1 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
3 the manual for details; for now, package locks can be disabled by
4 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
5 if no major problems are found then it is likely that they will be
6 compiled in unconditionally.
7 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
8 unsupported operators on all platforms. To load a shared library
9 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
10 object file, link it into a shared library outside of SBCL and
11 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
13 * bug workaround: ROOM T and the GC were not working together
14 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
15 reporting facility aren't preserved. That reporting has been
16 disabled, so now until and unless someone figures out how to make it
17 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
18 * minor incompatible change: as threatened around sbcl-0.8.0, the
19 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
20 The symbols are also exported from SB-PCL for backwards
21 compatibility, but more so than before SB-PCL should be treated as
22 an implementation-internal package.
23 * the SB-SPROF contrib now works on (most) non-x86 architectures.
24 It is known as of this release not to work on the Alpha, however.
25 * fixed bug #167: errors signalled due to illegal syntax in method
26 bodies are now more legible.
27 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
28 designators and can hence be used with TYPEP.
29 * fixed bug #333: CHECK-TYPE now ensures that the type error
30 signalled, if any, has the right object to be accessed by
31 TYPE-ERROR-DATUM. (reported by Tony Martinez)
32 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
33 5.1.2.3 for multiple-value place subforms. (reported by Kalle
35 * fixed bug #334: programmatic addition of slots using specialized
36 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
37 and :ALLOCATION :CLASS slots.
38 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
39 float exponents. (rereported by Peter Seibel)
40 * fixed a bug: #\Space (and other whitespace characters) are no
41 longer considered to be macro characters in standard syntax by
43 * fixed bug: initialization of condition class metaobjects no longer
44 causes an instance of the condition to be created. (reported by
46 * fixed bug: it is now possible to have slots such that
47 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
48 description is neither :INSTANCE nor :CLASS.
49 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
50 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
51 SLOT-VALUE-USING-CLASS), and no longer causes errors with
52 non-standard SLOT-DEFINITION-ALLOCATION values.
53 * fixed bugs: various race conditions handling exiting threads.
54 CL-PPCRE's thread torture test now passes.
55 * fixed arguably-a-bug: GC time proportional to number of threads in
56 system even when most of them are idle
57 * optimization: improved performance of BIT and SBIT on bit-vectors.
59 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
60 * minor incompatible change: the system no longer provides
61 optimization hints (conditions which are TYPEP
62 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
63 mode; these hints are still emitted when higher SPEED optimization
65 * new contrib module: a sampling profiler (profiling by statistical
66 sampling, rather than by instrumenting functions) is available as
67 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
68 * the behaviour of the standard function ED is now customizeable by
69 third parties through a hook variable: see ED's documentation
70 string for information on the protocol.
71 * the compiler no longer emits efficiency notes for (FUNCALL X)
72 when the type of X is uncertain under default optimization
74 * fixed bug 276: mutating a binding of a specialized parameter to a
75 method to something that is not TYPEP the specializer is now
77 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
78 exhibit the required behaviour on the x86 platform. (thanks to
79 Peter van Eynde, Eric Marsden and Bruno Haible)
80 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
81 even for difficult arguments. (reported by Peter Graves)
82 * fixed bug 141a: the backquote printer now descends quoted
84 * fixed another bug in backquote printing: no more destructive
85 modification of the form's list structure. (reported by Brian
87 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
88 move between its address being taken and the call to
89 interrupt_thread, fixing a crashing race condition.
90 * the SB-POSIX contrib implementation has been adjusted so that it
91 no longer exhibits ridiculously poor performance when constructing
92 instances corresponding to C structs.
94 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
95 * minor incompatible change: the sb-grovel contrib now treats C
96 structures as alien (in the sense of SB-ALIEN) objects rather than
97 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
98 has implications for memory management of client code
99 (sb-grovel-returned objects must now be manually managed) and for
100 type safety (alien objects now have full types).
101 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
102 used to control emission of compiler diagnostics, rather than the
103 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
104 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
105 quality should be considered deprecated.
106 * install.sh now installs the user manual as well
107 * (not quite a new documentable feature, but worth considering in
108 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
109 beginnings of a semantically meaningful condition hierarchy is
110 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
112 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
113 designator as the defaults argument.
114 * fixed bug: Displaced arrays whose displaced-to array has become
115 too small now cause ARRAY-DIMENSION to signal an error, providing
116 for safer bounds-checking. (reported by Bruno Haible)
117 * fixed bug: DEFCLASS slot definitions with identical :READER and
118 :WRITER names now signal a reasonable error. (reported by Thomas
120 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
122 * fixed bug: Invalid slot specification errors now print correctly.
123 (thanks to Zach Beane)
124 * fixed bug 320: Shared to local slot value transfers in class
125 redefinitions now happen corrently from superclasses as
126 well. (reported by Bruno Haible)
127 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
129 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
130 improper lists correctly. (reported by Bruno Haible)
131 * fixed bug 313: source-transform for <fun-name> was erroneously
132 applied to a call of a value of a variable with name <fun-name>.
133 (reported by Antonio Menezes Leitao)
134 * fixed bug 307: The obsolete instance protocol ensures that
135 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
136 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
137 remove dead unknown-values globs from the middle of the stack.
138 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
139 easier to resume long computations after using *BREAK-ON-SIGNALS*
140 to diagnose and fix failures. (thanks to Nikodemus Siivola)
141 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
142 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
143 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
144 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
145 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
147 * fixed a (fairly theoretical) bug in string printing: if
148 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
149 does not have array-element-type equal to the most general string
151 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
152 function name. (thanks to Nikodemus Siivola)
153 * fixed bug: compile-time format string checker failed on
154 non-closed ~{. (reported by Thomas F Burdick)
155 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
156 don't warn on legal loop code involving more than one aggregate
157 boolean. (thanks to Nikodemus Siivola)
158 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
159 in keyword position to MAKE-PACKAGE was not being reported as
161 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
162 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
164 * fixed bugs 280 and 312: the checking for multiple definitions in a
165 file is less likely to become confused by uses of inline
167 * fixed bug: the #S reader macro performs the keyword coercion
168 specified for slot names. (reported by Kalle Niemitalo)
169 * fixed bug: lambda lists may contain symbols whose names start with
170 & but are not lambda-list-keywords; their occurrence triggers a
172 * fixed bug 321: define-method-combination argument lambda lists do
173 not cause invalid code to be generated when &optional or &aux
174 variables have default values. (reported by Bruno Haible)
175 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
176 have CLOS classes; this appears also to have fixed reported
177 instability in stack exhaustion detection.
