1 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2 * new feature: on platforms where "dladdr" is available foreign
3 function names now appear in backtraces. (based on Helmut Eller's
5 * bug fix: backtraces involving undefined functions or assembly
6 routines are more informative. (thanks to Brian Downing)
7 * bug fix: mutually referent alien structures now work correctly.
8 (reported by Rick Taube)
9 * bug fix: structures defined by WITH-ALIEN can be referred to
10 within other definitions in the same WITH-ALIEN.
11 * bug fix: division operators (MOD, TRUNCATE and the like) with
12 constant zero divisors and integer dividends no longer generate
15 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
16 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
17 the manual for details; for now, package locks can be disabled by
18 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
19 if no major problems are found then it is likely that they will be
20 compiled in unconditionally.
21 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
22 unsupported operators on all platforms. To load a shared library
23 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
24 object file, link it into a shared library outside of SBCL and
25 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
27 * bug workaround: ROOM T and the GC were not working together
28 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
29 reporting facility aren't preserved. That reporting has been
30 disabled, so now until and unless someone figures out how to make it
31 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
32 * minor incompatible change: as threatened around sbcl-0.8.0, the
33 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
34 The symbols are also exported from SB-PCL for backwards
35 compatibility, but more so than before SB-PCL should be treated as
36 an implementation-internal package.
37 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
39 * the SB-SPROF contrib now works on (most) non-x86 architectures.
40 It is known as of this release not to work on the Alpha, however.
41 * fixed bug #167: errors signalled due to illegal syntax in method
42 bodies are now more legible.
43 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
44 designators and can hence be used with TYPEP.
45 * fixed bug #333: CHECK-TYPE now ensures that the type error
46 signalled, if any, has the right object to be accessed by
47 TYPE-ERROR-DATUM. (reported by Tony Martinez)
48 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
49 5.1.2.3 for multiple-value place subforms. (reported by Kalle
51 * fixed bug #334: programmatic addition of slots using specialized
52 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
53 and :ALLOCATION :CLASS slots.
54 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
55 float exponents. (rereported by Peter Seibel)
56 * fixed a bug: #\Space (and other whitespace characters) are no
57 longer considered to be macro characters in standard syntax by
59 * fixed bug: initialization of condition class metaobjects no longer
60 causes an instance of the condition to be created. (reported by
62 * fixed bug: it is now possible to have slots such that
63 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
64 description is neither :INSTANCE nor :CLASS.
65 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
66 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
67 SLOT-VALUE-USING-CLASS), and no longer causes errors with
68 non-standard SLOT-DEFINITION-ALLOCATION values.
69 * fixed bugs: various race conditions handling exiting threads.
70 CL-PPCRE's thread torture test now passes.
71 * fixed arguably-a-bug: GC time proportional to number of threads in
72 system even when most of them are idle
73 * optimization: improved performance of BIT and SBIT on bit-vectors.
74 * .fasl file incompatibility: The fasl file version number has
75 been incremented because of changes associated with package locks.
77 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
78 * minor incompatible change: the system no longer provides
79 optimization hints (conditions which are TYPEP
80 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
81 mode; these hints are still emitted when higher SPEED optimization
83 * new contrib module: a sampling profiler (profiling by statistical
84 sampling, rather than by instrumenting functions) is available as
85 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
86 * the behaviour of the standard function ED is now customizeable by
87 third parties through a hook variable: see ED's documentation
88 string for information on the protocol.
89 * the compiler no longer emits efficiency notes for (FUNCALL X)
90 when the type of X is uncertain under default optimization
92 * fixed bug 276: mutating a binding of a specialized parameter to a
93 method to something that is not TYPEP the specializer is now
95 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
96 exhibit the required behaviour on the x86 platform. (thanks to
97 Peter van Eynde, Eric Marsden and Bruno Haible)
98 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
99 even for difficult arguments. (reported by Peter Graves)
100 * fixed bug 141a: the backquote printer now descends quoted
102 * fixed another bug in backquote printing: no more destructive
103 modification of the form's list structure. (reported by Brian
105 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
106 move between its address being taken and the call to
107 interrupt_thread, fixing a crashing race condition.
108 * the SB-POSIX contrib implementation has been adjusted so that it
109 no longer exhibits ridiculously poor performance when constructing
110 instances corresponding to C structs.
112 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
113 * minor incompatible change: the sb-grovel contrib now treats C
114 structures as alien (in the sense of SB-ALIEN) objects rather than
115 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
116 has implications for memory management of client code
117 (sb-grovel-returned objects must now be manually managed) and for
118 type safety (alien objects now have full types).
119 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
120 used to control emission of compiler diagnostics, rather than the
121 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
122 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
123 quality should be considered deprecated.
124 * install.sh now installs the user manual as well
125 * (not quite a new documentable feature, but worth considering in
126 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
127 beginnings of a semantically meaningful condition hierarchy is
128 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
130 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
131 designator as the defaults argument.
132 * fixed bug: Displaced arrays whose displaced-to array has become
133 too small now cause ARRAY-DIMENSION to signal an error, providing
134 for safer bounds-checking. (reported by Bruno Haible)
135 * fixed bug: DEFCLASS slot definitions with identical :READER and
136 :WRITER names now signal a reasonable error. (reported by Thomas
138 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
140 * fixed bug: Invalid slot specification errors now print correctly.
141 (thanks to Zach Beane)
142 * fixed bug 320: Shared to local slot value transfers in class
143 redefinitions now happen corrently from superclasses as
144 well. (reported by Bruno Haible)
145 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
147 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
148 improper lists correctly. (reported by Bruno Haible)
149 * fixed bug 313: source-transform for <fun-name> was erroneously
150 applied to a call of a value of a variable with name <fun-name>.
151 (reported by Antonio Menezes Leitao)
152 * fixed bug 307: The obsolete instance protocol ensures that
153 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
154 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
155 remove dead unknown-values globs from the middle of the stack.
156 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
157 easier to resume long computations after using *BREAK-ON-SIGNALS*
158 to diagnose and fix failures. (thanks to Nikodemus Siivola)
159 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
160 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
161 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
162 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
163 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
165 * fixed a (fairly theoretical) bug in string printing: if
166 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
167 does not have array-element-type equal to the most general string
169 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
170 function name. (thanks to Nikodemus Siivola)
171 * fixed bug: compile-time format string checker failed on
172 non-closed ~{. (reported by Thomas F Burdick)
173 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
174 don't warn on legal loop code involving more than one aggregate
175 boolean. (thanks to Nikodemus Siivola)
176 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
177 in keyword position to MAKE-PACKAGE was not being reported as
179 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
180 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
182 * fixed bugs 280 and 312: the checking for multiple definitions in a
183 file is less likely to become confused by uses of inline
185 * fixed bug: the #S reader macro performs the keyword coercion
186 specified for slot names. (reported by Kalle Niemitalo)
187 * fixed bug: lambda lists may contain symbols whose names start with
188 & but are not lambda-list-keywords; their occurrence triggers a
190 * fixed bug 321: define-method-combination argument lambda lists do
191 not cause invalid code to be generated when &optional or &aux
192 variables have default values. (reported by Bruno Haible)
193 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
194 have CLOS classes; this appears also to have fixed reported
195 instability in stack exhaustion detection.
196 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
197 MUFFLE-WARNING when no associated restart is present is now
199 * optimization: rearranged the expansion of various defining macros
200 so that each expands into only one top-level form in a
201 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
203 * optimization: used a previously little-used slot in symbols to
204 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
206 * fixed some bugs revealed by Paul Dietz' test suite:
207 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
209 ** Fixed an optimization bug related to inheritance of initforms
210 from local to shared slots.
211 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
212 ** CAST optimizer forgot to flush argument derived type.
213 ** print/read consistency on floats is now orders of magnitude
214 more likely. (thanks also to Bruno Haible for a similar report
216 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
217 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
218 for objects of type REAL. Make it so.
219 ** SXHASH is specified (infelicitously) to respect similarity,
220 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
221 it so. (thanks to Markus Ziegler)
222 ** on the Alpha, the compiler succeeds in compiling functions
223 returning a known number of arguments greater than 63.
224 ** fixed handling of invalid NIL arguments in keyword position
226 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
228 ** provided a readably-printable representation for RANDOM-STATE
230 ** ensured that pathnames, where they have a namestring, always
231 print using #P"..." syntax.
233 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
234 * Support for the forthcoming 2.0 version of the NetBSD kernel
235 running on x86 hardware has been added. (thanks to Perry
236 E. Metzger most immediately, and others for their past work)
237 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
238 current release version; SBCL's OpenBSD support had been broken
239 since about the time of OpenBSD's switch to ELF binary format.)
240 * [placeholder for DX summary]
241 ** user code with &REST lists declared dynamic-extent, under high
242 speed or space and low safety and debug optimization policy.
243 * The manual has been converted to Texinfo format and the debugger
244 chapter from the cmucl manual has been added.
245 * A facility has been added to extract documentation strings from
246 sbcl and store them as Texinfo-formatted snippets for inclusion in
247 the manual (via Texinfo's @include directive)
248 * bug fix: compiler emitted division in optimized DEREF. (thanks for
249 the test case to Dave Roberts)
250 * bug fix: multidimensional simple arrays loaded from FASLs had fill
251 pointers. (reported by Sean Ross)
252 * bug fix: PROFILE output is printed nicely even for large numerical
253 values. (thanks to Zach Beane)
254 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
255 greater than 32 handle EOF correctly.
256 * bug fix: on X86 an immediate argument of the IMUL instruction is
257 correctly printed in disassembly. (thanks to Lutz Euler)
258 * bug fix: class slots in redefined classes preserve their old
259 values. (thanks to Bruno Haible and Nikodemus Siivola)
260 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
261 succeeds. (reported by Marco Baringer)
262 * fixed some bugs revealed by Paul Dietz' test suite:
263 ** READ-SEQUENCE now works on ECHO-STREAMs.
264 ** RATIONALIZE works more according to its specification. (thanks
267 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
268 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
269 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
270 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
271 should matter to you only if you rebind the printer control
272 variables and then find you want different bindings in the
273 debugger than in the ordinary execution of your program.)
274 * The runtime build system has been tweaked to support building
275 (on SPARC/SunOS) using a C compiler which invokes Sun's own
276 assembler and linker. (thanks to Nikodemus Siivola)
277 * Unbound, undefined, undeclared variables now trigger full
278 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
279 more often programmer error than deliberate exploitation of undefined
281 * optimization: the hash algorithm for strings has changed to one
282 that is less vulnerable to spurious collisions. (thanks to Juho
284 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
285 needless bounds checking. (thanks to Juho Snellman)
286 * optimization: implemented multiplication as a modular
287 (UNSIGNED-BYTE 32) operation on the PPC backend.
288 * fixed some bugs revealed by Paul Dietz' test suite:
289 ** ADJUST-ARRAY now copies the datum in a zero rank array if
291 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
292 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
293 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
295 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
297 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
299 ** ECHO-STREAMs no longer attempt to echo the end of file value to
300 their output stream on EOF from read.
301 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
302 have been read to end-of-file.
303 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
305 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
306 description of determination of which consecutive characters
308 ** Printing the "Space" character with escaping on now yields "#\\ ",
309 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
310 ** Reading floating-point numbers with *READ-BASE* set to a number
311 less than 10 works correctly.
312 ** Reading floating-point numbers with *READ-BASE* set to a number
313 more than 10 works correctly.
314 ** Printing with *PRINT-READABLY* targets the standard readtable, not
315 the readtable currently in effect.
317 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
318 * minor incompatible change: parsing of namestrings on a physical
319 (Unix) host has changed; numbers after the final #\. in a
320 namestring are no longer interpreted as a version field. This is
321 intented to be largely invisible to the user, except that the
322 meaning of the namestring "*.*.*" has changed: it now refers to a
323 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
324 should usually be replaced by
325 (make-pathname :name :wild :type :wild :version :wild)
326 with the added benefit that this is more likely to be portable.
327 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
328 option to OPEN now signals an error if the file being opened
329 exists; this may have an impact on existing code.
330 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
331 signals received, on the PowerPC platforms (both Linux and
332 Darwin). (thanks to Pierre Mai for pointing out the location of
334 * several fixes on OS X: The system now builds and runs cleanly on
335 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
336 (thanks to Brian Mastenbrook)
337 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
338 second-resolution: integer multiples of 1/3600 between -24 and 24.
339 (thanks to Vincent Arkesteijn)
340 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
341 type when called with 1 argument; PEEK-CHAR checked type of
342 PEEK-TYPE only after having read first character from a
343 stream. (reported by Peter Graves)
344 * bug fix: the garbage collector now has much better locality
345 behaviour, and in particular no longer treats all memory as being
346 exhausted when many small objects point to each other in a deeply
348 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
349 recognized as being TYPEP their class.
350 * bug fix: the PUSHNEW documentation string has been corrected.
351 (thanks to Vincent Arkesteijn)
352 * bug fix: defaulting of the value for the last of an atypically
353 large number of multiple values being bound was not being
354 performed correctly on the Alpha or PPC platforms
355 * optimization: implemented multiplication as a modular
356 (UNSIGNED-BYTE 32) operation on the x86 backend.
