1 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2 * incompatible change: the internal functions
3 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
4 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
5 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
6 instead of the old functions.
7 * new feature: on platforms where "dladdr" is available foreign
8 function names now appear in backtraces. (based on Helmut Eller's
10 * documentation: networking interface SB-BSD-SOCKETS' documentation
11 has been integrated into the user manual.
12 * bug fix: backtraces involving undefined functions or assembly
13 routines are more informative. (thanks to Brian Downing)
14 * bug fix: mutually referent alien structures now work correctly.
15 (reported by Rick Taube)
16 * bug fix: structures defined by WITH-ALIEN can be referred to
17 within other definitions in the same WITH-ALIEN.
18 * bug fix: division operators (MOD, TRUNCATE and the like) with
19 constant zero divisors and integer dividends no longer generate
21 * bug fix: provide default methods for INPUT-STREAM-P and
22 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
23 * optimization: in taking the GCD of bignums, reduce the two bignums
24 to approximately the same size (using Euclid's algorithm) before
25 applying the more sophisticated binary GCD. (thanks to Juho
28 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
29 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
30 the manual for details; for now, package locks can be disabled by
31 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
32 if no major problems are found then it is likely that they will be
33 compiled in unconditionally.
34 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
35 unsupported operators on all platforms. To load a shared library
36 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
37 object file, link it into a shared library outside of SBCL and
38 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
40 * bug workaround: ROOM T and the GC were not working together
41 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
42 reporting facility aren't preserved. That reporting has been
43 disabled, so now until and unless someone figures out how to make it
44 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
45 * minor incompatible change: as threatened around sbcl-0.8.0, the
46 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
47 The symbols are also exported from SB-PCL for backwards
48 compatibility, but more so than before SB-PCL should be treated as
49 an implementation-internal package.
50 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
52 * the SB-SPROF contrib now works on (most) non-x86 architectures.
53 It is known as of this release not to work on the Alpha, however.
54 * fixed bug #167: errors signalled due to illegal syntax in method
55 bodies are now more legible.
56 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
57 designators and can hence be used with TYPEP.
58 * fixed bug #333: CHECK-TYPE now ensures that the type error
59 signalled, if any, has the right object to be accessed by
60 TYPE-ERROR-DATUM. (reported by Tony Martinez)
61 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
62 5.1.2.3 for multiple-value place subforms. (reported by Kalle
64 * fixed bug #334: programmatic addition of slots using specialized
65 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
66 and :ALLOCATION :CLASS slots.
67 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
68 float exponents. (rereported by Peter Seibel)
69 * fixed a bug: #\Space (and other whitespace characters) are no
70 longer considered to be macro characters in standard syntax by
72 * fixed bug: initialization of condition class metaobjects no longer
73 causes an instance of the condition to be created. (reported by
75 * fixed bug: it is now possible to have slots such that
76 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
77 description is neither :INSTANCE nor :CLASS.
78 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
79 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
80 SLOT-VALUE-USING-CLASS), and no longer causes errors with
81 non-standard SLOT-DEFINITION-ALLOCATION values.
82 * fixed bugs: various race conditions handling exiting threads.
83 CL-PPCRE's thread torture test now passes.
84 * fixed arguably-a-bug: GC time proportional to number of threads in
85 system even when most of them are idle
86 * optimization: improved performance of BIT and SBIT on bit-vectors.
87 * .fasl file incompatibility: The fasl file version number has
88 been incremented because of changes associated with package locks.
90 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
91 * minor incompatible change: the system no longer provides
92 optimization hints (conditions which are TYPEP
93 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
94 mode; these hints are still emitted when higher SPEED optimization
96 * new contrib module: a sampling profiler (profiling by statistical
97 sampling, rather than by instrumenting functions) is available as
98 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
99 * the behaviour of the standard function ED is now customizeable by
100 third parties through a hook variable: see ED's documentation
101 string for information on the protocol.
102 * the compiler no longer emits efficiency notes for (FUNCALL X)
103 when the type of X is uncertain under default optimization
105 * fixed bug 276: mutating a binding of a specialized parameter to a
106 method to something that is not TYPEP the specializer is now
108 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
109 exhibit the required behaviour on the x86 platform. (thanks to
110 Peter van Eynde, Eric Marsden and Bruno Haible)
111 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
112 even for difficult arguments. (reported by Peter Graves)
113 * fixed bug 141a: the backquote printer now descends quoted
115 * fixed another bug in backquote printing: no more destructive
116 modification of the form's list structure. (reported by Brian
118 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
119 move between its address being taken and the call to
120 interrupt_thread, fixing a crashing race condition.
121 * the SB-POSIX contrib implementation has been adjusted so that it
122 no longer exhibits ridiculously poor performance when constructing
123 instances corresponding to C structs.
125 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
126 * minor incompatible change: the sb-grovel contrib now treats C
127 structures as alien (in the sense of SB-ALIEN) objects rather than
128 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
129 has implications for memory management of client code
130 (sb-grovel-returned objects must now be manually managed) and for
131 type safety (alien objects now have full types).
132 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
133 used to control emission of compiler diagnostics, rather than the
134 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
135 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
136 quality should be considered deprecated.
137 * install.sh now installs the user manual as well
138 * (not quite a new documentable feature, but worth considering in
139 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
140 beginnings of a semantically meaningful condition hierarchy is
141 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
143 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
144 designator as the defaults argument.
145 * fixed bug: Displaced arrays whose displaced-to array has become
146 too small now cause ARRAY-DIMENSION to signal an error, providing
147 for safer bounds-checking. (reported by Bruno Haible)
148 * fixed bug: DEFCLASS slot definitions with identical :READER and
149 :WRITER names now signal a reasonable error. (reported by Thomas
151 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
153 * fixed bug: Invalid slot specification errors now print correctly.
154 (thanks to Zach Beane)
155 * fixed bug 320: Shared to local slot value transfers in class
156 redefinitions now happen corrently from superclasses as
157 well. (reported by Bruno Haible)
158 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
160 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
161 improper lists correctly. (reported by Bruno Haible)
162 * fixed bug 313: source-transform for <fun-name> was erroneously
163 applied to a call of a value of a variable with name <fun-name>.
164 (reported by Antonio Menezes Leitao)
165 * fixed bug 307: The obsolete instance protocol ensures that
166 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
167 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
168 remove dead unknown-values globs from the middle of the stack.
169 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
170 easier to resume long computations after using *BREAK-ON-SIGNALS*
171 to diagnose and fix failures. (thanks to Nikodemus Siivola)
172 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
173 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
174 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
175 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
176 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
178 * fixed a (fairly theoretical) bug in string printing: if
179 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
180 does not have array-element-type equal to the most general string
182 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
183 function name. (thanks to Nikodemus Siivola)
184 * fixed bug: compile-time format string checker failed on
185 non-closed ~{. (reported by Thomas F Burdick)
186 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
187 don't warn on legal loop code involving more than one aggregate
188 boolean. (thanks to Nikodemus Siivola)
189 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
190 in keyword position to MAKE-PACKAGE was not being reported as
192 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
193 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
195 * fixed bugs 280 and 312: the checking for multiple definitions in a
196 file is less likely to become confused by uses of inline
198 * fixed bug: the #S reader macro performs the keyword coercion
199 specified for slot names. (reported by Kalle Niemitalo)
200 * fixed bug: lambda lists may contain symbols whose names start with
201 & but are not lambda-list-keywords; their occurrence triggers a
203 * fixed bug 321: define-method-combination argument lambda lists do
204 not cause invalid code to be generated when &optional or &aux
205 variables have default values. (reported by Bruno Haible)
206 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
207 have CLOS classes; this appears also to have fixed reported
208 instability in stack exhaustion detection.
209 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
210 MUFFLE-WARNING when no associated restart is present is now
212 * optimization: rearranged the expansion of various defining macros
213 so that each expands into only one top-level form in a
214 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
216 * optimization: used a previously little-used slot in symbols to
217 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
219 * fixed some bugs revealed by Paul Dietz' test suite:
220 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
222 ** Fixed an optimization bug related to inheritance of initforms
223 from local to shared slots.
224 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
225 ** CAST optimizer forgot to flush argument derived type.
226 ** print/read consistency on floats is now orders of magnitude
227 more likely. (thanks also to Bruno Haible for a similar report
229 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
230 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
231 for objects of type REAL. Make it so.
232 ** SXHASH is specified (infelicitously) to respect similarity,
233 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
234 it so. (thanks to Markus Ziegler)
235 ** on the Alpha, the compiler succeeds in compiling functions
236 returning a known number of arguments greater than 63.
237 ** fixed handling of invalid NIL arguments in keyword position
239 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
241 ** provided a readably-printable representation for RANDOM-STATE
243 ** ensured that pathnames, where they have a namestring, always
244 print using #P"..." syntax.
246 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
247 * Support for the forthcoming 2.0 version of the NetBSD kernel
248 running on x86 hardware has been added. (thanks to Perry
249 E. Metzger most immediately, and others for their past work)
250 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
251 current release version; SBCL's OpenBSD support had been broken
252 since about the time of OpenBSD's switch to ELF binary format.)
253 * [placeholder for DX summary]
254 ** user code with &REST lists declared dynamic-extent, under high
255 speed or space and low safety and debug optimization policy.
256 * The manual has been converted to Texinfo format and the debugger
257 chapter from the cmucl manual has been added.
258 * A facility has been added to extract documentation strings from
259 sbcl and store them as Texinfo-formatted snippets for inclusion in
260 the manual (via Texinfo's @include directive)
261 * bug fix: compiler emitted division in optimized DEREF. (thanks for
262 the test case to Dave Roberts)
263 * bug fix: multidimensional simple arrays loaded from FASLs had fill
264 pointers. (reported by Sean Ross)
265 * bug fix: PROFILE output is printed nicely even for large numerical
266 values. (thanks to Zach Beane)
267 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
268 greater than 32 handle EOF correctly.
269 * bug fix: on X86 an immediate argument of the IMUL instruction is
270 correctly printed in disassembly. (thanks to Lutz Euler)
271 * bug fix: class slots in redefined classes preserve their old
272 values. (thanks to Bruno Haible and Nikodemus Siivola)
273 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
274 succeeds. (reported by Marco Baringer)
275 * fixed some bugs revealed by Paul Dietz' test suite:
276 ** READ-SEQUENCE now works on ECHO-STREAMs.
277 ** RATIONALIZE works more according to its specification. (thanks
280 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
281 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
282 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
283 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
284 should matter to you only if you rebind the printer control
285 variables and then find you want different bindings in the
286 debugger than in the ordinary execution of your program.)
287 * The runtime build system has been tweaked to support building
288 (on SPARC/SunOS) using a C compiler which invokes Sun's own
289 assembler and linker. (thanks to Nikodemus Siivola)
290 * Unbound, undefined, undeclared variables now trigger full
291 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
292 more often programmer error than deliberate exploitation of undefined
294 * optimization: the hash algorithm for strings has changed to one
295 that is less vulnerable to spurious collisions. (thanks to Juho
297 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
298 needless bounds checking. (thanks to Juho Snellman)
299 * optimization: implemented multiplication as a modular
300 (UNSIGNED-BYTE 32) operation on the PPC backend.
301 * fixed some bugs revealed by Paul Dietz' test suite:
302 ** ADJUST-ARRAY now copies the datum in a zero rank array if
304 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
305 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
306 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
308 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
310 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
312 ** ECHO-STREAMs no longer attempt to echo the end of file value to
313 their output stream on EOF from read.
314 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
315 have been read to end-of-file.
316 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
318 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
319 description of determination of which consecutive characters
321 ** Printing the "Space" character with escaping on now yields "#\\ ",
322 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
323 ** Reading floating-point numbers with *READ-BASE* set to a number
324 less than 10 works correctly.
325 ** Reading floating-point numbers with *READ-BASE* set to a number
326 more than 10 works correctly.
327 ** Printing with *PRINT-READABLY* targets the standard readtable, not
328 the readtable currently in effect.
330 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
331 * minor incompatible change: parsing of namestrings on a physical
332 (Unix) host has changed; numbers after the final #\. in a
333 namestring are no longer interpreted as a version field. This is
334 intented to be largely invisible to the user, except that the
335 meaning of the namestring "*.*.*" has changed: it now refers to a
336 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
337 should usually be replaced by
338 (make-pathname :name :wild :type :wild :version :wild)
339 with the added benefit that this is more likely to be portable.
340 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
341 option to OPEN now signals an error if the file being opened
342 exists; this may have an impact on existing code.
343 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
344 signals received, on the PowerPC platforms (both Linux and
345 Darwin). (thanks to Pierre Mai for pointing out the location of
347 * several fixes on OS X: The system now builds and runs cleanly on
348 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
349 (thanks to Brian Mastenbrook)
350 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
351 second-resolution: integer multiples of 1/3600 between -24 and 24.
352 (thanks to Vincent Arkesteijn)
353 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
354 type when called with 1 argument; PEEK-CHAR checked type of
355 PEEK-TYPE only after having read first character from a
356 stream. (reported by Peter Graves)
357 * bug fix: the garbage collector now has much better locality
358 behaviour, and in particular no longer treats all memory as being
359 exhausted when many small objects point to each other in a deeply
361 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
362 recognized as being TYPEP their class.