178 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
179 MUFFLE-WARNING when no associated restart is present is now
181 * optimization: rearranged the expansion of various defining macros
182 so that each expands into only one top-level form in a
183 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
185 * optimization: used a previously little-used slot in symbols to
186 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
188 * fixed some bugs revealed by Paul Dietz' test suite:
189 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
191 ** Fixed an optimization bug related to inheritance of initforms
192 from local to shared slots.
193 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
194 ** CAST optimizer forgot to flush argument derived type.
195 ** print/read consistency on floats is now orders of magnitude
196 more likely. (thanks also to Bruno Haible for a similar report
198 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
199 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
200 for objects of type REAL. Make it so.
201 ** SXHASH is specified (infelicitously) to respect similarity,
202 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
203 it so. (thanks to Markus Ziegler)
204 ** on the Alpha, the compiler succeeds in compiling functions
205 returning a known number of arguments greater than 63.
206 ** fixed handling of invalid NIL arguments in keyword position
208 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
210 ** provided a readably-printable representation for RANDOM-STATE
212 ** ensured that pathnames, where they have a namestring, always
213 print using #P"..." syntax.
215 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
216 * Support for the forthcoming 2.0 version of the NetBSD kernel
217 running on x86 hardware has been added. (thanks to Perry
218 E. Metzger most immediately, and others for their past work)
219 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
220 current release version; SBCL's OpenBSD support had been broken
221 since about the time of OpenBSD's switch to ELF binary format.)
222 * [placeholder for DX summary]
223 ** user code with &REST lists declared dynamic-extent, under high
224 speed or space and low safety and debug optimization policy.
225 * The manual has been converted to Texinfo format and the debugger
226 chapter from the cmucl manual has been added.
227 * A facility has been added to extract documentation strings from
228 sbcl and store them as Texinfo-formatted snippets for inclusion in
229 the manual (via Texinfo's @include directive)
230 * bug fix: compiler emitted division in optimized DEREF. (thanks for
231 the test case to Dave Roberts)
232 * bug fix: multidimensional simple arrays loaded from FASLs had fill
233 pointers. (reported by Sean Ross)
234 * bug fix: PROFILE output is printed nicely even for large numerical
235 values. (thanks to Zach Beane)
236 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
237 greater than 32 handle EOF correctly.
238 * bug fix: on X86 an immediate argument of the IMUL instruction is
239 correctly printed in disassembly. (thanks to Lutz Euler)
240 * bug fix: class slots in redefined classes preserve their old
241 values. (thanks to Bruno Haible and Nikodemus Siivola)
242 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
243 succeeds. (reported by Marco Baringer)
244 * fixed some bugs revealed by Paul Dietz' test suite:
245 ** READ-SEQUENCE now works on ECHO-STREAMs.
246 ** RATIONALIZE works more according to its specification. (thanks
249 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
250 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
251 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
252 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
253 should matter to you only if you rebind the printer control
254 variables and then find you want different bindings in the
255 debugger than in the ordinary execution of your program.)
256 * The runtime build system has been tweaked to support building
257 (on SPARC/SunOS) using a C compiler which invokes Sun's own
258 assembler and linker. (thanks to Nikodemus Siivola)
259 * Unbound, undefined, undeclared variables now trigger full
260 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
261 more often programmer error than deliberate exploitation of undefined
263 * optimization: the hash algorithm for strings has changed to one
264 that is less vulnerable to spurious collisions. (thanks to Juho
266 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
267 needless bounds checking. (thanks to Juho Snellman)
268 * optimization: implemented multiplication as a modular
269 (UNSIGNED-BYTE 32) operation on the PPC backend.
270 * fixed some bugs revealed by Paul Dietz' test suite:
271 ** ADJUST-ARRAY now copies the datum in a zero rank array if
273 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
274 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
275 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
277 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
279 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
281 ** ECHO-STREAMs no longer attempt to echo the end of file value to
282 their output stream on EOF from read.
283 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
284 have been read to end-of-file.
285 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
287 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
288 description of determination of which consecutive characters
290 ** Printing the "Space" character with escaping on now yields "#\\ ",
291 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
292 ** Reading floating-point numbers with *READ-BASE* set to a number
293 less than 10 works correctly.
294 ** Reading floating-point numbers with *READ-BASE* set to a number
295 more than 10 works correctly.
296 ** Printing with *PRINT-READABLY* targets the standard readtable, not
297 the readtable currently in effect.
299 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
300 * minor incompatible change: parsing of namestrings on a physical
301 (Unix) host has changed; numbers after the final #\. in a
302 namestring are no longer interpreted as a version field. This is
303 intented to be largely invisible to the user, except that the
304 meaning of the namestring "*.*.*" has changed: it now refers to a
305 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
306 should usually be replaced by
307 (make-pathname :name :wild :type :wild :version :wild)
308 with the added benefit that this is more likely to be portable.
309 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
310 option to OPEN now signals an error if the file being opened
311 exists; this may have an impact on existing code.
312 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
313 signals received, on the PowerPC platforms (both Linux and
314 Darwin). (thanks to Pierre Mai for pointing out the location of
316 * several fixes on OS X: The system now builds and runs cleanly on
317 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
318 (thanks to Brian Mastenbrook)
319 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
320 second-resolution: integer multiples of 1/3600 between -24 and 24.
321 (thanks to Vincent Arkesteijn)
322 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
323 type when called with 1 argument; PEEK-CHAR checked type of
324 PEEK-TYPE only after having read first character from a
325 stream. (reported by Peter Graves)
326 * bug fix: the garbage collector now has much better locality
327 behaviour, and in particular no longer treats all memory as being
328 exhausted when many small objects point to each other in a deeply
330 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
331 recognized as being TYPEP their class.
332 * bug fix: the PUSHNEW documentation string has been corrected.
333 (thanks to Vincent Arkesteijn)
334 * bug fix: defaulting of the value for the last of an atypically
335 large number of multiple values being bound was not being
336 performed correctly on the Alpha or PPC platforms
337 * optimization: implemented multiplication as a modular
338 (UNSIGNED-BYTE 32) operation on the x86 backend.
339 * optimization: SEARCH on simple-base-strings can now be open-coded.
340 (see also contrib/compiler-extras.lisp for inspiration for
341 teaching the compiler about the Boyer-Moore algorithm).
