiling in make.sh: each of
2278 the phases of make.sh has its own script. (This should be transparent
2279 to people doing ordinary, non-cross-compile builds.)
2280 * Since my laptop doesn't have hundreds of megabytes of memory like
2281 my desktop machine, I became more motivated to do some items on
2282 my to-do list in order to reduce the size of the system a little:
2283 ** Arrange for various needed-only-at-cold-init things to be
2284 uninterned after cold init. To support this, those things have
2285 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2286 symbols with such names are now uninterned after cold init).
2287 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2288 fasl files for cold load.
2289 ** Remove the old compiler structure pooling code (which used to
2290 be conditional on the target feature :SB-ALLOC) completely.
2291 ** Redo the representation of some data in cold init to be more compact.
2292 (I also looked into supporting byte compiled code at bootstrap time,
2293 which would probably reduce the size of the system a lot, but that
2294 looked too complicated, so I punted for now.)
2295 * The maximum signal nesting depth in the src/runtime/ support code has
2296 been reduced from 4096 to 256. (I don't know any reason for the very
2297 large old value. If the new smaller value turns out to break something,
2298 I'll probably just bump it back up.)
2299 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2301 * Many, many bugs reported by Peter Van Eynde have been added to
2302 the BUGS list; some have even been fixed.
2303 * While enabling dynamic object file loading, I tried to make the
2304 code easier to understand, renaming various functions and variables
2305 with less ambiguous names, and changing some function calling
2306 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2307 * While trying to figure out how to do the OpenBSD port, I tried to
2308 clean up some of the code in src/runtime/. In particular, I dropped
2309 support for non-POSIX signal handling, added various comments,
2310 tweaked the code to reduce the number of compilation warnings, and
2311 renamed some files to increase consistency.
2312 * To support the new automatic configuration functionality in make.sh,
2313 the source file target-features.lisp-expr has been replaced with the
2314 source file base-target-features.lisp-expr and the machine-generated
2315 file local-target-features.lisp-expr.
2316 * fixed a stupid quoting error in make.sh so that using CMU CL
2317 "lisp -batch" as cross-compilation host works again
2319 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2321 * DESCRIBE no longer tries to call itself recursively to describe
2322 bound/fbound values, so that it no longer fails on symbols which are
2323 bound to themselves (like keywords, T, and NIL).
2324 * DESCRIBE now works on generic functions.
2325 * The printer now prints less-screwed-up representations of closures
2326 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2327 * A private symbol is used instead of the :EMPTY keyword previously
2328 used to mark empty slots in hash tables. Thus
2329 (DEFVAR *HT* (MAKE-HASH-TABLE))
2330 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2331 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2332 now does what ANSI says that it should. (You can still get
2333 similar noncompliant behavior if bang on the hash table
2334 implementation with all the symbols you get back from
2335 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2336 This breaks binary compatibility, since tests for equality to
2337 :EMPTY are wired into things like the macroexpansion of
2338 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2340 * There's now a minimal placeholder implementation for CL:STEP,
2341 as required by ANSI.
2342 * An obscure bug in the interaction of the normal compiler, the byte
2343 compiler, inlining, and structure predicates has been patched
2344 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2345 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2347 * Missing ordinary arguments in a macro call are now detected even
2348 when the macro lambda list contains &KEY or &REST.
2349 * The debugger no longer complains about encountering the top of the
2350 stack when you type "FRAME 0" to explicitly instruct it to go to
2351 the top of the stack. And it now prints the frame you request even
2352 if it's the current frame (instead of saying "You are here.").
2353 * As specified by ANSI, the system now always prints keywords
2354 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2356 * The default initial SIZE of HASH-TABLEs is now smaller.
2357 * Type information from CLOS class dispatch is now propagated
2358 into DEFMETHOD bodies, so that e.g.
2359 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2361 is now basically equivalent to
2362 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2363 (DECLARE (TYPE SINGLE-FLOAT X))
2365 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2366 operation, without having to do run-time type dispatch.
2367 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2368 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2369 * Since I don't seem to be making big file reorganizations very often
2370 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2371 is overflowing my ability to conveniently back them up), I've finally
2372 checked the system into CVS. (The CVS repository is on my home system,
2373 not at SourceForge -- putting it on SourceForge might come later.)
2374 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2375 high-level GC-NOTIFY-FOO functions send their output. (There's
2376 still very little control of where low-level verbose GC functions
2377 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2378 less than it used to -- the GC-NOTIFY-FOO functions are now under
2379 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2380 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2381 in only one place in the source code, and propagates it automatically
2382 everywhere that it's needed. Thus e.g. when I bump the version from
2383 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2384 * The C source files now include boilerplate legalese and documentation
2385 at the head of each file (just as the Lisp source files already did).