363 * bug fix: the PUSHNEW documentation string has been corrected.
364 (thanks to Vincent Arkesteijn)
365 * bug fix: defaulting of the value for the last of an atypically
366 large number of multiple values being bound was not being
367 performed correctly on the Alpha or PPC platforms
368 * optimization: implemented multiplication as a modular
369 (UNSIGNED-BYTE 32) operation on the x86 backend.
370 * optimization: SEARCH on simple-base-strings can now be open-coded.
371 (see also contrib/compiler-extras.lisp for inspiration for
372 teaching the compiler about the Boyer-Moore algorithm).
373 * value, returned by MAX (and MIN) called with several EQUALP, but
374 not EQL, arguments now does not depend on compiler settings.
375 * fixed some bugs revealed by Paul Dietz' test suite:
376 ** in stack analysis liveness information is propagated from
377 non-local entry points.
378 ** pathwise CAST removing failed when the CAST node did not start
380 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
381 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
383 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
384 host is already defined.
385 ** RENAME-FILE works on streams instead of signalling an internal
387 ** PEEK-CHAR uses the current readtable when determining whether
388 or not a character is whitespace.
389 ** MERGE-PATHNAMES handles the case when the pathname does not
390 specify a name while the default-pathname specifies a version.
391 ** Pathnames now stand a better chance of respecting print/read
393 ** Attempting to use standardized file system operators with a
394 pathname with invalid :DIRECTORY components signals a
396 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
398 ** DIRECTORY on logical pathnames is more correct.
399 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
400 signal a TYPE-ERROR if their argument is not a stream.
401 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
402 designator argument does not designate a stream.
403 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
404 examining the synonym.
405 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
407 ** OPEN and WITH-OPEN-STREAM allow opening streams with
408 element-type larger than ([UN]SIGNED-BYTE 32).
410 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
411 * When built with the :SB-FUTEX feature, threaded builds now take
412 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
413 for faster/more reliable mutex and condition variable support.
414 * Incompatible change (but one you probably shouldn't have been using
415 anyway): the interface and code for arbitrating between multiple
416 threads in the same user session has been redesigned.
417 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
418 error when called without an explicit environment argument.
419 (thanks to Nikodemus Siivola)
420 * bug fix: buffered :DIRECTION :IO streams are less likely to become
421 confused about their position. (thanks to Adam Warner and Gerd
423 * bug fix: Pretty printing backquoted forms with unquotations in the
424 argument list position of various code constructs such as LAMBDA
425 now works correctly. (reported by Paul Dietz)
426 * bug fix: Pretty printing unquotations no longer loses all
427 stream position information.
428 * optimization: performance of string output streams is now less
429 poor for multiple small sequence writes.
430 * optimization: performance of CSUBTYPEP in the presence of complex
431 expressions involving CONS and NOT many times has been improved.
432 (reported by Paul Dietz)
433 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
435 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
436 streams for the REPL, for the debugger, and so on. (thanks to
438 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
439 * fixed some bugs revealed by Paul Dietz' test suite:
440 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
441 is ignored if it is too small, rather than propagating through
442 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
443 ** extremely complex negations of CONS types were not being
444 sufficiently canonicalized, leading to inconsistencies in
446 ** VALUES tranformer lost derived type.
448 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
449 * fixed a bootstrapping bug: the build process no longer assumes
450 that the various BOOLE-related constants have the same value in
451 host and target lisps. (noted by Paul Dietz' test suite on an
452 SBCL binary built from CLISP)
453 * The system can now be dynamically linked on the MIPS platform,
454 which enables dynamic loading of foreign code from Lisp. (thanks
455 to Ralf Baechle for discussions on the MIPS ABI)
456 * The system now records debugging information for its own source
457 files in a filesystem-position-independent manner, relative to
458 the "SYS" logical host.
459 * fixed a compiler bug: MV-LET convertion did not check references
460 to the "max args" entry point. (reported by Brian Downing)
461 * tweaked disassembly notes to be less confident about proclaiming
462 some instruction as an LRA. (thanks to Brian Downing)
463 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
464 now each have their own history, command character, and other
465 characteristics. (thanks to David Lichteblau)
466 * fixed some bugs revealed by Paul Dietz' test suite:
467 ** compiler failure in compiling LOGAND expressions including a
469 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
470 shift greater than 32.
471 ** FLUSH-DEST did not mark blocks for type check regeneration.
472 ** HANDLER-CASE failed to accept declarations in handler clauses
473 in some circumstances.
475 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
476 * New code in contrib/sb-introspect (still probably not entirely
477 stable yet) provides some support for smart Lisp development
478 environments like SLIME.
479 * The conditions signalled for errors occurring when loading .fasl
480 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
481 in a way which should help ASDF recover gracefully.
482 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
483 stuff has been cleaned up. If you code contrib/ stuff, this might
484 affect you, and you can look at contrib/README, contrib/STANDARDS,
485 and/or the 0.8.4.27 diff to check.
486 * In full calls the compiler now does not generate checks for declared
487 argument types for all arguments.
488 * various threading fixes
489 ** and some experimental patches which didn't make it into
490 the main tree for this release, but which are shipped in
491 contrib/experimental-thread.patch as a possible fix for some
492 failures (deadlock, spinning...) in GC-intensive multithreaded
494 * fixed PPC build problem (source code incompatibility of different
495 library versions): added offsetof() hackery which attempts to divine
496 where glibc maintainers put uc_mcontext today
497 * fixed bug 282: compiler does not trust type assertions while passing
498 arguments to a full call.
499 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
500 &OPTIONAL VALUES type parameter.
501 * fix bug 214: algorithm for noting rejected templates is now more
502 similar to that of template seletion. (also reported by rydis on
504 * fixed bug 141b: printing backquoted information readably and prettily
505 inserts a space where necessary.
506 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
507 (CHARACTER 10) are now reported as errors, rather than propagated
508 as unknown types. (reported by piso on #lisp)
509 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
510 with values NIL and :ERROR. (thanks to Milan Zamazal)
511 * fixed bug 191c: CLOS now does proper keyword argument checking as
512 described in CLHS 7.6.5 and 7.6.5.1.
513 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
514 longer raise an error; further, using a list as a for-as-arithmetic
515 counter now raises a meaningful error.
516 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
517 now understood by sequence creation functions such as MAKE-SEQUENCE
519 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
520 a STRING-INPUT-STREAM.
521 * compiler enhancement: SIGNUM is now better able to derive the type
523 * type declarations inside WITH-SLOTS are checked. (reported by
525 * fixed some bugs revealed by Paul Dietz' test suite:
526 ** incorrect optimization of TRUNCATE for a positive first
527 argument and negative second.
528 ** compiler failure in let-convertion during flushing dead code.
529 ** compiler failure while deriving type of TRUNCATE on an
530 interval, containing 0.
531 ** ASH of a negative bignum by a negative bignum count now returns
533 ** intersection of CONS types now canonicalizes properly, fixing
534 inconsistencies in SUBTYPEP.
536 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
537 * incompatible change: The --disable-debugger command line
538 option now clobbers the debugger at a more fundamental
539 level, by redefining #'INVOKE-DEBUGGER instead of by
540 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
541 is specified by ANSI to ignore *DEBUGGER-HOOK* and
542 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
543 enter the debugger REPL and then suffer recursive errors
544 because *DEBUG-IO* is also messed up in --disable-debugger mode;
545 while under the new system, BREAK in --disable-debugger mode
546 terminates the system just as an unhandled error would.
547 * fixed compiler performance when processing loops with a step >1;
548 * bug fix: DOCUMENTATION now retrieves generic function
549 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
550 support has been systematized, and now supports the methods
551 specified by ANSI, along with a default method and a method for
552 slot documentation. (reported by Nathan Froyd)
553 * bug fix: effective methods associated with a generic function are
554 no longer cached over a change of that generic function's method
555 combination. (reported by Andreas Fuchs)
556 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
557 names a non-existent directory. (thanks to Andreas Fuchs)
558 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
559 the PPC platform, lead to stack corruption; this has been fixed.
560 (reported by Rainer Joswig)
561 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
562 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
563 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
564 arithmetic in general, is now much more reliable on the PPC
566 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
567 platform now returns the right answer.
568 * optimization: restored some effective method precomputation in
569 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
570 precomputation is now tunable.
571 * optimization: compiler-internal data structure use has been
572 reviewed, and changes have been made that should improve the
573 performance of the compiler by about 20%.
574 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
575 simple-base-strings and simple-bit-vectors is improved.
576 * optimization: the optimization of 32-bit logical and arithmetic
577 functions introduced in version 0.8.3 on the x86 has been
578 implemented on the mips, ppc and sparc platforms; an
579 implementation of the same facility, but for 64-bit arithmetic,
580 has been added for the alpha.
581 * microoptimization: the compiler is better able to make use of the
582 x86 LEA instruction for multiplication by constants.
583 * bug fix: in some situations compiler did not report usage of
584 generic arithmetic in (SPEED 3) policy.
585 * bug 145b fix: compiler used wrong type specifier while converting
586 MEMBER-types to numeric.
587 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
589 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
590 index argument now works properly on the Alpha platform.
591 * bug fix: floating point exception treatment on the Alpha platform
593 * bug fix: FILE-POSITION works much better on string input and
594 output streams. (thanks to Nikodemus Siivola)
595 * bug fix: many threading/garbage collection symptoms sorted.
596 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
597 might be pseudo-atomic.
598 * internal change: Stopping for GC is now done with signals not
599 ptrace. GC is now done in whichever thread wanted it, instead of
601 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
603 * bug fix: result form in DO is not contained in the implicit
605 * incompatible change: ICR structure is changed; the value part of
606 CONTINUATION is now called LVAR; corresponding functions are
607 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
608 * added type deriver for ISQRT (thanks to Robert E. Brown).
609 * bug fix: better support for loading from the command line when an
610 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
612 * fixed some bugs revealed by Paul Dietz' test suite:
613 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
614 ** ROUND and FROUND now give the right answer when given very
615 small float arguments.
616 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
618 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
619 ** (GCD 0 <negative-integer>) returned <negative-integer>.
620 ** LCM should return a non-negative integer.
621 ** PARSE-INTEGER returned the index of a terminator instead of the
622 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
623 ** PARSE-INTEGER returned an incorrect index being applied to a
625 ** LCM with two arguments of 0 returns 0 rather than signalling
627 ** unsigned addition of a 32-bit constant with the high bit set no
628 longer causes an internal compiler error on the x86.
629 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
630 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
631 with negative last argument.
632 ** byte specifiers with zero size and position no longer cause
633 an error during type derivation.
634 ** bignum multiplication on the Alpha platform now returns the
636 * porting: The system now builds on SuSE AMD64, although it still
637 generates a 32-bit binary.
638 * .fasl file incompatibility: The fasl file version number has
639 been incremented (because of the changes to internal compiler
640 data structures referred to above).
642 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
643 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
644 more accurately, on the Darwin kernel running on PowerPC hardware.
645 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
646 * Compiler code deletion notes now signal a condition of type
647 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
648 an associated MUFFLE-WARNING restart.
649 * The compiler now performs limited argument count validation of
650 constant format strings in FORMAT, and where appropriate in ERROR,
651 CERROR and WARN. (thanks to Gerd Moellmann)
652 * New ASDF-INSTALL contrib can be used for automatic download and
653 installation of third-party Lisp code from CCLAN or other sites
655 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
656 which forces another thread to execute a function supplied by the
658 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
659 now accept and act upon their :ELEMENT-TYPE keyword argument.
660 (reported by Martin Atzmueller, Edi Weitz)
661 * bug fix: FILE-POSITION now accepts position designators up to
662 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
663 is the greater. (thanks to Patrik Nordebo)
664 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
665 Kalle Olavi Niemitalo)
666 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
667 printable. (reported by Eric Marsden)
668 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
669 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
670 * x86 bug fix in control stack exhaustion checking: now shows backtrace
671 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
672 (thanks to Stig Sandoe)
673 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
674 UNREAD-CHAR now correctly includes the unread character in the
675 target sequence. (thanks to Gerd Moellmann)
676 * bug fix in threaded builds: the system can now be suspended and
677 resumed by shell job control with minimal disruption.
678 * bug fixes in times and timezones >2038AD
679 * better handling of "where is GNU make?" problem in build scripts
680 (thanks to Nikodemus Siivola)
681 * new optimization: inside a named function any reference to a
682 function with the same name is considered to be a self-reference;
683 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
684 optimization quality.
685 * new optimization on x86: logical functions and + now have
686 optimized (UNSIGNED-BYTE 32) versions, which are automatically
687 used when the result is truncated to 32 bits.
688 * VALUES declaration is partially enabled.
689 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
690 * fixed some bugs revealed by Paul Dietz' test suite:
691 ** The system now obeys the constraint imposed by
692 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
693 types form a lattice under type intersection.
694 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
695 ** ASSOC now ignores NIL elements in an alist.
696 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
697 and (1+ MOST-POSITIVE-FIXNUM) answers.