357 * optimization: SEARCH on simple-base-strings can now be open-coded.
358 (see also contrib/compiler-extras.lisp for inspiration for
359 teaching the compiler about the Boyer-Moore algorithm).
360 * value, returned by MAX (and MIN) called with several EQUALP, but
361 not EQL, arguments now does not depend on compiler settings.
362 * fixed some bugs revealed by Paul Dietz' test suite:
363 ** in stack analysis liveness information is propagated from
364 non-local entry points.
365 ** pathwise CAST removing failed when the CAST node did not start
367 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
368 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
370 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
371 host is already defined.
372 ** RENAME-FILE works on streams instead of signalling an internal
374 ** PEEK-CHAR uses the current readtable when determining whether
375 or not a character is whitespace.
376 ** MERGE-PATHNAMES handles the case when the pathname does not
377 specify a name while the default-pathname specifies a version.
378 ** Pathnames now stand a better chance of respecting print/read
380 ** Attempting to use standardized file system operators with a
381 pathname with invalid :DIRECTORY components signals a
383 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
385 ** DIRECTORY on logical pathnames is more correct.
386 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
387 signal a TYPE-ERROR if their argument is not a stream.
388 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
389 designator argument does not designate a stream.
390 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
391 examining the synonym.
392 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
394 ** OPEN and WITH-OPEN-STREAM allow opening streams with
395 element-type larger than ([UN]SIGNED-BYTE 32).
397 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
398 * When built with the :SB-FUTEX feature, threaded builds now take
399 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
400 for faster/more reliable mutex and condition variable support.
401 * Incompatible change (but one you probably shouldn't have been using
402 anyway): the interface and code for arbitrating between multiple
403 threads in the same user session has been redesigned.
404 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
405 error when called without an explicit environment argument.
406 (thanks to Nikodemus Siivola)
407 * bug fix: buffered :DIRECTION :IO streams are less likely to become
408 confused about their position. (thanks to Adam Warner and Gerd
410 * bug fix: Pretty printing backquoted forms with unquotations in the
411 argument list position of various code constructs such as LAMBDA
412 now works correctly. (reported by Paul Dietz)
413 * bug fix: Pretty printing unquotations no longer loses all
414 stream position information.
415 * optimization: performance of string output streams is now less
416 poor for multiple small sequence writes.
417 * optimization: performance of CSUBTYPEP in the presence of complex
418 expressions involving CONS and NOT many times has been improved.
419 (reported by Paul Dietz)
420 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
422 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
423 streams for the REPL, for the debugger, and so on. (thanks to
425 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
426 * fixed some bugs revealed by Paul Dietz' test suite:
427 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
428 is ignored if it is too small, rather than propagating through
429 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
430 ** extremely complex negations of CONS types were not being
431 sufficiently canonicalized, leading to inconsistencies in
433 ** VALUES tranformer lost derived type.
435 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
436 * fixed a bootstrapping bug: the build process no longer assumes
437 that the various BOOLE-related constants have the same value in
438 host and target lisps. (noted by Paul Dietz' test suite on an
439 SBCL binary built from CLISP)
440 * The system can now be dynamically linked on the MIPS platform,
441 which enables dynamic loading of foreign code from Lisp. (thanks
442 to Ralf Baechle for discussions on the MIPS ABI)
443 * The system now records debugging information for its own source
444 files in a filesystem-position-independent manner, relative to
445 the "SYS" logical host.
446 * fixed a compiler bug: MV-LET convertion did not check references
447 to the "max args" entry point. (reported by Brian Downing)
448 * tweaked disassembly notes to be less confident about proclaiming
449 some instruction as an LRA. (thanks to Brian Downing)
450 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
451 now each have their own history, command character, and other
452 characteristics. (thanks to David Lichteblau)
453 * fixed some bugs revealed by Paul Dietz' test suite:
454 ** compiler failure in compiling LOGAND expressions including a
456 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
457 shift greater than 32.
458 ** FLUSH-DEST did not mark blocks for type check regeneration.
459 ** HANDLER-CASE failed to accept declarations in handler clauses
460 in some circumstances.
462 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
463 * New code in contrib/sb-introspect (still probably not entirely
464 stable yet) provides some support for smart Lisp development
465 environments like SLIME.
466 * The conditions signalled for errors occurring when loading .fasl
467 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
468 in a way which should help ASDF recover gracefully.
469 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
470 stuff has been cleaned up. If you code contrib/ stuff, this might
471 affect you, and you can look at contrib/README, contrib/STANDARDS,
472 and/or the 0.8.4.27 diff to check.
473 * In full calls the compiler now does not generate checks for declared
474 argument types for all arguments.
475 * various threading fixes
476 ** and some experimental patches which didn't make it into
477 the main tree for this release, but which are shipped in
478 contrib/experimental-thread.patch as a possible fix for some
479 failures (deadlock, spinning...) in GC-intensive multithreaded
481 * fixed PPC build problem (source code incompatibility of different
482 library versions): added offsetof() hackery which attempts to divine
483 where glibc maintainers put uc_mcontext today
484 * fixed bug 282: compiler does not trust type assertions while passing
485 arguments to a full call.
486 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
487 &OPTIONAL VALUES type parameter.
488 * fix bug 214: algorithm for noting rejected templates is now more
489 similar to that of template seletion. (also reported by rydis on
491 * fixed bug 141b: printing backquoted information readably and prettily
492 inserts a space where necessary.
493 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
494 (CHARACTER 10) are now reported as errors, rather than propagated
495 as unknown types. (reported by piso on #lisp)
496 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
497 with values NIL and :ERROR. (thanks to Milan Zamazal)
498 * fixed bug 191c: CLOS now does proper keyword argument checking as
499 described in CLHS 7.6.5 and 7.6.5.1.
500 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
501 longer raise an error; further, using a list as a for-as-arithmetic
502 counter now raises a meaningful error.
503 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
504 now understood by sequence creation functions such as MAKE-SEQUENCE
506 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
507 a STRING-INPUT-STREAM.
508 * compiler enhancement: SIGNUM is now better able to derive the type
510 * type declarations inside WITH-SLOTS are checked. (reported by
512 * fixed some bugs revealed by Paul Dietz' test suite:
513 ** incorrect optimization of TRUNCATE for a positive first
514 argument and negative second.
515 ** compiler failure in let-convertion during flushing dead code.
516 ** compiler failure while deriving type of TRUNCATE on an
517 interval, containing 0.
518 ** ASH of a negative bignum by a negative bignum count now returns
520 ** intersection of CONS types now canonicalizes properly, fixing
521 inconsistencies in SUBTYPEP.
523 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
524 * incompatible change: The --disable-debugger command line
525 option now clobbers the debugger at a more fundamental
526 level, by redefining #'INVOKE-DEBUGGER instead of by
527 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
528 is specified by ANSI to ignore *DEBUGGER-HOOK* and
529 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
530 enter the debugger REPL and then suffer recursive errors
531 because *DEBUG-IO* is also messed up in --disable-debugger mode;
532 while under the new system, BREAK in --disable-debugger mode
533 terminates the system just as an unhandled error would.
534 * fixed compiler performance when processing loops with a step >1;
535 * bug fix: DOCUMENTATION now retrieves generic function
536 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
537 support has been systematized, and now supports the methods
538 specified by ANSI, along with a default method and a method for
539 slot documentation. (reported by Nathan Froyd)
540 * bug fix: effective methods associated with a generic function are
541 no longer cached over a change of that generic function's method
542 combination. (reported by Andreas Fuchs)
543 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
544 names a non-existent directory. (thanks to Andreas Fuchs)
545 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
546 the PPC platform, lead to stack corruption; this has been fixed.
547 (reported by Rainer Joswig)
548 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
549 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
550 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
551 arithmetic in general, is now much more reliable on the PPC
553 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
554 platform now returns the right answer.
555 * optimization: restored some effective method precomputation in
556 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
557 precomputation is now tunable.
558 * optimization: compiler-internal data structure use has been
559 reviewed, and changes have been made that should improve the
560 performance of the compiler by about 20%.
561 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
562 simple-base-strings and simple-bit-vectors is improved.
563 * optimization: the optimization of 32-bit logical and arithmetic
564 functions introduced in version 0.8.3 on the x86 has been
565 implemented on the mips, ppc and sparc platforms; an
566 implementation of the same facility, but for 64-bit arithmetic,
567 has been added for the alpha.
568 * microoptimization: the compiler is better able to make use of the
569 x86 LEA instruction for multiplication by constants.
570 * bug fix: in some situations compiler did not report usage of
571 generic arithmetic in (SPEED 3) policy.
572 * bug 145b fix: compiler used wrong type specifier while converting
573 MEMBER-types to numeric.
574 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
576 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
577 index argument now works properly on the Alpha platform.
578 * bug fix: floating point exception treatment on the Alpha platform
580 * bug fix: FILE-POSITION works much better on string input and
581 output streams. (thanks to Nikodemus Siivola)
582 * bug fix: many threading/garbage collection symptoms sorted.
583 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
584 might be pseudo-atomic.
585 * internal change: Stopping for GC is now done with signals not
586 ptrace. GC is now done in whichever thread wanted it, instead of
588 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
590 * bug fix: result form in DO is not contained in the implicit
592 * incompatible change: ICR structure is changed; the value part of
593 CONTINUATION is now called LVAR; corresponding functions are
594 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
595 * added type deriver for ISQRT (thanks to Robert E. Brown).
596 * bug fix: better support for loading from the command line when an
597 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
599 * fixed some bugs revealed by Paul Dietz' test suite:
600 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
601 ** ROUND and FROUND now give the right answer when given very
602 small float arguments.
603 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
605 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
606 ** (GCD 0 <negative-integer>) returned <negative-integer>.
607 ** LCM should return a non-negative integer.
608 ** PARSE-INTEGER returned the index of a terminator instead of the
609 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
610 ** PARSE-INTEGER returned an incorrect index being applied to a
612 ** LCM with two arguments of 0 returns 0 rather than signalling
614 ** unsigned addition of a 32-bit constant with the high bit set no
615 longer causes an internal compiler error on the x86.
616 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
617 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
618 with negative last argument.
619 ** byte specifiers with zero size and position no longer cause
620 an error during type derivation.
621 ** bignum multiplication on the Alpha platform now returns the
623 * porting: The system now builds on SuSE AMD64, although it still
624 generates a 32-bit binary.
625 * .fasl file incompatibility: The fasl file version number has
626 been incremented (because of the changes to internal compiler
627 data structures referred to above).
629 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
630 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
631 more accurately, on the Darwin kernel running on PowerPC hardware.
632 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
633 * Compiler code deletion notes now signal a condition of type
634 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
635 an associated MUFFLE-WARNING restart.
636 * The compiler now performs limited argument count validation of
637 constant format strings in FORMAT, and where appropriate in ERROR,
638 CERROR and WARN. (thanks to Gerd Moellmann)
639 * New ASDF-INSTALL contrib can be used for automatic download and
640 installation of third-party Lisp code from CCLAN or other sites
642 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
643 which forces another thread to execute a function supplied by the
645 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
646 now accept and act upon their :ELEMENT-TYPE keyword argument.
647 (reported by Martin Atzmueller, Edi Weitz)
648 * bug fix: FILE-POSITION now accepts position designators up to
649 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
650 is the greater. (thanks to Patrik Nordebo)
651 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
652 Kalle Olavi Niemitalo)
653 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
654 printable. (reported by Eric Marsden)
655 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
656 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
657 * x86 bug fix in control stack exhaustion checking: now shows backtrace
658 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
659 (thanks to Stig Sandoe)
660 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
661 UNREAD-CHAR now correctly includes the unread character in the
662 target sequence. (thanks to Gerd Moellmann)
663 * bug fix in threaded builds: the system can now be suspended and
664 resumed by shell job control with minimal disruption.
665 * bug fixes in times and timezones >2038AD
666 * better handling of "where is GNU make?" problem in build scripts
667 (thanks to Nikodemus Siivola)
668 * new optimization: inside a named function any reference to a
669 function with the same name is considered to be a self-reference;
670 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
671 optimization quality.
672 * new optimization on x86: logical functions and + now have
673 optimized (UNSIGNED-BYTE 32) versions, which are automatically
674 used when the result is truncated to 32 bits.
675 * VALUES declaration is partially enabled.
676 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
677 * fixed some bugs revealed by Paul Dietz' test suite:
678 ** The system now obeys the constraint imposed by
679 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
680 types form a lattice under type intersection.
681 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
682 ** ASSOC now ignores NIL elements in an alist.
683 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
684 and (1+ MOST-POSITIVE-FIXNUM) answers.
685 ** The addition of a method with invalid qualifiers to a generic
686 function does not cause an error to be signalled immediately;
687 a warning is signalled, and the error is generated only on
688 calling the generic function.
689 * changed .fasl file version number, in part to add type codes for
690 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
691 obscure ANSI requirements
693 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
694 * fixed bug 148: failure to inline-expand a local function left
695 garbage, confusing the compiler.