342 * value, returned by MAX (and MIN) called with several EQUALP, but
343 not EQL, arguments now does not depend on compiler settings.
344 * fixed some bugs revealed by Paul Dietz' test suite:
345 ** in stack analysis liveness information is propagated from
346 non-local entry points.
347 ** pathwise CAST removing failed when the CAST node did not start
349 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
350 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
352 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
353 host is already defined.
354 ** RENAME-FILE works on streams instead of signalling an internal
356 ** PEEK-CHAR uses the current readtable when determining whether
357 or not a character is whitespace.
358 ** MERGE-PATHNAMES handles the case when the pathname does not
359 specify a name while the default-pathname specifies a version.
360 ** Pathnames now stand a better chance of respecting print/read
362 ** Attempting to use standardized file system operators with a
363 pathname with invalid :DIRECTORY components signals a
365 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
367 ** DIRECTORY on logical pathnames is more correct.
368 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
369 signal a TYPE-ERROR if their argument is not a stream.
370 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
371 designator argument does not designate a stream.
372 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
373 examining the synonym.
374 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
376 ** OPEN and WITH-OPEN-STREAM allow opening streams with
377 element-type larger than ([UN]SIGNED-BYTE 32).
379 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
380 * When built with the :SB-FUTEX feature, threaded builds now take
381 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
382 for faster/more reliable mutex and condition variable support.
383 * Incompatible change (but one you probably shouldn't have been using
384 anyway): the interface and code for arbitrating between multiple
385 threads in the same user session has been redesigned.
386 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
387 error when called without an explicit environment argument.
388 (thanks to Nikodemus Siivola)
389 * bug fix: buffered :DIRECTION :IO streams are less likely to become
390 confused about their position. (thanks to Adam Warner and Gerd
392 * bug fix: Pretty printing backquoted forms with unquotations in the
393 argument list position of various code constructs such as LAMBDA
394 now works correctly. (reported by Paul Dietz)
395 * bug fix: Pretty printing unquotations no longer loses all
396 stream position information.
397 * optimization: performance of string output streams is now less
398 poor for multiple small sequence writes.
399 * optimization: performance of CSUBTYPEP in the presence of complex
400 expressions involving CONS and NOT many times has been improved.
401 (reported by Paul Dietz)
402 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
404 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
405 streams for the REPL, for the debugger, and so on. (thanks to
407 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
408 * fixed some bugs revealed by Paul Dietz' test suite:
409 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
410 is ignored if it is too small, rather than propagating through
411 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
412 ** extremely complex negations of CONS types were not being
413 sufficiently canonicalized, leading to inconsistencies in
415 ** VALUES tranformer lost derived type.
417 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
418 * fixed a bootstrapping bug: the build process no longer assumes
419 that the various BOOLE-related constants have the same value in
420 host and target lisps. (noted by Paul Dietz' test suite on an
421 SBCL binary built from CLISP)
422 * The system can now be dynamically linked on the MIPS platform,
423 which enables dynamic loading of foreign code from Lisp. (thanks
424 to Ralf Baechle for discussions on the MIPS ABI)
425 * The system now records debugging information for its own source
426 files in a filesystem-position-independent manner, relative to
427 the "SYS" logical host.
428 * fixed a compiler bug: MV-LET convertion did not check references
429 to the "max args" entry point. (reported by Brian Downing)
430 * tweaked disassembly notes to be less confident about proclaiming
431 some instruction as an LRA. (thanks to Brian Downing)
432 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
433 now each have their own history, command character, and other
434 characteristics. (thanks to David Lichteblau)
435 * fixed some bugs revealed by Paul Dietz' test suite:
436 ** compiler failure in compiling LOGAND expressions including a
438 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
439 shift greater than 32.
440 ** FLUSH-DEST did not mark blocks for type check regeneration.
441 ** HANDLER-CASE failed to accept declarations in handler clauses
442 in some circumstances.
444 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
445 * New code in contrib/sb-introspect (still probably not entirely
446 stable yet) provides some support for smart Lisp development
447 environments like SLIME.
448 * The conditions signalled for errors occurring when loading .fasl
449 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
450 in a way which should help ASDF recover gracefully.
451 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
452 stuff has been cleaned up. If you code contrib/ stuff, this might
453 affect you, and you can look at contrib/README, contrib/STANDARDS,
454 and/or the 0.8.4.27 diff to check.
455 * In full calls the compiler now does not generate checks for declared
456 argument types for all arguments.
457 * various threading fixes
458 ** and some experimental patches which didn't make it into
459 the main tree for this release, but which are shipped in
460 contrib/experimental-thread.patch as a possible fix for some
461 failures (deadlock, spinning...) in GC-intensive multithreaded
463 * fixed PPC build problem (source code incompatibility of different
464 library versions): added offsetof() hackery which attempts to divine
465 where glibc maintainers put uc_mcontext today
466 * fixed bug 282: compiler does not trust type assertions while passing
467 arguments to a full call.
468 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
469 &OPTIONAL VALUES type parameter.
470 * fix bug 214: algorithm for noting rejected templates is now more
471 similar to that of template seletion. (also reported by rydis on
473 * fixed bug 141b: printing backquoted information readably and prettily
474 inserts a space where necessary.
475 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
476 (CHARACTER 10) are now reported as errors, rather than propagated
477 as unknown types. (reported by piso on #lisp)
478 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
479 with values NIL and :ERROR. (thanks to Milan Zamazal)
480 * fixed bug 191c: CLOS now does proper keyword argument checking as
481 described in CLHS 7.6.5 and 7.6.5.1.
482 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
483 longer raise an error; further, using a list as a for-as-arithmetic
484 counter now raises a meaningful error.
485 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
486 now understood by sequence creation functions such as MAKE-SEQUENCE
488 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
489 a STRING-INPUT-STREAM.
490 * compiler enhancement: SIGNUM is now better able to derive the type
492 * type declarations inside WITH-SLOTS are checked. (reported by
494 * fixed some bugs revealed by Paul Dietz' test suite:
495 ** incorrect optimization of TRUNCATE for a positive first
496 argument and negative second.
497 ** compiler failure in let-convertion during flushing dead code.
498 ** compiler failure while deriving type of TRUNCATE on an
499 interval, containing 0.
500 ** ASH of a negative bignum by a negative bignum count now returns
502 ** intersection of CONS types now canonicalizes properly, fixing
503 inconsistencies in SUBTYPEP.