698 ** The addition of a method with invalid qualifiers to a generic
699 function does not cause an error to be signalled immediately;
700 a warning is signalled, and the error is generated only on
701 calling the generic function.
702 * changed .fasl file version number, in part to add type codes for
703 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
704 obscure ANSI requirements
706 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
707 * fixed bug 148: failure to inline-expand a local function left
708 garbage, confusing the compiler.
709 * fixed bugs 3cd: structure slot readers perform type check if the
710 slot can have an invalid value (i.e. it is either not initialized
711 or can be written with a less specific slot writer).
712 * bug fix: the compiler now traps array references to elements off
713 the end of an array; previously, the bounds checking in some
714 circumstances could go off-by-one.
715 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
717 * type declarations for array element types now obey the description
718 on the CLHS page "Declaration TYPE", as per discussions on
719 sbcl-help around 2003-05-08. This means that a declaration
720 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
721 declaration, all references to BAR will be asserted or assumed
722 (with THE, so dependent on compiler policy) to involve objects of
723 type FOO. Note that no such declaration is implied in
724 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
725 * declared types of functions from the "Conditions"
726 chapter. (reported by Paul Dietz)
727 * bug fix: CERROR accepts a function as its first argument.
728 * bug fix: NTH an NTHCDR accept a bignum as index
729 arguments. (reported by Adam Warner)
730 * optimization: character compare routines now optimize comparing
731 against a constant character. (reported by Gilbert Baumann)
732 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
733 argument now works properly on the MIPS platform.
734 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
735 * repeated evaluation of the same DEFSTRUCT, a slot of which is
736 declared to have a functional type, does not cause an error
738 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
739 many :DEFAULT-INITARGS. (reported by Istvan Marko)
740 * fixed bug: if last continuation of a deleted block has a
741 destination, this destination should be deleted too. (reported by
743 * fixed a bug in the bootstrap process: the host compiler's values
744 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
745 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
746 test case from Patrik Nordebo)
747 * improved the ability of the disassembler on the PPC platform to
748 provide helpful disassembly notes.
749 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
750 the class in more cases than previously.
751 * bug fix: FILE-POSITION now understands :START and :END for
752 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
753 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
754 8) data. (thanks to David Lichteblau)
755 * bug fix: it is possible to add a method to a generic function
757 * bug fix: reader failed to signal END-OF-FILE inside an
758 object representation. (reported by Nikodemus Siivola)
759 * fixed some bugs revealed by Paul Dietz' test suite:
760 ** LAST and [N]BUTLAST should accept a bignum.
761 ** condition slot accessors are methods.
762 ** (VECTOR NIL) is a subtype of STRING.
764 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
765 * minor incompatible change: some nonsensical specialized lambda
766 lists (used in DEFMETHOD) which were previously ignored now signal
768 * minor incompatible change: the system is now aware of the types of
769 variables in the COMMON-LISP package, and will signal errors for
770 most violations of these type constraints (where previously they
771 were silently accepted).
772 * minor incompatible change: COMPILE-FILE now uses the freedom
773 afforded (ANSI 3.2.2.3) to use derived function types for
774 functions defined in the same file. This also permits the system
775 to warn on static type mismatches and function
776 redefinition. (Currently it does not work with high DEBUG level.)
777 * minor incompatible change: VALUES declaration is disabled.
778 * When issuing notes, the compiler now signals a condition of type
779 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
780 restart for use in user handlers. It is expected that the
781 COMPILER-NOTE condition will eventually become a condition
782 supertype to a hierarchy of note types, which will then be
783 handleable in a similar fashion. However, at the moment, no such
784 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
785 but it's an implementation detail, not a classification for the
787 * Changes in type checking closed the following bugs:
788 ** type checking of unused values (192b, 194d, 203);
789 ** template selection based on unsafe type assertions (192c, 236);
790 ** type checking in branches (194bc).
791 * A short form of VALUES type specifier has ANSI meaning (it has
792 increased the number of situations when SBCL cannot perform type
794 * fixed bug in DEFSTRUCT: once again, naming structure slots with
795 keywords or constants is permissible.
796 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
797 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
798 classes. (thanks to Antonio Martinez)
799 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
800 outputs FRESH-LINE or TERPRI, and no longer converts its stream
801 argument to a pretty-print stream. Instead, it leaves any such
802 operations to DESCRIBE-OBJECT methods.
803 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
805 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
806 lambda lists. (thanks to David Lichteblau)
807 * bug fix: Defining a generic function with a :METHOD-CLASS being a
808 subclass of STANDARD-METHOD no longer causes stack exhaustion.
809 (thanks to Gerd Moellmann)
810 * fixed bug 246: increased compilation speed of long
811 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
813 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
815 * DEFCONSTANT now throws a condition of type
816 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
817 constant to a non-EQL value; CONTINUE and ABORT restarts
818 respectively change and preserve the value.
819 * fixed bug 63: The code walker, part of the implementation of CLOS,
820 is now better at handling symbol macros.
821 * bug fix: There is no longer an internal implementation type named
822 CL:LENGTH. (reported by Raymond Toy)
823 * bug fix: In macro-like defining macros/special operators the
824 implicit block does not enclose the lambda list.
825 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
826 enclosed in parentheses) are not suitable as type specifiers, and
827 their use properly signals an error now.
828 * bug fix: An argument count mismatch for a type specifier in code
829 being compiled no longer causes an unhandled error at compile
830 time, but signals a compile-time warning.
831 * fixed simple vector readable printing
832 * bug fix: DESCRIBE takes more care over whether the class
833 precedence list slot of a class is bound before accessing it.
834 (reported by Markus Krummenacker)
835 * bug fix: FORMATTER can successfully compile pretty-printer format
836 strings which use variants of the ~* directive inside.
837 * bug fix: SEARCH now applies its TEST predicate to the elements of
838 the arguments in the correct order. (thanks to Wolfhard Buss)
839 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
840 in type inference. (thanks to Robert E. Brown)
841 * bug fix: Reading in symbols with an explicit package name of ""
842 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
843 * fixed some bugs revealed by Paul Dietz' test suite:
844 ** NIL is now allowed as a structure slot name.
845 ** Arbitrary numbers, not just REALs, are allowed in certain
846 circumstances in LOOP for-as-arithmetic clauses.
847 ** Multiple class redefinitions before slot access no longer
849 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
850 association between the name and a class.
851 ** Generic functions with non-standard method-combination and over
852 five methods all of which return constants no longer return NIL
853 after the first few invocations. (thanks to Gerd Moellmann)
854 ** CALL-NEXT-METHOD with no arguments now passes the original
855 values of the arguments, even in the presence of assignment.
856 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
858 ** DISASSEMBLE works with closures and funcallable instances.
859 ** ADD-METHOD now returns the generic function, not the new method.
860 ** FIND-METHOD signals an error if the lengths of the specializers
861 is incompatible with the generic function, even if the ERRORP
863 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
864 which its argument is a member.
865 ** DEFCLASS only redefines the class named by its class-name
866 argument if that name is the proper name of the class;
867 otherwise, it creates a new class.
868 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
869 of the UNBOUND-SLOT condition to the name of the slot.
870 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
871 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
872 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
873 SLOT-MAKUNBOUND in the specified fashion.
875 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
876 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
877 cross-compilation host. As a consequence, we can now bootstrap our
878 way up to SBCL starting with a bare gcc toolchain and human-readable
879 source code (first the source to CLISP, then the source to SBCL).
880 * A contributed module containing a partial implementation of the
881 simple-streams interface has been included. (thanks to Rudi
883 * A contributed module implementing the RFC1321 Message Digest
884 Algorithm, known as MD5, has been included.
885 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
886 no longer has any effect, as the code controlled by this feature
887 has been deleted. (As far as we know, no-one has ever built using
888 this feature, and its semantics were confused in any case).
889 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
890 work on conditions (as required by the ANSI specification),
891 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
892 expected behaviour on conditions. Users should note, however,
893 that such behaviour is not required by the ANSI specification,
894 and so use of this behaviour may render their code unportable.
895 * fixed some bugs revealed by Paul Dietz' test suite:
896 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
898 ** &ENVIRONMENT parameter in macro lambda list is bound first.
899 ** SXHASH on condition objects no longer returns NIL.
900 ** :ALLOCATION :CLASS slots are better treated; their values are
901 updated on class redefinition, and initforms inherited from
902 superclasses are applied.
903 ** REMOVE-METHOD returns its generic function argument even when
904 no method was removed.
905 ** SHARED-INITIALIZE now initializes the values of the requested
906 slots, including those with :ALLOCATION :CLASS.
907 ** ALLOCATE-INSTANCE now works on structure classes defined via
908 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
910 ** SLOT-EXISTS-P now works on conditions, as well as structures
912 ** MAKE-LOAD-FORM now has the required methods on
913 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
914 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
915 keyword, and now implements the SLOT-NAMES argument.
916 ** methods with &OPTIONAL arguments no longer allow too many
917 arguments to be passed in the call without error.
918 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
919 option is consistent with the required arguments of the generic
920 function lambda list.
921 * bug fix: REQUIRE accepts a string designator. (Thanks to
923 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
924 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
925 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
926 * bug fix: The compiler now checks for duplicated variables in macro
928 * bug fix: SETQ on globals returns the correct value.
929 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
930 not a COMPILER-ERROR (followed by some other strange error on
931 choosing the CONTINUE restart).
932 * bug fix: make.sh and friends are now more consistent in the way that
933 they look for GNU "make".
935 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
936 * experimental native threads support (on x86 Linux >=2.4 only).
937 This is not compiled in by default: you need to add :SB-THREAD to
938 the target features. See the "Beyond ANSI" chapter of the manual
940 * fix for longstanding nonANSIism: The old distinction between
941 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
942 The return value from CL:FIND-CLASS is now a CLOS class, and
943 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
944 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
945 * An interface to the MetaObject Protocol, as described in Kiczales,
946 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
947 MIT Press, 1991, is available from the SB-MOP package.
948 * incompatible change: the SB-PCL package should now be considered
949 a private implementation detail, and no longer a semi-private MOP
951 * minor incompatible change: due to rearrangement for threads, the
952 control stack and binding stack are now allocated at arbitrary
953 addresses instead of being hardcoded per-port. Users affected by
954 this probably have to be doing advanced things with shared
955 libraries, and will know who they are.
956 * minor incompatible change: Previously, all --eval forms used were
957 processed with READ before any of them were processed with EVAL.
958 Now each --eval form is processed with both READ and EVAL before
959 the next --eval form is processed. (Thus package operations like
960 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
961 work as the user might reasonably expect.)
962 * minor incompatible change: *STANDARD-INPUT* is now only an
963 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
965 * minor incompatible change: Y-OR-N-P is now character-oriented,
966 not line oriented. Also, YES-OR-NO-P now works without errors.
967 (thanks to Antonio Martinez)
968 * sb-aclrepl module improvements: an integrated inspector, added
969 repl features, and a bug fix to :trace command.
970 * Known functions, which cannot be open coded by the backend, are
971 considered to be able to check types of their arguments. (fixing
972 a bug report by Nathan J. Froyd)
973 * fixed a bug in computing method discriminating functions: It is
974 now possible to define methods specialized on classes which have
975 forward-referenced superclasses. (thanks to Gerd Moellmann)
976 * fixed evaluation order in optional entries (reported by Gilbert
978 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
979 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
980 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
981 required-by-AMOP NAME argument, as well as CLASS and
982 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
983 * fixed bug 20: DEFMETHOD can define methods using names that are
984 not the proper names of classes to designate class specializers.
985 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
986 Unix file descriptors, instead of blowing up. (thanks to Antonio
988 * Garbage collection refactoring: user-visible change is that a
989 call to the GC function during WITHOUT-GCING will not do garbage
990 collection until the end of the WITHOUT-GCING. If you were doing
991 this you were probably losing anyway.
992 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
993 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
994 TYPEP the latter but not the former.
995 * The compiler issues a full WARNING for calls to undefined functions
996 with names from the CL package.
997 * MAP-INTO for a vector destination is open coded. (reported by
998 Brian Downing on c.l.l)
999 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1000 documentation string.
1001 * fixed some bugs revealed by Paul Dietz' test suite:
1002 ** COPY-ALIST now signals an error if its argument is a dotted
1004 ** Condition slots are now accessed more correctly in the presence
1005 of multiple initargs for a given slot.
1006 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1007 exclude restarts of the same name associated with a different
1009 ** DEFCLASS of forward-referenced classes with another
1010 forward-referenced class in the superclasses list no longer
1012 ** Condition slots are now initialized once each, not multiple
1013 times. (thanks to Gerd Moellmann)
1014 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1016 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1017 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1018 arguments contain duplicated elements.
1019 ** RESTART-CASE understands local macros.
1020 ** RESTART-CASE associates exactly its own restarts with a condition.
1021 ** ENDP in safe mode checks its argument to be of type LIST.
1022 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1023 in question is unbound.
1024 ** Optimized MAKE-INSTANCE functions no longer cause internal
1025 assertion failures in the presence of duplicate initargs.
1026 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1027 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1028 propagated even in the ctor optimized implementation of
1030 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1032 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1033 * a better implementation of SXHASH on (simple) bit vectors,
1034 measured both in execution speed and in distribution of results
1035 over the positive fixnums, has been installed. Likewise, a better
1036 implementation of EQUAL for simple bit vectors is now available.