2386 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2387 to his page will be replaced with a link to his new CLiki service.
2389 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2391 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2392 * The build process now looks for GNU make under the default name "gmake",
2393 instead of "make" as it used to. If GNU make is not available as "gmake"
2394 on your system, you can change this default behavior by setting the
2395 GNUMAKE environment variable.
2396 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2397 can build without error under CMU CL.
2399 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2401 * There is now a partial SBCL user manual (with some new text and some
2402 text cribbed from the CMU CL manual).
2403 * The beginnings of a profiler have been added (starting with the
2404 CMU CL profiler and simplifying and cleaning up). Eventually the
2405 main interface should be through the TRACE macro, but for now,
2406 it's still accessed through vaguely CMU-CL-style functions and macros
2407 exported from the package SB-PROFILE.
2408 * Some problems left over from porting CMU CL to the new
2409 cross-compilation bootstrap process have been cleaned up:
2410 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2411 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2412 part when getting rid of the compiler *BACKEND* stuff.)
2413 ** The value of *NULL-TYPE* was screwed up, because it was
2414 being initialized before the type system knew the final
2415 definition of the 'NULL type. This screwed up several key
2416 optimizations in the compiler, causing inefficiency in all sorts
2417 of places. (I found it because I wanted to understand why
2418 GET-INTERNAL-RUN-TIME was consing.)
2419 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2420 PROCLAIM FTYPE information. Unfortunately this broke binary
2421 compatibility again, since now the forms output by DEFGENERIC
2422 to refer to functions which didn't exist in 0.6.3.
2423 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2424 can use the new (as of 0.6.3) transform for SOME into MAP into
2426 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2427 values to (INTEGER 0 1000000), so that the time code will no longer
2428 occasionally get blown up by Linux returning 1000000 microseconds
2429 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2430 its output conform to the ANSI spec. (Alas, this makes its output
2431 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2432 * A full call to MAP NIL with a single sequence argument no longer conses.
2433 * fixes to problems pointed out by Martin Atzmueller:
2434 * The manual page no longer talks about multiprocessing as though
2435 it were currently supported.
2436 * The ILISP support patches have been removed from the distribution,
2437 because as of version 5.10.1, ILISP now supports SBCL without us
2438 having to maintain patches.
2439 * added a modified version of Raymond Toy's recent CMU CL patch for
2440 EQUALP comparison of HASH-TABLE
2442 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2444 * The system still can't cross-compile itself with
2445 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2446 entails), but at least it can compile more of itself that way
2447 than it used to be able to, and various buglets which were uncovered
2448 by trying to cross-compile itself that way have now been fixed.
2449 * This release breaks binary compatibility again. This time
2450 at least I've incremented the FASL file format version to 2, so that the
2451 problem can be detected reliably instead of just causing weird errors.
2452 * various new style warnings:
2453 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2454 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2455 ** using the lexical binding of a variable named in the *FOO* style
2456 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2457 as specified by ANSI.
2458 * *RANDOM-STATE* is no longer automatically initialized from
2459 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2460 default behavior of the system is to repeat its behavior every time
2461 it's run. If you'd like to change this behavior, you can always
2462 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2463 old convention there was no comparably easy way to get the system to
2464 repeat its behavior every time it was run.
2465 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2466 FTYPE declarations has been removed, in favor of their ANSI
2467 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2468 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2469 the types of their sequence arguments can be determined at compile time.
2470 This is done through a new open code expansion for MAP which eliminates
2471 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2472 argument types can be determined at compile time.
2473 * The optimizer now transforms COERCE into an identity operation when it
2474 can prove that the coerced object is already of the correct type. (This
2475 can be a win for machine generated code, including the output of other
2476 optimization transforms, such as the MAP transform above.)
2477 * Credit information has been moved from source file headers into CREDITS.
2478 * Source file headers have been made more standard.
2479 * The CASE macro now compiles without complaining even when it has
2482 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2484 * (Note that the way that the PCL macroexpansions were rewritten
2485 to accommodate the change in DEFGENERIC below breaks binary
2486 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2487 not run under sbcl-0.6.2. Once we get out of alpha releases,
2488 i.e. hit release 1.0.0, we'll probably try to maintain binary
2489 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2490 and sbcl-1.4.4. Until then, however, it might be fairly common
2491 for maintenance releases to break binary compatibility.)
2492 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2494 * The reporting of unhandled signals has been changed to print some
2495 explanatory text as well as the report form. (Previously only
2496 the report form was printed.)