505 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
506 * incompatible change: The --disable-debugger command line
507 option now clobbers the debugger at a more fundamental
508 level, by redefining #'INVOKE-DEBUGGER instead of by
509 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
510 is specified by ANSI to ignore *DEBUGGER-HOOK* and
511 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
512 enter the debugger REPL and then suffer recursive errors
513 because *DEBUG-IO* is also messed up in --disable-debugger mode;
514 while under the new system, BREAK in --disable-debugger mode
515 terminates the system just as an unhandled error would.
516 * fixed compiler performance when processing loops with a step >1;
517 * bug fix: DOCUMENTATION now retrieves generic function
518 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
519 support has been systematized, and now supports the methods
520 specified by ANSI, along with a default method and a method for
521 slot documentation. (reported by Nathan Froyd)
522 * bug fix: effective methods associated with a generic function are
523 no longer cached over a change of that generic function's method
524 combination. (reported by Andreas Fuchs)
525 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
526 names a non-existent directory. (thanks to Andreas Fuchs)
527 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
528 the PPC platform, lead to stack corruption; this has been fixed.
529 (reported by Rainer Joswig)
530 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
531 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
532 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
533 arithmetic in general, is now much more reliable on the PPC
535 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
536 platform now returns the right answer.
537 * optimization: restored some effective method precomputation in
538 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
539 precomputation is now tunable.
540 * optimization: compiler-internal data structure use has been
541 reviewed, and changes have been made that should improve the
542 performance of the compiler by about 20%.
543 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
544 simple-base-strings and simple-bit-vectors is improved.
545 * optimization: the optimization of 32-bit logical and arithmetic
546 functions introduced in version 0.8.3 on the x86 has been
547 implemented on the mips, ppc and sparc platforms; an
548 implementation of the same facility, but for 64-bit arithmetic,
549 has been added for the alpha.
550 * microoptimization: the compiler is better able to make use of the
551 x86 LEA instruction for multiplication by constants.
552 * bug fix: in some situations compiler did not report usage of
553 generic arithmetic in (SPEED 3) policy.
554 * bug 145b fix: compiler used wrong type specifier while converting
555 MEMBER-types to numeric.
556 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
558 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
559 index argument now works properly on the Alpha platform.
560 * bug fix: floating point exception treatment on the Alpha platform
562 * bug fix: FILE-POSITION works much better on string input and
563 output streams. (thanks to Nikodemus Siivola)
564 * bug fix: many threading/garbage collection symptoms sorted.
565 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
566 might be pseudo-atomic.
567 * internal change: Stopping for GC is now done with signals not
568 ptrace. GC is now done in whichever thread wanted it, instead of
570 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
572 * bug fix: result form in DO is not contained in the implicit
574 * incompatible change: ICR structure is changed; the value part of
575 CONTINUATION is now called LVAR; corresponding functions are
576 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
577 * added type deriver for ISQRT (thanks to Robert E. Brown).
578 * bug fix: better support for loading from the command line when an
579 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
581 * fixed some bugs revealed by Paul Dietz' test suite:
582 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
583 ** ROUND and FROUND now give the right answer when given very
584 small float arguments.
585 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
587 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
588 ** (GCD 0 <negative-integer>) returned <negative-integer>.
589 ** LCM should return a non-negative integer.
590 ** PARSE-INTEGER returned the index of a terminator instead of the
591 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
592 ** PARSE-INTEGER returned an incorrect index being applied to a
594 ** LCM with two arguments of 0 returns 0 rather than signalling
596 ** unsigned addition of a 32-bit constant with the high bit set no
597 longer causes an internal compiler error on the x86.
598 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
599 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
600 with negative last argument.
601 ** byte specifiers with zero size and position no longer cause
602 an error during type derivation.
603 ** bignum multiplication on the Alpha platform now returns the
605 * porting: The system now builds on SuSE AMD64, although it still
606 generates a 32-bit binary.
607 * .fasl file incompatibility: The fasl file version number has
608 been incremented (because of the changes to internal compiler
609 data structures referred to above).
611 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
612 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
613 more accurately, on the Darwin kernel running on PowerPC hardware.
614 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
615 * Compiler code deletion notes now signal a condition of type
616 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
617 an associated MUFFLE-WARNING restart.
618 * The compiler now performs limited argument count validation of
619 constant format strings in FORMAT, and where appropriate in ERROR,
620 CERROR and WARN. (thanks to Gerd Moellmann)
621 * New ASDF-INSTALL contrib can be used for automatic download and
622 installation of third-party Lisp code from CCLAN or other sites
624 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
625 which forces another thread to execute a function supplied by the
627 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
628 now accept and act upon their :ELEMENT-TYPE keyword argument.
629 (reported by Martin Atzmueller, Edi Weitz)
630 * bug fix: FILE-POSITION now accepts position designators up to
631 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
632 is the greater. (thanks to Patrik Nordebo)
633 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
634 Kalle Olavi Niemitalo)
635 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
636 printable. (reported by Eric Marsden)
637 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
638 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
639 * x86 bug fix in control stack exhaustion checking: now shows backtrace
640 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
641 (thanks to Stig Sandoe)
642 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
643 UNREAD-CHAR now correctly includes the unread character in the
644 target sequence. (thanks to Gerd Moellmann)
645 * bug fix in threaded builds: the system can now be suspended and
646 resumed by shell job control with minimal disruption.
647 * bug fixes in times and timezones >2038AD
648 * better handling of "where is GNU make?" problem in build scripts
649 (thanks to Nikodemus Siivola)
650 * new optimization: inside a named function any reference to a
651 function with the same name is considered to be a self-reference;
652 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
653 optimization quality.
654 * new optimization on x86: logical functions and + now have
655 optimized (UNSIGNED-BYTE 32) versions, which are automatically
656 used when the result is truncated to 32 bits.
657 * VALUES declaration is partially enabled.
658 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
659 * fixed some bugs revealed by Paul Dietz' test suite:
660 ** The system now obeys the constraint imposed by
661 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
662 types form a lattice under type intersection.
663 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
664 ** ASSOC now ignores NIL elements in an alist.
665 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
666 and (1+ MOST-POSITIVE-FIXNUM) answers.
667 ** The addition of a method with invalid qualifiers to a generic
668 function does not cause an error to be signalled immediately;
669 a warning is signalled, and the error is generated only on
670 calling the generic function.