1037 * fixed CEILING optimization for a divisor of form 2^k.
1038 * fixed bug 240 (emitting extra style warnings "using the lexical
1039 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1040 by Antonio Martinez)
1041 * fixed SXHASH, giving different results for NIL depending on type
1042 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1043 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1044 variable. (found by Rolf Wester)
1045 * SBCL does not ignore type declarations for special
1046 variables. (reported by rif on c.l.l 2003-03-05)
1047 * some bug fixes in contrib/sb-aclrepl/
1048 * fixed some bugs revealed by Paul Dietz' test suite:
1049 ** a bug in the CONS type specifier, whereby the CAR and CDR
1050 types got intertwined, has been fixed;
1051 ** the type system is now able to reason about the interaction
1052 between INTEGER and RATIO types more completely;
1053 ** APPEND, [N]REVERSE and NRECONC check that those their
1054 arguments, which must be proper lists, are really so;
1055 ** An array specialized to be unable to hold elements has been
1056 implemented, as required -- yes, really -- by ANSI;
1057 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1058 on malformed property lists;
1060 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1061 * incompatible packaging change: in line with Unix convention,
1062 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1063 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1064 * REQUIRE and PROVIDE are now optionally capable of doing something
1065 useful. See the documentation string for REQUIRE.
1066 * infrastructure for a managed SBCL contrib system: contributed
1067 modules in this release include:
1068 ** the ASDF system definition facility;
1069 ** an interface to the BSD Sockets API;
1070 ** an ACL-like convenience interface to the repl;
1071 (thanks to Kevin Rosenberg)
1072 ** an implementation of ROTATE-BYTE, with efficient implementation
1074 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1075 gives a closer approximation to the right answer than previously.
1076 (thanks to Raymond Toy)
1077 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1078 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1079 optional environment arguments, as required by ANSI.
1080 * fixed bugs in other functions taking environment objects, allowing
1081 calls with an explicit NIL environment argument to be compiled
1083 * fixed bug 228: primary return values from
1084 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1085 COMPILE or FUNCTION.
1086 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1087 :TYPE will no longer signal errors on innocuous objects.
1088 * fixed bug 231b: SETQ is better at respecting type declarations in
1089 the lexical environment.
1090 * fixed a bug in DEFCLASS: classes named by symbols with no or
1091 unprintable packages can now be defined.
1092 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1093 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1094 * fixed compiler failure related to checking types of functions.
1095 (reported by Robert E. Brown)
1096 * the compiler is now much more consistent in its error-checking
1097 treatment of bounding index arguments to sequence functions: in
1098 (SAFETY 3) code, errors will be signalled in almost all cases if
1099 invalid sequence bounding indices are passed to functions defined
1100 by ANSI to operate on sequences.
1101 * fixed a bug in the build procedure: documentation of SBCL-specific
1102 packages is now preserved and available in the final Lisp image.
1103 * lifted FDEFINITION lookup out of loops in the implementation of
1104 many list operations. (thanks to Robert E. Brown)
1105 * fixed a bug in the reader: the #n# reader macro now works for
1106 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1107 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1108 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1109 E. Brown and Tony Martinez respectively)
1110 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1111 SLOT-BOUNDP: in particular, it is now able to optimize them much
1112 better, and is now not vulnerable to having packages renamed.
1113 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1114 always called when a slot is not present in an instance. (thanks
1116 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1117 longer signal an error on unreading a character following EOF on
1118 the previous constituent stream. (thanks to Tony Martinez)
1119 * fixed some bugs revealed by Paul Dietz' test suite:
1120 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1121 not just nonnegative fixnums;
1122 ** the logical bit-array operators such as BIT-AND now accept an
1123 explicit NIL for their "opt-arg" argument (to indicate a
1124 freshly-consed result bit-array);
1125 ** ELT now signals an error on an invalid sequence index in safe
1127 ** the type system is now cleverer about negations of numeric
1128 types, and consequently understands the BIGNUM and RATIO types
1130 ** the type system is now cleverer about the interaction between
1131 INTEGER and RATIO types: while bugs still remain, many more
1132 cases are accurately computed;
1133 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1134 if it is in the last clause;
1135 ** CONSTANTLY now correctly returns a side-effect-free function in
1137 ** DECLARE is no longer treated as a special-operator; in
1138 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1139 * incremented fasl file version number due to the change in the
1140 DEFSTRUCT-SLOT-DESCRIPTION structure.
1142 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1143 * minor incompatible change: code processed by the "interpreter" or
1144 EVAL now has a compilation optimization policy of (DEBUG 2)
1145 (changed from (DEBUG 1)) to improve debuggability of interactive
1146 development, and to allow the use of the debug RETURN command in
1148 * an experimental implementation of the RETURN command for the
1149 debugger has been included. (thanks to Frederik Kuivinen)
1150 * fixed bug 62: constraints were not propagated into a loop.
1151 * fixed bug in embedded calls of SORT (reported and investigated by
1153 * fixed some bugs revealed by Paul Dietz' test suite:
1154 ** printing and reading of arrays with some dimensions having
1155 length 0 (thanks to Gerd Moellmann);
1156 ** BOA constructor with &AUX argument without a default value does
1157 not cause a type error;
1158 ** CONSTANTP now returns true for all self-evaluating objects.
1160 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1161 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1162 accessors that are related by inheritance, as specified in the
1163 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1164 Valtteri Vuorikoski)
1165 * The compiler is now able to inline functions that were defined in
1166 a complex lexical environment (e.g. inside a MACROLET).
1167 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1169 * Support for the upcoming FreeBSD-5.0 release has been included.
1170 (thanks to Dag-Erling Smorgrav)
1171 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1172 effect when it is not in a toplevel context.
1173 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1174 stand a better chance of being correct. (thanks to Gerd
1176 * fixed bug in COERCE, which now signals an error on coercing a
1177 rational to a bounded real type which excludes the expected
1179 * The compiler is now able to derive types more accurately from the
1180 COERCE and COMPILE functions.
1181 * fixed bug 223: functional binding is considered to be constant
1182 only for symbols in the CL package.
1183 * fixed bug 231: SETQ did not check the type of a variable being set
1184 (reported by Robert E. Brown)
1185 * A new optimization for MAKE-INSTANCE has been included, fixing
1186 various bugs (including relating to :ALLOCATION :CLASS slots and
1187 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1189 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1190 ** As required by ANSI, LOOP now disallows anonymous collection
1191 clauses such as COLLECT I in conjunction with aggregate boolean
1192 clauses such as THEREIS (= I 1);
1193 ** LOOP now signals an error when any variable is reused in the
1194 same loop (including the potentially useful construct analogous
1195 to WITH A = 1 WITH A = (1+ A);
1196 ** IT is only a special loop symbol within the first clause of a
1197 conditional loop clause;
1198 ** LOOP with a typed iteration variable over a hashtable now
1199 signals a type error iff it should.
1200 * fixed some other bugs revealed by Paul Dietz' test suite:
1201 ** FILE-STREAM now names the class previously known as FD-STREAM;
1202 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1203 argument) no longer signals an error;
1204 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1205 of the slot symbol, rather than using the current package
1206 ((:CONC-NAME "") continues to intern the slot's name in the
1208 * incremented fasl file version number, because of the incompatible
1209 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1210 of the new implementation of DEFINE-COMPILER-MACRO.
1212 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1213 * Support for building SBCL for MIPS platforms running in
1214 little-endian mode has now been checked in, and basic
1215 functionality on said platforms verified.
1216 * minor incompatible change: PCL now records the pathname of a file
1217 in which methods and the like are defined, rather than its
1219 * minor incompatible change: TRUENAME now considers the truename of
1220 a file naming a directory to be the pathname with :DIRECTORY
1221 component indicating that directory.
1222 * minor incompatible change: a NAMED clause in the extended form of
1223 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1224 reason for the previous behaviour is unclear.
1225 * more systematization and improvement of CLOS and MOP conformance
1226 in PCL (thanks to Gerd Moellman and Pierre Mai):
1227 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1229 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1230 primary methods with no specializers;
1231 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1233 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1234 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1235 FORWARD-REFERENCED-CLASSes; error reporting on
1236 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1238 ** SXHASH on CLOS instances now uses a slot internal to the
1239 instance to return different numbers on distinct instances,
1240 while preserving the same return value through invocations of
1242 ** DEFMETHOD signals errors when methods with longer incongruent
1243 lambda lists are added to generic functions;
1244 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1245 CLASS, as specified in AMOP;
1246 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1247 on the order returned by the primary method for classes of
1248 class STANDARD-CLASS;
1249 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1250 * fixed some bugs shown by Paul Dietz' test suite:
1251 ** DOLIST puts its body in TAGBODY;
1252 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1254 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1255 value producing form;
1256 ** if more variables are given to PROGV than values, extra
1257 variables are bound and made to have no value;
1258 ** NSUBSTITUTE on list arguments gets the right answer with
1260 ** ELT signals an error of type TYPE-ERROR when the index argument
1261 is not a valid sequence index;
1262 ** LOOP signals (at macroexpansion time) an error of type
1263 PROGRAM-ERROR when duplicate variable names are found;
1264 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1265 ** FUNCALL of special-operators now cause an error of type
1267 ** PSETQ now works as required in the presence of side-effecting
1268 symbol-macro places;
1269 ** NCONC accepts any object as its last argument;
1270 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1272 ** loop-for-as-package does not require a package to be explicitely
1274 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1276 * fixed bug 166: compiler preserves "there is a way to go"
1277 invariant when deleting code.
1278 * fixed bug 172: macro lambda lists with required arguments after
1279 &REST arguments now cause an error to be signalled. (thanks to
1281 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1282 bug. (thanks to Matthew Danish)
1283 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1285 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1286 arguments are lexically rebound. (thanks to Gerd Moellmann and
1288 * fixed bug 194: error messages are now more informative when there
1289 is no primary method applicable in a call to a generic
1290 function. (thanks to Gerd Moellmann)
1291 * fixed bug in command line argument checking (thanks to Julian
1293 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1294 (thanks to Matthew Danish)
1295 * incremented fasl file version number, because of the
1296 SXHASH-related changes in the layout of CLOS data structures
1298 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1299 * minor incompatible change: The runtime (the Unix executable named
1300 "sbcl") is now much pickier about the .core files it will load.
1301 Essentially it now requires .core files to descend from the same
1302 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1303 as the runtime does. (The intent is to prevent the crashes which
1304 can occur, and which can even be reported as mysterious failures,
1305 when people patch the sources or change the build parameters
1306 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1307 sbcl and .core files.)
1308 * fixed bug: VALUES-LIST is no longer optimized away.
1309 * fixed bug 142: The FFI conversion of C string values to Lisp
1310 string values no longer conses excessively. (thanks to Nathan
1311 Froyd porting Raymond Toy's fix to CMU CL)
1312 * began to systematize and improve MOP conformance in PCL (thanks to
1313 Nathan Froyd, Gerd Moellman and Pierre Mai):
1314 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1316 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1317 ** FINALIZE-INHERITANCE is now called on class finalization;
1318 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1319 argument precedence order.
1320 * fixed bug 202: The compiler no longer fails on functions whose
1321 derived types contradict their declared type.
1322 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1323 so it can be non-toplevel.
1324 * The fasl file version number has changed (because of the new
1325 implementation of DEFMACRO).
1326 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1327 safe code, that any length requirement by their type-specifier
1328 argument is valid. The exceptions to this are described in bug
1330 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1331 functions now check the types of their inputs as required by ANSI.
1332 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1333 introduced symbol is DECLAREd to be SPECIAL.
1334 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1335 * fixed bug: :COUNT argument to sequence functions may be negative
1336 * fixed bug: body of DO-SYMBOLS may contain declarations
1337 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1338 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1339 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1340 symbol macro only once
1341 * fixed printing of call frame when argument list is unavailable
1342 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1343 * compiler no longer signals WARNING on unknown keyword
1346 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1347 * A beta-quality port to the mips architecture running Linux,
1348 based on the old CMUCL backend, has been made. It has been tested
1349 on a big-endian kernel, and works sufficiently well to be able to
1350 rebuild itself; it has not been tested in little-endian mode.
1351 * fixed an inconsistency between gencgc.c and purify.c which made
1352 dumping/loading .core files unreliable
1353 * fixed bug 120a: The compiler now deals correctly with IFs where
1354 the consequent is the same as the alternative, instead of
1355 misderiving the return type. (thanks to Alexey Dejneka)
1356 * fixed bug 113: Logical pathnames are now dumpable (the logical
1357 host is resolved at load-time, throwing an error if it is not
1359 * fixed bug 174: FORMAT's error message is slightly clearer when a
1360 non-printing character is used in a format directive.
1361 * fixed several bugs in compiler checking of type declarations, i.e.