696 * fixed bugs 3cd: structure slot readers perform type check if the
697 slot can have an invalid value (i.e. it is either not initialized
698 or can be written with a less specific slot writer).
699 * bug fix: the compiler now traps array references to elements off
700 the end of an array; previously, the bounds checking in some
701 circumstances could go off-by-one.
702 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
704 * type declarations for array element types now obey the description
705 on the CLHS page "Declaration TYPE", as per discussions on
706 sbcl-help around 2003-05-08. This means that a declaration
707 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
708 declaration, all references to BAR will be asserted or assumed
709 (with THE, so dependent on compiler policy) to involve objects of
710 type FOO. Note that no such declaration is implied in
711 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
712 * declared types of functions from the "Conditions"
713 chapter. (reported by Paul Dietz)
714 * bug fix: CERROR accepts a function as its first argument.
715 * bug fix: NTH an NTHCDR accept a bignum as index
716 arguments. (reported by Adam Warner)
717 * optimization: character compare routines now optimize comparing
718 against a constant character. (reported by Gilbert Baumann)
719 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
720 argument now works properly on the MIPS platform.
721 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
722 * repeated evaluation of the same DEFSTRUCT, a slot of which is
723 declared to have a functional type, does not cause an error
725 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
726 many :DEFAULT-INITARGS. (reported by Istvan Marko)
727 * fixed bug: if last continuation of a deleted block has a
728 destination, this destination should be deleted too. (reported by
730 * fixed a bug in the bootstrap process: the host compiler's values
731 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
732 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
733 test case from Patrik Nordebo)
734 * improved the ability of the disassembler on the PPC platform to
735 provide helpful disassembly notes.
736 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
737 the class in more cases than previously.
738 * bug fix: FILE-POSITION now understands :START and :END for
739 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
740 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
741 8) data. (thanks to David Lichteblau)
742 * bug fix: it is possible to add a method to a generic function
744 * bug fix: reader failed to signal END-OF-FILE inside an
745 object representation. (reported by Nikodemus Siivola)
746 * fixed some bugs revealed by Paul Dietz' test suite:
747 ** LAST and [N]BUTLAST should accept a bignum.
748 ** condition slot accessors are methods.
749 ** (VECTOR NIL) is a subtype of STRING.
751 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
752 * minor incompatible change: some nonsensical specialized lambda
753 lists (used in DEFMETHOD) which were previously ignored now signal
755 * minor incompatible change: the system is now aware of the types of
756 variables in the COMMON-LISP package, and will signal errors for
757 most violations of these type constraints (where previously they
758 were silently accepted).
759 * minor incompatible change: COMPILE-FILE now uses the freedom
760 afforded (ANSI 3.2.2.3) to use derived function types for
761 functions defined in the same file. This also permits the system
762 to warn on static type mismatches and function
763 redefinition. (Currently it does not work with high DEBUG level.)
764 * minor incompatible change: VALUES declaration is disabled.
765 * When issuing notes, the compiler now signals a condition of type
766 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
767 restart for use in user handlers. It is expected that the
768 COMPILER-NOTE condition will eventually become a condition
769 supertype to a hierarchy of note types, which will then be
770 handleable in a similar fashion. However, at the moment, no such
771 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
772 but it's an implementation detail, not a classification for the
774 * Changes in type checking closed the following bugs:
775 ** type checking of unused values (192b, 194d, 203);
776 ** template selection based on unsafe type assertions (192c, 236);
777 ** type checking in branches (194bc).
778 * A short form of VALUES type specifier has ANSI meaning (it has
779 increased the number of situations when SBCL cannot perform type
781 * fixed bug in DEFSTRUCT: once again, naming structure slots with
782 keywords or constants is permissible.
783 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
784 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
785 classes. (thanks to Antonio Martinez)
786 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
787 outputs FRESH-LINE or TERPRI, and no longer converts its stream
788 argument to a pretty-print stream. Instead, it leaves any such
789 operations to DESCRIBE-OBJECT methods.
790 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
792 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
793 lambda lists. (thanks to David Lichteblau)
794 * bug fix: Defining a generic function with a :METHOD-CLASS being a
795 subclass of STANDARD-METHOD no longer causes stack exhaustion.
796 (thanks to Gerd Moellmann)
797 * fixed bug 246: increased compilation speed of long
798 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
800 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
802 * DEFCONSTANT now throws a condition of type
803 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
804 constant to a non-EQL value; CONTINUE and ABORT restarts
805 respectively change and preserve the value.
806 * fixed bug 63: The code walker, part of the implementation of CLOS,
807 is now better at handling symbol macros.
808 * bug fix: There is no longer an internal implementation type named
809 CL:LENGTH. (reported by Raymond Toy)
810 * bug fix: In macro-like defining macros/special operators the
811 implicit block does not enclose the lambda list.
812 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
813 enclosed in parentheses) are not suitable as type specifiers, and
814 their use properly signals an error now.
815 * bug fix: An argument count mismatch for a type specifier in code
816 being compiled no longer causes an unhandled error at compile
817 time, but signals a compile-time warning.
818 * fixed simple vector readable printing
819 * bug fix: DESCRIBE takes more care over whether the class
820 precedence list slot of a class is bound before accessing it.
821 (reported by Markus Krummenacker)
822 * bug fix: FORMATTER can successfully compile pretty-printer format
823 strings which use variants of the ~* directive inside.
824 * bug fix: SEARCH now applies its TEST predicate to the elements of
825 the arguments in the correct order. (thanks to Wolfhard Buss)
826 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
827 in type inference. (thanks to Robert E. Brown)
828 * bug fix: Reading in symbols with an explicit package name of ""
829 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
830 * fixed some bugs revealed by Paul Dietz' test suite:
831 ** NIL is now allowed as a structure slot name.
832 ** Arbitrary numbers, not just REALs, are allowed in certain
833 circumstances in LOOP for-as-arithmetic clauses.
834 ** Multiple class redefinitions before slot access no longer
836 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
837 association between the name and a class.
838 ** Generic functions with non-standard method-combination and over
839 five methods all of which return constants no longer return NIL
840 after the first few invocations. (thanks to Gerd Moellmann)
841 ** CALL-NEXT-METHOD with no arguments now passes the original
842 values of the arguments, even in the presence of assignment.
843 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
845 ** DISASSEMBLE works with closures and funcallable instances.
846 ** ADD-METHOD now returns the generic function, not the new method.
847 ** FIND-METHOD signals an error if the lengths of the specializers
848 is incompatible with the generic function, even if the ERRORP
850 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
851 which its argument is a member.
852 ** DEFCLASS only redefines the class named by its class-name
853 argument if that name is the proper name of the class;
854 otherwise, it creates a new class.
855 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
856 of the UNBOUND-SLOT condition to the name of the slot.
857 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
858 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
859 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
860 SLOT-MAKUNBOUND in the specified fashion.
862 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
863 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
864 cross-compilation host. As a consequence, we can now bootstrap our
865 way up to SBCL starting with a bare gcc toolchain and human-readable
866 source code (first the source to CLISP, then the source to SBCL).
867 * A contributed module containing a partial implementation of the
868 simple-streams interface has been included. (thanks to Rudi
870 * A contributed module implementing the RFC1321 Message Digest
871 Algorithm, known as MD5, has been included.
872 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
873 no longer has any effect, as the code controlled by this feature
874 has been deleted. (As far as we know, no-one has ever built using
875 this feature, and its semantics were confused in any case).
876 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
877 work on conditions (as required by the ANSI specification),
878 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
879 expected behaviour on conditions. Users should note, however,
880 that such behaviour is not required by the ANSI specification,
881 and so use of this behaviour may render their code unportable.
882 * fixed some bugs revealed by Paul Dietz' test suite:
883 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
885 ** &ENVIRONMENT parameter in macro lambda list is bound first.
886 ** SXHASH on condition objects no longer returns NIL.
887 ** :ALLOCATION :CLASS slots are better treated; their values are
888 updated on class redefinition, and initforms inherited from
889 superclasses are applied.
890 ** REMOVE-METHOD returns its generic function argument even when
891 no method was removed.
892 ** SHARED-INITIALIZE now initializes the values of the requested
893 slots, including those with :ALLOCATION :CLASS.
894 ** ALLOCATE-INSTANCE now works on structure classes defined via
895 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
897 ** SLOT-EXISTS-P now works on conditions, as well as structures
899 ** MAKE-LOAD-FORM now has the required methods on
900 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
901 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
902 keyword, and now implements the SLOT-NAMES argument.
903 ** methods with &OPTIONAL arguments no longer allow too many
904 arguments to be passed in the call without error.
905 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
906 option is consistent with the required arguments of the generic
907 function lambda list.
908 * bug fix: REQUIRE accepts a string designator. (Thanks to
910 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
911 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
912 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
913 * bug fix: The compiler now checks for duplicated variables in macro
915 * bug fix: SETQ on globals returns the correct value.
916 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
917 not a COMPILER-ERROR (followed by some other strange error on
918 choosing the CONTINUE restart).
919 * bug fix: make.sh and friends are now more consistent in the way that
920 they look for GNU "make".
922 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
923 * experimental native threads support (on x86 Linux >=2.4 only).
924 This is not compiled in by default: you need to add :SB-THREAD to
925 the target features. See the "Beyond ANSI" chapter of the manual
927 * fix for longstanding nonANSIism: The old distinction between
928 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
929 The return value from CL:FIND-CLASS is now a CLOS class, and
930 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
931 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
932 * An interface to the MetaObject Protocol, as described in Kiczales,
933 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
934 MIT Press, 1991, is available from the SB-MOP package.
935 * incompatible change: the SB-PCL package should now be considered
936 a private implementation detail, and no longer a semi-private MOP
938 * minor incompatible change: due to rearrangement for threads, the
939 control stack and binding stack are now allocated at arbitrary
940 addresses instead of being hardcoded per-port. Users affected by
941 this probably have to be doing advanced things with shared
942 libraries, and will know who they are.
943 * minor incompatible change: Previously, all --eval forms used were
944 processed with READ before any of them were processed with EVAL.
945 Now each --eval form is processed with both READ and EVAL before
946 the next --eval form is processed. (Thus package operations like
947 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
948 work as the user might reasonably expect.)
949 * minor incompatible change: *STANDARD-INPUT* is now only an
950 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
952 * minor incompatible change: Y-OR-N-P is now character-oriented,
953 not line oriented. Also, YES-OR-NO-P now works without errors.
954 (thanks to Antonio Martinez)
955 * sb-aclrepl module improvements: an integrated inspector, added
956 repl features, and a bug fix to :trace command.
957 * Known functions, which cannot be open coded by the backend, are
958 considered to be able to check types of their arguments. (fixing
959 a bug report by Nathan J. Froyd)
960 * fixed a bug in computing method discriminating functions: It is
961 now possible to define methods specialized on classes which have
962 forward-referenced superclasses. (thanks to Gerd Moellmann)
963 * fixed evaluation order in optional entries (reported by Gilbert
965 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
966 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
967 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
968 required-by-AMOP NAME argument, as well as CLASS and
969 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
970 * fixed bug 20: DEFMETHOD can define methods using names that are
971 not the proper names of classes to designate class specializers.
972 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
973 Unix file descriptors, instead of blowing up. (thanks to Antonio
975 * Garbage collection refactoring: user-visible change is that a
976 call to the GC function during WITHOUT-GCING will not do garbage
977 collection until the end of the WITHOUT-GCING. If you were doing
978 this you were probably losing anyway.
979 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
980 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
981 TYPEP the latter but not the former.
982 * The compiler issues a full WARNING for calls to undefined functions
983 with names from the CL package.
984 * MAP-INTO for a vector destination is open coded. (reported by
985 Brian Downing on c.l.l)
986 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
987 documentation string.
988 * fixed some bugs revealed by Paul Dietz' test suite:
989 ** COPY-ALIST now signals an error if its argument is a dotted
991 ** Condition slots are now accessed more correctly in the presence
992 of multiple initargs for a given slot.
993 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
994 exclude restarts of the same name associated with a different
996 ** DEFCLASS of forward-referenced classes with another
997 forward-referenced class in the superclasses list no longer
999 ** Condition slots are now initialized once each, not multiple
1000 times. (thanks to Gerd Moellmann)
1001 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1003 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1004 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1005 arguments contain duplicated elements.
1006 ** RESTART-CASE understands local macros.
1007 ** RESTART-CASE associates exactly its own restarts with a condition.
1008 ** ENDP in safe mode checks its argument to be of type LIST.
1009 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1010 in question is unbound.
1011 ** Optimized MAKE-INSTANCE functions no longer cause internal
1012 assertion failures in the presence of duplicate initargs.
1013 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1014 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1015 propagated even in the ctor optimized implementation of
1017 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1019 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1020 * a better implementation of SXHASH on (simple) bit vectors,
1021 measured both in execution speed and in distribution of results
1022 over the positive fixnums, has been installed. Likewise, a better
1023 implementation of EQUAL for simple bit vectors is now available.
1024 * fixed CEILING optimization for a divisor of form 2^k.