671 * changed .fasl file version number, in part to add type codes for
672 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
673 obscure ANSI requirements
675 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
676 * fixed bug 148: failure to inline-expand a local function left
677 garbage, confusing the compiler.
678 * fixed bugs 3cd: structure slot readers perform type check if the
679 slot can have an invalid value (i.e. it is either not initialized
680 or can be written with a less specific slot writer).
681 * bug fix: the compiler now traps array references to elements off
682 the end of an array; previously, the bounds checking in some
683 circumstances could go off-by-one.
684 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
686 * type declarations for array element types now obey the description
687 on the CLHS page "Declaration TYPE", as per discussions on
688 sbcl-help around 2003-05-08. This means that a declaration
689 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
690 declaration, all references to BAR will be asserted or assumed
691 (with THE, so dependent on compiler policy) to involve objects of
692 type FOO. Note that no such declaration is implied in
693 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
694 * declared types of functions from the "Conditions"
695 chapter. (reported by Paul Dietz)
696 * bug fix: CERROR accepts a function as its first argument.
697 * bug fix: NTH an NTHCDR accept a bignum as index
698 arguments. (reported by Adam Warner)
699 * optimization: character compare routines now optimize comparing
700 against a constant character. (reported by Gilbert Baumann)
701 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
702 argument now works properly on the MIPS platform.
703 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
704 * repeated evaluation of the same DEFSTRUCT, a slot of which is
705 declared to have a functional type, does not cause an error
707 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
708 many :DEFAULT-INITARGS. (reported by Istvan Marko)
709 * fixed bug: if last continuation of a deleted block has a
710 destination, this destination should be deleted too. (reported by
712 * fixed a bug in the bootstrap process: the host compiler's values
713 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
714 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
715 test case from Patrik Nordebo)
716 * improved the ability of the disassembler on the PPC platform to
717 provide helpful disassembly notes.
718 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
719 the class in more cases than previously.
720 * bug fix: FILE-POSITION now understands :START and :END for
721 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
722 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
723 8) data. (thanks to David Lichteblau)
724 * bug fix: it is possible to add a method to a generic function
726 * bug fix: reader failed to signal END-OF-FILE inside an
727 object representation. (reported by Nikodemus Siivola)
728 * fixed some bugs revealed by Paul Dietz' test suite:
729 ** LAST and [N]BUTLAST should accept a bignum.
730 ** condition slot accessors are methods.
731 ** (VECTOR NIL) is a subtype of STRING.
733 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
734 * minor incompatible change: some nonsensical specialized lambda
735 lists (used in DEFMETHOD) which were previously ignored now signal
737 * minor incompatible change: the system is now aware of the types of
738 variables in the COMMON-LISP package, and will signal errors for
739 most violations of these type constraints (where previously they
740 were silently accepted).
741 * minor incompatible change: COMPILE-FILE now uses the freedom
742 afforded (ANSI 3.2.2.3) to use derived function types for
743 functions defined in the same file. This also permits the system
744 to warn on static type mismatches and function
745 redefinition. (Currently it does not work with high DEBUG level.)
746 * minor incompatible change: VALUES declaration is disabled.
747 * When issuing notes, the compiler now signals a condition of type
748 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
749 restart for use in user handlers. It is expected that the
750 COMPILER-NOTE condition will eventually become a condition
751 supertype to a hierarchy of note types, which will then be
752 handleable in a similar fashion. However, at the moment, no such
753 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
754 but it's an implementation detail, not a classification for the
756 * Changes in type checking closed the following bugs:
757 ** type checking of unused values (192b, 194d, 203);
758 ** template selection based on unsafe type assertions (192c, 236);
759 ** type checking in branches (194bc).
760 * A short form of VALUES type specifier has ANSI meaning (it has
761 increased the number of situations when SBCL cannot perform type
763 * fixed bug in DEFSTRUCT: once again, naming structure slots with
764 keywords or constants is permissible.
765 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
766 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
767 classes. (thanks to Antonio Martinez)
768 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
769 outputs FRESH-LINE or TERPRI, and no longer converts its stream
770 argument to a pretty-print stream. Instead, it leaves any such
771 operations to DESCRIBE-OBJECT methods.
772 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
774 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
775 lambda lists. (thanks to David Lichteblau)
776 * bug fix: Defining a generic function with a :METHOD-CLASS being a
777 subclass of STANDARD-METHOD no longer causes stack exhaustion.
778 (thanks to Gerd Moellmann)
779 * fixed bug 246: increased compilation speed of long
780 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
782 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
784 * DEFCONSTANT now throws a condition of type
785 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
786 constant to a non-EQL value; CONTINUE and ABORT restarts
787 respectively change and preserve the value.
788 * fixed bug 63: The code walker, part of the implementation of CLOS,
789 is now better at handling symbol macros.
790 * bug fix: There is no longer an internal implementation type named
791 CL:LENGTH. (reported by Raymond Toy)
792 * bug fix: In macro-like defining macros/special operators the
793 implicit block does not enclose the lambda list.
794 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
795 enclosed in parentheses) are not suitable as type specifiers, and
796 their use properly signals an error now.
797 * bug fix: An argument count mismatch for a type specifier in code
798 being compiled no longer causes an unhandled error at compile
799 time, but signals a compile-time warning.
800 * fixed simple vector readable printing
801 * bug fix: DESCRIBE takes more care over whether the class
802 precedence list slot of a class is bound before accessing it.
803 (reported by Markus Krummenacker)
804 * bug fix: FORMATTER can successfully compile pretty-printer format
805 strings which use variants of the ~* directive inside.
806 * bug fix: SEARCH now applies its TEST predicate to the elements of
807 the arguments in the correct order. (thanks to Wolfhard Buss)
808 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
809 in type inference. (thanks to Robert E. Brown)
810 * bug fix: Reading in symbols with an explicit package name of ""
811 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
812 * fixed some bugs revealed by Paul Dietz' test suite:
813 ** NIL is now allowed as a structure slot name.
814 ** Arbitrary numbers, not just REALs, are allowed in certain
815 circumstances in LOOP for-as-arithmetic clauses.
816 ** Multiple class redefinitions before slot access no longer
818 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
819 association between the name and a class.
820 ** Generic functions with non-standard method-combination and over
821 five methods all of which return constants no longer return NIL
822 after the first few invocations. (thanks to Gerd Moellmann)
823 ** CALL-NEXT-METHOD with no arguments now passes the original
824 values of the arguments, even in the presence of assignment.