2497 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2498 it defines, so that the compiler no longer issues undefined function
2499 warnings for compiled-but-not-yet-loaded generic functions.
2500 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2501 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2502 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2503 section of the ANSI spec are supported.
2504 * The "" nickname for the "KEYWORD" package has been removed.
2505 The reader still handles symbol tokens which begin with a package marker
2506 as keywords, but it doesn't expose its mechanism for doing so in the
2507 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2508 * The system now issues STYLE-WARNINGs for contradictory TYPE
2509 proclamations. (Warnings for contradictory FTYPE proclamations would
2510 be nice too, but those can't be done usefully unless the type system
2511 is made smarter about FUNCTION types.)
2512 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2513 systematized, so that "*target-*.lisp is supposed to exist only on the
2514 target and imply that there's a related file which exists on the
2515 host, and *host-*.lisp is supposed to exist only on the host and imply
2516 that there's a related file which exists on the target. This involves a
2517 lot of renaming. Hopefully the acute confusion caused by the renaming
2518 will be justified by the reduction in chronic confusion..
2519 ** runtime-type.lisp -> early-target-type.lisp
2520 ** target-type.lisp -> late-target-type.lisp
2521 ** early-host-format.lisp -> early-format.lisp
2522 ** late-host-format.lisp -> late-format.lisp
2523 ** host-error.lisp -> misc-error.lisp
2524 ** early-error.lisp -> early-target-error.lisp
2525 ** late-error.lisp -> late-target-error.lisp
2526 ** host-defboot.lisp -> early-defboot.lisp
2527 ** code/misc.lisp -> code/target-misc.lisp
2528 ** code/host-misc.lisp -> code/misc.lisp
2529 ** code/numbers.lisp -> code/target-numbers.lisp
2530 ** code/early-numbers.lisp -> numbers.lisp
2531 ** early-host-type.lisp -> early-type.lisp
2532 ** late-host-type.lisp -> late-type.lisp
2533 ** host-typep.lisp -> typep.lisp
2534 ** load.lisp -> target-load.lisp
2535 ** host-load.lisp -> load.lisp
2536 ** host-disassem.lisp -> disassem.lisp
2537 ** host-insts.lisp -> insts.lisp
2538 ** byte-comp.lisp -> target-byte-comp.lisp
2539 ** host-byte-comp.lisp -> byte-comp.lisp
2540 ** host-signal.lisp -> signal.lisp
2541 ** host-defstruct.lisp -> defstruct.lisp
2542 ** late-target-type.lisp -> deftypes-for-target.lisp
2543 Furthermore, several other previously target-only files foo.lisp (e.g.
2544 hash-table.lisp and random.lisp) have been split into a target-and-host
2545 foo.lisp file and a target-only target-foo.lisp file, with their key type
2546 definitions in the target-and-host part, so that the cross-compiler will
2547 know more about target types.
2548 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2549 gone away. In their place are various *BACKEND-FOO* variables
2550 corresponding to the slots of the old structure.
2551 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2552 into the target SBCL has been fixed.
2553 * The chill.lisp system for loading cold code into a running SBCL
2555 * Support for the CMU CL "scavenger hook" extension has been removed.
2556 (It was undocumented and unused in the CMU CL sources that SBCL was
2557 derived from, and stale in sbcl-0.6.1.)
2558 * Various errors in the cross-compiler type system were detected
2559 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2560 (enabling various consistency checks). Many of them were fixed,
2561 but some hard problems remain, so the compiler is back to
2562 running without *TYPE-SYSTEM-INITIALIZED* for now.
2563 * As part of the cross-compiler type system cleanup, I implemented
2564 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2565 * I have started adding UNCROSS calls throughout the type system
2566 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2567 on cross-compiler input files will be able to go away, and various
2569 * CONSTANTP now returns true for quoted forms (as explicitly required
2572 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2574 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2575 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2576 (MAKE-LIST -1). In the longer term, it ought to become true
2577 that declarations are assertions even at SAFETY 1. For now, it's not
2578 quite true even at SAFETY 3, but it's at least more nearly true..
2579 (Note that this change seems to increases the size of the system by
2580 O(5%) and to decrease the speed of the compiler by 20% or more.)
2581 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2582 for the problem of printing dozens of lines of distracting information
2583 about low-level system machinery as part of the top stack frame
2584 on entry to the debugger when an undefined function was called.
2585 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2586 is not reset to COMMON-LISP-USER.
2587 * Compilation of stuff related to dyncount.lisp has been made conditional
2588 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2589 smaller. The various dyncount-related symbols have been moved into
2590 a new "SB-DYNCOUNT" package.
