1 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2 * bug fix: incorrect expansion of defgeneric that caused
3 a style warning. (thanks for Zach Beane)
5 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
6 * incompatible change: the internal functions
7 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
8 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
9 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
10 instead of the old functions.
11 * new feature: on platforms where "dladdr" is available foreign
12 function names now appear in backtraces. (based on Helmut Eller's
14 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
15 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
17 * documentation: ASDF manual is now distributed with SBCL.
18 * bug fix: SBCL can now load its contributed modules with REQUIRE
19 even if the system-provided entries have been removed from
20 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
22 * bug fix: backtraces involving undefined functions or assembly
23 routines are more informative. (thanks to Brian Downing)
24 * bug fix: mutually referent alien structures now work correctly.
25 (reported by Rick Taube)
26 * bug fix: structures defined by WITH-ALIEN can be referred to
27 within other definitions in the same WITH-ALIEN.
28 * bug fix: division operators (MOD, TRUNCATE and the like) with
29 constant zero divisors and integer dividends no longer generate
31 * bug fix: provide default methods for INPUT-STREAM-P and
32 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
33 * bug fix: improve the detection and resolution of MOP metacycles
34 (where effective method computation depends on the generic function
35 having its effective method computed). (reported by Bruno Haible)
36 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
37 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
38 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
39 * bug fix: pathnames with relative directory components can be
40 represented relative to default pathnames.
41 * optimization: in taking the GCD of bignums, reduce the two bignums
42 to approximately the same size (using Euclid's algorithm) before
43 applying the more sophisticated binary GCD. (thanks to Juho
45 * optimization: COUNT on bitvectors now operates word-at-a-time.
46 * optimization: ASH with a positive, but not necessarily constant,
47 (leftwards) shift, when in a modular context, is compiled to a
49 * fixed some bugs revealed by Paul Dietz' test suite:
50 ** FORMAT variable parameters ("~V<char>") are defaulted properly
51 if the corresponding argument is NIL.
52 ** FORMAT directives accepting a minpad parameter treat negative
54 ** literal commas as character parameters to FORMAT directives are
56 ** literal spaces directly after ~<Newline> directives within a
57 format-logical-block (~:< ~@:>) do not induce :FILL-style
59 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
60 as specified: it no longer includes conditional newlines).
61 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
63 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
64 :PER-LINE-PREFIX argument does not evaluate to a string.
66 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
67 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
68 the manual for details; for now, package locks can be disabled by
69 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
70 if no major problems are found then it is likely that they will be
71 compiled in unconditionally.
72 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
73 unsupported operators on all platforms. To load a shared library
74 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
75 object file, link it into a shared library outside of SBCL and
76 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
78 * bug workaround: ROOM T and the GC were not working together
79 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
80 reporting facility aren't preserved. That reporting has been
81 disabled, so now until and unless someone figures out how to make it
82 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
83 * minor incompatible change: as threatened around sbcl-0.8.0, the
84 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
85 The symbols are also exported from SB-PCL for backwards
86 compatibility, but more so than before SB-PCL should be treated as
87 an implementation-internal package.
88 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
90 * the SB-SPROF contrib now works on (most) non-x86 architectures.
91 It is known as of this release not to work on the Alpha, however.
92 * fixed bug #167: errors signalled due to illegal syntax in method
93 bodies are now more legible.
94 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
95 designators and can hence be used with TYPEP.
96 * fixed bug #333: CHECK-TYPE now ensures that the type error
97 signalled, if any, has the right object to be accessed by
98 TYPE-ERROR-DATUM. (reported by Tony Martinez)
99 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
100 5.1.2.3 for multiple-value place subforms. (reported by Kalle
102 * fixed bug #334: programmatic addition of slots using specialized
103 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
104 and :ALLOCATION :CLASS slots.
105 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
106 float exponents. (rereported by Peter Seibel)
107 * fixed a bug: #\Space (and other whitespace characters) are no
108 longer considered to be macro characters in standard syntax by
110 * fixed bug: initialization of condition class metaobjects no longer
111 causes an instance of the condition to be created. (reported by
113 * fixed bug: it is now possible to have slots such that
114 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
115 description is neither :INSTANCE nor :CLASS.
116 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
117 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
118 SLOT-VALUE-USING-CLASS), and no longer causes errors with
119 non-standard SLOT-DEFINITION-ALLOCATION values.
120 * fixed bugs: various race conditions handling exiting threads.
121 CL-PPCRE's thread torture test now passes.
122 * fixed arguably-a-bug: GC time proportional to number of threads in
123 system even when most of them are idle
124 * optimization: improved performance of BIT and SBIT on bit-vectors.
125 * .fasl file incompatibility: The fasl file version number has
126 been incremented because of changes associated with package locks.
128 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
129 * minor incompatible change: the system no longer provides
130 optimization hints (conditions which are TYPEP
131 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
132 mode; these hints are still emitted when higher SPEED optimization
134 * new contrib module: a sampling profiler (profiling by statistical
135 sampling, rather than by instrumenting functions) is available as
136 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
137 * the behaviour of the standard function ED is now customizeable by
138 third parties through a hook variable: see ED's documentation
139 string for information on the protocol.
140 * the compiler no longer emits efficiency notes for (FUNCALL X)
141 when the type of X is uncertain under default optimization
143 * fixed bug 276: mutating a binding of a specialized parameter to a
144 method to something that is not TYPEP the specializer is now
146 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
147 exhibit the required behaviour on the x86 platform. (thanks to
148 Peter van Eynde, Eric Marsden and Bruno Haible)
149 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
150 even for difficult arguments. (reported by Peter Graves)
151 * fixed bug 141a: the backquote printer now descends quoted
153 * fixed another bug in backquote printing: no more destructive
154 modification of the form's list structure. (reported by Brian
156 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
157 move between its address being taken and the call to
158 interrupt_thread, fixing a crashing race condition.
159 * the SB-POSIX contrib implementation has been adjusted so that it
160 no longer exhibits ridiculously poor performance when constructing
161 instances corresponding to C structs.
163 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
164 * minor incompatible change: the sb-grovel contrib now treats C
165 structures as alien (in the sense of SB-ALIEN) objects rather than
166 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
167 has implications for memory management of client code
168 (sb-grovel-returned objects must now be manually managed) and for
169 type safety (alien objects now have full types).
170 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
171 used to control emission of compiler diagnostics, rather than the
172 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
173 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
174 quality should be considered deprecated.
175 * install.sh now installs the user manual as well
176 * (not quite a new documentable feature, but worth considering in
177 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
178 beginnings of a semantically meaningful condition hierarchy is
179 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
181 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
182 designator as the defaults argument.
183 * fixed bug: Displaced arrays whose displaced-to array has become
184 too small now cause ARRAY-DIMENSION to signal an error, providing
185 for safer bounds-checking. (reported by Bruno Haible)
186 * fixed bug: DEFCLASS slot definitions with identical :READER and
187 :WRITER names now signal a reasonable error. (reported by Thomas
189 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
191 * fixed bug: Invalid slot specification errors now print correctly.
192 (thanks to Zach Beane)
193 * fixed bug 320: Shared to local slot value transfers in class
194 redefinitions now happen corrently from superclasses as
195 well. (reported by Bruno Haible)
196 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
198 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
199 improper lists correctly. (reported by Bruno Haible)
200 * fixed bug 313: source-transform for <fun-name> was erroneously
201 applied to a call of a value of a variable with name <fun-name>.
202 (reported by Antonio Menezes Leitao)
203 * fixed bug 307: The obsolete instance protocol ensures that
204 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
205 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
206 remove dead unknown-values globs from the middle of the stack.
207 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
208 easier to resume long computations after using *BREAK-ON-SIGNALS*
209 to diagnose and fix failures. (thanks to Nikodemus Siivola)
210 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
211 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
212 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
213 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
214 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
216 * fixed a (fairly theoretical) bug in string printing: if
217 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
218 does not have array-element-type equal to the most general string
220 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
221 function name. (thanks to Nikodemus Siivola)
222 * fixed bug: compile-time format string checker failed on
223 non-closed ~{. (reported by Thomas F Burdick)
224 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
225 don't warn on legal loop code involving more than one aggregate
226 boolean. (thanks to Nikodemus Siivola)
227 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
228 in keyword position to MAKE-PACKAGE was not being reported as
230 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
231 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
233 * fixed bugs 280 and 312: the checking for multiple definitions in a
234 file is less likely to become confused by uses of inline
236 * fixed bug: the #S reader macro performs the keyword coercion
237 specified for slot names. (reported by Kalle Niemitalo)
238 * fixed bug: lambda lists may contain symbols whose names start with
239 & but are not lambda-list-keywords; their occurrence triggers a
241 * fixed bug 321: define-method-combination argument lambda lists do
242 not cause invalid code to be generated when &optional or &aux
243 variables have default values. (reported by Bruno Haible)
244 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
245 have CLOS classes; this appears also to have fixed reported
246 instability in stack exhaustion detection.
247 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
248 MUFFLE-WARNING when no associated restart is present is now
250 * optimization: rearranged the expansion of various defining macros
251 so that each expands into only one top-level form in a
252 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
254 * optimization: used a previously little-used slot in symbols to
255 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
257 * fixed some bugs revealed by Paul Dietz' test suite:
258 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
260 ** Fixed an optimization bug related to inheritance of initforms
261 from local to shared slots.
262 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
263 ** CAST optimizer forgot to flush argument derived type.
264 ** print/read consistency on floats is now orders of magnitude
265 more likely. (thanks also to Bruno Haible for a similar report
267 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
268 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
269 for objects of type REAL. Make it so.
270 ** SXHASH is specified (infelicitously) to respect similarity,
271 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
272 it so. (thanks to Markus Ziegler)
273 ** on the Alpha, the compiler succeeds in compiling functions
274 returning a known number of arguments greater than 63.
275 ** fixed handling of invalid NIL arguments in keyword position
277 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
279 ** provided a readably-printable representation for RANDOM-STATE
281 ** ensured that pathnames, where they have a namestring, always
282 print using #P"..." syntax.
284 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
285 * Support for the forthcoming 2.0 version of the NetBSD kernel
286 running on x86 hardware has been added. (thanks to Perry
287 E. Metzger most immediately, and others for their past work)
288 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
289 current release version; SBCL's OpenBSD support had been broken
290 since about the time of OpenBSD's switch to ELF binary format.)
291 * [placeholder for DX summary]
292 ** user code with &REST lists declared dynamic-extent, under high
293 speed or space and low safety and debug optimization policy.
294 * The manual has been converted to Texinfo format and the debugger
295 chapter from the cmucl manual has been added.
296 * A facility has been added to extract documentation strings from
297 sbcl and store them as Texinfo-formatted snippets for inclusion in
298 the manual (via Texinfo's @include directive)
299 * bug fix: compiler emitted division in optimized DEREF. (thanks for
300 the test case to Dave Roberts)
301 * bug fix: multidimensional simple arrays loaded from FASLs had fill
302 pointers. (reported by Sean Ross)
303 * bug fix: PROFILE output is printed nicely even for large numerical
304 values. (thanks to Zach Beane)
305 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
306 greater than 32 handle EOF correctly.
307 * bug fix: on X86 an immediate argument of the IMUL instruction is
308 correctly printed in disassembly. (thanks to Lutz Euler)
309 * bug fix: class slots in redefined classes preserve their old
310 values. (thanks to Bruno Haible and Nikodemus Siivola)
311 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
312 succeeds. (reported by Marco Baringer)
313 * fixed some bugs revealed by Paul Dietz' test suite:
314 ** READ-SEQUENCE now works on ECHO-STREAMs.
315 ** RATIONALIZE works more according to its specification. (thanks
318 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
319 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
320 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
321 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
322 should matter to you only if you rebind the printer control
323 variables and then find you want different bindings in the
324 debugger than in the ordinary execution of your program.)
325 * The runtime build system has been tweaked to support building
326 (on SPARC/SunOS) using a C compiler which invokes Sun's own
327 assembler and linker. (thanks to Nikodemus Siivola)
328 * Unbound, undefined, undeclared variables now trigger full
329 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
330 more often programmer error than deliberate exploitation of undefined
332 * optimization: the hash algorithm for strings has changed to one
333 that is less vulnerable to spurious collisions. (thanks to Juho
335 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
336 needless bounds checking. (thanks to Juho Snellman)
337 * optimization: implemented multiplication as a modular
338 (UNSIGNED-BYTE 32) operation on the PPC backend.
339 * fixed some bugs revealed by Paul Dietz' test suite:
340 ** ADJUST-ARRAY now copies the datum in a zero rank array if
342 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
343 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
344 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
346 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
348 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
350 ** ECHO-STREAMs no longer attempt to echo the end of file value to
351 their output stream on EOF from read.
352 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
353 have been read to end-of-file.
354 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
356 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
357 description of determination of which consecutive characters
359 ** Printing the "Space" character with escaping on now yields "#\\ ",
360 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
361 ** Reading floating-point numbers with *READ-BASE* set to a number
362 less than 10 works correctly.
363 ** Reading floating-point numbers with *READ-BASE* set to a number
364 more than 10 works correctly.
365 ** Printing with *PRINT-READABLY* targets the standard readtable, not
366 the readtable currently in effect.