1025 * fixed bug 240 (emitting extra style warnings "using the lexical
1026 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1027 by Antonio Martinez)
1028 * fixed SXHASH, giving different results for NIL depending on type
1029 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1030 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1031 variable. (found by Rolf Wester)
1032 * SBCL does not ignore type declarations for special
1033 variables. (reported by rif on c.l.l 2003-03-05)
1034 * some bug fixes in contrib/sb-aclrepl/
1035 * fixed some bugs revealed by Paul Dietz' test suite:
1036 ** a bug in the CONS type specifier, whereby the CAR and CDR
1037 types got intertwined, has been fixed;
1038 ** the type system is now able to reason about the interaction
1039 between INTEGER and RATIO types more completely;
1040 ** APPEND, [N]REVERSE and NRECONC check that those their
1041 arguments, which must be proper lists, are really so;
1042 ** An array specialized to be unable to hold elements has been
1043 implemented, as required -- yes, really -- by ANSI;
1044 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1045 on malformed property lists;
1047 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1048 * incompatible packaging change: in line with Unix convention,
1049 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1050 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1051 * REQUIRE and PROVIDE are now optionally capable of doing something
1052 useful. See the documentation string for REQUIRE.
1053 * infrastructure for a managed SBCL contrib system: contributed
1054 modules in this release include:
1055 ** the ASDF system definition facility;
1056 ** an interface to the BSD Sockets API;
1057 ** an ACL-like convenience interface to the repl;
1058 (thanks to Kevin Rosenberg)
1059 ** an implementation of ROTATE-BYTE, with efficient implementation
1061 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1062 gives a closer approximation to the right answer than previously.
1063 (thanks to Raymond Toy)
1064 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1065 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1066 optional environment arguments, as required by ANSI.
1067 * fixed bugs in other functions taking environment objects, allowing
1068 calls with an explicit NIL environment argument to be compiled
1070 * fixed bug 228: primary return values from
1071 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1072 COMPILE or FUNCTION.
1073 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1074 :TYPE will no longer signal errors on innocuous objects.
1075 * fixed bug 231b: SETQ is better at respecting type declarations in
1076 the lexical environment.
1077 * fixed a bug in DEFCLASS: classes named by symbols with no or
1078 unprintable packages can now be defined.
1079 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1080 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1081 * fixed compiler failure related to checking types of functions.
1082 (reported by Robert E. Brown)
1083 * the compiler is now much more consistent in its error-checking
1084 treatment of bounding index arguments to sequence functions: in
1085 (SAFETY 3) code, errors will be signalled in almost all cases if
1086 invalid sequence bounding indices are passed to functions defined
1087 by ANSI to operate on sequences.
1088 * fixed a bug in the build procedure: documentation of SBCL-specific
1089 packages is now preserved and available in the final Lisp image.
1090 * lifted FDEFINITION lookup out of loops in the implementation of
1091 many list operations. (thanks to Robert E. Brown)
1092 * fixed a bug in the reader: the #n# reader macro now works for
1093 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1094 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1095 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1096 E. Brown and Tony Martinez respectively)
1097 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1098 SLOT-BOUNDP: in particular, it is now able to optimize them much
1099 better, and is now not vulnerable to having packages renamed.
1100 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1101 always called when a slot is not present in an instance. (thanks
1103 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1104 longer signal an error on unreading a character following EOF on
1105 the previous constituent stream. (thanks to Tony Martinez)
1106 * fixed some bugs revealed by Paul Dietz' test suite:
1107 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1108 not just nonnegative fixnums;
1109 ** the logical bit-array operators such as BIT-AND now accept an
1110 explicit NIL for their "opt-arg" argument (to indicate a
1111 freshly-consed result bit-array);
1112 ** ELT now signals an error on an invalid sequence index in safe
1114 ** the type system is now cleverer about negations of numeric
1115 types, and consequently understands the BIGNUM and RATIO types
1117 ** the type system is now cleverer about the interaction between
1118 INTEGER and RATIO types: while bugs still remain, many more
1119 cases are accurately computed;
1120 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1121 if it is in the last clause;
1122 ** CONSTANTLY now correctly returns a side-effect-free function in
1124 ** DECLARE is no longer treated as a special-operator; in
1125 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1126 * incremented fasl file version number due to the change in the
1127 DEFSTRUCT-SLOT-DESCRIPTION structure.
1129 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1130 * minor incompatible change: code processed by the "interpreter" or
1131 EVAL now has a compilation optimization policy of (DEBUG 2)
1132 (changed from (DEBUG 1)) to improve debuggability of interactive
1133 development, and to allow the use of the debug RETURN command in
1135 * an experimental implementation of the RETURN command for the
1136 debugger has been included. (thanks to Frederik Kuivinen)
1137 * fixed bug 62: constraints were not propagated into a loop.
1138 * fixed bug in embedded calls of SORT (reported and investigated by
1140 * fixed some bugs revealed by Paul Dietz' test suite:
1141 ** printing and reading of arrays with some dimensions having
1142 length 0 (thanks to Gerd Moellmann);
1143 ** BOA constructor with &AUX argument without a default value does
1144 not cause a type error;
1145 ** CONSTANTP now returns true for all self-evaluating objects.
1147 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1148 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1149 accessors that are related by inheritance, as specified in the
1150 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1151 Valtteri Vuorikoski)
1152 * The compiler is now able to inline functions that were defined in
1153 a complex lexical environment (e.g. inside a MACROLET).
1154 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1156 * Support for the upcoming FreeBSD-5.0 release has been included.
1157 (thanks to Dag-Erling Smorgrav)
1158 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1159 effect when it is not in a toplevel context.
1160 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1161 stand a better chance of being correct. (thanks to Gerd
1163 * fixed bug in COERCE, which now signals an error on coercing a
1164 rational to a bounded real type which excludes the expected
1166 * The compiler is now able to derive types more accurately from the
1167 COERCE and COMPILE functions.
1168 * fixed bug 223: functional binding is considered to be constant
1169 only for symbols in the CL package.
1170 * fixed bug 231: SETQ did not check the type of a variable being set
1171 (reported by Robert E. Brown)
1172 * A new optimization for MAKE-INSTANCE has been included, fixing
1173 various bugs (including relating to :ALLOCATION :CLASS slots and
1174 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1176 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1177 ** As required by ANSI, LOOP now disallows anonymous collection
1178 clauses such as COLLECT I in conjunction with aggregate boolean
1179 clauses such as THEREIS (= I 1);
1180 ** LOOP now signals an error when any variable is reused in the
1181 same loop (including the potentially useful construct analogous
1182 to WITH A = 1 WITH A = (1+ A);
1183 ** IT is only a special loop symbol within the first clause of a
1184 conditional loop clause;
1185 ** LOOP with a typed iteration variable over a hashtable now
1186 signals a type error iff it should.
1187 * fixed some other bugs revealed by Paul Dietz' test suite:
1188 ** FILE-STREAM now names the class previously known as FD-STREAM;
1189 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1190 argument) no longer signals an error;
1191 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1192 of the slot symbol, rather than using the current package
1193 ((:CONC-NAME "") continues to intern the slot's name in the
1195 * incremented fasl file version number, because of the incompatible
1196 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1197 of the new implementation of DEFINE-COMPILER-MACRO.
1199 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1200 * Support for building SBCL for MIPS platforms running in
1201 little-endian mode has now been checked in, and basic
1202 functionality on said platforms verified.
1203 * minor incompatible change: PCL now records the pathname of a file
1204 in which methods and the like are defined, rather than its
1206 * minor incompatible change: TRUENAME now considers the truename of
1207 a file naming a directory to be the pathname with :DIRECTORY
1208 component indicating that directory.
1209 * minor incompatible change: a NAMED clause in the extended form of
1210 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1211 reason for the previous behaviour is unclear.
1212 * more systematization and improvement of CLOS and MOP conformance
1213 in PCL (thanks to Gerd Moellman and Pierre Mai):
1214 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1216 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1217 primary methods with no specializers;
1218 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1220 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1221 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1222 FORWARD-REFERENCED-CLASSes; error reporting on
1223 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1225 ** SXHASH on CLOS instances now uses a slot internal to the
1226 instance to return different numbers on distinct instances,
1227 while preserving the same return value through invocations of
1229 ** DEFMETHOD signals errors when methods with longer incongruent
1230 lambda lists are added to generic functions;
1231 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1232 CLASS, as specified in AMOP;
1233 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1234 on the order returned by the primary method for classes of
1235 class STANDARD-CLASS;
1236 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1237 * fixed some bugs shown by Paul Dietz' test suite:
1238 ** DOLIST puts its body in TAGBODY;
1239 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1241 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1242 value producing form;
1243 ** if more variables are given to PROGV than values, extra
1244 variables are bound and made to have no value;
1245 ** NSUBSTITUTE on list arguments gets the right answer with
1247 ** ELT signals an error of type TYPE-ERROR when the index argument
1248 is not a valid sequence index;
1249 ** LOOP signals (at macroexpansion time) an error of type
1250 PROGRAM-ERROR when duplicate variable names are found;
1251 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1252 ** FUNCALL of special-operators now cause an error of type
1254 ** PSETQ now works as required in the presence of side-effecting
1255 symbol-macro places;
1256 ** NCONC accepts any object as its last argument;
1257 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1259 ** loop-for-as-package does not require a package to be explicitely
1261 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1263 * fixed bug 166: compiler preserves "there is a way to go"
1264 invariant when deleting code.
1265 * fixed bug 172: macro lambda lists with required arguments after
1266 &REST arguments now cause an error to be signalled. (thanks to
1268 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1269 bug. (thanks to Matthew Danish)
1270 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1272 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1273 arguments are lexically rebound. (thanks to Gerd Moellmann and
1275 * fixed bug 194: error messages are now more informative when there
1276 is no primary method applicable in a call to a generic
1277 function. (thanks to Gerd Moellmann)
1278 * fixed bug in command line argument checking (thanks to Julian
1280 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1281 (thanks to Matthew Danish)
1282 * incremented fasl file version number, because of the
1283 SXHASH-related changes in the layout of CLOS data structures
1285 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1286 * minor incompatible change: The runtime (the Unix executable named
1287 "sbcl") is now much pickier about the .core files it will load.
1288 Essentially it now requires .core files to descend from the same
1289 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1290 as the runtime does. (The intent is to prevent the crashes which
1291 can occur, and which can even be reported as mysterious failures,
1292 when people patch the sources or change the build parameters
1293 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1294 sbcl and .core files.)
1295 * fixed bug: VALUES-LIST is no longer optimized away.
1296 * fixed bug 142: The FFI conversion of C string values to Lisp
1297 string values no longer conses excessively. (thanks to Nathan
1298 Froyd porting Raymond Toy's fix to CMU CL)
1299 * began to systematize and improve MOP conformance in PCL (thanks to
1300 Nathan Froyd, Gerd Moellman and Pierre Mai):
1301 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1303 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1304 ** FINALIZE-INHERITANCE is now called on class finalization;
1305 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1306 argument precedence order.
1307 * fixed bug 202: The compiler no longer fails on functions whose
1308 derived types contradict their declared type.
1309 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1310 so it can be non-toplevel.
1311 * The fasl file version number has changed (because of the new
1312 implementation of DEFMACRO).
1313 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1314 safe code, that any length requirement by their type-specifier
1315 argument is valid. The exceptions to this are described in bug
1317 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1318 functions now check the types of their inputs as required by ANSI.
1319 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1320 introduced symbol is DECLAREd to be SPECIAL.
1321 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1322 * fixed bug: :COUNT argument to sequence functions may be negative
1323 * fixed bug: body of DO-SYMBOLS may contain declarations
1324 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1325 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1326 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1327 symbol macro only once
1328 * fixed printing of call frame when argument list is unavailable
1329 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1330 * compiler no longer signals WARNING on unknown keyword
1333 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1334 * A beta-quality port to the mips architecture running Linux,
1335 based on the old CMUCL backend, has been made. It has been tested
1336 on a big-endian kernel, and works sufficiently well to be able to
1337 rebuild itself; it has not been tested in little-endian mode.
1338 * fixed an inconsistency between gencgc.c and purify.c which made
1339 dumping/loading .core files unreliable
1340 * fixed bug 120a: The compiler now deals correctly with IFs where
1341 the consequent is the same as the alternative, instead of
1342 misderiving the return type. (thanks to Alexey Dejneka)
1343 * fixed bug 113: Logical pathnames are now dumpable (the logical
1344 host is resolved at load-time, throwing an error if it is not
1346 * fixed bug 174: FORMAT's error message is slightly clearer when a
1347 non-printing character is used in a format directive.
1348 * fixed several bugs in compiler checking of type declarations, i.e.
1349 violations of the Python "declarations are assertions" principle
1350 (thanks to Alexey Dejneka)
1351 * fixed several bugs in PCL's error checking (thanks to Gerd
1353 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1355 * fixed bug in compilation of functions as first class values
1356 (thanks to Antonio Martinez-Shotton)
1357 * The compiler's handling TYPE-ERRORs which it can prove will
1358 inevitably happen at runtime has been cleaned up and corrected
1359 in several ways. (thanks to Alexey Dejneka)
1360 * improved argument type checking for various basic arithmetic
1361 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1362 lavished on them in the past that they can be compiled in many
1363 ways in different special cases
1364 * fixed bug 181: compiler checks validity of user supplied type
1366 * cleaned up code flushing in optimization: Function calls which
1367 should signal errors for safety purposes (e.g. which ANSI says
1368 should signal errors when their arguments are of incorrect type)
1369 are no longer optimized away.