1362 violations of the Python "declarations are assertions" principle
1363 (thanks to Alexey Dejneka)
1364 * fixed several bugs in PCL's error checking (thanks to Gerd
1366 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1368 * fixed bug in compilation of functions as first class values
1369 (thanks to Antonio Martinez-Shotton)
1370 * The compiler's handling TYPE-ERRORs which it can prove will
1371 inevitably happen at runtime has been cleaned up and corrected
1372 in several ways. (thanks to Alexey Dejneka)
1373 * improved argument type checking for various basic arithmetic
1374 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1375 lavished on them in the past that they can be compiled in many
1376 ways in different special cases
1377 * fixed bug 181: compiler checks validity of user supplied type
1379 * cleaned up code flushing in optimization: Function calls which
1380 should signal errors for safety purposes (e.g. which ANSI says
1381 should signal errors when their arguments are of incorrect type)
1382 are no longer optimized away.
1383 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1384 * incremented fasl file version number, because changes in the
1385 implementation of sequence functions like COERCE caused
1386 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1387 in old inline expansions) to become undefined. (Actually these
1388 changes were later undone, so we might very well be binary
1389 compatible with 0.7.7 after all, but leaving the version number
1390 incremented seemed like the simplest and most conservative
1393 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1394 * An alpha-quality port to the parisc architecture running Linux,
1395 based on the old CMUCL backend, has been made. This, even more so
1396 than the other backends, should be considered still a work in
1397 progress; known problems include that the Linux kernel in 64-bit
1398 mode does not propagate the correct sigcontext structure to
1399 userspace, and consequently SBCL on a parisc64 kernel will not
1401 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1402 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1403 suppressed.) Also now that the compiler is looking at declarations
1404 in the environment, it checks optimization declarations as well,
1405 and suppresses inlining when (> DEBUG SPEED).
1406 * More fixes have been made to treatment of floating point exception
1407 treatment and other Unix signals. In particular, floating point
1408 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1409 * The detection and handling of control stack exhaustion (infinite
1410 or very deeply nested recursion) has changed. Stack exhaustion
1411 detection is now done by write-protecting pages at the OS level
1412 and applies at all optimization settings; when found, a
1413 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1414 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1415 be caught using IGNORE-ERRORS.
1416 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1417 correctly when transforming recursive LABELS functions to LETs.
1418 (thanks to Alexey Dejneka)
1419 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1420 that are names of constants or global variables.
1421 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1422 alien routines with docstrings.
1423 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1424 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1426 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1427 (thanks to Eric Marsden)
1428 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1429 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1430 * Bug fix: The SPARC backend can now compile functions involving
1431 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1432 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1433 to a functions expecting a PATHNAME-DESIGNATOR.
1434 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1435 lambda lists. (thanks to Alexey Dejneka)
1436 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1437 functions so that EXPORTing the name of the class doesn't cause
1438 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1439 to Antonio Martinez for reporting this)
1440 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1441 constant-folded without failing an assertion. (thanks to Einar
1442 Floystad Dorum for reporting this)
1443 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1444 types are now dealt with more correctly by the compiler.
1445 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1446 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1447 behaviour analogous to RENAME-FILE. This puts its behaviour more
1448 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1450 * The fasl file version number has changed again. (because of the
1451 bug fix involving the names of PCL MAKE-INSTANCE functions)
1453 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1454 * bug fix: Floating point exceptions are treated much more
1455 consistently on the x86/Linux and PPC/Linux platforms.
1456 * Array initialization with :INITIAL-ELEMENT is now much faster for
1457 cases when the compiler cannot open code the array creation, but
1458 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1459 array accesses have also seen a speed increase.
1460 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1461 specified. (This was at the root of some bad interactions between
1462 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1464 * bug fix: Internal error arguments for undefined functions are now
1465 computed correctly on the PPC/Linux platform.
1466 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1467 Raymond Toy's patch for CMU CL)
1468 * Support for the Solaris 9 operating environment has been included
1469 (thanks to Daniel Merritt)
1470 * A very ugly but hopefully complete draft of the missing FFI chapter
1471 of the manual has been created by reformatting the corresponding
1472 CMU CL manual chapter into (currently very ugly and incoherent)
1473 DocBook and bringing it up to date for SBCL behavior. Thus, the
1474 manual is now essentially complete, at least by my extreme
1475 once-and-only-once standards, whereby it's acceptable to refer to
1476 the doc strings of SB-EXT functions as the primary documentation.
1477 * The fasl file version number has changed again, due to cleanup of
1478 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1479 is no longer a static symbol.)
1481 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1482 * SBCL now builds with OpenMCL (version 0.12) as the
1483 cross-compilation host; also, more progress has been made toward
1484 bootstrapping under CLISP.
1485 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1487 * bug 158 fixed: The compiler can now deal with integer loop
1488 increments different from 1; fixing this turned out also to fix
1490 * bug 169 fixed: no more bogus warnings about using lexical bindings
1491 despite the presence of perfectly good SPECIAL declarations (thanks
1492 to David Lichteblau)
1493 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1494 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1495 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1497 * bug 180 fixed: Method combination specifications no longer ignore
1498 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1499 * bug fix: Structure type predicate functions now check their argument
1500 count as they should.
1501 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1502 correctly. (thanks to Pierre Mai)
1503 * minor incompatible change: The --noprogrammer option is deprecated
1504 in favor of the new --disable-debugger option, which is very similar.
1505 (The major difference is that it takes effect at a slightly different
1506 time at startup, causing handling of errors in --sysinit and
1507 --userinit files will be affected differently.) The
1508 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1509 been added to allow this functionality to be controlled from ordinary
1510 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1511 maintainers, who might want to run non-interactive scripts to
1512 build SBCL cores which will later be used interactively.)
1513 * minor incompatible change: The LOAD function no longer, when given
1514 a wild pathname to load, loads all files matching that pathname.
1515 Instead, an error of type FILE-ERROR is signalled.
1517 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1518 * bug 147 fixed: The compiler preserves its block link/count
1519 invariants more correctly now so that it doesn't crash. (thanks
1521 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1523 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1524 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1525 no-value-for-this-character as NIL (as specified by ANSI).
1526 * HOST-NAMESTRING on physical pathnames now returns a string that is
1527 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1528 (thanks to Christophe Rhodes)
1529 * The Alpha port handles icache flushing more correctly. (thanks to
1531 * More progress has been made toward bootstrapping under CLISP. (thanks
1532 to Christophe Rhodes)
1533 * The fasl file format has changed again, because dynamic loading
1534 on OpenBSD (which has non-ELF object files) motivated some cleanups
1535 in the way that foreign symbols are transformed and passed around.
1536 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1537 is no longer treated as whitespace by the reader, but instead as
1538 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1539 |A
\7fB|, instead of A as it used to.
1541 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1542 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1543 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1544 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1545 this as of 0.7.1.45, but was left out of the previous news section
1546 (thanks to Dan Barlow)
1547 * SBCL now runs on the Solaris operating system on SPARC architectures
1548 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1549 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1550 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1551 * SPARC backend cleanups, allowing builds of cores optimized for V8
1552 and V9 SPARCS, and also emission of code targeted to a particular
1553 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1555 * SBCL is closer to bootstrapping under CLISP, thanks to various
1556 fixes by Christophe Rhodes.
1557 * The fasl file format has changed again, to allow the compiler's
1558 INFO database to support symbol macros.
1559 * The user manual (in doc/) is formatted into HTML more nicely.
1560 (thanks to coreythomas)
1561 * The system is smarter about SUBTYPEP relationships, especially
1562 those involving NOT types (including types such as ATOM which are
1563 represented internally using NOT types). Thus SUBTYPEP is less
1564 likely to return (VALUES NIL NIL) in general, and in particular
1565 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1567 * The fasl file format has changed again, because the internal
1568 representation of types now includes a new slot to support the new
1569 SUBTYPEP-of-NOT-types logic.
1570 * (not a change in the main branch of SBCL, but a related prototype
1571 which can hopefully be merged into the main branch of SBCL in the
1572 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1573 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1574 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1575 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1576 dispatch macro character combinations. (thanks to Alexey Dejneka)
1577 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1578 that superficially look like logical namestrings correctly.
1579 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1581 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1582 default supplied-p) for &optional and &key arguments. (thanks to
1585 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1586 * incompatible change: The compiler is now less aggressive about
1587 tail call optimization, doing it only when (> SPACE DEBUG) or
1588 (> SPEED DEBUG). (This is an incompatible change because there are
1589 programs which relied on the old CMU-CL-style behavior to optimize
1590 away their unbounded recursion which will now die of stack overflow.)
1591 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1592 for non-GENCGC systems has been increased to 20M (since that
1593 seems much closer to the likely performance optimum for modern
1594 systems than the old 4M value was)
1595 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1596 and *DEBUG-PRINT-LEVEL*
1597 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1598 of CMU CL's support for SPARC, and various endianness and other
1599 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1600 * new syntactic sugar for the Unix command line: --load foo.bar is now
1601 an alternate notation for --eval '(load "foo.bar")'.
1603 ** The system now detects stack overflow and handles it gracefully,
1604 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1605 optimization settings. (This is a good thing in general, and
1606 its introduction in this version should be particularly timely
1607 for anyone whose code fails because of suppression of tail
1609 ** The system now hunts for the C variable "environ" in a more
1610 devious way, to avoid segfaults when the C library version differs
1611 between compile time and run time. (thanks to Christophe Rhodes)
1612 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1613 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1614 ** The compiler no longer issues bogus style warnings for undefined
1615 classes in the same source file as the DEFCLASSes which defined
1616 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1618 ** fixes in CONDITION class precedence list for undefined function
1619 errors (thanks to Alexei Dejneka)
1620 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1621 correctly. (thanks to Dan Barlow)
1622 ** portability fixes aiming at bootstrapping under CLISP (thanks
1623 to Dave McDonald and Christophe Rhodes)
1624 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1625 ** fixes in type translation and and type inference (thanks to
1627 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1628 ** various fixes in the new ports (thanks to Dan Barlow)
1629 * several changes related to debugging:
1630 ** suppression of tail recursion, as noted above
1631 ** stack overflow detection, as noted above
1632 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1633 is now the default. (For some time encapsulation has been more
1634 reliable than the breakpoint-based :ENCAPSULATE NIL
1635 implementation, at least on X86 systems; and I just noticed that
1636 encapsulation also seems closer to the spirit of the ANSI
1639 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1641 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1642 up properly again. (There was a packaging bug in 0.7.0 which
1643 left their definitions in SB-SYS::LOAD-FOREIGN and
1644 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1645 vital for most things which interface to C-level interfaces,
1646 like extensions working with sockets or databases or
1647 Perl-compatible regexes or whatever, and the need to fix
1648 this bug is the main reason that 0.7.1 was released so
1650 ** DEFGENERIC is now choosier about the methods it redefines, so that
1651 reLOADing a previously-LOADed file containing DEFGENERICs does
1652 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1653 cycle now works as it should. (thanks to Alexey Dejneka)
1654 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1655 fixed by Christophe Rhodes. (He actually submitted this patch
1656 months ago, and I delayed until after 0.7.0.)
1657 ** Bug 111 (internal compiler confusion about runtime checks on
1658 FUNCTION types) was fixed by Alexey Dejneka.
1659 * Some internal cleanups (getting rid of variables which aren't
1660 needed now that the byte interpreter is gone) caused the fasl
1661 file format number to change again.
1663 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1664 * major incompatible change: The default fasl file extension, i.e. the
1665 default extension for files produced by COMPILE-FILE, has changed
1666 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1668 ** There are many changes in the implementation of the compiler.
1669 SBCL is now essentially a compiler-only implementation of ANSI
1670 Common Lisp. EVAL still "interprets" a few special cases, but
1671 almost all the interesting cases are handled by creating
1672 a LAMBDA expression, calling COMPILE on it, then calling
1673 FUNCALL on the result.
1674 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1675 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1676 Since the code is newer, there might still be some new bugs
1677 (though not as many as before Martin Atzmueller's fixes:-). But
1678 the new code is substantially simpler and clearer, and hopefully
1679 any remaining bugs will be simpler, less fundamental, and more
1680 fixable then the bugs in the old code.
1681 ** The revised compiler is still a little unsteady on its feet.
1683 *** The debugging information it produces (particularly the names
1684 of FUNCTION objects) is sometimes much less useful than what
1685 the old compiler produced.
1686 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1687 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1688 within a MACROLET) has been temporarily weakened.
1689 ** There are new compiler optimizations for various functions:
1690 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1691 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1692 *** the math functions TRUNCATE, FLOOR, and CEILING
1693 *** the function-of-all-trades COERCE
1694 Mostly these should be transparent, but there's one
1695 potentially-annoying problem (bug 117): when the compiler
1696 inline-expands a function and does type analysis on the result,
1697 it can create control paths which have type mismatches, and
1698 when it can't prove that those control paths aren't taken,
1699 it will issue WARNINGs about the type mismatches. This is
1700 a particular problem in practice for the new sequence functions.
1701 It's not clear how this should be fixed, and for now, a
1702 workaround is given in the entry for 117 in the BUGS file.
1703 ** (Because of the interaction between the two previous items --
1704 occasional inlining problems and new inline expansions -- some
1705 of the new sequence function optimizations won't really kick in
1706 completely until debugging information, and then inlining, are
1707 straightened out in some future version.)