368 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
369 * minor incompatible change: parsing of namestrings on a physical
370 (Unix) host has changed; numbers after the final #\. in a
371 namestring are no longer interpreted as a version field. This is
372 intented to be largely invisible to the user, except that the
373 meaning of the namestring "*.*.*" has changed: it now refers to a
374 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
375 should usually be replaced by
376 (make-pathname :name :wild :type :wild :version :wild)
377 with the added benefit that this is more likely to be portable.
378 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
379 option to OPEN now signals an error if the file being opened
380 exists; this may have an impact on existing code.
381 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
382 signals received, on the PowerPC platforms (both Linux and
383 Darwin). (thanks to Pierre Mai for pointing out the location of
385 * several fixes on OS X: The system now builds and runs cleanly on
386 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
387 (thanks to Brian Mastenbrook)
388 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
389 second-resolution: integer multiples of 1/3600 between -24 and 24.
390 (thanks to Vincent Arkesteijn)
391 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
392 type when called with 1 argument; PEEK-CHAR checked type of
393 PEEK-TYPE only after having read first character from a
394 stream. (reported by Peter Graves)
395 * bug fix: the garbage collector now has much better locality
396 behaviour, and in particular no longer treats all memory as being
397 exhausted when many small objects point to each other in a deeply
399 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
400 recognized as being TYPEP their class.
401 * bug fix: the PUSHNEW documentation string has been corrected.
402 (thanks to Vincent Arkesteijn)
403 * bug fix: defaulting of the value for the last of an atypically
404 large number of multiple values being bound was not being
405 performed correctly on the Alpha or PPC platforms
406 * optimization: implemented multiplication as a modular
407 (UNSIGNED-BYTE 32) operation on the x86 backend.
408 * optimization: SEARCH on simple-base-strings can now be open-coded.
409 (see also contrib/compiler-extras.lisp for inspiration for
410 teaching the compiler about the Boyer-Moore algorithm).
411 * value, returned by MAX (and MIN) called with several EQUALP, but
412 not EQL, arguments now does not depend on compiler settings.
413 * fixed some bugs revealed by Paul Dietz' test suite:
414 ** in stack analysis liveness information is propagated from
415 non-local entry points.
416 ** pathwise CAST removing failed when the CAST node did not start
418 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
419 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
421 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
422 host is already defined.
423 ** RENAME-FILE works on streams instead of signalling an internal
425 ** PEEK-CHAR uses the current readtable when determining whether
426 or not a character is whitespace.
427 ** MERGE-PATHNAMES handles the case when the pathname does not
428 specify a name while the default-pathname specifies a version.
429 ** Pathnames now stand a better chance of respecting print/read
431 ** Attempting to use standardized file system operators with a
432 pathname with invalid :DIRECTORY components signals a
434 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
436 ** DIRECTORY on logical pathnames is more correct.
437 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
438 signal a TYPE-ERROR if their argument is not a stream.
439 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
440 designator argument does not designate a stream.
441 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
442 examining the synonym.
443 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
445 ** OPEN and WITH-OPEN-STREAM allow opening streams with
446 element-type larger than ([UN]SIGNED-BYTE 32).
448 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
449 * When built with the :SB-FUTEX feature, threaded builds now take
450 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
451 for faster/more reliable mutex and condition variable support.
452 * Incompatible change (but one you probably shouldn't have been using
453 anyway): the interface and code for arbitrating between multiple
454 threads in the same user session has been redesigned.
455 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
456 error when called without an explicit environment argument.
457 (thanks to Nikodemus Siivola)
458 * bug fix: buffered :DIRECTION :IO streams are less likely to become
459 confused about their position. (thanks to Adam Warner and Gerd
461 * bug fix: Pretty printing backquoted forms with unquotations in the
462 argument list position of various code constructs such as LAMBDA
463 now works correctly. (reported by Paul Dietz)
464 * bug fix: Pretty printing unquotations no longer loses all
465 stream position information.
466 * optimization: performance of string output streams is now less
467 poor for multiple small sequence writes.
468 * optimization: performance of CSUBTYPEP in the presence of complex
469 expressions involving CONS and NOT many times has been improved.
470 (reported by Paul Dietz)
471 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
473 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
474 streams for the REPL, for the debugger, and so on. (thanks to
476 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
477 * fixed some bugs revealed by Paul Dietz' test suite:
478 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
479 is ignored if it is too small, rather than propagating through
480 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
481 ** extremely complex negations of CONS types were not being
482 sufficiently canonicalized, leading to inconsistencies in
484 ** VALUES tranformer lost derived type.
486 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
487 * fixed a bootstrapping bug: the build process no longer assumes
488 that the various BOOLE-related constants have the same value in
489 host and target lisps. (noted by Paul Dietz' test suite on an
490 SBCL binary built from CLISP)
491 * The system can now be dynamically linked on the MIPS platform,
492 which enables dynamic loading of foreign code from Lisp. (thanks
493 to Ralf Baechle for discussions on the MIPS ABI)
494 * The system now records debugging information for its own source
495 files in a filesystem-position-independent manner, relative to
496 the "SYS" logical host.
497 * fixed a compiler bug: MV-LET convertion did not check references
498 to the "max args" entry point. (reported by Brian Downing)
499 * tweaked disassembly notes to be less confident about proclaiming
500 some instruction as an LRA. (thanks to Brian Downing)
501 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
502 now each have their own history, command character, and other
503 characteristics. (thanks to David Lichteblau)
504 * fixed some bugs revealed by Paul Dietz' test suite:
505 ** compiler failure in compiling LOGAND expressions including a
507 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
508 shift greater than 32.
509 ** FLUSH-DEST did not mark blocks for type check regeneration.
510 ** HANDLER-CASE failed to accept declarations in handler clauses
511 in some circumstances.
513 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
514 * New code in contrib/sb-introspect (still probably not entirely
515 stable yet) provides some support for smart Lisp development
516 environments like SLIME.
517 * The conditions signalled for errors occurring when loading .fasl
518 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
519 in a way which should help ASDF recover gracefully.
520 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
521 stuff has been cleaned up. If you code contrib/ stuff, this might
522 affect you, and you can look at contrib/README, contrib/STANDARDS,
523 and/or the 0.8.4.27 diff to check.
524 * In full calls the compiler now does not generate checks for declared
525 argument types for all arguments.
526 * various threading fixes
527 ** and some experimental patches which didn't make it into
528 the main tree for this release, but which are shipped in
529 contrib/experimental-thread.patch as a possible fix for some
530 failures (deadlock, spinning...) in GC-intensive multithreaded
532 * fixed PPC build problem (source code incompatibility of different
533 library versions): added offsetof() hackery which attempts to divine
534 where glibc maintainers put uc_mcontext today
535 * fixed bug 282: compiler does not trust type assertions while passing
536 arguments to a full call.
537 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
538 &OPTIONAL VALUES type parameter.
539 * fix bug 214: algorithm for noting rejected templates is now more
540 similar to that of template seletion. (also reported by rydis on
542 * fixed bug 141b: printing backquoted information readably and prettily
543 inserts a space where necessary.
544 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
545 (CHARACTER 10) are now reported as errors, rather than propagated
546 as unknown types. (reported by piso on #lisp)
547 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
548 with values NIL and :ERROR. (thanks to Milan Zamazal)
549 * fixed bug 191c: CLOS now does proper keyword argument checking as
550 described in CLHS 7.6.5 and 7.6.5.1.
551 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
552 longer raise an error; further, using a list as a for-as-arithmetic
553 counter now raises a meaningful error.
554 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
555 now understood by sequence creation functions such as MAKE-SEQUENCE
557 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
558 a STRING-INPUT-STREAM.
559 * compiler enhancement: SIGNUM is now better able to derive the type
561 * type declarations inside WITH-SLOTS are checked. (reported by
563 * fixed some bugs revealed by Paul Dietz' test suite:
564 ** incorrect optimization of TRUNCATE for a positive first
565 argument and negative second.
566 ** compiler failure in let-convertion during flushing dead code.
567 ** compiler failure while deriving type of TRUNCATE on an
568 interval, containing 0.
569 ** ASH of a negative bignum by a negative bignum count now returns
571 ** intersection of CONS types now canonicalizes properly, fixing
572 inconsistencies in SUBTYPEP.
574 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
575 * incompatible change: The --disable-debugger command line
576 option now clobbers the debugger at a more fundamental
577 level, by redefining #'INVOKE-DEBUGGER instead of by
578 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
579 is specified by ANSI to ignore *DEBUGGER-HOOK* and
580 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
581 enter the debugger REPL and then suffer recursive errors
582 because *DEBUG-IO* is also messed up in --disable-debugger mode;
583 while under the new system, BREAK in --disable-debugger mode
584 terminates the system just as an unhandled error would.
585 * fixed compiler performance when processing loops with a step >1;
586 * bug fix: DOCUMENTATION now retrieves generic function
587 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
588 support has been systematized, and now supports the methods
589 specified by ANSI, along with a default method and a method for
590 slot documentation. (reported by Nathan Froyd)
591 * bug fix: effective methods associated with a generic function are
592 no longer cached over a change of that generic function's method
593 combination. (reported by Andreas Fuchs)
594 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
595 names a non-existent directory. (thanks to Andreas Fuchs)
596 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
597 the PPC platform, lead to stack corruption; this has been fixed.
598 (reported by Rainer Joswig)
599 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
600 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
601 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
602 arithmetic in general, is now much more reliable on the PPC
604 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
605 platform now returns the right answer.
606 * optimization: restored some effective method precomputation in
607 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
608 precomputation is now tunable.
609 * optimization: compiler-internal data structure use has been
610 reviewed, and changes have been made that should improve the
611 performance of the compiler by about 20%.
612 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
613 simple-base-strings and simple-bit-vectors is improved.
614 * optimization: the optimization of 32-bit logical and arithmetic
615 functions introduced in version 0.8.3 on the x86 has been
616 implemented on the mips, ppc and sparc platforms; an
617 implementation of the same facility, but for 64-bit arithmetic,
618 has been added for the alpha.
619 * microoptimization: the compiler is better able to make use of the
620 x86 LEA instruction for multiplication by constants.
621 * bug fix: in some situations compiler did not report usage of
622 generic arithmetic in (SPEED 3) policy.
623 * bug 145b fix: compiler used wrong type specifier while converting
624 MEMBER-types to numeric.
625 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
627 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
628 index argument now works properly on the Alpha platform.
629 * bug fix: floating point exception treatment on the Alpha platform
631 * bug fix: FILE-POSITION works much better on string input and
632 output streams. (thanks to Nikodemus Siivola)
633 * bug fix: many threading/garbage collection symptoms sorted.
634 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
635 might be pseudo-atomic.
636 * internal change: Stopping for GC is now done with signals not
637 ptrace. GC is now done in whichever thread wanted it, instead of
639 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
641 * bug fix: result form in DO is not contained in the implicit
643 * incompatible change: ICR structure is changed; the value part of
644 CONTINUATION is now called LVAR; corresponding functions are
645 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
646 * added type deriver for ISQRT (thanks to Robert E. Brown).
647 * bug fix: better support for loading from the command line when an
648 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
650 * fixed some bugs revealed by Paul Dietz' test suite:
651 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
652 ** ROUND and FROUND now give the right answer when given very
653 small float arguments.
654 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
656 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
657 ** (GCD 0 <negative-integer>) returned <negative-integer>.
658 ** LCM should return a non-negative integer.
659 ** PARSE-INTEGER returned the index of a terminator instead of the
660 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
661 ** PARSE-INTEGER returned an incorrect index being applied to a
663 ** LCM with two arguments of 0 returns 0 rather than signalling
665 ** unsigned addition of a 32-bit constant with the high bit set no
666 longer causes an internal compiler error on the x86.
667 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
668 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
669 with negative last argument.
670 ** byte specifiers with zero size and position no longer cause
671 an error during type derivation.
672 ** bignum multiplication on the Alpha platform now returns the
674 * porting: The system now builds on SuSE AMD64, although it still
675 generates a 32-bit binary.
676 * .fasl file incompatibility: The fasl file version number has
677 been incremented (because of the changes to internal compiler
678 data structures referred to above).
680 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
681 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
682 more accurately, on the Darwin kernel running on PowerPC hardware.
683 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
684 * Compiler code deletion notes now signal a condition of type
685 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
686 an associated MUFFLE-WARNING restart.
687 * The compiler now performs limited argument count validation of
688 constant format strings in FORMAT, and where appropriate in ERROR,
689 CERROR and WARN. (thanks to Gerd Moellmann)
690 * New ASDF-INSTALL contrib can be used for automatic download and
691 installation of third-party Lisp code from CCLAN or other sites
693 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
694 which forces another thread to execute a function supplied by the
696 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
697 now accept and act upon their :ELEMENT-TYPE keyword argument.
698 (reported by Martin Atzmueller, Edi Weitz)
699 * bug fix: FILE-POSITION now accepts position designators up to
700 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
701 is the greater. (thanks to Patrik Nordebo)
702 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
703 Kalle Olavi Niemitalo)
704 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
705 printable. (reported by Eric Marsden)
706 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
707 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
708 * x86 bug fix in control stack exhaustion checking: now shows backtrace
709 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
710 (thanks to Stig Sandoe)
711 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
712 UNREAD-CHAR now correctly includes the unread character in the
713 target sequence. (thanks to Gerd Moellmann)
714 * bug fix in threaded builds: the system can now be suspended and
715 resumed by shell job control with minimal disruption.