1370 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1371 * incremented fasl file version number, because changes in the
1372 implementation of sequence functions like COERCE caused
1373 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1374 in old inline expansions) to become undefined. (Actually these
1375 changes were later undone, so we might very well be binary
1376 compatible with 0.7.7 after all, but leaving the version number
1377 incremented seemed like the simplest and most conservative
1380 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1381 * An alpha-quality port to the parisc architecture running Linux,
1382 based on the old CMUCL backend, has been made. This, even more so
1383 than the other backends, should be considered still a work in
1384 progress; known problems include that the Linux kernel in 64-bit
1385 mode does not propagate the correct sigcontext structure to
1386 userspace, and consequently SBCL on a parisc64 kernel will not
1388 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1389 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1390 suppressed.) Also now that the compiler is looking at declarations
1391 in the environment, it checks optimization declarations as well,
1392 and suppresses inlining when (> DEBUG SPEED).
1393 * More fixes have been made to treatment of floating point exception
1394 treatment and other Unix signals. In particular, floating point
1395 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1396 * The detection and handling of control stack exhaustion (infinite
1397 or very deeply nested recursion) has changed. Stack exhaustion
1398 detection is now done by write-protecting pages at the OS level
1399 and applies at all optimization settings; when found, a
1400 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1401 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1402 be caught using IGNORE-ERRORS.
1403 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1404 correctly when transforming recursive LABELS functions to LETs.
1405 (thanks to Alexey Dejneka)
1406 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1407 that are names of constants or global variables.
1408 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1409 alien routines with docstrings.
1410 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1411 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1413 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1414 (thanks to Eric Marsden)
1415 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1416 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1417 * Bug fix: The SPARC backend can now compile functions involving
1418 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1419 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1420 to a functions expecting a PATHNAME-DESIGNATOR.
1421 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1422 lambda lists. (thanks to Alexey Dejneka)
1423 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1424 functions so that EXPORTing the name of the class doesn't cause
1425 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1426 to Antonio Martinez for reporting this)
1427 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1428 constant-folded without failing an assertion. (thanks to Einar
1429 Floystad Dorum for reporting this)
1430 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1431 types are now dealt with more correctly by the compiler.
1432 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1433 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1434 behaviour analogous to RENAME-FILE. This puts its behaviour more
1435 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1437 * The fasl file version number has changed again. (because of the
1438 bug fix involving the names of PCL MAKE-INSTANCE functions)
1440 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1441 * bug fix: Floating point exceptions are treated much more
1442 consistently on the x86/Linux and PPC/Linux platforms.
1443 * Array initialization with :INITIAL-ELEMENT is now much faster for
1444 cases when the compiler cannot open code the array creation, but
1445 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1446 array accesses have also seen a speed increase.
1447 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1448 specified. (This was at the root of some bad interactions between
1449 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1451 * bug fix: Internal error arguments for undefined functions are now
1452 computed correctly on the PPC/Linux platform.
1453 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1454 Raymond Toy's patch for CMU CL)
1455 * Support for the Solaris 9 operating environment has been included
1456 (thanks to Daniel Merritt)
1457 * A very ugly but hopefully complete draft of the missing FFI chapter
1458 of the manual has been created by reformatting the corresponding
1459 CMU CL manual chapter into (currently very ugly and incoherent)
1460 DocBook and bringing it up to date for SBCL behavior. Thus, the
1461 manual is now essentially complete, at least by my extreme
1462 once-and-only-once standards, whereby it's acceptable to refer to
1463 the doc strings of SB-EXT functions as the primary documentation.
1464 * The fasl file version number has changed again, due to cleanup of
1465 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1466 is no longer a static symbol.)
1468 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1469 * SBCL now builds with OpenMCL (version 0.12) as the
1470 cross-compilation host; also, more progress has been made toward
1471 bootstrapping under CLISP.
1472 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1474 * bug 158 fixed: The compiler can now deal with integer loop
1475 increments different from 1; fixing this turned out also to fix
1477 * bug 169 fixed: no more bogus warnings about using lexical bindings
1478 despite the presence of perfectly good SPECIAL declarations (thanks
1479 to David Lichteblau)
1480 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1481 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1482 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1484 * bug 180 fixed: Method combination specifications no longer ignore
1485 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1486 * bug fix: Structure type predicate functions now check their argument
1487 count as they should.
1488 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1489 correctly. (thanks to Pierre Mai)
1490 * minor incompatible change: The --noprogrammer option is deprecated
1491 in favor of the new --disable-debugger option, which is very similar.
1492 (The major difference is that it takes effect at a slightly different
1493 time at startup, causing handling of errors in --sysinit and
1494 --userinit files will be affected differently.) The
1495 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1496 been added to allow this functionality to be controlled from ordinary
1497 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1498 maintainers, who might want to run non-interactive scripts to
1499 build SBCL cores which will later be used interactively.)
1500 * minor incompatible change: The LOAD function no longer, when given
1501 a wild pathname to load, loads all files matching that pathname.
1502 Instead, an error of type FILE-ERROR is signalled.
1504 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1505 * bug 147 fixed: The compiler preserves its block link/count
1506 invariants more correctly now so that it doesn't crash. (thanks
1508 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1510 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1511 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1512 no-value-for-this-character as NIL (as specified by ANSI).
1513 * HOST-NAMESTRING on physical pathnames now returns a string that is
1514 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1515 (thanks to Christophe Rhodes)
1516 * The Alpha port handles icache flushing more correctly. (thanks to
1518 * More progress has been made toward bootstrapping under CLISP. (thanks
1519 to Christophe Rhodes)
1520 * The fasl file format has changed again, because dynamic loading
1521 on OpenBSD (which has non-ELF object files) motivated some cleanups
1522 in the way that foreign symbols are transformed and passed around.
1523 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1524 is no longer treated as whitespace by the reader, but instead as
1525 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1526 |A
\7fB|, instead of A as it used to.
1528 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1529 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1530 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1531 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1532 this as of 0.7.1.45, but was left out of the previous news section
1533 (thanks to Dan Barlow)
1534 * SBCL now runs on the Solaris operating system on SPARC architectures
1535 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1536 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1537 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1538 * SPARC backend cleanups, allowing builds of cores optimized for V8
1539 and V9 SPARCS, and also emission of code targeted to a particular
1540 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1542 * SBCL is closer to bootstrapping under CLISP, thanks to various
1543 fixes by Christophe Rhodes.
1544 * The fasl file format has changed again, to allow the compiler's
1545 INFO database to support symbol macros.
1546 * The user manual (in doc/) is formatted into HTML more nicely.
1547 (thanks to coreythomas)
1548 * The system is smarter about SUBTYPEP relationships, especially
1549 those involving NOT types (including types such as ATOM which are
1550 represented internally using NOT types). Thus SUBTYPEP is less
1551 likely to return (VALUES NIL NIL) in general, and in particular
1552 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1554 * The fasl file format has changed again, because the internal
1555 representation of types now includes a new slot to support the new
1556 SUBTYPEP-of-NOT-types logic.
1557 * (not a change in the main branch of SBCL, but a related prototype
1558 which can hopefully be merged into the main branch of SBCL in the
1559 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1560 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1561 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1562 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1563 dispatch macro character combinations. (thanks to Alexey Dejneka)
1564 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1565 that superficially look like logical namestrings correctly.
1566 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1568 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1569 default supplied-p) for &optional and &key arguments. (thanks to
1572 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1573 * incompatible change: The compiler is now less aggressive about
1574 tail call optimization, doing it only when (> SPACE DEBUG) or
1575 (> SPEED DEBUG). (This is an incompatible change because there are
1576 programs which relied on the old CMU-CL-style behavior to optimize
1577 away their unbounded recursion which will now die of stack overflow.)
1578 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1579 for non-GENCGC systems has been increased to 20M (since that
1580 seems much closer to the likely performance optimum for modern
1581 systems than the old 4M value was)
1582 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1583 and *DEBUG-PRINT-LEVEL*
1584 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1585 of CMU CL's support for SPARC, and various endianness and other
1586 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1587 * new syntactic sugar for the Unix command line: --load foo.bar is now
1588 an alternate notation for --eval '(load "foo.bar")'.
1590 ** The system now detects stack overflow and handles it gracefully,
1591 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1592 optimization settings. (This is a good thing in general, and
1593 its introduction in this version should be particularly timely
1594 for anyone whose code fails because of suppression of tail
1596 ** The system now hunts for the C variable "environ" in a more
1597 devious way, to avoid segfaults when the C library version differs
1598 between compile time and run time. (thanks to Christophe Rhodes)
1599 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1600 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1601 ** The compiler no longer issues bogus style warnings for undefined
1602 classes in the same source file as the DEFCLASSes which defined
1603 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1605 ** fixes in CONDITION class precedence list for undefined function
1606 errors (thanks to Alexei Dejneka)
1607 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1608 correctly. (thanks to Dan Barlow)
1609 ** portability fixes aiming at bootstrapping under CLISP (thanks
1610 to Dave McDonald and Christophe Rhodes)
1611 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1612 ** fixes in type translation and and type inference (thanks to
1614 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1615 ** various fixes in the new ports (thanks to Dan Barlow)
1616 * several changes related to debugging:
1617 ** suppression of tail recursion, as noted above
1618 ** stack overflow detection, as noted above
1619 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1620 is now the default. (For some time encapsulation has been more
1621 reliable than the breakpoint-based :ENCAPSULATE NIL
1622 implementation, at least on X86 systems; and I just noticed that
1623 encapsulation also seems closer to the spirit of the ANSI
1626 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1628 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1629 up properly again. (There was a packaging bug in 0.7.0 which
1630 left their definitions in SB-SYS::LOAD-FOREIGN and
1631 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1632 vital for most things which interface to C-level interfaces,
1633 like extensions working with sockets or databases or
1634 Perl-compatible regexes or whatever, and the need to fix
1635 this bug is the main reason that 0.7.1 was released so
1637 ** DEFGENERIC is now choosier about the methods it redefines, so that
1638 reLOADing a previously-LOADed file containing DEFGENERICs does
1639 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1640 cycle now works as it should. (thanks to Alexey Dejneka)
1641 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1642 fixed by Christophe Rhodes. (He actually submitted this patch
1643 months ago, and I delayed until after 0.7.0.)
1644 ** Bug 111 (internal compiler confusion about runtime checks on
1645 FUNCTION types) was fixed by Alexey Dejneka.
1646 * Some internal cleanups (getting rid of variables which aren't
1647 needed now that the byte interpreter is gone) caused the fasl
1648 file format number to change again.
1650 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1651 * major incompatible change: The default fasl file extension, i.e. the
1652 default extension for files produced by COMPILE-FILE, has changed
1653 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1655 ** There are many changes in the implementation of the compiler.
1656 SBCL is now essentially a compiler-only implementation of ANSI
1657 Common Lisp. EVAL still "interprets" a few special cases, but
1658 almost all the interesting cases are handled by creating
1659 a LAMBDA expression, calling COMPILE on it, then calling
1660 FUNCALL on the result.
1661 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1662 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1663 Since the code is newer, there might still be some new bugs
1664 (though not as many as before Martin Atzmueller's fixes:-). But
1665 the new code is substantially simpler and clearer, and hopefully
1666 any remaining bugs will be simpler, less fundamental, and more
1667 fixable then the bugs in the old code.
1668 ** The revised compiler is still a little unsteady on its feet.
1670 *** The debugging information it produces (particularly the names
1671 of FUNCTION objects) is sometimes much less useful than what
1672 the old compiler produced.
1673 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1674 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1675 within a MACROLET) has been temporarily weakened.
1676 ** There are new compiler optimizations for various functions:
1677 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1678 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1679 *** the math functions TRUNCATE, FLOOR, and CEILING
1680 *** the function-of-all-trades COERCE
1681 Mostly these should be transparent, but there's one
1682 potentially-annoying problem (bug 117): when the compiler
1683 inline-expands a function and does type analysis on the result,
1684 it can create control paths which have type mismatches, and
1685 when it can't prove that those control paths aren't taken,
1686 it will issue WARNINGs about the type mismatches. This is
1687 a particular problem in practice for the new sequence functions.
1688 It's not clear how this should be fixed, and for now, a
1689 workaround is given in the entry for 117 in the BUGS file.
1690 ** (Because of the interaction between the two previous items --
1691 occasional inlining problems and new inline expansions -- some
1692 of the new sequence function optimizations won't really kick in
1693 completely until debugging information, and then inlining, are
1694 straightened out in some future version.)