825 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
827 ** DISASSEMBLE works with closures and funcallable instances.
828 ** ADD-METHOD now returns the generic function, not the new method.
829 ** FIND-METHOD signals an error if the lengths of the specializers
830 is incompatible with the generic function, even if the ERRORP
832 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
833 which its argument is a member.
834 ** DEFCLASS only redefines the class named by its class-name
835 argument if that name is the proper name of the class;
836 otherwise, it creates a new class.
837 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
838 of the UNBOUND-SLOT condition to the name of the slot.
839 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
840 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
841 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
842 SLOT-MAKUNBOUND in the specified fashion.
844 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
845 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
846 cross-compilation host. As a consequence, we can now bootstrap our
847 way up to SBCL starting with a bare gcc toolchain and human-readable
848 source code (first the source to CLISP, then the source to SBCL).
849 * A contributed module containing a partial implementation of the
850 simple-streams interface has been included. (thanks to Rudi
852 * A contributed module implementing the RFC1321 Message Digest
853 Algorithm, known as MD5, has been included.
854 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
855 no longer has any effect, as the code controlled by this feature
856 has been deleted. (As far as we know, no-one has ever built using
857 this feature, and its semantics were confused in any case).
858 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
859 work on conditions (as required by the ANSI specification),
860 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
861 expected behaviour on conditions. Users should note, however,
862 that such behaviour is not required by the ANSI specification,
863 and so use of this behaviour may render their code unportable.
864 * fixed some bugs revealed by Paul Dietz' test suite:
865 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
867 ** &ENVIRONMENT parameter in macro lambda list is bound first.
868 ** SXHASH on condition objects no longer returns NIL.
869 ** :ALLOCATION :CLASS slots are better treated; their values are
870 updated on class redefinition, and initforms inherited from
871 superclasses are applied.
872 ** REMOVE-METHOD returns its generic function argument even when
873 no method was removed.
874 ** SHARED-INITIALIZE now initializes the values of the requested
875 slots, including those with :ALLOCATION :CLASS.
876 ** ALLOCATE-INSTANCE now works on structure classes defined via
877 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
879 ** SLOT-EXISTS-P now works on conditions, as well as structures
881 ** MAKE-LOAD-FORM now has the required methods on
882 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
883 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
884 keyword, and now implements the SLOT-NAMES argument.
885 ** methods with &OPTIONAL arguments no longer allow too many
886 arguments to be passed in the call without error.
887 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
888 option is consistent with the required arguments of the generic
889 function lambda list.
890 * bug fix: REQUIRE accepts a string designator. (Thanks to
892 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
893 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
894 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
895 * bug fix: The compiler now checks for duplicated variables in macro
897 * bug fix: SETQ on globals returns the correct value.
898 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
899 not a COMPILER-ERROR (followed by some other strange error on
900 choosing the CONTINUE restart).
901 * bug fix: make.sh and friends are now more consistent in the way that
902 they look for GNU "make".
904 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
905 * experimental native threads support (on x86 Linux >=2.4 only).
906 This is not compiled in by default: you need to add :SB-THREAD to
907 the target features. See the "Beyond ANSI" chapter of the manual
909 * fix for longstanding nonANSIism: The old distinction between
910 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
911 The return value from CL:FIND-CLASS is now a CLOS class, and
912 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
913 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
914 * An interface to the MetaObject Protocol, as described in Kiczales,
915 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
916 MIT Press, 1991, is available from the SB-MOP package.
917 * incompatible change: the SB-PCL package should now be considered
918 a private implementation detail, and no longer a semi-private MOP
920 * minor incompatible change: due to rearrangement for threads, the
921 control stack and binding stack are now allocated at arbitrary
922 addresses instead of being hardcoded per-port. Users affected by
923 this probably have to be doing advanced things with shared
924 libraries, and will know who they are.
925 * minor incompatible change: Previously, all --eval forms used were
926 processed with READ before any of them were processed with EVAL.
927 Now each --eval form is processed with both READ and EVAL before
928 the next --eval form is processed. (Thus package operations like
929 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
930 work as the user might reasonably expect.)
931 * minor incompatible change: *STANDARD-INPUT* is now only an
932 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
934 * minor incompatible change: Y-OR-N-P is now character-oriented,
935 not line oriented. Also, YES-OR-NO-P now works without errors.
936 (thanks to Antonio Martinez)
937 * sb-aclrepl module improvements: an integrated inspector, added
938 repl features, and a bug fix to :trace command.
939 * Known functions, which cannot be open coded by the backend, are
940 considered to be able to check types of their arguments. (fixing
941 a bug report by Nathan J. Froyd)
942 * fixed a bug in computing method discriminating functions: It is
943 now possible to define methods specialized on classes which have
944 forward-referenced superclasses. (thanks to Gerd Moellmann)
945 * fixed evaluation order in optional entries (reported by Gilbert
947 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
948 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
949 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
950 required-by-AMOP NAME argument, as well as CLASS and
951 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
952 * fixed bug 20: DEFMETHOD can define methods using names that are
953 not the proper names of classes to designate class specializers.
954 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
955 Unix file descriptors, instead of blowing up. (thanks to Antonio
957 * Garbage collection refactoring: user-visible change is that a
958 call to the GC function during WITHOUT-GCING will not do garbage
959 collection until the end of the WITHOUT-GCING. If you were doing
960 this you were probably losing anyway.
961 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
962 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
963 TYPEP the latter but not the former.
964 * The compiler issues a full WARNING for calls to undefined functions
965 with names from the CL package.
966 * MAP-INTO for a vector destination is open coded. (reported by
967 Brian Downing on c.l.l)
968 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
969 documentation string.
970 * fixed some bugs revealed by Paul Dietz' test suite:
971 ** COPY-ALIST now signals an error if its argument is a dotted
973 ** Condition slots are now accessed more correctly in the presence
974 of multiple initargs for a given slot.
975 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
976 exclude restarts of the same name associated with a different
978 ** DEFCLASS of forward-referenced classes with another
979 forward-referenced class in the superclasses list no longer
981 ** Condition slots are now initialized once each, not multiple
982 times. (thanks to Gerd Moellmann)
983 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
985 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
986 ** NSET-EXCLUSIVE-OR does not return extra elements when its
987 arguments contain duplicated elements.
988 ** RESTART-CASE understands local macros.
989 ** RESTART-CASE associates exactly its own restarts with a condition.