716 * bug fixes in times and timezones >2038AD
717 * better handling of "where is GNU make?" problem in build scripts
718 (thanks to Nikodemus Siivola)
719 * new optimization: inside a named function any reference to a
720 function with the same name is considered to be a self-reference;
721 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
722 optimization quality.
723 * new optimization on x86: logical functions and + now have
724 optimized (UNSIGNED-BYTE 32) versions, which are automatically
725 used when the result is truncated to 32 bits.
726 * VALUES declaration is partially enabled.
727 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
728 * fixed some bugs revealed by Paul Dietz' test suite:
729 ** The system now obeys the constraint imposed by
730 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
731 types form a lattice under type intersection.
732 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
733 ** ASSOC now ignores NIL elements in an alist.
734 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
735 and (1+ MOST-POSITIVE-FIXNUM) answers.
736 ** The addition of a method with invalid qualifiers to a generic
737 function does not cause an error to be signalled immediately;
738 a warning is signalled, and the error is generated only on
739 calling the generic function.
740 * changed .fasl file version number, in part to add type codes for
741 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
742 obscure ANSI requirements
744 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
745 * fixed bug 148: failure to inline-expand a local function left
746 garbage, confusing the compiler.
747 * fixed bugs 3cd: structure slot readers perform type check if the
748 slot can have an invalid value (i.e. it is either not initialized
749 or can be written with a less specific slot writer).
750 * bug fix: the compiler now traps array references to elements off
751 the end of an array; previously, the bounds checking in some
752 circumstances could go off-by-one.
753 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
755 * type declarations for array element types now obey the description
756 on the CLHS page "Declaration TYPE", as per discussions on
757 sbcl-help around 2003-05-08. This means that a declaration
758 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
759 declaration, all references to BAR will be asserted or assumed
760 (with THE, so dependent on compiler policy) to involve objects of
761 type FOO. Note that no such declaration is implied in
762 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
763 * declared types of functions from the "Conditions"
764 chapter. (reported by Paul Dietz)
765 * bug fix: CERROR accepts a function as its first argument.
766 * bug fix: NTH an NTHCDR accept a bignum as index
767 arguments. (reported by Adam Warner)
768 * optimization: character compare routines now optimize comparing
769 against a constant character. (reported by Gilbert Baumann)
770 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
771 argument now works properly on the MIPS platform.
772 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
773 * repeated evaluation of the same DEFSTRUCT, a slot of which is
774 declared to have a functional type, does not cause an error
776 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
777 many :DEFAULT-INITARGS. (reported by Istvan Marko)
778 * fixed bug: if last continuation of a deleted block has a
779 destination, this destination should be deleted too. (reported by
781 * fixed a bug in the bootstrap process: the host compiler's values
782 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
783 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
784 test case from Patrik Nordebo)
785 * improved the ability of the disassembler on the PPC platform to
786 provide helpful disassembly notes.
787 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
788 the class in more cases than previously.
789 * bug fix: FILE-POSITION now understands :START and :END for
790 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
791 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
792 8) data. (thanks to David Lichteblau)
793 * bug fix: it is possible to add a method to a generic function
795 * bug fix: reader failed to signal END-OF-FILE inside an
796 object representation. (reported by Nikodemus Siivola)
797 * fixed some bugs revealed by Paul Dietz' test suite:
798 ** LAST and [N]BUTLAST should accept a bignum.
799 ** condition slot accessors are methods.
800 ** (VECTOR NIL) is a subtype of STRING.
802 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
803 * minor incompatible change: some nonsensical specialized lambda
804 lists (used in DEFMETHOD) which were previously ignored now signal
806 * minor incompatible change: the system is now aware of the types of
807 variables in the COMMON-LISP package, and will signal errors for
808 most violations of these type constraints (where previously they
809 were silently accepted).
810 * minor incompatible change: COMPILE-FILE now uses the freedom
811 afforded (ANSI 3.2.2.3) to use derived function types for
812 functions defined in the same file. This also permits the system
813 to warn on static type mismatches and function
814 redefinition. (Currently it does not work with high DEBUG level.)
815 * minor incompatible change: VALUES declaration is disabled.
816 * When issuing notes, the compiler now signals a condition of type
817 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
818 restart for use in user handlers. It is expected that the
819 COMPILER-NOTE condition will eventually become a condition
820 supertype to a hierarchy of note types, which will then be
821 handleable in a similar fashion. However, at the moment, no such
822 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
823 but it's an implementation detail, not a classification for the
825 * Changes in type checking closed the following bugs:
826 ** type checking of unused values (192b, 194d, 203);
827 ** template selection based on unsafe type assertions (192c, 236);
828 ** type checking in branches (194bc).
829 * A short form of VALUES type specifier has ANSI meaning (it has
830 increased the number of situations when SBCL cannot perform type
832 * fixed bug in DEFSTRUCT: once again, naming structure slots with
833 keywords or constants is permissible.
834 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
835 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
836 classes. (thanks to Antonio Martinez)
837 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
838 outputs FRESH-LINE or TERPRI, and no longer converts its stream
839 argument to a pretty-print stream. Instead, it leaves any such
840 operations to DESCRIBE-OBJECT methods.
841 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
843 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
844 lambda lists. (thanks to David Lichteblau)
845 * bug fix: Defining a generic function with a :METHOD-CLASS being a
846 subclass of STANDARD-METHOD no longer causes stack exhaustion.
847 (thanks to Gerd Moellmann)
848 * fixed bug 246: increased compilation speed of long
849 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
851 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
853 * DEFCONSTANT now throws a condition of type
854 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
855 constant to a non-EQL value; CONTINUE and ABORT restarts
856 respectively change and preserve the value.
857 * fixed bug 63: The code walker, part of the implementation of CLOS,
858 is now better at handling symbol macros.
859 * bug fix: There is no longer an internal implementation type named
860 CL:LENGTH. (reported by Raymond Toy)
861 * bug fix: In macro-like defining macros/special operators the
862 implicit block does not enclose the lambda list.
863 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
864 enclosed in parentheses) are not suitable as type specifiers, and
865 their use properly signals an error now.
866 * bug fix: An argument count mismatch for a type specifier in code
867 being compiled no longer causes an unhandled error at compile
868 time, but signals a compile-time warning.
869 * fixed simple vector readable printing
870 * bug fix: DESCRIBE takes more care over whether the class
871 precedence list slot of a class is bound before accessing it.
872 (reported by Markus Krummenacker)
873 * bug fix: FORMATTER can successfully compile pretty-printer format
874 strings which use variants of the ~* directive inside.
875 * bug fix: SEARCH now applies its TEST predicate to the elements of
876 the arguments in the correct order. (thanks to Wolfhard Buss)
877 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
878 in type inference. (thanks to Robert E. Brown)
879 * bug fix: Reading in symbols with an explicit package name of ""
880 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
881 * fixed some bugs revealed by Paul Dietz' test suite:
882 ** NIL is now allowed as a structure slot name.
883 ** Arbitrary numbers, not just REALs, are allowed in certain
884 circumstances in LOOP for-as-arithmetic clauses.
885 ** Multiple class redefinitions before slot access no longer
887 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
888 association between the name and a class.
889 ** Generic functions with non-standard method-combination and over
890 five methods all of which return constants no longer return NIL
891 after the first few invocations. (thanks to Gerd Moellmann)
892 ** CALL-NEXT-METHOD with no arguments now passes the original
893 values of the arguments, even in the presence of assignment.
894 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
896 ** DISASSEMBLE works with closures and funcallable instances.
897 ** ADD-METHOD now returns the generic function, not the new method.
898 ** FIND-METHOD signals an error if the lengths of the specializers
899 is incompatible with the generic function, even if the ERRORP
901 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
902 which its argument is a member.
903 ** DEFCLASS only redefines the class named by its class-name
904 argument if that name is the proper name of the class;
905 otherwise, it creates a new class.
906 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
907 of the UNBOUND-SLOT condition to the name of the slot.
908 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
909 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
910 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
911 SLOT-MAKUNBOUND in the specified fashion.
913 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
914 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
915 cross-compilation host. As a consequence, we can now bootstrap our
916 way up to SBCL starting with a bare gcc toolchain and human-readable
917 source code (first the source to CLISP, then the source to SBCL).
918 * A contributed module containing a partial implementation of the
919 simple-streams interface has been included. (thanks to Rudi
921 * A contributed module implementing the RFC1321 Message Digest
922 Algorithm, known as MD5, has been included.
923 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
924 no longer has any effect, as the code controlled by this feature
925 has been deleted. (As far as we know, no-one has ever built using
926 this feature, and its semantics were confused in any case).
927 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
928 work on conditions (as required by the ANSI specification),
929 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
930 expected behaviour on conditions. Users should note, however,
931 that such behaviour is not required by the ANSI specification,
932 and so use of this behaviour may render their code unportable.
933 * fixed some bugs revealed by Paul Dietz' test suite:
934 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
936 ** &ENVIRONMENT parameter in macro lambda list is bound first.
937 ** SXHASH on condition objects no longer returns NIL.
938 ** :ALLOCATION :CLASS slots are better treated; their values are
939 updated on class redefinition, and initforms inherited from
940 superclasses are applied.
941 ** REMOVE-METHOD returns its generic function argument even when
942 no method was removed.
943 ** SHARED-INITIALIZE now initializes the values of the requested
944 slots, including those with :ALLOCATION :CLASS.
945 ** ALLOCATE-INSTANCE now works on structure classes defined via
946 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
948 ** SLOT-EXISTS-P now works on conditions, as well as structures
950 ** MAKE-LOAD-FORM now has the required methods on
951 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
952 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
953 keyword, and now implements the SLOT-NAMES argument.
954 ** methods with &OPTIONAL arguments no longer allow too many
955 arguments to be passed in the call without error.
956 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
957 option is consistent with the required arguments of the generic
958 function lambda list.
959 * bug fix: REQUIRE accepts a string designator. (Thanks to
961 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
962 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
963 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
964 * bug fix: The compiler now checks for duplicated variables in macro
966 * bug fix: SETQ on globals returns the correct value.
967 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
968 not a COMPILER-ERROR (followed by some other strange error on
969 choosing the CONTINUE restart).
970 * bug fix: make.sh and friends are now more consistent in the way that
971 they look for GNU "make".
973 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
974 * experimental native threads support (on x86 Linux >=2.4 only).
975 This is not compiled in by default: you need to add :SB-THREAD to
976 the target features. See the "Beyond ANSI" chapter of the manual
978 * fix for longstanding nonANSIism: The old distinction between
979 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
980 The return value from CL:FIND-CLASS is now a CLOS class, and
981 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
982 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
983 * An interface to the MetaObject Protocol, as described in Kiczales,
984 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
985 MIT Press, 1991, is available from the SB-MOP package.
986 * incompatible change: the SB-PCL package should now be considered
987 a private implementation detail, and no longer a semi-private MOP
989 * minor incompatible change: due to rearrangement for threads, the
990 control stack and binding stack are now allocated at arbitrary
991 addresses instead of being hardcoded per-port. Users affected by
992 this probably have to be doing advanced things with shared
993 libraries, and will know who they are.
994 * minor incompatible change: Previously, all --eval forms used were
995 processed with READ before any of them were processed with EVAL.
996 Now each --eval form is processed with both READ and EVAL before
997 the next --eval form is processed. (Thus package operations like
998 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
999 work as the user might reasonably expect.)
1000 * minor incompatible change: *STANDARD-INPUT* is now only an
1001 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1003 * minor incompatible change: Y-OR-N-P is now character-oriented,
1004 not line oriented. Also, YES-OR-NO-P now works without errors.
1005 (thanks to Antonio Martinez)
1006 * sb-aclrepl module improvements: an integrated inspector, added
1007 repl features, and a bug fix to :trace command.
1008 * Known functions, which cannot be open coded by the backend, are
1009 considered to be able to check types of their arguments. (fixing
1010 a bug report by Nathan J. Froyd)
1011 * fixed a bug in computing method discriminating functions: It is
1012 now possible to define methods specialized on classes which have
1013 forward-referenced superclasses. (thanks to Gerd Moellmann)
1014 * fixed evaluation order in optional entries (reported by Gilbert
1016 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1017 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1018 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1019 required-by-AMOP NAME argument, as well as CLASS and
1020 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1021 * fixed bug 20: DEFMETHOD can define methods using names that are
1022 not the proper names of classes to designate class specializers.
1023 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1024 Unix file descriptors, instead of blowing up. (thanks to Antonio
1026 * Garbage collection refactoring: user-visible change is that a
1027 call to the GC function during WITHOUT-GCING will not do garbage
1028 collection until the end of the WITHOUT-GCING. If you were doing
1029 this you were probably losing anyway.
1030 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1031 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1032 TYPEP the latter but not the former.
1033 * The compiler issues a full WARNING for calls to undefined functions
1034 with names from the CL package.
1035 * MAP-INTO for a vector destination is open coded. (reported by
1036 Brian Downing on c.l.l)
1037 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1038 documentation string.
1039 * fixed some bugs revealed by Paul Dietz' test suite:
1040 ** COPY-ALIST now signals an error if its argument is a dotted
1042 ** Condition slots are now accessed more correctly in the presence
1043 of multiple initargs for a given slot.