1708 * minor incompatible changes:
1709 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1710 DIRECTORY no longer implicitly promotes NIL slots of its
1711 pathname argument to :WILD. In particular, when you ask for the
1712 contents of a directory (which you used to be able to do without
1713 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1714 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1715 ** changes in behavior that ANSI explicitly defines to be
1716 implementation dependent:
1717 *** The new compiler-only implementation still conforms with ANSI,
1718 but acts a little different than before. Besides the obvious
1719 changes in performance tradeoffs (that the cost per form passed
1720 to EVAL has gone up, and the cost per form executed by EVAL
1721 has gone down), the behavior of the system changes a little
1722 because there are no longer any interpreted function objects.
1723 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1724 e.g. doing COMPILE on the output of interactive DEFUN is
1726 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1728 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1729 has changed from (:CL) to NIL.
1730 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1731 e.g. #\Newline and #\Tab, don't have names specified in the
1732 ANSI Common Lisp standard, is now based on their ASCII symbolic
1733 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1734 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1735 are no longer used for output.
1736 ** changes in internal implementation constants:
1737 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1738 4 million. (If your application spends a lot of time GCing and
1739 you have a lot of RAM, you might want to experiment with
1740 increasing it even more.)
1741 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1742 However, almost all old code should still continue to work without
1743 immediate update, as SB-C-CALL is now a (deprecated) nickname
1745 ** Old operator names in the style DEF-FOO are now deprecated in
1746 favor of new corresponding names DEFINE-FOO, for consistency with
1747 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1748 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1749 internal symbols, but a few supported extensions like
1750 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1751 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1752 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1753 etc. as you get deeper into recursive calls to the debugger
1754 command loop, instead of the old "5]", "5]]", "5]]]"
1755 sequence. (I was motivated to do this when squabbles between
1756 ILISP and SBCL left me very deeply nested in the debugger. In the
1757 short term, this change will probably provoke more ILISP/SBCL
1758 squabbles, but hopefully it will be an improvement in the long run.)
1759 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1760 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1761 function, since declaiming return types involving aliens is
1762 (1) annoyingly messy to do by hand and (2) vital to efficient
1763 compilation of code which calls such functions.
1764 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1765 longer reexported by the SB-EXT package. They're solely useful
1766 for alien code, so it seems more logical that you should get
1767 them from the SB-ALIEN package, not in SB-EXT.
1768 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1769 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1770 features. Instead, the code that they used to control is always
1771 built into the system.
1772 * many other bug fixes
1773 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1774 advantage of the new EVAL-WHEN stuff and to clean them up in
1775 general, and they are now more ANSI-compliant in a number of
1776 ways. Martin Atzmueller is responsible for a lot of this.
1777 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1779 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1780 *** correct ERROR type for various file operations
1781 *** some fixes for Lisp streams
1782 *** DEFMETHOD syntax checking
1783 *** changing old weird representation of debug information as
1784 strings (which, among their other deficiencies, don't transform
1785 correctly when you rename packages, and don't change their
1786 print representation when you change things like *PACKAGE*
1787 and *PRINT-LENGTH*) to symbols and lists of symbols
1788 He also made several improvements and fixed several bugs in DESCRIBE.
1789 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1791 *** misbehavior of WRITE-STRING/WRITE-LINE
1792 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1793 and several other LOOP problems as well
1794 *** DIRECTORY when similar filenames are present
1795 *** DEFGENERIC with :METHOD options
1796 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1797 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1798 *** argument ordering in FIND with :TEST option
1799 *** mishandled package designator argument in APROPOS-LIST
1800 *** various problems in the backquote readmacro
1801 *** a bug in APROPOS
1802 *** probably some others that I'm not describing very well here,
1803 since the CVS log documents them by reference to sbcl-devel
1804 messages, and the SourceForge archives aren't working well.:-(
1805 ** Dan Barlow improved the Alpha port (and is making progress on the
1806 PPC port, for those of you who think different).
1807 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
1808 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
1809 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
1810 continues to work on the SPARC port (for those of you in a position
1811 to look down upon our little PC-compatible boxes from a great height).
1812 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
1813 and uses that copy, rather than the current dynamic value, when
1814 it's trying to decide whether to truncate output. Thus e.g.
1815 (let ((*print-lines* 50))
1816 (pprint-logical-block (stream nil)
1818 (let ((*print-lines* 8))
1819 (print (aref possiblybigthings i) stream)))))
1820 should now truncate the logical block only at 50 lines, instead of
1821 often truncating it at 8 lines, as it did before.
1822 * The doc/cmucl/ directory, containing old CMU CL documentation
1823 from the time of the fork, is no longer part of the base system.
1824 SourceForge has shut down its anonymous FTP service, and with it
1825 my original plan for distributing the old CMU CL documentation
1826 there. For now, if you need these files you can download an old
1827 SBCL source release and extract them from it.
1828 * The fasl file version number changed again, for dozens of reasons,
1829 some of which are apparent above.
1831 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
1832 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
1833 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
1834 patch, so that DISASSEMBLE output is much nicer.
1835 * The code in the SB-PROFILE package now seems reasonably stable.
1836 I still haven't decided what the final interface should look like
1837 (I'd like PROFILE to interact cleanly with TRACE, since both
1838 facilities use function encapsulation) but if you have a need
1839 for profiling now, you can probably use it successfully with
1840 the current CMU-CL-style interface.
1841 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
1842 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
1843 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
1844 link now returns the dangling link itself, and for similar
1845 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
1846 link itself. (In these cases the old code signalled an error and
1847 looped endlessly, respectively.) Thus, DIRECTORY now works even
1848 in the presence of dangling and cyclic symbolic links.
1849 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
1850 is now a supported extension again, since the consensus on
1851 sbcl-devel was that it can be useful for ordinary development
1852 work, not just for debugging SBCL itself.
1853 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
1854 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
1855 that currently-defined functions might be redefined later with
1856 different return types.
1857 * Hash tables can be printed readably, as inspired by CMU CL code
1858 of Eric Marsden and SBCL code of Martin Atzmueller.
1859 * better error handling in CLOS method combination, thanks to
1860 Martin Atzmueller porting Pierre Mai's CMU CL patches
1861 * more overflow fixes for >16Mbyte I/O buffers
1862 * A bug in READ has been fixed, so that now a single Ctrl-D
1863 character suffices to cause end-of-file on character streams.
1864 In particular, now you only need one Ctrl-D at the command
1865 line (not two) to exit SBCL.
1866 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
1868 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
1869 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
1870 does the right thing.
1871 * The implementation of some type tests, especially for CONDITION
1872 types, is now tidier and maybe faster, due to CMU CL code
1873 originally by Douglas Crosher, ported by Martin Atzmueller.
1874 * Some math functions have been fixed, and there are new
1875 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
1876 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
1877 * (There are also some new optimizers in contrib/*-extras.lisp. Those
1878 aren't built into sbcl-0.6.13, but are a sneak preview of what's
1879 likely to be built into sbcl-0.7.0.)
1880 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
1881 CL, ported to SBCL by Martin Atzmueller)
1882 * DESCRIBE now gives more information in some cases. (Pierre Mai's
1883 patch to CMU CL, ported to SBCL by Martin Atzmueller)
1884 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
1885 * There's a new slam.sh hack to shorten the edit/compile/debug
1886 cycle for low-level changes to SBCL itself, and a new
1887 :SB-AFTER-XC-CORE target feature to control the generation of
1888 the after-xc.core file needed by slam.sh.
1889 * minor incompatible change: The ENTRY-POINTS &KEY argument to
1890 COMPILE-FILE is no longer supported, so that now every function
1891 gets an entry point, so that block compilation looks a little
1892 more like the plain vanilla ANSI section 3.2.2.3 scheme.
1893 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
1894 returns the number of bytes consed since the system started,
1895 rather than the number consed since the first time the function
1896 was called. (The new definition parallels ANSI functions like
1897 CL:GET-INTERNAL-RUN-TIME.)
1898 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
1899 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
1900 supported. Now DIRECTORY always does the abstract Common-Lisp-y
1901 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
1902 * Fasl file version numbers are now independent of the target CPU,
1903 since historically most system changes which required version
1904 number changes have affected all CPUs equally. Similarly,
1905 the byte fasl file version is now equal to the ordinary
1908 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
1909 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
1910 is no longer recognized. I apologize for this, because it was
1911 listed in SB-EXT as a supported extension, but I found that
1912 its existing behavior was poorly specified, as well as incorrectly
1913 specified, and it looked like too much of a mess to straighten it
1914 out. I have enough on my hands trying to get ANSI stuff to work..
1915 * many patches ported from CMU CL by Martin Atzmueller, with
1916 half a dozen bug fixes in pretty-printing and the debugger, and
1917 half a dozen others elsewhere
1918 * fixed bug 13: Floating point infinities are now supported again.
1919 They might still be a little bit flaky, but thanks to bug reports
1920 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
1921 as flaky as they were.
1922 * The --noprogrammer command line option is now supported. (Its
1923 behavior is slightly different in detail from what the old man
1924 page claimed it would do, but it's still appropriate under the
1925 same circumstances that the man page talks about.)
1926 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
1927 are now supported, and enabled by default. Thus, the compiler can
1928 handle many floating point and complex operations much less
1929 inefficiently. (Thus e.g. you can implement a complex FFT
1931 * The compiler now detects type mismatches between DECLAIM FTYPE
1932 and DEFUN better, and implements CHECK-TYPE more correctly, and
1933 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
1934 thanks to patches from Martin Atzmueller.
1935 * various fixes to make the cross-compiler more portable to
1936 ANSI-conforming-but-different cross-compilation hosts (notably
1937 Lispworks for Windows, following bug reports from Arthur Lemmens)
1938 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
1939 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
1940 fixed thanks to Pierre Mai's CMU CL patches.
1941 * fixes to keep the system from overflowing internal counters when
1942 it tries to use i/o buffers larger than 16M bytes
1943 * fixed bug 45a: Various internal functions required to support
1944 complex special functions have been merged from CMU CL sources.
1945 (When I was first setting up SBCL, I misunderstood a compile-time
1946 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
1947 * improved support for type intersection and union, fixing bug 12
1948 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
1949 more obscure bugs as well
1950 * some steps toward byte-compiling non-performance-critical
1951 parts of the system, courtesy of patches from Martin Atzmueller
1952 * Christophe Rhodes has made some debian packages of sbcl at
1953 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
1954 From his sbcl-devel e-mail of 2001-04-08 they're not completely
1955 stable, but are nonetheless usable. When he's ready, I'd be happy
1956 to add them to the SourceForge "File Releases" section. (And if
1957 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
1958 * new fasl file format version number (because of changes in
1959 internal representation of (OR ..) types to accommodate the new
1960 support for (AND ..) types, among other things)
1962 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
1963 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
1965 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
1967 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
1968 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
1969 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
1970 * DESCRIBE now works on structure objects again.
1971 * Most function call argument type mismatches are now handled as
1972 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
1973 know whether the function will be redefined before the call is
1974 executed. (The compiler could flag local calls with full WARNINGs,
1975 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
1976 it doesn't keep track of enough information to know whether calls
1977 are local in this sense.)
1978 * Compiler output is now more verbose, with messages truncated
1979 later than before. (There should be some supported way for users
1980 to override the default verbosity, but I haven't decided how to
1981 provide it yet, so this behavior is still controlled by the internal
1982 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
1983 src/compiler/ir1util.lisp.)
1984 * Fasl file format version numbers have increased again, because
1985 support for the Gray streams extension changes the layout of the
1986 system's STREAM objects.
1987 * The Gray subclassable streams extension now works, thanks to a
1988 patch from Martin Atzmueller.
1989 * The full LOAD-FOREIGN extension (not just the primitive
1990 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
1991 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
1992 but like most other programs, it defaults to copying the Unix
1993 environment from the original process instead of starting the
1994 new process in an empty environment.
1995 * Extensions which manipulate the Unix environment now support
1996 an :ENVIRONMENT keyword option which doesn't smash case or
1997 do other bad things. The CMU-CL-style :ENV option is retained
1998 for porting convenience.
1999 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2000 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2002 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2004 * A patch from Martin Atzmueller seems to have solved the SIGINT
2005 problem, and as far as we know, signal-handling now works cleanly.
2006 (If you find any new bugs, please report them!)
2007 * The system no longer defaults Lisp source file names to types
2008 ".l", ".cl", or ".lsp", but only to ".lisp".
2009 * The compiler no longer uses special default file extensions for
2010 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2011 seems to expect a single default extension for all compiled code,
2012 and there's no compelling reason to try to stretch the standard
2013 to allow two different extensions.) Instead, byte-compiled files
2014 default to the same extension as native-compiled files.
2015 * Fasl file format version numbers have increased again, because
2016 a rearrangement of internal implementation packages made some
2017 dumped symbols in old fasl files unreadable in new cores.
2018 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2019 many fewer weird special cases.
2020 * Bug #17 (differing COMPILE-FILE behavior between logical and
2021 physical pathnames) has been fixed, and some related misbehavior too,
2022 thanks to a patch from Martin Atzmueller.
2023 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2024 by Tim Moore, ported to SBCL by Martin Atzmueller.