990 ** ENDP in safe mode checks its argument to be of type LIST.
991 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
992 in question is unbound.
993 ** Optimized MAKE-INSTANCE functions no longer cause internal
994 assertion failures in the presence of duplicate initargs.
995 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
996 ** Side-effectful :DEFAULT-INITARGS have their side-effects
997 propagated even in the ctor optimized implementation of
999 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1001 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1002 * a better implementation of SXHASH on (simple) bit vectors,
1003 measured both in execution speed and in distribution of results
1004 over the positive fixnums, has been installed. Likewise, a better
1005 implementation of EQUAL for simple bit vectors is now available.
1006 * fixed CEILING optimization for a divisor of form 2^k.
1007 * fixed bug 240 (emitting extra style warnings "using the lexical
1008 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1009 by Antonio Martinez)
1010 * fixed SXHASH, giving different results for NIL depending on type
1011 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1012 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1013 variable. (found by Rolf Wester)
1014 * SBCL does not ignore type declarations for special
1015 variables. (reported by rif on c.l.l 2003-03-05)
1016 * some bug fixes in contrib/sb-aclrepl/
1017 * fixed some bugs revealed by Paul Dietz' test suite:
1018 ** a bug in the CONS type specifier, whereby the CAR and CDR
1019 types got intertwined, has been fixed;
1020 ** the type system is now able to reason about the interaction
1021 between INTEGER and RATIO types more completely;
1022 ** APPEND, [N]REVERSE and NRECONC check that those their
1023 arguments, which must be proper lists, are really so;
1024 ** An array specialized to be unable to hold elements has been
1025 implemented, as required -- yes, really -- by ANSI;
1026 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1027 on malformed property lists;
1029 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1030 * incompatible packaging change: in line with Unix convention,
1031 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1032 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1033 * REQUIRE and PROVIDE are now optionally capable of doing something
1034 useful. See the documentation string for REQUIRE.
1035 * infrastructure for a managed SBCL contrib system: contributed
1036 modules in this release include:
1037 ** the ASDF system definition facility;
1038 ** an interface to the BSD Sockets API;
1039 ** an ACL-like convenience interface to the repl;
1040 (thanks to Kevin Rosenberg)
1041 ** an implementation of ROTATE-BYTE, with efficient implementation
1043 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1044 gives a closer approximation to the right answer than previously.
1045 (thanks to Raymond Toy)
1046 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1047 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1048 optional environment arguments, as required by ANSI.
1049 * fixed bugs in other functions taking environment objects, allowing
1050 calls with an explicit NIL environment argument to be compiled
1052 * fixed bug 228: primary return values from
1053 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1054 COMPILE or FUNCTION.
1055 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1056 :TYPE will no longer signal errors on innocuous objects.
1057 * fixed bug 231b: SETQ is better at respecting type declarations in
1058 the lexical environment.
1059 * fixed a bug in DEFCLASS: classes named by symbols with no or
1060 unprintable packages can now be defined.
1061 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1062 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1063 * fixed compiler failure related to checking types of functions.
1064 (reported by Robert E. Brown)
1065 * the compiler is now much more consistent in its error-checking
1066 treatment of bounding index arguments to sequence functions: in
1067 (SAFETY 3) code, errors will be signalled in almost all cases if
1068 invalid sequence bounding indices are passed to functions defined
1069 by ANSI to operate on sequences.
1070 * fixed a bug in the build procedure: documentation of SBCL-specific
1071 packages is now preserved and available in the final Lisp image.
1072 * lifted FDEFINITION lookup out of loops in the implementation of
1073 many list operations. (thanks to Robert E. Brown)
1074 * fixed a bug in the reader: the #n# reader macro now works for
1075 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1076 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1077 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1078 E. Brown and Tony Martinez respectively)
1079 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1080 SLOT-BOUNDP: in particular, it is now able to optimize them much
1081 better, and is now not vulnerable to having packages renamed.
1082 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1083 always called when a slot is not present in an instance. (thanks
1085 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1086 longer signal an error on unreading a character following EOF on
1087 the previous constituent stream. (thanks to Tony Martinez)
1088 * fixed some bugs revealed by Paul Dietz' test suite:
1089 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1090 not just nonnegative fixnums;
1091 ** the logical bit-array operators such as BIT-AND now accept an
1092 explicit NIL for their "opt-arg" argument (to indicate a
1093 freshly-consed result bit-array);
1094 ** ELT now signals an error on an invalid sequence index in safe
1096 ** the type system is now cleverer about negations of numeric
1097 types, and consequently understands the BIGNUM and RATIO types
1099 ** the type system is now cleverer about the interaction between
1100 INTEGER and RATIO types: while bugs still remain, many more
1101 cases are accurately computed;
1102 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1103 if it is in the last clause;
1104 ** CONSTANTLY now correctly returns a side-effect-free function in
1106 ** DECLARE is no longer treated as a special-operator; in
1107 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1108 * incremented fasl file version number due to the change in the
1109 DEFSTRUCT-SLOT-DESCRIPTION structure.
1111 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1112 * minor incompatible change: code processed by the "interpreter" or
1113 EVAL now has a compilation optimization policy of (DEBUG 2)
1114 (changed from (DEBUG 1)) to improve debuggability of interactive
1115 development, and to allow the use of the debug RETURN command in
1117 * an experimental implementation of the RETURN command for the
1118 debugger has been included. (thanks to Frederik Kuivinen)
1119 * fixed bug 62: constraints were not propagated into a loop.
1120 * fixed bug in embedded calls of SORT (reported and investigated by
1122 * fixed some bugs revealed by Paul Dietz' test suite:
1123 ** printing and reading of arrays with some dimensions having
1124 length 0 (thanks to Gerd Moellmann);
1125 ** BOA constructor with &AUX argument without a default value does
1126 not cause a type error;
1127 ** CONSTANTP now returns true for all self-evaluating objects.
1129 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1130 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1131 accessors that are related by inheritance, as specified in the
1132 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1133 Valtteri Vuorikoski)
1134 * The compiler is now able to inline functions that were defined in
1135 a complex lexical environment (e.g. inside a MACROLET).
1136 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1138 * Support for the upcoming FreeBSD-5.0 release has been included.
1139 (thanks to Dag-Erling Smorgrav)
1140 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1141 effect when it is not in a toplevel context.