1044 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1045 exclude restarts of the same name associated with a different
1047 ** DEFCLASS of forward-referenced classes with another
1048 forward-referenced class in the superclasses list no longer
1050 ** Condition slots are now initialized once each, not multiple
1051 times. (thanks to Gerd Moellmann)
1052 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1054 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1055 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1056 arguments contain duplicated elements.
1057 ** RESTART-CASE understands local macros.
1058 ** RESTART-CASE associates exactly its own restarts with a condition.
1059 ** ENDP in safe mode checks its argument to be of type LIST.
1060 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1061 in question is unbound.
1062 ** Optimized MAKE-INSTANCE functions no longer cause internal
1063 assertion failures in the presence of duplicate initargs.
1064 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1065 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1066 propagated even in the ctor optimized implementation of
1068 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1070 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1071 * a better implementation of SXHASH on (simple) bit vectors,
1072 measured both in execution speed and in distribution of results
1073 over the positive fixnums, has been installed. Likewise, a better
1074 implementation of EQUAL for simple bit vectors is now available.
1075 * fixed CEILING optimization for a divisor of form 2^k.
1076 * fixed bug 240 (emitting extra style warnings "using the lexical
1077 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1078 by Antonio Martinez)
1079 * fixed SXHASH, giving different results for NIL depending on type
1080 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1081 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1082 variable. (found by Rolf Wester)
1083 * SBCL does not ignore type declarations for special
1084 variables. (reported by rif on c.l.l 2003-03-05)
1085 * some bug fixes in contrib/sb-aclrepl/
1086 * fixed some bugs revealed by Paul Dietz' test suite:
1087 ** a bug in the CONS type specifier, whereby the CAR and CDR
1088 types got intertwined, has been fixed;
1089 ** the type system is now able to reason about the interaction
1090 between INTEGER and RATIO types more completely;
1091 ** APPEND, [N]REVERSE and NRECONC check that those their
1092 arguments, which must be proper lists, are really so;
1093 ** An array specialized to be unable to hold elements has been
1094 implemented, as required -- yes, really -- by ANSI;
1095 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1096 on malformed property lists;
1098 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1099 * incompatible packaging change: in line with Unix convention,
1100 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1101 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1102 * REQUIRE and PROVIDE are now optionally capable of doing something
1103 useful. See the documentation string for REQUIRE.
1104 * infrastructure for a managed SBCL contrib system: contributed
1105 modules in this release include:
1106 ** the ASDF system definition facility;
1107 ** an interface to the BSD Sockets API;
1108 ** an ACL-like convenience interface to the repl;
1109 (thanks to Kevin Rosenberg)
1110 ** an implementation of ROTATE-BYTE, with efficient implementation
1112 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1113 gives a closer approximation to the right answer than previously.
1114 (thanks to Raymond Toy)
1115 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1116 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1117 optional environment arguments, as required by ANSI.
1118 * fixed bugs in other functions taking environment objects, allowing
1119 calls with an explicit NIL environment argument to be compiled
1121 * fixed bug 228: primary return values from
1122 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1123 COMPILE or FUNCTION.
1124 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1125 :TYPE will no longer signal errors on innocuous objects.
1126 * fixed bug 231b: SETQ is better at respecting type declarations in
1127 the lexical environment.
1128 * fixed a bug in DEFCLASS: classes named by symbols with no or
1129 unprintable packages can now be defined.
1130 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1131 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1132 * fixed compiler failure related to checking types of functions.
1133 (reported by Robert E. Brown)
1134 * the compiler is now much more consistent in its error-checking
1135 treatment of bounding index arguments to sequence functions: in
1136 (SAFETY 3) code, errors will be signalled in almost all cases if
1137 invalid sequence bounding indices are passed to functions defined
1138 by ANSI to operate on sequences.
1139 * fixed a bug in the build procedure: documentation of SBCL-specific
1140 packages is now preserved and available in the final Lisp image.
1141 * lifted FDEFINITION lookup out of loops in the implementation of
1142 many list operations. (thanks to Robert E. Brown)
1143 * fixed a bug in the reader: the #n# reader macro now works for
1144 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1145 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1146 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1147 E. Brown and Tony Martinez respectively)
1148 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1149 SLOT-BOUNDP: in particular, it is now able to optimize them much
1150 better, and is now not vulnerable to having packages renamed.
1151 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1152 always called when a slot is not present in an instance. (thanks
1154 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1155 longer signal an error on unreading a character following EOF on
1156 the previous constituent stream. (thanks to Tony Martinez)
1157 * fixed some bugs revealed by Paul Dietz' test suite:
1158 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1159 not just nonnegative fixnums;
1160 ** the logical bit-array operators such as BIT-AND now accept an
1161 explicit NIL for their "opt-arg" argument (to indicate a
1162 freshly-consed result bit-array);
1163 ** ELT now signals an error on an invalid sequence index in safe
1165 ** the type system is now cleverer about negations of numeric
1166 types, and consequently understands the BIGNUM and RATIO types
1168 ** the type system is now cleverer about the interaction between
1169 INTEGER and RATIO types: while bugs still remain, many more
1170 cases are accurately computed;
1171 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1172 if it is in the last clause;
1173 ** CONSTANTLY now correctly returns a side-effect-free function in
1175 ** DECLARE is no longer treated as a special-operator; in
1176 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1177 * incremented fasl file version number due to the change in the
1178 DEFSTRUCT-SLOT-DESCRIPTION structure.
1180 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1181 * minor incompatible change: code processed by the "interpreter" or
1182 EVAL now has a compilation optimization policy of (DEBUG 2)
1183 (changed from (DEBUG 1)) to improve debuggability of interactive
1184 development, and to allow the use of the debug RETURN command in
1186 * an experimental implementation of the RETURN command for the
1187 debugger has been included. (thanks to Frederik Kuivinen)
1188 * fixed bug 62: constraints were not propagated into a loop.
1189 * fixed bug in embedded calls of SORT (reported and investigated by
1191 * fixed some bugs revealed by Paul Dietz' test suite:
1192 ** printing and reading of arrays with some dimensions having
1193 length 0 (thanks to Gerd Moellmann);
1194 ** BOA constructor with &AUX argument without a default value does
1195 not cause a type error;
1196 ** CONSTANTP now returns true for all self-evaluating objects.
1198 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1199 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1200 accessors that are related by inheritance, as specified in the
1201 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1202 Valtteri Vuorikoski)
1203 * The compiler is now able to inline functions that were defined in
1204 a complex lexical environment (e.g. inside a MACROLET).
1205 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1207 * Support for the upcoming FreeBSD-5.0 release has been included.
1208 (thanks to Dag-Erling Smorgrav)
1209 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1210 effect when it is not in a toplevel context.
1211 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1212 stand a better chance of being correct. (thanks to Gerd
1214 * fixed bug in COERCE, which now signals an error on coercing a
1215 rational to a bounded real type which excludes the expected
1217 * The compiler is now able to derive types more accurately from the
1218 COERCE and COMPILE functions.
1219 * fixed bug 223: functional binding is considered to be constant
1220 only for symbols in the CL package.
1221 * fixed bug 231: SETQ did not check the type of a variable being set
1222 (reported by Robert E. Brown)
1223 * A new optimization for MAKE-INSTANCE has been included, fixing
1224 various bugs (including relating to :ALLOCATION :CLASS slots and
1225 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1227 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1228 ** As required by ANSI, LOOP now disallows anonymous collection
1229 clauses such as COLLECT I in conjunction with aggregate boolean
1230 clauses such as THEREIS (= I 1);
1231 ** LOOP now signals an error when any variable is reused in the
1232 same loop (including the potentially useful construct analogous
1233 to WITH A = 1 WITH A = (1+ A);
1234 ** IT is only a special loop symbol within the first clause of a
1235 conditional loop clause;
1236 ** LOOP with a typed iteration variable over a hashtable now
1237 signals a type error iff it should.
1238 * fixed some other bugs revealed by Paul Dietz' test suite:
1239 ** FILE-STREAM now names the class previously known as FD-STREAM;
1240 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1241 argument) no longer signals an error;
1242 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1243 of the slot symbol, rather than using the current package
1244 ((:CONC-NAME "") continues to intern the slot's name in the
1246 * incremented fasl file version number, because of the incompatible
1247 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1248 of the new implementation of DEFINE-COMPILER-MACRO.
1250 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1251 * Support for building SBCL for MIPS platforms running in
1252 little-endian mode has now been checked in, and basic
1253 functionality on said platforms verified.
1254 * minor incompatible change: PCL now records the pathname of a file
1255 in which methods and the like are defined, rather than its
1257 * minor incompatible change: TRUENAME now considers the truename of
1258 a file naming a directory to be the pathname with :DIRECTORY
1259 component indicating that directory.
1260 * minor incompatible change: a NAMED clause in the extended form of
1261 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1262 reason for the previous behaviour is unclear.
1263 * more systematization and improvement of CLOS and MOP conformance
1264 in PCL (thanks to Gerd Moellman and Pierre Mai):
1265 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1267 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1268 primary methods with no specializers;
1269 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1271 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1272 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1273 FORWARD-REFERENCED-CLASSes; error reporting on
1274 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1276 ** SXHASH on CLOS instances now uses a slot internal to the
1277 instance to return different numbers on distinct instances,
1278 while preserving the same return value through invocations of
1280 ** DEFMETHOD signals errors when methods with longer incongruent
1281 lambda lists are added to generic functions;
1282 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1283 CLASS, as specified in AMOP;
1284 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1285 on the order returned by the primary method for classes of
1286 class STANDARD-CLASS;
1287 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1288 * fixed some bugs shown by Paul Dietz' test suite:
1289 ** DOLIST puts its body in TAGBODY;
1290 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1292 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1293 value producing form;
1294 ** if more variables are given to PROGV than values, extra
1295 variables are bound and made to have no value;
1296 ** NSUBSTITUTE on list arguments gets the right answer with
1298 ** ELT signals an error of type TYPE-ERROR when the index argument
1299 is not a valid sequence index;
1300 ** LOOP signals (at macroexpansion time) an error of type
1301 PROGRAM-ERROR when duplicate variable names are found;
1302 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1303 ** FUNCALL of special-operators now cause an error of type
1305 ** PSETQ now works as required in the presence of side-effecting
1306 symbol-macro places;
1307 ** NCONC accepts any object as its last argument;
1308 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1310 ** loop-for-as-package does not require a package to be explicitely
1312 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1314 * fixed bug 166: compiler preserves "there is a way to go"
1315 invariant when deleting code.
1316 * fixed bug 172: macro lambda lists with required arguments after
1317 &REST arguments now cause an error to be signalled. (thanks to
1319 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1320 bug. (thanks to Matthew Danish)
1321 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1323 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1324 arguments are lexically rebound. (thanks to Gerd Moellmann and
1326 * fixed bug 194: error messages are now more informative when there
1327 is no primary method applicable in a call to a generic
1328 function. (thanks to Gerd Moellmann)
1329 * fixed bug in command line argument checking (thanks to Julian
1331 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1332 (thanks to Matthew Danish)
1333 * incremented fasl file version number, because of the
1334 SXHASH-related changes in the layout of CLOS data structures
1336 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1337 * minor incompatible change: The runtime (the Unix executable named
1338 "sbcl") is now much pickier about the .core files it will load.
1339 Essentially it now requires .core files to descend from the same
1340 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1341 as the runtime does. (The intent is to prevent the crashes which
1342 can occur, and which can even be reported as mysterious failures,
1343 when people patch the sources or change the build parameters
1344 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1345 sbcl and .core files.)
1346 * fixed bug: VALUES-LIST is no longer optimized away.
1347 * fixed bug 142: The FFI conversion of C string values to Lisp
1348 string values no longer conses excessively. (thanks to Nathan
1349 Froyd porting Raymond Toy's fix to CMU CL)
1350 * began to systematize and improve MOP conformance in PCL (thanks to
1351 Nathan Froyd, Gerd Moellman and Pierre Mai):
1352 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1354 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1355 ** FINALIZE-INHERITANCE is now called on class finalization;
1356 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1357 argument precedence order.
1358 * fixed bug 202: The compiler no longer fails on functions whose
1359 derived types contradict their declared type.
1360 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1361 so it can be non-toplevel.
1362 * The fasl file version number has changed (because of the new
1363 implementation of DEFMACRO).
1364 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1365 safe code, that any length requirement by their type-specifier
1366 argument is valid. The exceptions to this are described in bug
1368 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1369 functions now check the types of their inputs as required by ANSI.
1370 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1371 introduced symbol is DECLAREd to be SPECIAL.
1372 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1373 * fixed bug: :COUNT argument to sequence functions may be negative
1374 * fixed bug: body of DO-SYMBOLS may contain declarations
1375 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1376 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1377 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1378 symbol macro only once
1379 * fixed printing of call frame when argument list is unavailable
1380 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1381 * compiler no longer signals WARNING on unknown keyword
1384 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1385 * A beta-quality port to the mips architecture running Linux,
1386 based on the old CMUCL backend, has been made. It has been tested
1387 on a big-endian kernel, and works sufficiently well to be able to
1388 rebuild itself; it has not been tested in little-endian mode.