2025 * Martin Atzmueller fixed several filesystem-related problems,
2026 including bug #36, in part by porting CMU CL patches, which were
2027 written in part by Paul Werkowski.
2028 * More compiler warnings in src/runtime/ are gone, thanks to
2029 more patches from Martin Atzmueller.
2030 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2033 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2035 * DESCRIBE now works on CONDITION objects.
2036 * The debugger now handles errors which arise when trying to print
2037 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2039 * The build system now uses an additional file, customize-target-features.lisp,
2040 to allow local modifications to the target *FEATURES* list. (The point of
2041 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2042 debugging features enabled, without having to worry about propagating it
2043 into everyone's system when I do a "cvs update".) When no
2044 customize-target-features.lisp file exists, the target *FEATURES* list
2045 should be constructed the same way as before.
2046 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2047 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2048 warn about behavior which is undefined under the ANSI spec. Note: This
2049 is specified by ANSI, but it's not very popular with programmers.
2050 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2051 macro in the SBCL sources for an example of a workaround which you
2052 might use to make portable ANSI-standard code which does what you want.
2053 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2054 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2055 when it's used as a non-toplevel form. (This is required in order
2056 to implement the DEFCONSTANT-EQX macro.)
2057 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2058 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2059 funny things, due to bugs in the implementation of EVAL-WHEN
2060 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2061 non-toplevel DEFCONSTANTs are uncommon.)
2062 * The core file version number and fasl file version number have been
2063 incremented, because the old noncompliant DEFCONSTANT behavior involved
2064 calling functions which no longer exist, and because I also took the
2065 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2066 * fixed bug 1 (error handling before read-eval-print loop starts), and
2067 redid debugger restarts and related debugger commands somewhat while
2069 ** The QUIT debugger command is gone, since it did something
2070 rather different than the SB-EXT:QUIT command, and since it never
2071 worked properly outside the main toplevel read/eval/print loop.
2072 Invoking the new TOPLEVEL restart provides the same functionality.
2073 ** The GO debugger command is also gone, since you can just invoke
2074 the CONTINUE restart directly instead.
2075 ** The TOP debugger command is also gone, since it's redundant with the
2076 FRAME 0 command, and since it interfered with abbreviations for the
2078 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2080 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2081 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2082 some porting work by Martin Atzmueller)
2083 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2084 with a boatload of patches which he ported to SBCL. Now that those
2086 ** The system tries to make sure that its low-priority messages
2087 are prefixed by semicolons, to help people who like to use
2088 syntax highlighting in their ILISP buffer. (This patch
2089 was originally due to Raymond Toy.)
2090 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2091 patches originally written by Raymond Toy.
2092 ** The compiler understands coercion between single-value and
2093 multiple-VALUES type expressions better, getting rid of some very
2094 weird behavior, thanks to patches originally by Robert MacLachlan
2095 and Douglas Crosher.
2096 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2097 lambda lists, thanks to a patch originally by Pierre Mai.
2098 ** The system no longer bogusly warns about "abbreviated type
2100 ** The compiler gets less confused by inlining and RETURN-FROM,
2101 thanks to some patches originally by Tim Moore.
2102 ** The system no longer hangs when dumping circular lists to fasl
2103 files, thanks to a patch originally from Douglas Crosher.
2104 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2105 undefined function error.
2106 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2107 in structure slot initforms) for now, documented workaround instead:-|
2108 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2109 * fixed bug 5: added stubs for various Gray stream functions called
2110 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2111 installed, at least appropriate type errors are generated
2112 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2113 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2114 consistently in DEFMETHOD forms.
2115 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2116 it doesn't seem to affect SBCL after all
2117 * The C runtime system now builds with better optimization and many
2118 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2120 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2122 * The system is now under CVS at SourceForge (instead of the
2123 CVS repository on my home machine).
2124 * The new signal handling code has been tweaked to treat register
2125 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2126 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2127 downstream, e.g. in debug-int.lisp, has implicit dependencies
2128 on the unsignedness of integer representation of machine words,
2129 and that caused the system to bomb out with infinite regress
2130 when trying to recover from type errors involving signed values,
2131 e.g. (BUTLAST '(1 2 3) -1).)
2132 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2133 (This was one of the bugs Peter Van Eynde reported back in July.)
2134 * The system now uses code inspired by Colin Walters' O(N)
2135 implementation of MAP (from the cmucl-imp@cons.org mailing
2136 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2137 inline the MAP operation, and there is more than one
2138 sequence argument to the MAP call (so that it can't just
2139 do ETYPECASE once and for all based on the type of the
2140 single sequence argument). (The old non-inline implementation
2141 of the general M-argument sequence-of-length-N case required
2142 O(M*N*N) time when any of the sequence arguments were LISTs.)
2143 * The QUIT :UNIX-CODE keyword argument has been renamed to
2144 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2145 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2147 * Raymond Wiker's patches to port dynamic loading from Linux to
2148 FreeBSD have been added.
2149 * The BUGS file is now more nearly up to date, thanks in large part
2150 to Martin Atzmueller's review of it.
2151 * The debugger now flushes standard output streams before it begins
2152 its output ("debugger invoked" and so forth).
2153 * The core version number and fasl file version number have both
2154 been incremented, because of incompatible changes in the layout
2156 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2157 used to not be called for a saved Lisp image.)
2158 * Martin Atzmueller's version of a patch to fix a compiler crash,
2159 as posted on sbcl-devel 13 September 2000, has been installed.
2160 * Instead of installing Martin Atzmueller's patch for the
2161 compiler transform for SUBSEQ, I deleted the compiler transform,
2162 and transforms for some similar consing operations.
2163 * A bug in signal handling which kept TRACE from working on OpenBSD
2165 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2166 away by constant folding
2167 * The system now defines its address space constants in one place
2168 (in the Lisp sources), and propagates them automatically elsewhere
2169 (through GENESIS and the sbcl.h file). Therefore, patching the
2170 address map is less unnecessarily tedious and error-prone. The
2171 Lisp names of address space constants have also been systematized.
2172 * CVS tags like dollar-Header-dollar have been removed from
2173 the sources, because they have never saved me trouble and
2174 they've been source of trouble working with patches and other
2175 diff-related operations.
2176 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2177 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2179 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2181 * The system has been ported to OpenBSD.
2182 * The system now compiles with a simple "sh make.sh" on the systems
2183 that it's supported on. I.e., now you no longer need to tweak
2184 text in the target-features.lisp-expr and symlinks in src/runtime/
2185 by hand, the make.sh takes care of it for you.
2186 * The system is no longer so grossly inefficient when compiling code
2187 involving vectors implemented as general (not simple) vectors (VECTOR T),
2188 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2189 which dares to use the various sequence functions on non-simple
2190 vectors, takes less of a performance hit.
2191 * There is now a primitive type predicate VECTOR-T-P
2192 to test for the (VECTOR T) type, so that e.g.
2193 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2194 can now be compiled with some semblance of efficiency. (The old code
2195 turned the type declaration into a full call to %TYPEP at runtime!)
2196 * AREF on (VECTOR T) is still not fast, since it's still compiled
2197 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2198 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2199 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2200 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2201 without making full calls to SUBTYPEP at runtime.
2202 (Some analogous efficiency issues for non-simple vectors specialized to
2203 element types other than T, or for non-simple multidimensional arrays,
2204 have not been addressed. They could almost certainly be handled the
2205 same way if anyone is motivated to do so.)
2206 * The changes in array handling break binary compatibility, so
2207 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2208 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2209 instead of (VALUES T T).
2210 * By following the instructions that Dan Barlow posted to sbcl-devel
2211 on 2 July 2000, I was able to enable primitive dynamic object
2212 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2213 functionality is not implemented (since it calls ld to resolve
2214 library references automatically, requiring RUN-PROGRAM for its
2215 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2216 try to resolve library references) is now supported.
2217 * The system now flushes the standard output streams when it terminates,
2218 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2219 them at several other probably-convenient times, e.g. in each pass of
2220 the toplevel read-eval-print loop, and after evaluating a form given
2221 as an "--eval" command-line option. (These changes were motivated by a
2222 discussion of stream flushing issues on cmucl-imp in August 2000.)
2223 * The source transform for TYPEP of array types no longer assumes
2224 that an array whose element type is a not-yet-defined type
2225 is implemented as an array of T, but instead punts, so that the
2226 type will be interpreted at runtime.
2227 * There is now some support for cross-compiling in make.sh: each of
2228 the phases of make.sh has its own script. (This should be transparent
2229 to people doing ordinary, non-cross-compile builds.)
2230 * Since my laptop doesn't have hundreds of megabytes of memory like
2231 my desktop machine, I became more motivated to do some items on
2232 my to-do list in order to reduce the size of the system a little:
2233 ** Arrange for various needed-only-at-cold-init things to be
2234 uninterned after cold init. To support this, those things have
2235 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2236 symbols with such names are now uninterned after cold init).
2237 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2238 fasl files for cold load.
2239 ** Remove the old compiler structure pooling code (which used to
2240 be conditional on the target feature :SB-ALLOC) completely.
2241 ** Redo the representation of some data in cold init to be more compact.
2242 (I also looked into supporting byte compiled code at bootstrap time,
2243 which would probably reduce the size of the system a lot, but that
2244 looked too complicated, so I punted for now.)
2245 * The maximum signal nesting depth in the src/runtime/ support code has
2246 been reduced from 4096 to 256. (I don't know any reason for the very
2247 large old value. If the new smaller value turns out to break something,
2248 I'll probably just bump it back up.)
2249 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2251 * Many, many bugs reported by Peter Van Eynde have been added to
2252 the BUGS list; some have even been fixed.
2253 * While enabling dynamic object file loading, I tried to make the
2254 code easier to understand, renaming various functions and variables
2255 with less ambiguous names, and changing some function calling
2256 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2257 * While trying to figure out how to do the OpenBSD port, I tried to
2258 clean up some of the code in src/runtime/. In particular, I dropped
2259 support for non-POSIX signal handling, added various comments,
2260 tweaked the code to reduce the number of compilation warnings, and
2261 renamed some files to increase consistency.
2262 * To support the new automatic configuration functionality in make.sh,
2263 the source file target-features.lisp-expr has been replaced with the
2264 source file base-target-features.lisp-expr and the machine-generated
2265 file local-target-features.lisp-expr.
2266 * fixed a stupid quoting error in make.sh so that using CMU CL
2267 "lisp -batch" as cross-compilation host works again
2269 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2271 * DESCRIBE no longer tries to call itself recursively to describe
2272 bound/fbound values, so that it no longer fails on symbols which are
2273 bound to themselves (like keywords, T, and NIL).
2274 * DESCRIBE now works on generic functions.
2275 * The printer now prints less-screwed-up representations of closures
2276 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2277 * A private symbol is used instead of the :EMPTY keyword previously
2278 used to mark empty slots in hash tables. Thus
2279 (DEFVAR *HT* (MAKE-HASH-TABLE))
2280 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2281 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2282 now does what ANSI says that it should. (You can still get
2283 similar noncompliant behavior if bang on the hash table
2284 implementation with all the symbols you get back from
2285 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2286 This breaks binary compatibility, since tests for equality to
2287 :EMPTY are wired into things like the macroexpansion of
2288 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2290 * There's now a minimal placeholder implementation for CL:STEP,
2291 as required by ANSI.
2292 * An obscure bug in the interaction of the normal compiler, the byte
2293 compiler, inlining, and structure predicates has been patched
2294 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2295 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2297 * Missing ordinary arguments in a macro call are now detected even
2298 when the macro lambda list contains &KEY or &REST.
2299 * The debugger no longer complains about encountering the top of the
2300 stack when you type "FRAME 0" to explicitly instruct it to go to
2301 the top of the stack. And it now prints the frame you request even
2302 if it's the current frame (instead of saying "You are here.").
2303 * As specified by ANSI, the system now always prints keywords
2304 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2306 * The default initial SIZE of HASH-TABLEs is now smaller.
2307 * Type information from CLOS class dispatch is now propagated
2308 into DEFMETHOD bodies, so that e.g.
2309 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2311 is now basically equivalent to
2312 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2313 (DECLARE (TYPE SINGLE-FLOAT X))
2315 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2316 operation, without having to do run-time type dispatch.
2317 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2318 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2319 * Since I don't seem to be making big file reorganizations very often
2320 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2321 is overflowing my ability to conveniently back them up), I've finally
2322 checked the system into CVS. (The CVS repository is on my home system,
2323 not at SourceForge -- putting it on SourceForge might come later.)
2324 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2325 high-level GC-NOTIFY-FOO functions send their output. (There's
2326 still very little control of where low-level verbose GC functions
2327 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2328 less than it used to -- the GC-NOTIFY-FOO functions are now under
2329 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2330 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2331 in only one place in the source code, and propagates it automatically
2332 everywhere that it's needed. Thus e.g. when I bump the version from
2333 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2334 * The C source files now include boilerplate legalese and documentation
2335 at the head of each file (just as the Lisp source files already did).
2336 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2337 to his page will be replaced with a link to his new CLiki service.
2339 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2341 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2342 * The build process now looks for GNU make under the default name "gmake",
2343 instead of "make" as it used to. If GNU make is not available as "gmake"
2344 on your system, you can change this default behavior by setting the
2345 GNUMAKE environment variable.
2346 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2347 can build without error under CMU CL.