1695 * minor incompatible changes:
1696 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1697 DIRECTORY no longer implicitly promotes NIL slots of its
1698 pathname argument to :WILD. In particular, when you ask for the
1699 contents of a directory (which you used to be able to do without
1700 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1701 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1702 ** changes in behavior that ANSI explicitly defines to be
1703 implementation dependent:
1704 *** The new compiler-only implementation still conforms with ANSI,
1705 but acts a little different than before. Besides the obvious
1706 changes in performance tradeoffs (that the cost per form passed
1707 to EVAL has gone up, and the cost per form executed by EVAL
1708 has gone down), the behavior of the system changes a little
1709 because there are no longer any interpreted function objects.
1710 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1711 e.g. doing COMPILE on the output of interactive DEFUN is
1713 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1715 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1716 has changed from (:CL) to NIL.
1717 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1718 e.g. #\Newline and #\Tab, don't have names specified in the
1719 ANSI Common Lisp standard, is now based on their ASCII symbolic
1720 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1721 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1722 are no longer used for output.
1723 ** changes in internal implementation constants:
1724 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1725 4 million. (If your application spends a lot of time GCing and
1726 you have a lot of RAM, you might want to experiment with
1727 increasing it even more.)
1728 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1729 However, almost all old code should still continue to work without
1730 immediate update, as SB-C-CALL is now a (deprecated) nickname
1732 ** Old operator names in the style DEF-FOO are now deprecated in
1733 favor of new corresponding names DEFINE-FOO, for consistency with
1734 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1735 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1736 internal symbols, but a few supported extensions like
1737 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1738 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1739 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1740 etc. as you get deeper into recursive calls to the debugger
1741 command loop, instead of the old "5]", "5]]", "5]]]"
1742 sequence. (I was motivated to do this when squabbles between
1743 ILISP and SBCL left me very deeply nested in the debugger. In the
1744 short term, this change will probably provoke more ILISP/SBCL
1745 squabbles, but hopefully it will be an improvement in the long run.)
1746 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1747 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1748 function, since declaiming return types involving aliens is
1749 (1) annoyingly messy to do by hand and (2) vital to efficient
1750 compilation of code which calls such functions.
1751 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1752 longer reexported by the SB-EXT package. They're solely useful
1753 for alien code, so it seems more logical that you should get
1754 them from the SB-ALIEN package, not in SB-EXT.
1755 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1756 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1757 features. Instead, the code that they used to control is always
1758 built into the system.
1759 * many other bug fixes
1760 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1761 advantage of the new EVAL-WHEN stuff and to clean them up in
1762 general, and they are now more ANSI-compliant in a number of
1763 ways. Martin Atzmueller is responsible for a lot of this.
1764 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1766 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1767 *** correct ERROR type for various file operations
1768 *** some fixes for Lisp streams
1769 *** DEFMETHOD syntax checking
1770 *** changing old weird representation of debug information as
1771 strings (which, among their other deficiencies, don't transform
1772 correctly when you rename packages, and don't change their
1773 print representation when you change things like *PACKAGE*
1774 and *PRINT-LENGTH*) to symbols and lists of symbols
1775 He also made several improvements and fixed several bugs in DESCRIBE.
1776 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1778 *** misbehavior of WRITE-STRING/WRITE-LINE
1779 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1780 and several other LOOP problems as well
1781 *** DIRECTORY when similar filenames are present
1782 *** DEFGENERIC with :METHOD options
1783 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1784 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1785 *** argument ordering in FIND with :TEST option
1786 *** mishandled package designator argument in APROPOS-LIST
1787 *** various problems in the backquote readmacro
1788 *** a bug in APROPOS
1789 *** probably some others that I'm not describing very well here,
1790 since the CVS log documents them by reference to sbcl-devel
1791 messages, and the SourceForge archives aren't working well.:-(
1792 ** Dan Barlow improved the Alpha port (and is making progress on the
1793 PPC port, for those of you who think different).
1794 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
1795 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
1796 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
1797 continues to work on the SPARC port (for those of you in a position
1798 to look down upon our little PC-compatible boxes from a great height).
1799 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
1800 and uses that copy, rather than the current dynamic value, when
1801 it's trying to decide whether to truncate output. Thus e.g.
1802 (let ((*print-lines* 50))
1803 (pprint-logical-block (stream nil)
1805 (let ((*print-lines* 8))
1806 (print (aref possiblybigthings i) stream)))))
1807 should now truncate the logical block only at 50 lines, instead of
1808 often truncating it at 8 lines, as it did before.
1809 * The doc/cmucl/ directory, containing old CMU CL documentation
1810 from the time of the fork, is no longer part of the base system.
1811 SourceForge has shut down its anonymous FTP service, and with it
1812 my original plan for distributing the old CMU CL documentation
1813 there. For now, if you need these files you can download an old
1814 SBCL source release and extract them from it.
1815 * The fasl file version number changed again, for dozens of reasons,
1816 some of which are apparent above.
1818 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
1819 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
1820 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
1821 patch, so that DISASSEMBLE output is much nicer.
1822 * The code in the SB-PROFILE package now seems reasonably stable.
1823 I still haven't decided what the final interface should look like
1824 (I'd like PROFILE to interact cleanly with TRACE, since both
1825 facilities use function encapsulation) but if you have a need
1826 for profiling now, you can probably use it successfully with
1827 the current CMU-CL-style interface.
1828 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
1829 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
1830 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
1831 link now returns the dangling link itself, and for similar
1832 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
1833 link itself. (In these cases the old code signalled an error and
1834 looped endlessly, respectively.) Thus, DIRECTORY now works even
1835 in the presence of dangling and cyclic symbolic links.
1836 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
1837 is now a supported extension again, since the consensus on
1838 sbcl-devel was that it can be useful for ordinary development
1839 work, not just for debugging SBCL itself.
1840 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
1841 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
1842 that currently-defined functions might be redefined later with
1843 different return types.
1844 * Hash tables can be printed readably, as inspired by CMU CL code
1845 of Eric Marsden and SBCL code of Martin Atzmueller.
1846 * better error handling in CLOS method combination, thanks to
1847 Martin Atzmueller porting Pierre Mai's CMU CL patches
1848 * more overflow fixes for >16Mbyte I/O buffers
1849 * A bug in READ has been fixed, so that now a single Ctrl-D
1850 character suffices to cause end-of-file on character streams.
1851 In particular, now you only need one Ctrl-D at the command
1852 line (not two) to exit SBCL.
1853 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
1855 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
1856 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
1857 does the right thing.
1858 * The implementation of some type tests, especially for CONDITION
1859 types, is now tidier and maybe faster, due to CMU CL code
1860 originally by Douglas Crosher, ported by Martin Atzmueller.
1861 * Some math functions have been fixed, and there are new
1862 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
1863 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
1864 * (There are also some new optimizers in contrib/*-extras.lisp. Those
1865 aren't built into sbcl-0.6.13, but are a sneak preview of what's
1866 likely to be built into sbcl-0.7.0.)
1867 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
1868 CL, ported to SBCL by Martin Atzmueller)
1869 * DESCRIBE now gives more information in some cases. (Pierre Mai's
1870 patch to CMU CL, ported to SBCL by Martin Atzmueller)
1871 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
1872 * There's a new slam.sh hack to shorten the edit/compile/debug
1873 cycle for low-level changes to SBCL itself, and a new
1874 :SB-AFTER-XC-CORE target feature to control the generation of
1875 the after-xc.core file needed by slam.sh.
1876 * minor incompatible change: The ENTRY-POINTS &KEY argument to
1877 COMPILE-FILE is no longer supported, so that now every function
1878 gets an entry point, so that block compilation looks a little
1879 more like the plain vanilla ANSI section 3.2.2.3 scheme.
1880 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
1881 returns the number of bytes consed since the system started,
1882 rather than the number consed since the first time the function
1883 was called. (The new definition parallels ANSI functions like
1884 CL:GET-INTERNAL-RUN-TIME.)
1885 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
1886 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
1887 supported. Now DIRECTORY always does the abstract Common-Lisp-y
1888 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
1889 * Fasl file version numbers are now independent of the target CPU,
1890 since historically most system changes which required version
1891 number changes have affected all CPUs equally. Similarly,
1892 the byte fasl file version is now equal to the ordinary
1895 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
1896 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
1897 is no longer recognized. I apologize for this, because it was
1898 listed in SB-EXT as a supported extension, but I found that
1899 its existing behavior was poorly specified, as well as incorrectly
1900 specified, and it looked like too much of a mess to straighten it
1901 out. I have enough on my hands trying to get ANSI stuff to work..
1902 * many patches ported from CMU CL by Martin Atzmueller, with
1903 half a dozen bug fixes in pretty-printing and the debugger, and
1904 half a dozen others elsewhere
1905 * fixed bug 13: Floating point infinities are now supported again.
1906 They might still be a little bit flaky, but thanks to bug reports
1907 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
1908 as flaky as they were.
1909 * The --noprogrammer command line option is now supported. (Its
1910 behavior is slightly different in detail from what the old man
1911 page claimed it would do, but it's still appropriate under the
1912 same circumstances that the man page talks about.)
1913 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
1914 are now supported, and enabled by default. Thus, the compiler can
1915 handle many floating point and complex operations much less
1916 inefficiently. (Thus e.g. you can implement a complex FFT
1918 * The compiler now detects type mismatches between DECLAIM FTYPE
1919 and DEFUN better, and implements CHECK-TYPE more correctly, and
1920 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
1921 thanks to patches from Martin Atzmueller.
1922 * various fixes to make the cross-compiler more portable to
1923 ANSI-conforming-but-different cross-compilation hosts (notably
1924 Lispworks for Windows, following bug reports from Arthur Lemmens)
1925 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
1926 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
1927 fixed thanks to Pierre Mai's CMU CL patches.
1928 * fixes to keep the system from overflowing internal counters when
1929 it tries to use i/o buffers larger than 16M bytes
1930 * fixed bug 45a: Various internal functions required to support
1931 complex special functions have been merged from CMU CL sources.
1932 (When I was first setting up SBCL, I misunderstood a compile-time
1933 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
1934 * improved support for type intersection and union, fixing bug 12
1935 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
1936 more obscure bugs as well
1937 * some steps toward byte-compiling non-performance-critical
1938 parts of the system, courtesy of patches from Martin Atzmueller
1939 * Christophe Rhodes has made some debian packages of sbcl at
1940 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
1941 From his sbcl-devel e-mail of 2001-04-08 they're not completely
1942 stable, but are nonetheless usable. When he's ready, I'd be happy
1943 to add them to the SourceForge "File Releases" section. (And if
1944 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
1945 * new fasl file format version number (because of changes in
1946 internal representation of (OR ..) types to accommodate the new
1947 support for (AND ..) types, among other things)
1949 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
1950 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
1952 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
1954 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
1955 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
1956 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
1957 * DESCRIBE now works on structure objects again.
1958 * Most function call argument type mismatches are now handled as
1959 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
1960 know whether the function will be redefined before the call is
1961 executed. (The compiler could flag local calls with full WARNINGs,
1962 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
1963 it doesn't keep track of enough information to know whether calls
1964 are local in this sense.)
1965 * Compiler output is now more verbose, with messages truncated
1966 later than before. (There should be some supported way for users
1967 to override the default verbosity, but I haven't decided how to
1968 provide it yet, so this behavior is still controlled by the internal
1969 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
1970 src/compiler/ir1util.lisp.)
1971 * Fasl file format version numbers have increased again, because
1972 support for the Gray streams extension changes the layout of the
1973 system's STREAM objects.
1974 * The Gray subclassable streams extension now works, thanks to a
1975 patch from Martin Atzmueller.
1976 * The full LOAD-FOREIGN extension (not just the primitive
1977 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
1978 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
1979 but like most other programs, it defaults to copying the Unix
1980 environment from the original process instead of starting the
1981 new process in an empty environment.
1982 * Extensions which manipulate the Unix environment now support
1983 an :ENVIRONMENT keyword option which doesn't smash case or
1984 do other bad things. The CMU-CL-style :ENV option is retained
1985 for porting convenience.
1986 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
1987 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
1989 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
1991 * A patch from Martin Atzmueller seems to have solved the SIGINT
1992 problem, and as far as we know, signal-handling now works cleanly.
1993 (If you find any new bugs, please report them!)
1994 * The system no longer defaults Lisp source file names to types
1995 ".l", ".cl", or ".lsp", but only to ".lisp".
1996 * The compiler no longer uses special default file extensions for
1997 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
1998 seems to expect a single default extension for all compiled code,
1999 and there's no compelling reason to try to stretch the standard
2000 to allow two different extensions.) Instead, byte-compiled files
2001 default to the same extension as native-compiled files.
2002 * Fasl file format version numbers have increased again, because
2003 a rearrangement of internal implementation packages made some
2004 dumped symbols in old fasl files unreadable in new cores.
2005 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2006 many fewer weird special cases.
2007 * Bug #17 (differing COMPILE-FILE behavior between logical and
2008 physical pathnames) has been fixed, and some related misbehavior too,
2009 thanks to a patch from Martin Atzmueller.
2010 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2011 by Tim Moore, ported to SBCL by Martin Atzmueller.
2012 * Martin Atzmueller fixed several filesystem-related problems,
2013 including bug #36, in part by porting CMU CL patches, which were
2014 written in part by Paul Werkowski.