1389 * fixed an inconsistency between gencgc.c and purify.c which made
1390 dumping/loading .core files unreliable
1391 * fixed bug 120a: The compiler now deals correctly with IFs where
1392 the consequent is the same as the alternative, instead of
1393 misderiving the return type. (thanks to Alexey Dejneka)
1394 * fixed bug 113: Logical pathnames are now dumpable (the logical
1395 host is resolved at load-time, throwing an error if it is not
1397 * fixed bug 174: FORMAT's error message is slightly clearer when a
1398 non-printing character is used in a format directive.
1399 * fixed several bugs in compiler checking of type declarations, i.e.
1400 violations of the Python "declarations are assertions" principle
1401 (thanks to Alexey Dejneka)
1402 * fixed several bugs in PCL's error checking (thanks to Gerd
1404 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1406 * fixed bug in compilation of functions as first class values
1407 (thanks to Antonio Martinez-Shotton)
1408 * The compiler's handling TYPE-ERRORs which it can prove will
1409 inevitably happen at runtime has been cleaned up and corrected
1410 in several ways. (thanks to Alexey Dejneka)
1411 * improved argument type checking for various basic arithmetic
1412 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1413 lavished on them in the past that they can be compiled in many
1414 ways in different special cases
1415 * fixed bug 181: compiler checks validity of user supplied type
1417 * cleaned up code flushing in optimization: Function calls which
1418 should signal errors for safety purposes (e.g. which ANSI says
1419 should signal errors when their arguments are of incorrect type)
1420 are no longer optimized away.
1421 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1422 * incremented fasl file version number, because changes in the
1423 implementation of sequence functions like COERCE caused
1424 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1425 in old inline expansions) to become undefined. (Actually these
1426 changes were later undone, so we might very well be binary
1427 compatible with 0.7.7 after all, but leaving the version number
1428 incremented seemed like the simplest and most conservative
1431 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1432 * An alpha-quality port to the parisc architecture running Linux,
1433 based on the old CMUCL backend, has been made. This, even more so
1434 than the other backends, should be considered still a work in
1435 progress; known problems include that the Linux kernel in 64-bit
1436 mode does not propagate the correct sigcontext structure to
1437 userspace, and consequently SBCL on a parisc64 kernel will not
1439 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1440 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1441 suppressed.) Also now that the compiler is looking at declarations
1442 in the environment, it checks optimization declarations as well,
1443 and suppresses inlining when (> DEBUG SPEED).
1444 * More fixes have been made to treatment of floating point exception
1445 treatment and other Unix signals. In particular, floating point
1446 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1447 * The detection and handling of control stack exhaustion (infinite
1448 or very deeply nested recursion) has changed. Stack exhaustion
1449 detection is now done by write-protecting pages at the OS level
1450 and applies at all optimization settings; when found, a
1451 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1452 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1453 be caught using IGNORE-ERRORS.
1454 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1455 correctly when transforming recursive LABELS functions to LETs.
1456 (thanks to Alexey Dejneka)
1457 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1458 that are names of constants or global variables.
1459 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1460 alien routines with docstrings.
1461 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1462 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1464 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1465 (thanks to Eric Marsden)
1466 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1467 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1468 * Bug fix: The SPARC backend can now compile functions involving
1469 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1470 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1471 to a functions expecting a PATHNAME-DESIGNATOR.
1472 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1473 lambda lists. (thanks to Alexey Dejneka)
1474 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1475 functions so that EXPORTing the name of the class doesn't cause
1476 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1477 to Antonio Martinez for reporting this)
1478 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1479 constant-folded without failing an assertion. (thanks to Einar
1480 Floystad Dorum for reporting this)
1481 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1482 types are now dealt with more correctly by the compiler.
1483 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1484 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1485 behaviour analogous to RENAME-FILE. This puts its behaviour more
1486 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1488 * The fasl file version number has changed again. (because of the
1489 bug fix involving the names of PCL MAKE-INSTANCE functions)
1491 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1492 * bug fix: Floating point exceptions are treated much more
1493 consistently on the x86/Linux and PPC/Linux platforms.
1494 * Array initialization with :INITIAL-ELEMENT is now much faster for
1495 cases when the compiler cannot open code the array creation, but
1496 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1497 array accesses have also seen a speed increase.
1498 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1499 specified. (This was at the root of some bad interactions between
1500 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1502 * bug fix: Internal error arguments for undefined functions are now
1503 computed correctly on the PPC/Linux platform.
1504 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1505 Raymond Toy's patch for CMU CL)
1506 * Support for the Solaris 9 operating environment has been included
1507 (thanks to Daniel Merritt)
1508 * A very ugly but hopefully complete draft of the missing FFI chapter
1509 of the manual has been created by reformatting the corresponding
1510 CMU CL manual chapter into (currently very ugly and incoherent)
1511 DocBook and bringing it up to date for SBCL behavior. Thus, the
1512 manual is now essentially complete, at least by my extreme
1513 once-and-only-once standards, whereby it's acceptable to refer to
1514 the doc strings of SB-EXT functions as the primary documentation.
1515 * The fasl file version number has changed again, due to cleanup of
1516 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1517 is no longer a static symbol.)
1519 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1520 * SBCL now builds with OpenMCL (version 0.12) as the
1521 cross-compilation host; also, more progress has been made toward
1522 bootstrapping under CLISP.
1523 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1525 * bug 158 fixed: The compiler can now deal with integer loop
1526 increments different from 1; fixing this turned out also to fix
1528 * bug 169 fixed: no more bogus warnings about using lexical bindings
1529 despite the presence of perfectly good SPECIAL declarations (thanks
1530 to David Lichteblau)
1531 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1532 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1533 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1535 * bug 180 fixed: Method combination specifications no longer ignore
1536 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1537 * bug fix: Structure type predicate functions now check their argument
1538 count as they should.
1539 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1540 correctly. (thanks to Pierre Mai)
1541 * minor incompatible change: The --noprogrammer option is deprecated
1542 in favor of the new --disable-debugger option, which is very similar.
1543 (The major difference is that it takes effect at a slightly different
1544 time at startup, causing handling of errors in --sysinit and
1545 --userinit files will be affected differently.) The
1546 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1547 been added to allow this functionality to be controlled from ordinary
1548 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1549 maintainers, who might want to run non-interactive scripts to
1550 build SBCL cores which will later be used interactively.)
1551 * minor incompatible change: The LOAD function no longer, when given
1552 a wild pathname to load, loads all files matching that pathname.
1553 Instead, an error of type FILE-ERROR is signalled.
1555 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1556 * bug 147 fixed: The compiler preserves its block link/count
1557 invariants more correctly now so that it doesn't crash. (thanks
1559 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1561 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1562 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1563 no-value-for-this-character as NIL (as specified by ANSI).
1564 * HOST-NAMESTRING on physical pathnames now returns a string that is
1565 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1566 (thanks to Christophe Rhodes)
1567 * The Alpha port handles icache flushing more correctly. (thanks to
1569 * More progress has been made toward bootstrapping under CLISP. (thanks
1570 to Christophe Rhodes)
1571 * The fasl file format has changed again, because dynamic loading
1572 on OpenBSD (which has non-ELF object files) motivated some cleanups
1573 in the way that foreign symbols are transformed and passed around.
1574 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1575 is no longer treated as whitespace by the reader, but instead as
1576 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1577 |A
\7fB|, instead of A as it used to.
1579 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1580 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1581 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1582 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1583 this as of 0.7.1.45, but was left out of the previous news section
1584 (thanks to Dan Barlow)
1585 * SBCL now runs on the Solaris operating system on SPARC architectures
1586 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1587 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1588 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1589 * SPARC backend cleanups, allowing builds of cores optimized for V8
1590 and V9 SPARCS, and also emission of code targeted to a particular
1591 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1593 * SBCL is closer to bootstrapping under CLISP, thanks to various
1594 fixes by Christophe Rhodes.
1595 * The fasl file format has changed again, to allow the compiler's
1596 INFO database to support symbol macros.
1597 * The user manual (in doc/) is formatted into HTML more nicely.
1598 (thanks to coreythomas)
1599 * The system is smarter about SUBTYPEP relationships, especially
1600 those involving NOT types (including types such as ATOM which are
1601 represented internally using NOT types). Thus SUBTYPEP is less
1602 likely to return (VALUES NIL NIL) in general, and in particular
1603 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1605 * The fasl file format has changed again, because the internal
1606 representation of types now includes a new slot to support the new
1607 SUBTYPEP-of-NOT-types logic.
1608 * (not a change in the main branch of SBCL, but a related prototype
1609 which can hopefully be merged into the main branch of SBCL in the
1610 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1611 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1612 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1613 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1614 dispatch macro character combinations. (thanks to Alexey Dejneka)
1615 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1616 that superficially look like logical namestrings correctly.
1617 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1619 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1620 default supplied-p) for &optional and &key arguments. (thanks to
1623 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1624 * incompatible change: The compiler is now less aggressive about
1625 tail call optimization, doing it only when (> SPACE DEBUG) or
1626 (> SPEED DEBUG). (This is an incompatible change because there are
1627 programs which relied on the old CMU-CL-style behavior to optimize
1628 away their unbounded recursion which will now die of stack overflow.)
1629 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1630 for non-GENCGC systems has been increased to 20M (since that
1631 seems much closer to the likely performance optimum for modern
1632 systems than the old 4M value was)
1633 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1634 and *DEBUG-PRINT-LEVEL*
1635 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1636 of CMU CL's support for SPARC, and various endianness and other
1637 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1638 * new syntactic sugar for the Unix command line: --load foo.bar is now
1639 an alternate notation for --eval '(load "foo.bar")'.
1641 ** The system now detects stack overflow and handles it gracefully,
1642 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1643 optimization settings. (This is a good thing in general, and
1644 its introduction in this version should be particularly timely
1645 for anyone whose code fails because of suppression of tail
1647 ** The system now hunts for the C variable "environ" in a more
1648 devious way, to avoid segfaults when the C library version differs
1649 between compile time and run time. (thanks to Christophe Rhodes)
1650 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1651 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1652 ** The compiler no longer issues bogus style warnings for undefined
1653 classes in the same source file as the DEFCLASSes which defined
1654 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1656 ** fixes in CONDITION class precedence list for undefined function
1657 errors (thanks to Alexei Dejneka)
1658 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1659 correctly. (thanks to Dan Barlow)
1660 ** portability fixes aiming at bootstrapping under CLISP (thanks
1661 to Dave McDonald and Christophe Rhodes)
1662 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1663 ** fixes in type translation and and type inference (thanks to
1665 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1666 ** various fixes in the new ports (thanks to Dan Barlow)
1667 * several changes related to debugging:
1668 ** suppression of tail recursion, as noted above
1669 ** stack overflow detection, as noted above
1670 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1671 is now the default. (For some time encapsulation has been more
1672 reliable than the breakpoint-based :ENCAPSULATE NIL
1673 implementation, at least on X86 systems; and I just noticed that
1674 encapsulation also seems closer to the spirit of the ANSI
1677 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1679 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1680 up properly again. (There was a packaging bug in 0.7.0 which
1681 left their definitions in SB-SYS::LOAD-FOREIGN and
1682 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1683 vital for most things which interface to C-level interfaces,
1684 like extensions working with sockets or databases or
1685 Perl-compatible regexes or whatever, and the need to fix
1686 this bug is the main reason that 0.7.1 was released so
1688 ** DEFGENERIC is now choosier about the methods it redefines, so that
1689 reLOADing a previously-LOADed file containing DEFGENERICs does
1690 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1691 cycle now works as it should. (thanks to Alexey Dejneka)
1692 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1693 fixed by Christophe Rhodes. (He actually submitted this patch
1694 months ago, and I delayed until after 0.7.0.)
1695 ** Bug 111 (internal compiler confusion about runtime checks on
1696 FUNCTION types) was fixed by Alexey Dejneka.
1697 * Some internal cleanups (getting rid of variables which aren't
1698 needed now that the byte interpreter is gone) caused the fasl
1699 file format number to change again.
1701 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1702 * major incompatible change: The default fasl file extension, i.e. the
1703 default extension for files produced by COMPILE-FILE, has changed
1704 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1706 ** There are many changes in the implementation of the compiler.
1707 SBCL is now essentially a compiler-only implementation of ANSI
1708 Common Lisp. EVAL still "interprets" a few special cases, but
1709 almost all the interesting cases are handled by creating
1710 a LAMBDA expression, calling COMPILE on it, then calling
1711 FUNCALL on the result.
1712 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1713 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1714 Since the code is newer, there might still be some new bugs
1715 (though not as many as before Martin Atzmueller's fixes:-). But
1716 the new code is substantially simpler and clearer, and hopefully
1717 any remaining bugs will be simpler, less fundamental, and more
1718 fixable then the bugs in the old code.
1719 ** The revised compiler is still a little unsteady on its feet.
1721 *** The debugging information it produces (particularly the names
1722 of FUNCTION objects) is sometimes much less useful than what
1723 the old compiler produced.
1724 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1725 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1726 within a MACROLET) has been temporarily weakened.
1727 ** There are new compiler optimizations for various functions:
1728 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1729 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1730 *** the math functions TRUNCATE, FLOOR, and CEILING
1731 *** the function-of-all-trades COERCE
1732 Mostly these should be transparent, but there's one
1733 potentially-annoying problem (bug 117): when the compiler
1734 inline-expands a function and does type analysis on the result,
1735 it can create control paths which have type mismatches, and
1736 when it can't prove that those control paths aren't taken,
1737 it will issue WARNINGs about the type mismatches. This is
1738 a particular problem in practice for the new sequence functions.