2349 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2351 * There is now a partial SBCL user manual (with some new text and some
2352 text cribbed from the CMU CL manual).
2353 * The beginnings of a profiler have been added (starting with the
2354 CMU CL profiler and simplifying and cleaning up). Eventually the
2355 main interface should be through the TRACE macro, but for now,
2356 it's still accessed through vaguely CMU-CL-style functions and macros
2357 exported from the package SB-PROFILE.
2358 * Some problems left over from porting CMU CL to the new
2359 cross-compilation bootstrap process have been cleaned up:
2360 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2361 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2362 part when getting rid of the compiler *BACKEND* stuff.)
2363 ** The value of *NULL-TYPE* was screwed up, because it was
2364 being initialized before the type system knew the final
2365 definition of the 'NULL type. This screwed up several key
2366 optimizations in the compiler, causing inefficiency in all sorts
2367 of places. (I found it because I wanted to understand why
2368 GET-INTERNAL-RUN-TIME was consing.)
2369 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2370 PROCLAIM FTYPE information. Unfortunately this broke binary
2371 compatibility again, since now the forms output by DEFGENERIC
2372 to refer to functions which didn't exist in 0.6.3.
2373 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2374 can use the new (as of 0.6.3) transform for SOME into MAP into
2376 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2377 values to (INTEGER 0 1000000), so that the time code will no longer
2378 occasionally get blown up by Linux returning 1000000 microseconds
2379 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2380 its output conform to the ANSI spec. (Alas, this makes its output
2381 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2382 * A full call to MAP NIL with a single sequence argument no longer conses.
2383 * fixes to problems pointed out by Martin Atzmueller:
2384 * The manual page no longer talks about multiprocessing as though
2385 it were currently supported.
2386 * The ILISP support patches have been removed from the distribution,
2387 because as of version 5.10.1, ILISP now supports SBCL without us
2388 having to maintain patches.
2389 * added a modified version of Raymond Toy's recent CMU CL patch for
2390 EQUALP comparison of HASH-TABLE
2392 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2394 * The system still can't cross-compile itself with
2395 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2396 entails), but at least it can compile more of itself that way
2397 than it used to be able to, and various buglets which were uncovered
2398 by trying to cross-compile itself that way have now been fixed.
2399 * This release breaks binary compatibility again. This time
2400 at least I've incremented the FASL file format version to 2, so that the
2401 problem can be detected reliably instead of just causing weird errors.
2402 * various new style warnings:
2403 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2404 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2405 ** using the lexical binding of a variable named in the *FOO* style
2406 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2407 as specified by ANSI.
2408 * *RANDOM-STATE* is no longer automatically initialized from
2409 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2410 default behavior of the system is to repeat its behavior every time
2411 it's run. If you'd like to change this behavior, you can always
2412 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2413 old convention there was no comparably easy way to get the system to
2414 repeat its behavior every time it was run.
2415 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2416 FTYPE declarations has been removed, in favor of their ANSI
2417 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2418 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2419 the types of their sequence arguments can be determined at compile time.
2420 This is done through a new open code expansion for MAP which eliminates
2421 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2422 argument types can be determined at compile time.
2423 * The optimizer now transforms COERCE into an identity operation when it
2424 can prove that the coerced object is already of the correct type. (This
2425 can be a win for machine generated code, including the output of other
2426 optimization transforms, such as the MAP transform above.)
2427 * Credit information has been moved from source file headers into CREDITS.
2428 * Source file headers have been made more standard.
2429 * The CASE macro now compiles without complaining even when it has
2432 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2434 * (Note that the way that the PCL macroexpansions were rewritten
2435 to accommodate the change in DEFGENERIC below breaks binary
2436 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2437 not run under sbcl-0.6.2. Once we get out of alpha releases,
2438 i.e. hit release 1.0.0, we'll probably try to maintain binary
2439 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2440 and sbcl-1.4.4. Until then, however, it might be fairly common
2441 for maintenance releases to break binary compatibility.)
2442 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2444 * The reporting of unhandled signals has been changed to print some
2445 explanatory text as well as the report form. (Previously only
2446 the report form was printed.)
2447 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2448 it defines, so that the compiler no longer issues undefined function
2449 warnings for compiled-but-not-yet-loaded generic functions.
2450 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2451 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2452 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2453 section of the ANSI spec are supported.
2454 * The "" nickname for the "KEYWORD" package has been removed.
2455 The reader still handles symbol tokens which begin with a package marker
2456 as keywords, but it doesn't expose its mechanism for doing so in the
2457 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2458 * The system now issues STYLE-WARNINGs for contradictory TYPE
2459 proclamations. (Warnings for contradictory FTYPE proclamations would
2460 be nice too, but those can't be done usefully unless the type system
2461 is made smarter about FUNCTION types.)
2462 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2463 systematized, so that "*target-*.lisp is supposed to exist only on the
2464 target and imply that there's a related file which exists on the
2465 host, and *host-*.lisp is supposed to exist only on the host and imply
2466 that there's a related file which exists on the target. This involves a
2467 lot of renaming. Hopefully the acute confusion caused by the renaming
2468 will be justified by the reduction in chronic confusion..
2469 ** runtime-type.lisp -> early-target-type.lisp
2470 ** target-type.lisp -> late-target-type.lisp
2471 ** early-host-format.lisp -> early-format.lisp
2472 ** late-host-format.lisp -> late-format.lisp
2473 ** host-error.lisp -> misc-error.lisp
2474 ** early-error.lisp -> early-target-error.lisp
2475 ** late-error.lisp -> late-target-error.lisp
2476 ** host-defboot.lisp -> early-defboot.lisp
2477 ** code/misc.lisp -> code/target-misc.lisp
2478 ** code/host-misc.lisp -> code/misc.lisp
2479 ** code/numbers.lisp -> code/target-numbers.lisp
2480 ** code/early-numbers.lisp -> numbers.lisp
2481 ** early-host-type.lisp -> early-type.lisp
2482 ** late-host-type.lisp -> late-type.lisp
2483 ** host-typep.lisp -> typep.lisp
2484 ** load.lisp -> target-load.lisp
2485 ** host-load.lisp -> load.lisp
2486 ** host-disassem.lisp -> disassem.lisp
2487 ** host-insts.lisp -> insts.lisp
2488 ** byte-comp.lisp -> target-byte-comp.lisp
2489 ** host-byte-comp.lisp -> byte-comp.lisp
2490 ** host-signal.lisp -> signal.lisp
2491 ** host-defstruct.lisp -> defstruct.lisp
2492 ** late-target-type.lisp -> deftypes-for-target.lisp
2493 Furthermore, several other previously target-only files foo.lisp (e.g.
2494 hash-table.lisp and random.lisp) have been split into a target-and-host
2495 foo.lisp file and a target-only target-foo.lisp file, with their key type
2496 definitions in the target-and-host part, so that the cross-compiler will
2497 know more about target types.
2498 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2499 gone away. In their place are various *BACKEND-FOO* variables
2500 corresponding to the slots of the old structure.
2501 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2502 into the target SBCL has been fixed.
2503 * The chill.lisp system for loading cold code into a running SBCL
2505 * Support for the CMU CL "scavenger hook" extension has been removed.
2506 (It was undocumented and unused in the CMU CL sources that SBCL was
2507 derived from, and stale in sbcl-0.6.1.)
2508 * Various errors in the cross-compiler type system were detected
2509 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2510 (enabling various consistency checks). Many of them were fixed,
2511 but some hard problems remain, so the compiler is back to
2512 running without *TYPE-SYSTEM-INITIALIZED* for now.
2513 * As part of the cross-compiler type system cleanup, I implemented
2514 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2515 * I have started adding UNCROSS calls throughout the type system
2516 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2517 on cross-compiler input files will be able to go away, and various
2519 * CONSTANTP now returns true for quoted forms (as explicitly required
2522 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2524 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2525 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2526 (MAKE-LIST -1). In the longer term, it ought to become true
2527 that declarations are assertions even at SAFETY 1. For now, it's not
2528 quite true even at SAFETY 3, but it's at least more nearly true..
2529 (Note that this change seems to increases the size of the system by
2530 O(5%) and to decrease the speed of the compiler by 20% or more.)
2531 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2532 for the problem of printing dozens of lines of distracting information
2533 about low-level system machinery as part of the top stack frame
2534 on entry to the debugger when an undefined function was called.
2535 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2536 is not reset to COMMON-LISP-USER.
2537 * Compilation of stuff related to dyncount.lisp has been made conditional
2538 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2539 smaller. The various dyncount-related symbols have been moved into
2540 a new "SB-DYNCOUNT" package.
2541 * tty-inspect.lisp has been renamed to inspect.lisp.
2542 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2543 feature has gone away. (When we eventually port to other flavors of
2544 libc and/or Unix, we'll try to make the differences between flavors
2545 invisible at the user level.)
2546 * Various other *FEATURES* tags, and/or their associated conditionals,
2547 have been removed if obsolescent, or given better documentation, or
2548 sometimes given more-mnemonic names.
2550 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2552 * tidied up "make.sh" script
2553 * tidied up system directory structure
2554 * better "clean.sh" behavior
2555 * added doc/FOR-CMUCL-DEVELOPERS
2556 * many many small tweaks to output format, e.g. removing possibly-confusing
2557 trailing #\. character in DESCRIBE-INSTANCE
2558 * (EQUALP #\A 'A) no longer signals an error.
2559 * new hashing code, including EQUALP hashing
2560 * tidied up Lisp initialization and toplevel
2561 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2562 * command line argument processing
2563 * added POSIX-GETENV function to deal with Unix-ish environment variables
2564 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2565 terminating SBCL on EOF
2566 * non-verbose GC by default
2567 * There is no more "sbcl" shell script; the sbcl file is now the C
2568 runtime executable (just like CMU CL).
2569 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2571 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2572 symbol and package information as Lisp native symbol and package objects
2573 instead of strings naming symbols and strings naming packages. This way,
2574 whenever packages are renamed (as in warm init), debug information is
2575 transformed along with everything else.
2576 * tweaked the optimization policy declarations which control the building
2577 of SBCL itself. Now, among other things, the system no longer saves
2578 source location debugging information. (This helps two problems at once
2579 by reducing SBCL size and by keeping SBCL from trying to look for its
2580 sources -- which may not exist -- when reporting errors.)
2581 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2582 debugging and testing purposes
2583 * cleaned up printing, making the printer call PRINT-OBJECT for
2584 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2585 methods, giving nearly-ANSI behavior
2586 * converted almost all special variables to use *FOO* naming convention
2587 * deleted PARSE-TIME functionality, since it can be done portably
2588 * moved some files out of cold init into warm init
2589 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2591 * regularized formatting of source files
2592 * added an install.sh script
2593 * fixed ridiculous memory usage of cross-compiler by making
2594 compiler/alloc.lisp not try to do pooling unless it can hook
2595 itself into the GC of the cross-compilation host. Now the system
2596 builds nicely on my old laptop.
2597 * added :SB-ALLOC in target-features.lisp-expr
2598 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2599 was not implemented)
2600 * re-did condition handling and note reporting in the compiler. Notes
2601 are no longer handled by signalling conditions. Style warnings
2602 and warnings are handled more correctly and reported in such a way
2603 that it's easy to find one or the other in your output (so that you
2604 can e.g. figure out which of many problems caused COMPILE-FILE to
2606 * changed the severity of several compiler warnings from full WARNING
2607 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2608 compiler note reporting so that it doesn't use the condition system
2609 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2610 COMPILE-FILE command)
2611 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2612 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2613 no longer supported, since their implementation was deeply intertwingled
2614 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2615 * removed bogus "support" for compiler macros named (SETF FOO), and
2616 removed the compiler macro for SETF INFO (but only after making a fool
2617 of myself on the cmucl-imp mailing list by posting a bogus patch for
2618 DEFINE-COMPILER-MACRO..)
2619 * Compiled files containing forms which have side effects on the Lisp
2620 reader (such as DEFPACKAGE forms) are now handled more correctly.
2621 (Compiler queuing of top level lambdas has been suppressed by setting
2622 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2623 * deleted various currently-unused source files, e.g. gengc.lisp. They
2624 may be added back at some point e.g. when porting to other architectures,
2625 but until they are it's distracting to distribute them and to try to
2627 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2628 there were so many of them they're just distractions, and UNCROSS is
2629 known to be able to handle the current sources
2630 * moved PROFILE functionality into TRACE, so that it will be clear
2631 how the wrapping and unwrapping of functions when you profile them
2632 interacts with the wrapping and unwrapping of functions when you
2633 trace them. (Actually, the functionality isn't there yet, but at least
2634 the interface specification is there. Hopefully, the functionality will
2635 arrive with some future maintenance release.)
2636 * removed host-oops.lisp
2637 * changed signature of QUIT function to allow UNIX-CODE argument
2638 * fixed READ-SEQUENCE bug
2639 * tweaked verbose GC output so that it looks more like the progress
2640 output that ANSI specifies for functions like LOAD
2641 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2642 * added <http://sbcl.sourceforge.com> to the banner information printed by