2015 * More compiler warnings in src/runtime/ are gone, thanks to
2016 more patches from Martin Atzmueller.
2017 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2020 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2022 * DESCRIBE now works on CONDITION objects.
2023 * The debugger now handles errors which arise when trying to print
2024 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2026 * The build system now uses an additional file, customize-target-features.lisp,
2027 to allow local modifications to the target *FEATURES* list. (The point of
2028 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2029 debugging features enabled, without having to worry about propagating it
2030 into everyone's system when I do a "cvs update".) When no
2031 customize-target-features.lisp file exists, the target *FEATURES* list
2032 should be constructed the same way as before.
2033 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2034 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2035 warn about behavior which is undefined under the ANSI spec. Note: This
2036 is specified by ANSI, but it's not very popular with programmers.
2037 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2038 macro in the SBCL sources for an example of a workaround which you
2039 might use to make portable ANSI-standard code which does what you want.
2040 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2041 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2042 when it's used as a non-toplevel form. (This is required in order
2043 to implement the DEFCONSTANT-EQX macro.)
2044 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2045 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2046 funny things, due to bugs in the implementation of EVAL-WHEN
2047 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2048 non-toplevel DEFCONSTANTs are uncommon.)
2049 * The core file version number and fasl file version number have been
2050 incremented, because the old noncompliant DEFCONSTANT behavior involved
2051 calling functions which no longer exist, and because I also took the
2052 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2053 * fixed bug 1 (error handling before read-eval-print loop starts), and
2054 redid debugger restarts and related debugger commands somewhat while
2056 ** The QUIT debugger command is gone, since it did something
2057 rather different than the SB-EXT:QUIT command, and since it never
2058 worked properly outside the main toplevel read/eval/print loop.
2059 Invoking the new TOPLEVEL restart provides the same functionality.
2060 ** The GO debugger command is also gone, since you can just invoke
2061 the CONTINUE restart directly instead.
2062 ** The TOP debugger command is also gone, since it's redundant with the
2063 FRAME 0 command, and since it interfered with abbreviations for the
2065 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2067 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2068 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2069 some porting work by Martin Atzmueller)
2070 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2071 with a boatload of patches which he ported to SBCL. Now that those
2073 ** The system tries to make sure that its low-priority messages
2074 are prefixed by semicolons, to help people who like to use
2075 syntax highlighting in their ILISP buffer. (This patch
2076 was originally due to Raymond Toy.)
2077 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2078 patches originally written by Raymond Toy.
2079 ** The compiler understands coercion between single-value and
2080 multiple-VALUES type expressions better, getting rid of some very
2081 weird behavior, thanks to patches originally by Robert MacLachlan
2082 and Douglas Crosher.
2083 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2084 lambda lists, thanks to a patch originally by Pierre Mai.
2085 ** The system no longer bogusly warns about "abbreviated type
2087 ** The compiler gets less confused by inlining and RETURN-FROM,
2088 thanks to some patches originally by Tim Moore.
2089 ** The system no longer hangs when dumping circular lists to fasl
2090 files, thanks to a patch originally from Douglas Crosher.
2091 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2092 undefined function error.
2093 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2094 in structure slot initforms) for now, documented workaround instead:-|
2095 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2096 * fixed bug 5: added stubs for various Gray stream functions called
2097 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2098 installed, at least appropriate type errors are generated
2099 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2100 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2101 consistently in DEFMETHOD forms.
2102 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2103 it doesn't seem to affect SBCL after all
2104 * The C runtime system now builds with better optimization and many
2105 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2107 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2109 * The system is now under CVS at SourceForge (instead of the
2110 CVS repository on my home machine).
2111 * The new signal handling code has been tweaked to treat register
2112 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2113 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2114 downstream, e.g. in debug-int.lisp, has implicit dependencies
2115 on the unsignedness of integer representation of machine words,
2116 and that caused the system to bomb out with infinite regress
2117 when trying to recover from type errors involving signed values,
2118 e.g. (BUTLAST '(1 2 3) -1).)
2119 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2120 (This was one of the bugs Peter Van Eynde reported back in July.)
2121 * The system now uses code inspired by Colin Walters' O(N)
2122 implementation of MAP (from the cmucl-imp@cons.org mailing
2123 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2124 inline the MAP operation, and there is more than one
2125 sequence argument to the MAP call (so that it can't just
2126 do ETYPECASE once and for all based on the type of the
2127 single sequence argument). (The old non-inline implementation
2128 of the general M-argument sequence-of-length-N case required
2129 O(M*N*N) time when any of the sequence arguments were LISTs.)
2130 * The QUIT :UNIX-CODE keyword argument has been renamed to
2131 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2132 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2134 * Raymond Wiker's patches to port dynamic loading from Linux to
2135 FreeBSD have been added.
2136 * The BUGS file is now more nearly up to date, thanks in large part
2137 to Martin Atzmueller's review of it.
2138 * The debugger now flushes standard output streams before it begins
2139 its output ("debugger invoked" and so forth).
2140 * The core version number and fasl file version number have both
2141 been incremented, because of incompatible changes in the layout
2143 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2144 used to not be called for a saved Lisp image.)
2145 * Martin Atzmueller's version of a patch to fix a compiler crash,
2146 as posted on sbcl-devel 13 September 2000, has been installed.
2147 * Instead of installing Martin Atzmueller's patch for the
2148 compiler transform for SUBSEQ, I deleted the compiler transform,
2149 and transforms for some similar consing operations.
2150 * A bug in signal handling which kept TRACE from working on OpenBSD
2152 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2153 away by constant folding
2154 * The system now defines its address space constants in one place
2155 (in the Lisp sources), and propagates them automatically elsewhere
2156 (through GENESIS and the sbcl.h file). Therefore, patching the
2157 address map is less unnecessarily tedious and error-prone. The
2158 Lisp names of address space constants have also been systematized.
2159 * CVS tags like dollar-Header-dollar have been removed from
2160 the sources, because they have never saved me trouble and
2161 they've been source of trouble working with patches and other
2162 diff-related operations.
2163 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2164 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2166 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2168 * The system has been ported to OpenBSD.
2169 * The system now compiles with a simple "sh make.sh" on the systems
2170 that it's supported on. I.e., now you no longer need to tweak
2171 text in the target-features.lisp-expr and symlinks in src/runtime/
2172 by hand, the make.sh takes care of it for you.
2173 * The system is no longer so grossly inefficient when compiling code
2174 involving vectors implemented as general (not simple) vectors (VECTOR T),
2175 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2176 which dares to use the various sequence functions on non-simple
2177 vectors, takes less of a performance hit.
2178 * There is now a primitive type predicate VECTOR-T-P
2179 to test for the (VECTOR T) type, so that e.g.
2180 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2181 can now be compiled with some semblance of efficiency. (The old code
2182 turned the type declaration into a full call to %TYPEP at runtime!)
2183 * AREF on (VECTOR T) is still not fast, since it's still compiled
2184 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2185 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2186 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2187 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2188 without making full calls to SUBTYPEP at runtime.
2189 (Some analogous efficiency issues for non-simple vectors specialized to
2190 element types other than T, or for non-simple multidimensional arrays,
2191 have not been addressed. They could almost certainly be handled the
2192 same way if anyone is motivated to do so.)
2193 * The changes in array handling break binary compatibility, so
2194 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2195 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2196 instead of (VALUES T T).
2197 * By following the instructions that Dan Barlow posted to sbcl-devel
2198 on 2 July 2000, I was able to enable primitive dynamic object
2199 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2200 functionality is not implemented (since it calls ld to resolve
2201 library references automatically, requiring RUN-PROGRAM for its
2202 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2203 try to resolve library references) is now supported.
2204 * The system now flushes the standard output streams when it terminates,
2205 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2206 them at several other probably-convenient times, e.g. in each pass of
2207 the toplevel read-eval-print loop, and after evaluating a form given
2208 as an "--eval" command-line option. (These changes were motivated by a
2209 discussion of stream flushing issues on cmucl-imp in August 2000.)
2210 * The source transform for TYPEP of array types no longer assumes
2211 that an array whose element type is a not-yet-defined type
2212 is implemented as an array of T, but instead punts, so that the
2213 type will be interpreted at runtime.
2214 * There is now some support for cross-compiling in make.sh: each of
2215 the phases of make.sh has its own script. (This should be transparent
2216 to people doing ordinary, non-cross-compile builds.)
2217 * Since my laptop doesn't have hundreds of megabytes of memory like
2218 my desktop machine, I became more motivated to do some items on
2219 my to-do list in order to reduce the size of the system a little:
2220 ** Arrange for various needed-only-at-cold-init things to be
2221 uninterned after cold init. To support this, those things have
2222 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2223 symbols with such names are now uninterned after cold init).
2224 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2225 fasl files for cold load.
2226 ** Remove the old compiler structure pooling code (which used to
2227 be conditional on the target feature :SB-ALLOC) completely.
2228 ** Redo the representation of some data in cold init to be more compact.
2229 (I also looked into supporting byte compiled code at bootstrap time,
2230 which would probably reduce the size of the system a lot, but that
2231 looked too complicated, so I punted for now.)
2232 * The maximum signal nesting depth in the src/runtime/ support code has
2233 been reduced from 4096 to 256. (I don't know any reason for the very
2234 large old value. If the new smaller value turns out to break something,
2235 I'll probably just bump it back up.)
2236 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2238 * Many, many bugs reported by Peter Van Eynde have been added to
2239 the BUGS list; some have even been fixed.
2240 * While enabling dynamic object file loading, I tried to make the
2241 code easier to understand, renaming various functions and variables
2242 with less ambiguous names, and changing some function calling
2243 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2244 * While trying to figure out how to do the OpenBSD port, I tried to
2245 clean up some of the code in src/runtime/. In particular, I dropped
2246 support for non-POSIX signal handling, added various comments,
2247 tweaked the code to reduce the number of compilation warnings, and
2248 renamed some files to increase consistency.
2249 * To support the new automatic configuration functionality in make.sh,
2250 the source file target-features.lisp-expr has been replaced with the
2251 source file base-target-features.lisp-expr and the machine-generated
2252 file local-target-features.lisp-expr.
2253 * fixed a stupid quoting error in make.sh so that using CMU CL
2254 "lisp -batch" as cross-compilation host works again
2256 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2258 * DESCRIBE no longer tries to call itself recursively to describe
2259 bound/fbound values, so that it no longer fails on symbols which are
2260 bound to themselves (like keywords, T, and NIL).
2261 * DESCRIBE now works on generic functions.
2262 * The printer now prints less-screwed-up representations of closures
2263 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2264 * A private symbol is used instead of the :EMPTY keyword previously
2265 used to mark empty slots in hash tables. Thus
2266 (DEFVAR *HT* (MAKE-HASH-TABLE))
2267 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2268 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2269 now does what ANSI says that it should. (You can still get
2270 similar noncompliant behavior if bang on the hash table
2271 implementation with all the symbols you get back from
2272 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2273 This breaks binary compatibility, since tests for equality to
2274 :EMPTY are wired into things like the macroexpansion of
2275 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2277 * There's now a minimal placeholder implementation for CL:STEP,
2278 as required by ANSI.
2279 * An obscure bug in the interaction of the normal compiler, the byte
2280 compiler, inlining, and structure predicates has been patched
2281 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2282 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2284 * Missing ordinary arguments in a macro call are now detected even
2285 when the macro lambda list contains &KEY or &REST.
2286 * The debugger no longer complains about encountering the top of the
2287 stack when you type "FRAME 0" to explicitly instruct it to go to
2288 the top of the stack. And it now prints the frame you request even
2289 if it's the current frame (instead of saying "You are here.").
2290 * As specified by ANSI, the system now always prints keywords
2291 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2293 * The default initial SIZE of HASH-TABLEs is now smaller.
2294 * Type information from CLOS class dispatch is now propagated
2295 into DEFMETHOD bodies, so that e.g.
2296 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2298 is now basically equivalent to
2299 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2300 (DECLARE (TYPE SINGLE-FLOAT X))
2302 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2303 operation, without having to do run-time type dispatch.
2304 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2305 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2306 * Since I don't seem to be making big file reorganizations very often
2307 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2308 is overflowing my ability to conveniently back them up), I've finally
2309 checked the system into CVS. (The CVS repository is on my home system,
2310 not at SourceForge -- putting it on SourceForge might come later.)
2311 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2312 high-level GC-NOTIFY-FOO functions send their output. (There's
2313 still very little control of where low-level verbose GC functions
2314 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2315 less than it used to -- the GC-NOTIFY-FOO functions are now under
2316 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2317 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2318 in only one place in the source code, and propagates it automatically
2319 everywhere that it's needed. Thus e.g. when I bump the version from
2320 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2321 * The C source files now include boilerplate legalese and documentation
2322 at the head of each file (just as the Lisp source files already did).
2323 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2324 to his page will be replaced with a link to his new CLiki service.
2326 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2328 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2329 * The build process now looks for GNU make under the default name "gmake",
2330 instead of "make" as it used to. If GNU make is not available as "gmake"
2331 on your system, you can change this default behavior by setting the
2332 GNUMAKE environment variable.
2333 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2334 can build without error under CMU CL.