1739 It's not clear how this should be fixed, and for now, a
1740 workaround is given in the entry for 117 in the BUGS file.
1741 ** (Because of the interaction between the two previous items --
1742 occasional inlining problems and new inline expansions -- some
1743 of the new sequence function optimizations won't really kick in
1744 completely until debugging information, and then inlining, are
1745 straightened out in some future version.)
1746 * minor incompatible changes:
1747 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1748 DIRECTORY no longer implicitly promotes NIL slots of its
1749 pathname argument to :WILD. In particular, when you ask for the
1750 contents of a directory (which you used to be able to do without
1751 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1752 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1753 ** changes in behavior that ANSI explicitly defines to be
1754 implementation dependent:
1755 *** The new compiler-only implementation still conforms with ANSI,
1756 but acts a little different than before. Besides the obvious
1757 changes in performance tradeoffs (that the cost per form passed
1758 to EVAL has gone up, and the cost per form executed by EVAL
1759 has gone down), the behavior of the system changes a little
1760 because there are no longer any interpreted function objects.
1761 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1762 e.g. doing COMPILE on the output of interactive DEFUN is
1764 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1766 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1767 has changed from (:CL) to NIL.
1768 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1769 e.g. #\Newline and #\Tab, don't have names specified in the
1770 ANSI Common Lisp standard, is now based on their ASCII symbolic
1771 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1772 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1773 are no longer used for output.
1774 ** changes in internal implementation constants:
1775 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1776 4 million. (If your application spends a lot of time GCing and
1777 you have a lot of RAM, you might want to experiment with
1778 increasing it even more.)
1779 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1780 However, almost all old code should still continue to work without
1781 immediate update, as SB-C-CALL is now a (deprecated) nickname
1783 ** Old operator names in the style DEF-FOO are now deprecated in
1784 favor of new corresponding names DEFINE-FOO, for consistency with
1785 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1786 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1787 internal symbols, but a few supported extensions like
1788 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1789 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1790 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1791 etc. as you get deeper into recursive calls to the debugger
1792 command loop, instead of the old "5]", "5]]", "5]]]"
1793 sequence. (I was motivated to do this when squabbles between
1794 ILISP and SBCL left me very deeply nested in the debugger. In the
1795 short term, this change will probably provoke more ILISP/SBCL
1796 squabbles, but hopefully it will be an improvement in the long run.)
1797 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1798 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1799 function, since declaiming return types involving aliens is
1800 (1) annoyingly messy to do by hand and (2) vital to efficient
1801 compilation of code which calls such functions.
1802 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1803 longer reexported by the SB-EXT package. They're solely useful
1804 for alien code, so it seems more logical that you should get
1805 them from the SB-ALIEN package, not in SB-EXT.
1806 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1807 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1808 features. Instead, the code that they used to control is always
1809 built into the system.
1810 * many other bug fixes
1811 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1812 advantage of the new EVAL-WHEN stuff and to clean them up in
1813 general, and they are now more ANSI-compliant in a number of
1814 ways. Martin Atzmueller is responsible for a lot of this.
1815 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1817 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1818 *** correct ERROR type for various file operations
1819 *** some fixes for Lisp streams
1820 *** DEFMETHOD syntax checking
1821 *** changing old weird representation of debug information as
1822 strings (which, among their other deficiencies, don't transform
1823 correctly when you rename packages, and don't change their
1824 print representation when you change things like *PACKAGE*
1825 and *PRINT-LENGTH*) to symbols and lists of symbols
1826 He also made several improvements and fixed several bugs in DESCRIBE.
1827 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1829 *** misbehavior of WRITE-STRING/WRITE-LINE
1830 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1831 and several other LOOP problems as well
1832 *** DIRECTORY when similar filenames are present
1833 *** DEFGENERIC with :METHOD options
1834 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1835 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1836 *** argument ordering in FIND with :TEST option
1837 *** mishandled package designator argument in APROPOS-LIST
1838 *** various problems in the backquote readmacro
1839 *** a bug in APROPOS
1840 *** probably some others that I'm not describing very well here,
1841 since the CVS log documents them by reference to sbcl-devel
1842 messages, and the SourceForge archives aren't working well.:-(
1843 ** Dan Barlow improved the Alpha port (and is making progress on the
1844 PPC port, for those of you who think different).
1845 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
1846 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
1847 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
1848 continues to work on the SPARC port (for those of you in a position
1849 to look down upon our little PC-compatible boxes from a great height).
1850 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
1851 and uses that copy, rather than the current dynamic value, when
1852 it's trying to decide whether to truncate output. Thus e.g.
1853 (let ((*print-lines* 50))
1854 (pprint-logical-block (stream nil)
1856 (let ((*print-lines* 8))
1857 (print (aref possiblybigthings i) stream)))))
1858 should now truncate the logical block only at 50 lines, instead of
1859 often truncating it at 8 lines, as it did before.
1860 * The doc/cmucl/ directory, containing old CMU CL documentation
1861 from the time of the fork, is no longer part of the base system.
1862 SourceForge has shut down its anonymous FTP service, and with it
1863 my original plan for distributing the old CMU CL documentation
1864 there. For now, if you need these files you can download an old
1865 SBCL source release and extract them from it.
1866 * The fasl file version number changed again, for dozens of reasons,
1867 some of which are apparent above.
1869 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
1870 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
1871 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
1872 patch, so that DISASSEMBLE output is much nicer.
1873 * The code in the SB-PROFILE package now seems reasonably stable.
1874 I still haven't decided what the final interface should look like
1875 (I'd like PROFILE to interact cleanly with TRACE, since both
1876 facilities use function encapsulation) but if you have a need
1877 for profiling now, you can probably use it successfully with
1878 the current CMU-CL-style interface.
1879 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
1880 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
1881 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
1882 link now returns the dangling link itself, and for similar
1883 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
1884 link itself. (In these cases the old code signalled an error and
1885 looped endlessly, respectively.) Thus, DIRECTORY now works even
1886 in the presence of dangling and cyclic symbolic links.
1887 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
1888 is now a supported extension again, since the consensus on
1889 sbcl-devel was that it can be useful for ordinary development
1890 work, not just for debugging SBCL itself.
1891 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
1892 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
1893 that currently-defined functions might be redefined later with
1894 different return types.
1895 * Hash tables can be printed readably, as inspired by CMU CL code
1896 of Eric Marsden and SBCL code of Martin Atzmueller.
1897 * better error handling in CLOS method combination, thanks to
1898 Martin Atzmueller porting Pierre Mai's CMU CL patches
1899 * more overflow fixes for >16Mbyte I/O buffers
1900 * A bug in READ has been fixed, so that now a single Ctrl-D
1901 character suffices to cause end-of-file on character streams.
1902 In particular, now you only need one Ctrl-D at the command
1903 line (not two) to exit SBCL.
1904 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
1906 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
1907 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
1908 does the right thing.
1909 * The implementation of some type tests, especially for CONDITION
1910 types, is now tidier and maybe faster, due to CMU CL code
1911 originally by Douglas Crosher, ported by Martin Atzmueller.
1912 * Some math functions have been fixed, and there are new
1913 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
1914 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
1915 * (There are also some new optimizers in contrib/*-extras.lisp. Those
1916 aren't built into sbcl-0.6.13, but are a sneak preview of what's
1917 likely to be built into sbcl-0.7.0.)
1918 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
1919 CL, ported to SBCL by Martin Atzmueller)
1920 * DESCRIBE now gives more information in some cases. (Pierre Mai's
1921 patch to CMU CL, ported to SBCL by Martin Atzmueller)
1922 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
1923 * There's a new slam.sh hack to shorten the edit/compile/debug
1924 cycle for low-level changes to SBCL itself, and a new
1925 :SB-AFTER-XC-CORE target feature to control the generation of
1926 the after-xc.core file needed by slam.sh.
1927 * minor incompatible change: The ENTRY-POINTS &KEY argument to
1928 COMPILE-FILE is no longer supported, so that now every function
1929 gets an entry point, so that block compilation looks a little
1930 more like the plain vanilla ANSI section 3.2.2.3 scheme.
1931 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
1932 returns the number of bytes consed since the system started,
1933 rather than the number consed since the first time the function
1934 was called. (The new definition parallels ANSI functions like
1935 CL:GET-INTERNAL-RUN-TIME.)
1936 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
1937 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
1938 supported. Now DIRECTORY always does the abstract Common-Lisp-y
1939 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
1940 * Fasl file version numbers are now independent of the target CPU,
1941 since historically most system changes which required version
1942 number changes have affected all CPUs equally. Similarly,
1943 the byte fasl file version is now equal to the ordinary
1946 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
1947 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
1948 is no longer recognized. I apologize for this, because it was
1949 listed in SB-EXT as a supported extension, but I found that
1950 its existing behavior was poorly specified, as well as incorrectly
1951 specified, and it looked like too much of a mess to straighten it
1952 out. I have enough on my hands trying to get ANSI stuff to work..
1953 * many patches ported from CMU CL by Martin Atzmueller, with
1954 half a dozen bug fixes in pretty-printing and the debugger, and
1955 half a dozen others elsewhere
1956 * fixed bug 13: Floating point infinities are now supported again.
1957 They might still be a little bit flaky, but thanks to bug reports
1958 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
1959 as flaky as they were.
1960 * The --noprogrammer command line option is now supported. (Its
1961 behavior is slightly different in detail from what the old man
1962 page claimed it would do, but it's still appropriate under the
1963 same circumstances that the man page talks about.)
1964 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
1965 are now supported, and enabled by default. Thus, the compiler can
1966 handle many floating point and complex operations much less
1967 inefficiently. (Thus e.g. you can implement a complex FFT
1969 * The compiler now detects type mismatches between DECLAIM FTYPE
1970 and DEFUN better, and implements CHECK-TYPE more correctly, and
1971 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
1972 thanks to patches from Martin Atzmueller.
1973 * various fixes to make the cross-compiler more portable to
1974 ANSI-conforming-but-different cross-compilation hosts (notably
1975 Lispworks for Windows, following bug reports from Arthur Lemmens)
1976 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
1977 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
1978 fixed thanks to Pierre Mai's CMU CL patches.
1979 * fixes to keep the system from overflowing internal counters when
1980 it tries to use i/o buffers larger than 16M bytes
1981 * fixed bug 45a: Various internal functions required to support
1982 complex special functions have been merged from CMU CL sources.
1983 (When I was first setting up SBCL, I misunderstood a compile-time
1984 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
1985 * improved support for type intersection and union, fixing bug 12
1986 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
1987 more obscure bugs as well
1988 * some steps toward byte-compiling non-performance-critical
1989 parts of the system, courtesy of patches from Martin Atzmueller
1990 * Christophe Rhodes has made some debian packages of sbcl at
1991 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
1992 From his sbcl-devel e-mail of 2001-04-08 they're not completely
1993 stable, but are nonetheless usable. When he's ready, I'd be happy
1994 to add them to the SourceForge "File Releases" section. (And if
1995 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
1996 * new fasl file format version number (because of changes in
1997 internal representation of (OR ..) types to accommodate the new
1998 support for (AND ..) types, among other things)
2000 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2001 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2003 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2005 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2006 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2007 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2008 * DESCRIBE now works on structure objects again.
2009 * Most function call argument type mismatches are now handled as
2010 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2011 know whether the function will be redefined before the call is
2012 executed. (The compiler could flag local calls with full WARNINGs,
2013 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2014 it doesn't keep track of enough information to know whether calls
2015 are local in this sense.)
2016 * Compiler output is now more verbose, with messages truncated
2017 later than before. (There should be some supported way for users
2018 to override the default verbosity, but I haven't decided how to
2019 provide it yet, so this behavior is still controlled by the internal
2020 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2021 src/compiler/ir1util.lisp.)
2022 * Fasl file format version numbers have increased again, because
2023 support for the Gray streams extension changes the layout of the
2024 system's STREAM objects.
2025 * The Gray subclassable streams extension now works, thanks to a
2026 patch from Martin Atzmueller.
2027 * The full LOAD-FOREIGN extension (not just the primitive
2028 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2029 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2030 but like most other programs, it defaults to copying the Unix
2031 environment from the original process instead of starting the
2032 new process in an empty environment.
2033 * Extensions which manipulate the Unix environment now support
2034 an :ENVIRONMENT keyword option which doesn't smash case or
2035 do other bad things. The CMU-CL-style :ENV option is retained
2036 for porting convenience.
2037 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2038 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2040 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2042 * A patch from Martin Atzmueller seems to have solved the SIGINT
2043 problem, and as far as we know, signal-handling now works cleanly.
2044 (If you find any new bugs, please report them!)
2045 * The system no longer defaults Lisp source file names to types
2046 ".l", ".cl", or ".lsp", but only to ".lisp".
2047 * The compiler no longer uses special default file extensions for
2048 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2049 seems to expect a single default extension for all compiled code,
2050 and there's no compelling reason to try to stretch the standard
2051 to allow two different extensions.) Instead, byte-compiled files
2052 default to the same extension as native-compiled files.
2053 * Fasl file format version numbers have increased again, because
2054 a rearrangement of internal implementation packages made some
2055 dumped symbols in old fasl files unreadable in new cores.
2056 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2057 many fewer weird special cases.