2591 * tty-inspect.lisp has been renamed to inspect.lisp.
2592 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2593 feature has gone away. (When we eventually port to other flavors of
2594 libc and/or Unix, we'll try to make the differences between flavors
2595 invisible at the user level.)
2596 * Various other *FEATURES* tags, and/or their associated conditionals,
2597 have been removed if obsolescent, or given better documentation, or
2598 sometimes given more-mnemonic names.
2600 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2602 * tidied up "make.sh" script
2603 * tidied up system directory structure
2604 * better "clean.sh" behavior
2605 * added doc/FOR-CMUCL-DEVELOPERS
2606 * many many small tweaks to output format, e.g. removing possibly-confusing
2607 trailing #\. character in DESCRIBE-INSTANCE
2608 * (EQUALP #\A 'A) no longer signals an error.
2609 * new hashing code, including EQUALP hashing
2610 * tidied up Lisp initialization and toplevel
2611 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2612 * command line argument processing
2613 * added POSIX-GETENV function to deal with Unix-ish environment variables
2614 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2615 terminating SBCL on EOF
2616 * non-verbose GC by default
2617 * There is no more "sbcl" shell script; the sbcl file is now the C
2618 runtime executable (just like CMU CL).
2619 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2621 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2622 symbol and package information as Lisp native symbol and package objects
2623 instead of strings naming symbols and strings naming packages. This way,
2624 whenever packages are renamed (as in warm init), debug information is
2625 transformed along with everything else.
2626 * tweaked the optimization policy declarations which control the building
2627 of SBCL itself. Now, among other things, the system no longer saves
2628 source location debugging information. (This helps two problems at once
2629 by reducing SBCL size and by keeping SBCL from trying to look for its
2630 sources -- which may not exist -- when reporting errors.)
2631 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2632 debugging and testing purposes
2633 * cleaned up printing, making the printer call PRINT-OBJECT for
2634 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2635 methods, giving nearly-ANSI behavior
2636 * converted almost all special variables to use *FOO* naming convention
2637 * deleted PARSE-TIME functionality, since it can be done portably
2638 * moved some files out of cold init into warm init
2639 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2641 * regularized formatting of source files
2642 * added an install.sh script
2643 * fixed ridiculous memory usage of cross-compiler by making
2644 compiler/alloc.lisp not try to do pooling unless it can hook
2645 itself into the GC of the cross-compilation host. Now the system
2646 builds nicely on my old laptop.
2647 * added :SB-ALLOC in target-features.lisp-expr
2648 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2649 was not implemented)
2650 * re-did condition handling and note reporting in the compiler. Notes
2651 are no longer handled by signalling conditions. Style warnings
2652 and warnings are handled more correctly and reported in such a way
2653 that it's easy to find one or the other in your output (so that you
2654 can e.g. figure out which of many problems caused COMPILE-FILE to
2656 * changed the severity of several compiler warnings from full WARNING
2657 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2658 compiler note reporting so that it doesn't use the condition system
2659 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2660 COMPILE-FILE command)
2661 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2662 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2663 no longer supported, since their implementation was deeply intertwingled
2664 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2665 * removed bogus "support" for compiler macros named (SETF FOO), and
2666 removed the compiler macro for SETF INFO (but only after making a fool
2667 of myself on the cmucl-imp mailing list by posting a bogus patch for
2668 DEFINE-COMPILER-MACRO..)
2669 * Compiled files containing forms which have side effects on the Lisp
2670 reader (such as DEFPACKAGE forms) are now handled more correctly.
2671 (Compiler queuing of top level lambdas has been suppressed by setting
2672 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2673 * deleted various currently-unused source files, e.g. gengc.lisp. They
2674 may be added back at some point e.g. when porting to other architectures,
2675 but until they are it's distracting to distribute them and to try to
2677 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2678 there were so many of them they're just distractions, and UNCROSS is
2679 known to be able to handle the current sources
2680 * moved PROFILE functionality into TRACE, so that it will be clear
2681 how the wrapping and unwrapping of functions when you profile them
2682 interacts with the wrapping and unwrapping of functions when you
2683 trace them. (Actually, the functionality isn't there yet, but at least
2684 the interface specification is there. Hopefully, the functionality will
2685 arrive with some future maintenance release.)
2686 * removed host-oops.lisp
2687 * changed signature of QUIT function to allow UNIX-CODE argument
2688 * fixed READ-SEQUENCE bug
2689 * tweaked verbose GC output so that it looks more like the progress
2690 output that ANSI specifies for functions like LOAD
2691 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2692 * added <http://sbcl.sourceforge.com> to the banner information printed by