2336 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2338 * There is now a partial SBCL user manual (with some new text and some
2339 text cribbed from the CMU CL manual).
2340 * The beginnings of a profiler have been added (starting with the
2341 CMU CL profiler and simplifying and cleaning up). Eventually the
2342 main interface should be through the TRACE macro, but for now,
2343 it's still accessed through vaguely CMU-CL-style functions and macros
2344 exported from the package SB-PROFILE.
2345 * Some problems left over from porting CMU CL to the new
2346 cross-compilation bootstrap process have been cleaned up:
2347 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2348 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2349 part when getting rid of the compiler *BACKEND* stuff.)
2350 ** The value of *NULL-TYPE* was screwed up, because it was
2351 being initialized before the type system knew the final
2352 definition of the 'NULL type. This screwed up several key
2353 optimizations in the compiler, causing inefficiency in all sorts
2354 of places. (I found it because I wanted to understand why
2355 GET-INTERNAL-RUN-TIME was consing.)
2356 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2357 PROCLAIM FTYPE information. Unfortunately this broke binary
2358 compatibility again, since now the forms output by DEFGENERIC
2359 to refer to functions which didn't exist in 0.6.3.
2360 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2361 can use the new (as of 0.6.3) transform for SOME into MAP into
2363 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2364 values to (INTEGER 0 1000000), so that the time code will no longer
2365 occasionally get blown up by Linux returning 1000000 microseconds
2366 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2367 its output conform to the ANSI spec. (Alas, this makes its output
2368 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2369 * A full call to MAP NIL with a single sequence argument no longer conses.
2370 * fixes to problems pointed out by Martin Atzmueller:
2371 * The manual page no longer talks about multiprocessing as though
2372 it were currently supported.
2373 * The ILISP support patches have been removed from the distribution,
2374 because as of version 5.10.1, ILISP now supports SBCL without us
2375 having to maintain patches.
2376 * added a modified version of Raymond Toy's recent CMU CL patch for
2377 EQUALP comparison of HASH-TABLE
2379 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2381 * The system still can't cross-compile itself with
2382 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2383 entails), but at least it can compile more of itself that way
2384 than it used to be able to, and various buglets which were uncovered
2385 by trying to cross-compile itself that way have now been fixed.
2386 * This release breaks binary compatibility again. This time
2387 at least I've incremented the FASL file format version to 2, so that the
2388 problem can be detected reliably instead of just causing weird errors.
2389 * various new style warnings:
2390 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2391 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2392 ** using the lexical binding of a variable named in the *FOO* style
2393 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2394 as specified by ANSI.
2395 * *RANDOM-STATE* is no longer automatically initialized from
2396 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2397 default behavior of the system is to repeat its behavior every time
2398 it's run. If you'd like to change this behavior, you can always
2399 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2400 old convention there was no comparably easy way to get the system to
2401 repeat its behavior every time it was run.
2402 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2403 FTYPE declarations has been removed, in favor of their ANSI
2404 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2405 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2406 the types of their sequence arguments can be determined at compile time.
2407 This is done through a new open code expansion for MAP which eliminates
2408 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2409 argument types can be determined at compile time.
2410 * The optimizer now transforms COERCE into an identity operation when it
2411 can prove that the coerced object is already of the correct type. (This
2412 can be a win for machine generated code, including the output of other
2413 optimization transforms, such as the MAP transform above.)
2414 * Credit information has been moved from source file headers into CREDITS.
2415 * Source file headers have been made more standard.
2416 * The CASE macro now compiles without complaining even when it has
2419 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2421 * (Note that the way that the PCL macroexpansions were rewritten
2422 to accommodate the change in DEFGENERIC below breaks binary
2423 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2424 not run under sbcl-0.6.2. Once we get out of alpha releases,
2425 i.e. hit release 1.0.0, we'll probably try to maintain binary
2426 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2427 and sbcl-1.4.4. Until then, however, it might be fairly common
2428 for maintenance releases to break binary compatibility.)
2429 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2431 * The reporting of unhandled signals has been changed to print some
2432 explanatory text as well as the report form. (Previously only
2433 the report form was printed.)
2434 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2435 it defines, so that the compiler no longer issues undefined function
2436 warnings for compiled-but-not-yet-loaded generic functions.
2437 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2438 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2439 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2440 section of the ANSI spec are supported.
2441 * The "" nickname for the "KEYWORD" package has been removed.
2442 The reader still handles symbol tokens which begin with a package marker
2443 as keywords, but it doesn't expose its mechanism for doing so in the
2444 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2445 * The system now issues STYLE-WARNINGs for contradictory TYPE
2446 proclamations. (Warnings for contradictory FTYPE proclamations would
2447 be nice too, but those can't be done usefully unless the type system
2448 is made smarter about FUNCTION types.)
2449 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2450 systematized, so that "*target-*.lisp is supposed to exist only on the
2451 target and imply that there's a related file which exists on the
2452 host, and *host-*.lisp is supposed to exist only on the host and imply
2453 that there's a related file which exists on the target. This involves a
2454 lot of renaming. Hopefully the acute confusion caused by the renaming
2455 will be justified by the reduction in chronic confusion..
2456 ** runtime-type.lisp -> early-target-type.lisp
2457 ** target-type.lisp -> late-target-type.lisp
2458 ** early-host-format.lisp -> early-format.lisp
2459 ** late-host-format.lisp -> late-format.lisp
2460 ** host-error.lisp -> misc-error.lisp
2461 ** early-error.lisp -> early-target-error.lisp
2462 ** late-error.lisp -> late-target-error.lisp
2463 ** host-defboot.lisp -> early-defboot.lisp
2464 ** code/misc.lisp -> code/target-misc.lisp
2465 ** code/host-misc.lisp -> code/misc.lisp
2466 ** code/numbers.lisp -> code/target-numbers.lisp
2467 ** code/early-numbers.lisp -> numbers.lisp
2468 ** early-host-type.lisp -> early-type.lisp
2469 ** late-host-type.lisp -> late-type.lisp
2470 ** host-typep.lisp -> typep.lisp
2471 ** load.lisp -> target-load.lisp
2472 ** host-load.lisp -> load.lisp
2473 ** host-disassem.lisp -> disassem.lisp
2474 ** host-insts.lisp -> insts.lisp
2475 ** byte-comp.lisp -> target-byte-comp.lisp
2476 ** host-byte-comp.lisp -> byte-comp.lisp
2477 ** host-signal.lisp -> signal.lisp
2478 ** host-defstruct.lisp -> defstruct.lisp
2479 ** late-target-type.lisp -> deftypes-for-target.lisp
2480 Furthermore, several other previously target-only files foo.lisp (e.g.
2481 hash-table.lisp and random.lisp) have been split into a target-and-host
2482 foo.lisp file and a target-only target-foo.lisp file, with their key type
2483 definitions in the target-and-host part, so that the cross-compiler will
2484 know more about target types.
2485 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2486 gone away. In their place are various *BACKEND-FOO* variables
2487 corresponding to the slots of the old structure.
2488 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2489 into the target SBCL has been fixed.
2490 * The chill.lisp system for loading cold code into a running SBCL
2492 * Support for the CMU CL "scavenger hook" extension has been removed.
2493 (It was undocumented and unused in the CMU CL sources that SBCL was
2494 derived from, and stale in sbcl-0.6.1.)
2495 * Various errors in the cross-compiler type system were detected
2496 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2497 (enabling various consistency checks). Many of them were fixed,
2498 but some hard problems remain, so the compiler is back to
2499 running without *TYPE-SYSTEM-INITIALIZED* for now.
2500 * As part of the cross-compiler type system cleanup, I implemented
2501 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2502 * I have started adding UNCROSS calls throughout the type system
2503 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2504 on cross-compiler input files will be able to go away, and various
2506 * CONSTANTP now returns true for quoted forms (as explicitly required
2509 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2511 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2512 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2513 (MAKE-LIST -1). In the longer term, it ought to become true
2514 that declarations are assertions even at SAFETY 1. For now, it's not
2515 quite true even at SAFETY 3, but it's at least more nearly true..
2516 (Note that this change seems to increases the size of the system by
2517 O(5%) and to decrease the speed of the compiler by 20% or more.)
2518 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2519 for the problem of printing dozens of lines of distracting information
2520 about low-level system machinery as part of the top stack frame
2521 on entry to the debugger when an undefined function was called.
2522 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2523 is not reset to COMMON-LISP-USER.
2524 * Compilation of stuff related to dyncount.lisp has been made conditional
2525 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2526 smaller. The various dyncount-related symbols have been moved into
2527 a new "SB-DYNCOUNT" package.
2528 * tty-inspect.lisp has been renamed to inspect.lisp.
2529 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2530 feature has gone away. (When we eventually port to other flavors of
2531 libc and/or Unix, we'll try to make the differences between flavors
2532 invisible at the user level.)
2533 * Various other *FEATURES* tags, and/or their associated conditionals,
2534 have been removed if obsolescent, or given better documentation, or
2535 sometimes given more-mnemonic names.
2537 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2539 * tidied up "make.sh" script
2540 * tidied up system directory structure
2541 * better "clean.sh" behavior
2542 * added doc/FOR-CMUCL-DEVELOPERS
2543 * many many small tweaks to output format, e.g. removing possibly-confusing
2544 trailing #\. character in DESCRIBE-INSTANCE
2545 * (EQUALP #\A 'A) no longer signals an error.
2546 * new hashing code, including EQUALP hashing
2547 * tidied up Lisp initialization and toplevel
2548 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2549 * command line argument processing
2550 * added POSIX-GETENV function to deal with Unix-ish environment variables
2551 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2552 terminating SBCL on EOF
2553 * non-verbose GC by default
2554 * There is no more "sbcl" shell script; the sbcl file is now the C
2555 runtime executable (just like CMU CL).
2556 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2558 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2559 symbol and package information as Lisp native symbol and package objects
2560 instead of strings naming symbols and strings naming packages. This way,
2561 whenever packages are renamed (as in warm init), debug information is
2562 transformed along with everything else.
2563 * tweaked the optimization policy declarations which control the building
2564 of SBCL itself. Now, among other things, the system no longer saves
2565 source location debugging information. (This helps two problems at once
2566 by reducing SBCL size and by keeping SBCL from trying to look for its
2567 sources -- which may not exist -- when reporting errors.)
2568 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2569 debugging and testing purposes
2570 * cleaned up printing, making the printer call PRINT-OBJECT for
2571 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2572 methods, giving nearly-ANSI behavior
2573 * converted almost all special variables to use *FOO* naming convention
2574 * deleted PARSE-TIME functionality, since it can be done portably
2575 * moved some files out of cold init into warm init
2576 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2578 * regularized formatting of source files
2579 * added an install.sh script
2580 * fixed ridiculous memory usage of cross-compiler by making
2581 compiler/alloc.lisp not try to do pooling unless it can hook
2582 itself into the GC of the cross-compilation host. Now the system
2583 builds nicely on my old laptop.
2584 * added :SB-ALLOC in target-features.lisp-expr
2585 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2586 was not implemented)
2587 * re-did condition handling and note reporting in the compiler. Notes
2588 are no longer handled by signalling conditions. Style warnings
2589 and warnings are handled more correctly and reported in such a way
2590 that it's easy to find one or the other in your output (so that you
2591 can e.g. figure out which of many problems caused COMPILE-FILE to
2593 * changed the severity of several compiler warnings from full WARNING
2594 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2595 compiler note reporting so that it doesn't use the condition system
2596 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2597 COMPILE-FILE command)
2598 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2599 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2600 no longer supported, since their implementation was deeply intertwingled
2601 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2602 * removed bogus "support" for compiler macros named (SETF FOO), and
2603 removed the compiler macro for SETF INFO (but only after making a fool
2604 of myself on the cmucl-imp mailing list by posting a bogus patch for
2605 DEFINE-COMPILER-MACRO..)
2606 * Compiled files containing forms which have side effects on the Lisp
2607 reader (such as DEFPACKAGE forms) are now handled more correctly.
2608 (Compiler queuing of top level lambdas has been suppressed by setting
2609 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2610 * deleted various currently-unused source files, e.g. gengc.lisp. They
2611 may be added back at some point e.g. when porting to other architectures,
2612 but until they are it's distracting to distribute them and to try to
2614 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2615 there were so many of them they're just distractions, and UNCROSS is
2616 known to be able to handle the current sources
2617 * moved PROFILE functionality into TRACE, so that it will be clear
2618 how the wrapping and unwrapping of functions when you profile them
2619 interacts with the wrapping and unwrapping of functions when you
2620 trace them. (Actually, the functionality isn't there yet, but at least
2621 the interface specification is there. Hopefully, the functionality will
2622 arrive with some future maintenance release.)
2623 * removed host-oops.lisp
2624 * changed signature of QUIT function to allow UNIX-CODE argument
2625 * fixed READ-SEQUENCE bug
2626 * tweaked verbose GC output so that it looks more like the progress
2627 output that ANSI specifies for functions like LOAD
2628 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2629 * added <http://sbcl.sourceforge.com> to the banner information printed by