2058 * Bug #17 (differing COMPILE-FILE behavior between logical and
2059 physical pathnames) has been fixed, and some related misbehavior too,
2060 thanks to a patch from Martin Atzmueller.
2061 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2062 by Tim Moore, ported to SBCL by Martin Atzmueller.
2063 * Martin Atzmueller fixed several filesystem-related problems,
2064 including bug #36, in part by porting CMU CL patches, which were
2065 written in part by Paul Werkowski.
2066 * More compiler warnings in src/runtime/ are gone, thanks to
2067 more patches from Martin Atzmueller.
2068 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2071 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2073 * DESCRIBE now works on CONDITION objects.
2074 * The debugger now handles errors which arise when trying to print
2075 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2077 * The build system now uses an additional file, customize-target-features.lisp,
2078 to allow local modifications to the target *FEATURES* list. (The point of
2079 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2080 debugging features enabled, without having to worry about propagating it
2081 into everyone's system when I do a "cvs update".) When no
2082 customize-target-features.lisp file exists, the target *FEATURES* list
2083 should be constructed the same way as before.
2084 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2085 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2086 warn about behavior which is undefined under the ANSI spec. Note: This
2087 is specified by ANSI, but it's not very popular with programmers.
2088 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2089 macro in the SBCL sources for an example of a workaround which you
2090 might use to make portable ANSI-standard code which does what you want.
2091 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2092 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2093 when it's used as a non-toplevel form. (This is required in order
2094 to implement the DEFCONSTANT-EQX macro.)
2095 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2096 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2097 funny things, due to bugs in the implementation of EVAL-WHEN
2098 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2099 non-toplevel DEFCONSTANTs are uncommon.)
2100 * The core file version number and fasl file version number have been
2101 incremented, because the old noncompliant DEFCONSTANT behavior involved
2102 calling functions which no longer exist, and because I also took the
2103 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2104 * fixed bug 1 (error handling before read-eval-print loop starts), and
2105 redid debugger restarts and related debugger commands somewhat while
2107 ** The QUIT debugger command is gone, since it did something
2108 rather different than the SB-EXT:QUIT command, and since it never
2109 worked properly outside the main toplevel read/eval/print loop.
2110 Invoking the new TOPLEVEL restart provides the same functionality.
2111 ** The GO debugger command is also gone, since you can just invoke
2112 the CONTINUE restart directly instead.
2113 ** The TOP debugger command is also gone, since it's redundant with the
2114 FRAME 0 command, and since it interfered with abbreviations for the
2116 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2118 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2119 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2120 some porting work by Martin Atzmueller)
2121 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2122 with a boatload of patches which he ported to SBCL. Now that those
2124 ** The system tries to make sure that its low-priority messages
2125 are prefixed by semicolons, to help people who like to use
2126 syntax highlighting in their ILISP buffer. (This patch
2127 was originally due to Raymond Toy.)
2128 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2129 patches originally written by Raymond Toy.
2130 ** The compiler understands coercion between single-value and
2131 multiple-VALUES type expressions better, getting rid of some very
2132 weird behavior, thanks to patches originally by Robert MacLachlan
2133 and Douglas Crosher.
2134 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2135 lambda lists, thanks to a patch originally by Pierre Mai.
2136 ** The system no longer bogusly warns about "abbreviated type
2138 ** The compiler gets less confused by inlining and RETURN-FROM,
2139 thanks to some patches originally by Tim Moore.
2140 ** The system no longer hangs when dumping circular lists to fasl
2141 files, thanks to a patch originally from Douglas Crosher.
2142 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2143 undefined function error.
2144 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2145 in structure slot initforms) for now, documented workaround instead:-|
2146 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2147 * fixed bug 5: added stubs for various Gray stream functions called
2148 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2149 installed, at least appropriate type errors are generated
2150 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2151 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2152 consistently in DEFMETHOD forms.
2153 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2154 it doesn't seem to affect SBCL after all
2155 * The C runtime system now builds with better optimization and many
2156 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2158 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2160 * The system is now under CVS at SourceForge (instead of the
2161 CVS repository on my home machine).
2162 * The new signal handling code has been tweaked to treat register
2163 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2164 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2165 downstream, e.g. in debug-int.lisp, has implicit dependencies
2166 on the unsignedness of integer representation of machine words,
2167 and that caused the system to bomb out with infinite regress
2168 when trying to recover from type errors involving signed values,
2169 e.g. (BUTLAST '(1 2 3) -1).)
2170 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2171 (This was one of the bugs Peter Van Eynde reported back in July.)
2172 * The system now uses code inspired by Colin Walters' O(N)
2173 implementation of MAP (from the cmucl-imp@cons.org mailing
2174 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2175 inline the MAP operation, and there is more than one
2176 sequence argument to the MAP call (so that it can't just
2177 do ETYPECASE once and for all based on the type of the
2178 single sequence argument). (The old non-inline implementation
2179 of the general M-argument sequence-of-length-N case required
2180 O(M*N*N) time when any of the sequence arguments were LISTs.)
2181 * The QUIT :UNIX-CODE keyword argument has been renamed to
2182 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2183 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2185 * Raymond Wiker's patches to port dynamic loading from Linux to
2186 FreeBSD have been added.
2187 * The BUGS file is now more nearly up to date, thanks in large part
2188 to Martin Atzmueller's review of it.
2189 * The debugger now flushes standard output streams before it begins
2190 its output ("debugger invoked" and so forth).
2191 * The core version number and fasl file version number have both
2192 been incremented, because of incompatible changes in the layout
2194 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2195 used to not be called for a saved Lisp image.)
2196 * Martin Atzmueller's version of a patch to fix a compiler crash,
2197 as posted on sbcl-devel 13 September 2000, has been installed.
2198 * Instead of installing Martin Atzmueller's patch for the
2199 compiler transform for SUBSEQ, I deleted the compiler transform,
2200 and transforms for some similar consing operations.
2201 * A bug in signal handling which kept TRACE from working on OpenBSD
2203 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2204 away by constant folding
2205 * The system now defines its address space constants in one place
2206 (in the Lisp sources), and propagates them automatically elsewhere
2207 (through GENESIS and the sbcl.h file). Therefore, patching the
2208 address map is less unnecessarily tedious and error-prone. The
2209 Lisp names of address space constants have also been systematized.
2210 * CVS tags like dollar-Header-dollar have been removed from
2211 the sources, because they have never saved me trouble and
2212 they've been source of trouble working with patches and other
2213 diff-related operations.
2214 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2215 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2217 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2219 * The system has been ported to OpenBSD.
2220 * The system now compiles with a simple "sh make.sh" on the systems
2221 that it's supported on. I.e., now you no longer need to tweak
2222 text in the target-features.lisp-expr and symlinks in src/runtime/
2223 by hand, the make.sh takes care of it for you.
2224 * The system is no longer so grossly inefficient when compiling code
2225 involving vectors implemented as general (not simple) vectors (VECTOR T),
2226 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2227 which dares to use the various sequence functions on non-simple
2228 vectors, takes less of a performance hit.
2229 * There is now a primitive type predicate VECTOR-T-P
2230 to test for the (VECTOR T) type, so that e.g.
2231 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2232 can now be compiled with some semblance of efficiency. (The old code
2233 turned the type declaration into a full call to %TYPEP at runtime!)
2234 * AREF on (VECTOR T) is still not fast, since it's still compiled
2235 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2236 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2237 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2238 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2239 without making full calls to SUBTYPEP at runtime.
2240 (Some analogous efficiency issues for non-simple vectors specialized to
2241 element types other than T, or for non-simple multidimensional arrays,
2242 have not been addressed. They could almost certainly be handled the
2243 same way if anyone is motivated to do so.)
2244 * The changes in array handling break binary compatibility, so
2245 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2246 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2247 instead of (VALUES T T).
2248 * By following the instructions that Dan Barlow posted to sbcl-devel
2249 on 2 July 2000, I was able to enable primitive dynamic object
2250 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2251 functionality is not implemented (since it calls ld to resolve
2252 library references automatically, requiring RUN-PROGRAM for its
2253 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2254 try to resolve library references) is now supported.
2255 * The system now flushes the standard output streams when it terminates,
2256 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2257 them at several other probably-convenient times, e.g. in each pass of
2258 the toplevel read-eval-print loop, and after evaluating a form given
2259 as an "--eval" command-line option. (These changes were motivated by a
2260 discussion of stream flushing issues on cmucl-imp in August 2000.)
2261 * The source transform for TYPEP of array types no longer assumes
2262 that an array whose element type is a not-yet-defined type
2263 is implemented as an array of T, but instead punts, so that the
2264 type will be interpreted at runtime.
2265 * There is now some support for cross-compiling in make.sh: each of
2266 the phases of make.sh has its own script. (This should be transparent
2267 to people doing ordinary, non-cross-compile builds.)
2268 * Since my laptop doesn't have hundreds of megabytes of memory like
2269 my desktop machine, I became more motivated to do some items on
2270 my to-do list in order to reduce the size of the system a little:
2271 ** Arrange for various needed-only-at-cold-init things to be
2272 uninterned after cold init. To support this, those things have
2273 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2274 symbols with such names are now uninterned after cold init).
2275 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2276 fasl files for cold load.
2277 ** Remove the old compiler structure pooling code (which used to
2278 be conditional on the target feature :SB-ALLOC) completely.
2279 ** Redo the representation of some data in cold init to be more compact.
2280 (I also looked into supporting byte compiled code at bootstrap time,
2281 which would probably reduce the size of the system a lot, but that
2282 looked too complicated, so I punted for now.)
2283 * The maximum signal nesting depth in the src/runtime/ support code has
2284 been reduced from 4096 to 256. (I don't know any reason for the very
2285 large old value. If the new smaller value turns out to break something,
2286 I'll probably just bump it back up.)
2287 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2289 * Many, many bugs reported by Peter Van Eynde have been added to
2290 the BUGS list; some have even been fixed.
2291 * While enabling dynamic object file loading, I tried to make the
2292 code easier to understand, renaming various functions and variables
2293 with less ambiguous names, and changing some function calling
2294 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2295 * While trying to figure out how to do the OpenBSD port, I tried to
2296 clean up some of the code in src/runtime/. In particular, I dropped
2297 support for non-POSIX signal handling, added various comments,
2298 tweaked the code to reduce the number of compilation warnings, and
2299 renamed some files to increase consistency.
2300 * To support the new automatic configuration functionality in make.sh,
2301 the source file target-features.lisp-expr has been replaced with the
2302 source file base-target-features.lisp-expr and the machine-generated
2303 file local-target-features.lisp-expr.
2304 * fixed a stupid quoting error in make.sh so that using CMU CL
2305 "lisp -batch" as cross-compilation host works again
2307 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2309 * DESCRIBE no longer tries to call itself recursively to describe
2310 bound/fbound values, so that it no longer fails on symbols which are
2311 bound to themselves (like keywords, T, and NIL).
2312 * DESCRIBE now works on generic functions.
2313 * The printer now prints less-screwed-up representations of closures
2314 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2315 * A private symbol is used instead of the :EMPTY keyword previously
2316 used to mark empty slots in hash tables. Thus
2317 (DEFVAR *HT* (MAKE-HASH-TABLE))
2318 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2319 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2320 now does what ANSI says that it should. (You can still get
2321 similar noncompliant behavior if bang on the hash table
2322 implementation with all the symbols you get back from
2323 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2324 This breaks binary compatibility, since tests for equality to
2325 :EMPTY are wired into things like the macroexpansion of
2326 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2328 * There's now a minimal placeholder implementation for CL:STEP,
2329 as required by ANSI.
2330 * An obscure bug in the interaction of the normal compiler, the byte
2331 compiler, inlining, and structure predicates has been patched
2332 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2333 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2335 * Missing ordinary arguments in a macro call are now detected even
2336 when the macro lambda list contains &KEY or &REST.
2337 * The debugger no longer complains about encountering the top of the
2338 stack when you type "FRAME 0" to explicitly instruct it to go to
2339 the top of the stack. And it now prints the frame you request even
2340 if it's the current frame (instead of saying "You are here.").
2341 * As specified by ANSI, the system now always prints keywords
2342 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2344 * The default initial SIZE of HASH-TABLEs is now smaller.
2345 * Type information from CLOS class dispatch is now propagated
2346 into DEFMETHOD bodies, so that e.g.
2347 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2349 is now basically equivalent to
2350 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2351 (DECLARE (TYPE SINGLE-FLOAT X))
2353 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2354 operation, without having to do run-time type dispatch.
2355 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2356 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2357 * Since I don't seem to be making big file reorganizations very often
2358 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2359 is overflowing my ability to conveniently back them up), I've finally
2360 checked the system into CVS. (The CVS repository is on my home system,
2361 not at SourceForge -- putting it on SourceForge might come later.)
2362 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2363 high-level GC-NOTIFY-FOO functions send their output. (There's
2364 still very little control of where low-level verbose GC functions
2365 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2366 less than it used to -- the GC-NOTIFY-FOO functions are now under
2367 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2368 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2369 in only one place in the source code, and propagates it automatically
2370 everywhere that it's needed. Thus e.g. when I bump the version from
2371 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2372 * The C source files now include boilerplate legalese and documentation
2373 at the head of each file (just as the Lisp source files already did).
2374 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2375 to his page will be replaced with a link to his new CLiki service.
2377 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2379 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2380 * The build process now looks for GNU make under the default name "gmake",
2381 instead of "make" as it used to. If GNU make is not available as "gmake"
2382 on your system, you can change this default behavior by setting the
2383 GNUMAKE environment variable.
2384 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2385 can build without error under CMU CL.
2387 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2389 * There is now a partial SBCL user manual (with some new text and some
2390 text cribbed from the CMU CL manual).
2391 * The beginnings of a profiler have been added (starting with the
2392 CMU CL profiler and simplifying and cleaning up). Eventually the
2393 main interface should be through the TRACE macro, but for now,
2394 it's still accessed through vaguely CMU-CL-style functions and macros
2395 exported from the package SB-PROFILE.
2396 * Some problems left over from porting CMU CL to the new
2397 cross-compilation bootstrap process have been cleaned up:
2398 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2399 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2400 part when getting rid of the compiler *BACKEND* stuff.)
2401 ** The value of *NULL-TYPE* was screwed up, because it was
2402 being initialized before the type system knew the final
2403 definition of the 'NULL type. This screwed up several key
2404 optimizations in the compiler, causing inefficiency in all sorts
2405 of places. (I found it because I wanted to understand why
2406 GET-INTERNAL-RUN-TIME was consing.)
2407 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2408 PROCLAIM FTYPE information. Unfortunately this broke binary
2409 compatibility again, since now the forms output by DEFGENERIC
2410 to refer to functions which didn't exist in 0.6.3.
2411 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2412 can use the new (as of 0.6.3) transform for SOME into MAP into
2414 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2415 values to (INTEGER 0 1000000), so that the time code will no longer
2416 occasionally get blown up by Linux returning 1000000 microseconds
2417 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2418 its output conform to the ANSI spec. (Alas, this makes its output
2419 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2420 * A full call to MAP NIL with a single sequence argument no longer conses.
2421 * fixes to problems pointed out by Martin Atzmueller:
2422 * The manual page no longer talks about multiprocessing as though
2423 it were currently supported.
2424 * The ILISP support patches have been removed from the distribution,
2425 because as of version 5.10.1, ILISP now supports SBCL without us
2426 having to maintain patches.
2427 * added a modified version of Raymond Toy's recent CMU CL patch for
2428 EQUALP comparison of HASH-TABLE
2430 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2432 * The system still can't cross-compile itself with
2433 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2434 entails), but at least it can compile more of itself that way
2435 than it used to be able to, and various buglets which were uncovered
2436 by trying to cross-compile itself that way have now been fixed.
2437 * This release breaks binary compatibility again. This time
2438 at least I've incremented the FASL file format version to 2, so that the
2439 problem can be detected reliably instead of just causing weird errors.
2440 * various new style warnings:
2441 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2442 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2443 ** using the lexical binding of a variable named in the *FOO* style
2444 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2445 as specified by ANSI.
2446 * *RANDOM-STATE* is no longer automatically initialized from
2447 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2448 default behavior of the system is to repeat its behavior every time
2449 it's run. If you'd like to change this behavior, you can always
2450 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2451 old convention there was no comparably easy way to get the system to
2452 repeat its behavior every time it was run.
2453 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2454 FTYPE declarations has been removed, in favor of their ANSI
2455 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2456 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2457 the types of their sequence arguments can be determined at compile time.
2458 This is done through a new open code expansion for MAP which eliminates
2459 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2460 argument types can be determined at compile time.
2461 * The optimizer now transforms COERCE into an identity operation when it
2462 can prove that the coerced object is already of the correct type. (This
2463 can be a win for machine generated code, including the output of other
2464 optimization transforms, such as the MAP transform above.)
2465 * Credit information has been moved from source file headers into CREDITS.
2466 * Source file headers have been made more standard.
2467 * The CASE macro now compiles without complaining even when it has
2470 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2472 * (Note that the way that the PCL macroexpansions were rewritten
2473 to accommodate the change in DEFGENERIC below breaks binary
2474 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2475 not run under sbcl-0.6.2. Once we get out of alpha releases,
2476 i.e. hit release 1.0.0, we'll probably try to maintain binary
2477 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2478 and sbcl-1.4.4. Until then, however, it might be fairly common
2479 for maintenance releases to break binary compatibility.)
2480 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2482 * The reporting of unhandled signals has been changed to print some
2483 explanatory text as well as the report form. (Previously only
2484 the report form was printed.)
2485 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2486 it defines, so that the compiler no longer issues undefined function
2487 warnings for compiled-but-not-yet-loaded generic functions.
2488 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2489 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2490 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2491 section of the ANSI spec are supported.
2492 * The "" nickname for the "KEYWORD" package has been removed.
2493 The reader still handles symbol tokens which begin with a package marker
2494 as keywords, but it doesn't expose its mechanism for doing so in the
2495 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2496 * The system now issues STYLE-WARNINGs for contradictory TYPE
2497 proclamations. (Warnings for contradictory FTYPE proclamations would
2498 be nice too, but those can't be done usefully unless the type system
2499 is made smarter about FUNCTION types.)
2500 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2501 systematized, so that "*target-*.lisp is supposed to exist only on the
2502 target and imply that there's a related file which exists on the
2503 host, and *host-*.lisp is supposed to exist only on the host and imply
2504 that there's a related file which exists on the target. This involves a
2505 lot of renaming. Hopefully the acute confusion caused by the renaming
2506 will be justified by the reduction in chronic confusion..
2507 ** runtime-type.lisp -> early-target-type.lisp
2508 ** target-type.lisp -> late-target-type.lisp
2509 ** early-host-format.lisp -> early-format.lisp
2510 ** late-host-format.lisp -> late-format.lisp
2511 ** host-error.lisp -> misc-error.lisp
2512 ** early-error.lisp -> early-target-error.lisp
2513 ** late-error.lisp -> late-target-error.lisp
2514 ** host-defboot.lisp -> early-defboot.lisp
2515 ** code/misc.lisp -> code/target-misc.lisp
2516 ** code/host-misc.lisp -> code/misc.lisp
2517 ** code/numbers.lisp -> code/target-numbers.lisp
2518 ** code/early-numbers.lisp -> numbers.lisp
2519 ** early-host-type.lisp -> early-type.lisp
2520 ** late-host-type.lisp -> late-type.lisp
2521 ** host-typep.lisp -> typep.lisp
2522 ** load.lisp -> target-load.lisp
2523 ** host-load.lisp -> load.lisp
2524 ** host-disassem.lisp -> disassem.lisp
2525 ** host-insts.lisp -> insts.lisp
2526 ** byte-comp.lisp -> target-byte-comp.lisp
2527 ** host-byte-comp.lisp -> byte-comp.lisp
2528 ** host-signal.lisp -> signal.lisp
2529 ** host-defstruct.lisp -> defstruct.lisp
2530 ** late-target-type.lisp -> deftypes-for-target.lisp
2531 Furthermore, several other previously target-only files foo.lisp (e.g.
2532 hash-table.lisp and random.lisp) have been split into a target-and-host
2533 foo.lisp file and a target-only target-foo.lisp file, with their key type
2534 definitions in the target-and-host part, so that the cross-compiler will
2535 know more about target types.
2536 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2537 gone away. In their place are various *BACKEND-FOO* variables
2538 corresponding to the slots of the old structure.
2539 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2540 into the target SBCL has been fixed.
2541 * The chill.lisp system for loading cold code into a running SBCL
2543 * Support for the CMU CL "scavenger hook" extension has been removed.
2544 (It was undocumented and unused in the CMU CL sources that SBCL was
2545 derived from, and stale in sbcl-0.6.1.)
2546 * Various errors in the cross-compiler type system were detected
2547 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2548 (enabling various consistency checks). Many of them were fixed,
2549 but some hard problems remain, so the compiler is back to
2550 running without *TYPE-SYSTEM-INITIALIZED* for now.
2551 * As part of the cross-compiler type system cleanup, I implemented
2552 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2553 * I have started adding UNCROSS calls throughout the type system
2554 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2555 on cross-compiler input files will be able to go away, and various
2557 * CONSTANTP now returns true for quoted forms (as explicitly required
2560 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2562 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2563 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2564 (MAKE-LIST -1). In the longer term, it ought to become true
2565 that declarations are assertions even at SAFETY 1. For now, it's not
2566 quite true even at SAFETY 3, but it's at least more nearly true..
2567 (Note that this change seems to increases the size of the system by
2568 O(5%) and to decrease the speed of the compiler by 20% or more.)
2569 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2570 for the problem of printing dozens of lines of distracting information
2571 about low-level system machinery as part of the top stack frame
2572 on entry to the debugger when an undefined function was called.
2573 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2574 is not reset to COMMON-LISP-USER.
2575 * Compilation of stuff related to dyncount.lisp has been made conditional
2576 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2577 smaller. The various dyncount-related symbols have been moved into
2578 a new "SB-DYNCOUNT" package.
2579 * tty-inspect.lisp has been renamed to inspect.lisp.
2580 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2581 feature has gone away. (When we eventually port to other flavors of
2582 libc and/or Unix, we'll try to make the differences between flavors
2583 invisible at the user level.)
2584 * Various other *FEATURES* tags, and/or their associated conditionals,
2585 have been removed if obsolescent, or given better documentation, or
2586 sometimes given more-mnemonic names.
2588 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2590 * tidied up "make.sh" script
2591 * tidied up system directory structure
2592 * better "clean.sh" behavior
2593 * added doc/FOR-CMUCL-DEVELOPERS
2594 * many many small tweaks to output format, e.g. removing possibly-confusing
2595 trailing #\. character in DESCRIBE-INSTANCE
2596 * (EQUALP #\A 'A) no longer signals an error.
2597 * new hashing code, including EQUALP hashing
2598 * tidied up Lisp initialization and toplevel
2599 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2600 * command line argument processing
2601 * added POSIX-GETENV function to deal with Unix-ish environment variables
2602 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2603 terminating SBCL on EOF
2604 * non-verbose GC by default
2605 * There is no more "sbcl" shell script; the sbcl file is now the C
2606 runtime executable (just like CMU CL).
2607 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2609 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2610 symbol and package information as Lisp native symbol and package objects
2611 instead of strings naming symbols and strings naming packages. This way,
2612 whenever packages are renamed (as in warm init), debug information is
2613 transformed along with everything else.
2614 * tweaked the optimization policy declarations which control the building
2615 of SBCL itself. Now, among other things, the system no longer saves
2616 source location debugging information. (This helps two problems at once
2617 by reducing SBCL size and by keeping SBCL from trying to look for its
2618 sources -- which may not exist -- when reporting errors.)
2619 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2620 debugging and testing purposes
2621 * cleaned up printing, making the printer call PRINT-OBJECT for
2622 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2623 methods, giving nearly-ANSI behavior
2624 * converted almost all special variables to use *FOO* naming convention
2625 * deleted PARSE-TIME functionality, since it can be done portably
2626 * moved some files out of cold init into warm init
2627 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2629 * regularized formatting of source files
2630 * added an install.sh script
2631 * fixed ridiculous memory usage of cross-compiler by making
2632 compiler/alloc.lisp not try to do pooling unless it can hook
2633 itself into the GC of the cross-compilation host. Now the system
2634 builds nicely on my old laptop.
2635 * added :SB-ALLOC in target-features.lisp-expr
2636 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2637 was not implemented)
2638 * re-did condition handling and note reporting in the compiler. Notes
2639 are no longer handled by signalling conditions. Style warnings
2640 and warnings are handled more correctly and reported in such a way
2641 that it's easy to find one or the other in your output (so that you
2642 can e.g. figure out which of many problems caused COMPILE-FILE to
2644 * changed the severity of several compiler warnings from full WARNING
2645 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2646 compiler note reporting so that it doesn't use the condition system
2647 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2648 COMPILE-FILE command)
2649 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2650 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2651 no longer supported, since their implementation was deeply intertwingled
2652 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2653 * removed bogus "support" for compiler macros named (SETF FOO), and
2654 removed the compiler macro for SETF INFO (but only after making a fool
2655 of myself on the cmucl-imp mailing list by posting a bogus patch for
2656 DEFINE-COMPILER-MACRO..)
2657 * Compiled files containing forms which have side effects on the Lisp
2658 reader (such as DEFPACKAGE forms) are now handled more correctly.
2659 (Compiler queuing of top level lambdas has been suppressed by setting
2660 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2661 * deleted various currently-unused source files, e.g. gengc.lisp. They
2662 may be added back at some point e.g. when porting to other architectures,
2663 but until they are it's distracting to distribute them and to try to
2665 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2666 there were so many of them they're just distractions, and UNCROSS is
2667 known to be able to handle the current sources
2668 * moved PROFILE functionality into TRACE, so that it will be clear
2669 how the wrapping and unwrapping of functions when you profile them
2670 interacts with the wrapping and unwrapping of functions when you
2671 trace them. (Actually, the functionality isn't there yet, but at least
2672 the interface specification is there. Hopefully, the functionality will
2673 arrive with some future maintenance release.)
2674 * removed host-oops.lisp
2675 * changed signature of QUIT function to allow UNIX-CODE argument
2676 * fixed READ-SEQUENCE bug
2677 * tweaked verbose GC output so that it looks more like the progress
2678 output that ANSI specifies for functions like LOAD
2679 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2680 * added <http://sbcl.sourceforge.com> to the banner information printed by