1142 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1143 stand a better chance of being correct. (thanks to Gerd
1145 * fixed bug in COERCE, which now signals an error on coercing a
1146 rational to a bounded real type which excludes the expected
1148 * The compiler is now able to derive types more accurately from the
1149 COERCE and COMPILE functions.
1150 * fixed bug 223: functional binding is considered to be constant
1151 only for symbols in the CL package.
1152 * fixed bug 231: SETQ did not check the type of a variable being set
1153 (reported by Robert E. Brown)
1154 * A new optimization for MAKE-INSTANCE has been included, fixing
1155 various bugs (including relating to :ALLOCATION :CLASS slots and
1156 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1158 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1159 ** As required by ANSI, LOOP now disallows anonymous collection
1160 clauses such as COLLECT I in conjunction with aggregate boolean
1161 clauses such as THEREIS (= I 1);
1162 ** LOOP now signals an error when any variable is reused in the
1163 same loop (including the potentially useful construct analogous
1164 to WITH A = 1 WITH A = (1+ A);
1165 ** IT is only a special loop symbol within the first clause of a
1166 conditional loop clause;
1167 ** LOOP with a typed iteration variable over a hashtable now
1168 signals a type error iff it should.
1169 * fixed some other bugs revealed by Paul Dietz' test suite:
1170 ** FILE-STREAM now names the class previously known as FD-STREAM;
1171 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1172 argument) no longer signals an error;
1173 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1174 of the slot symbol, rather than using the current package
1175 ((:CONC-NAME "") continues to intern the slot's name in the
1177 * incremented fasl file version number, because of the incompatible
1178 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1179 of the new implementation of DEFINE-COMPILER-MACRO.
1181 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1182 * Support for building SBCL for MIPS platforms running in
1183 little-endian mode has now been checked in, and basic
1184 functionality on said platforms verified.
1185 * minor incompatible change: PCL now records the pathname of a file
1186 in which methods and the like are defined, rather than its
1188 * minor incompatible change: TRUENAME now considers the truename of
1189 a file naming a directory to be the pathname with :DIRECTORY
1190 component indicating that directory.
1191 * minor incompatible change: a NAMED clause in the extended form of
1192 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1193 reason for the previous behaviour is unclear.
1194 * more systematization and improvement of CLOS and MOP conformance
1195 in PCL (thanks to Gerd Moellman and Pierre Mai):
1196 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1198 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1199 primary methods with no specializers;
1200 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1202 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1203 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1204 FORWARD-REFERENCED-CLASSes; error reporting on
1205 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1207 ** SXHASH on CLOS instances now uses a slot internal to the
1208 instance to return different numbers on distinct instances,
1209 while preserving the same return value through invocations of
1211 ** DEFMETHOD signals errors when methods with longer incongruent
1212 lambda lists are added to generic functions;
1213 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1214 CLASS, as specified in AMOP;
1215 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1216 on the order returned by the primary method for classes of
1217 class STANDARD-CLASS;
1218 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1219 * fixed some bugs shown by Paul Dietz' test suite:
1220 ** DOLIST puts its body in TAGBODY;
1221 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1223 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1224 value producing form;
1225 ** if more variables are given to PROGV than values, extra
1226 variables are bound and made to have no value;
1227 ** NSUBSTITUTE on list arguments gets the right answer with
1229 ** ELT signals an error of type TYPE-ERROR when the index argument
1230 is not a valid sequence index;
1231 ** LOOP signals (at macroexpansion time) an error of type
1232 PROGRAM-ERROR when duplicate variable names are found;
1233 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1234 ** FUNCALL of special-operators now cause an error of type
1236 ** PSETQ now works as required in the presence of side-effecting
1237 symbol-macro places;
1238 ** NCONC accepts any object as its last argument;
1239 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1241 ** loop-for-as-package does not require a package to be explicitely
1243 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1245 * fixed bug 166: compiler preserves "there is a way to go"
1246 invariant when deleting code.
1247 * fixed bug 172: macro lambda lists with required arguments after
1248 &REST arguments now cause an error to be signalled. (thanks to
1250 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1251 bug. (thanks to Matthew Danish)
1252 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1254 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1255 arguments are lexically rebound. (thanks to Gerd Moellmann and
1257 * fixed bug 194: error messages are now more informative when there
1258 is no primary method applicable in a call to a generic
1259 function. (thanks to Gerd Moellmann)
1260 * fixed bug in command line argument checking (thanks to Julian
1262 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1263 (thanks to Matthew Danish)
1264 * incremented fasl file version number, because of the
1265 SXHASH-related changes in the layout of CLOS data structures
1267 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1268 * minor incompatible change: The runtime (the Unix executable named
1269 "sbcl") is now much pickier about the .core files it will load.
1270 Essentially it now requires .core files to descend from the same
1271 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1272 as the runtime does. (The intent is to prevent the crashes which
1273 can occur, and which can even be reported as mysterious failures,
1274 when people patch the sources or change the build parameters
1275 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1276 sbcl and .core files.)
1277 * fixed bug: VALUES-LIST is no longer optimized away.
1278 * fixed bug 142: The FFI conversion of C string values to Lisp
1279 string values no longer conses excessively. (thanks to Nathan
1280 Froyd porting Raymond Toy's fix to CMU CL)
1281 * began to systematize and improve MOP conformance in PCL (thanks to
1282 Nathan Froyd, Gerd Moellman and Pierre Mai):
1283 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1285 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1286 ** FINALIZE-INHERITANCE is now called on class finalization;
1287 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1288 argument precedence order.
1289 * fixed bug 202: The compiler no longer fails on functions whose
1290 derived types contradict their declared type.
1291 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1292 so it can be non-toplevel.
1293 * The fasl file version number has changed (because of the new
1294 implementation of DEFMACRO).
1295 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1296 safe code, that any length requirement by their type-specifier
1297 argument is valid. The exceptions to this are described in bug
1299 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1300 functions now check the types of their inputs as required by ANSI.
1301 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1302 introduced symbol is DECLAREd to be SPECIAL.
1303 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1304 * fixed bug: :COUNT argument to sequence functions may be negative
1305 * fixed bug: body of DO-SYMBOLS may contain declarations
1306 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1307 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1308 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1309 symbol macro only once
1310 * fixed printing of call frame when argument list is unavailable
1311 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1312 * compiler no longer signals WARNING on unknown keyword
1315 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1316 * A beta-quality port to the mips architecture running Linux,
1317 based on the old CMUCL backend, has been made. It has been tested
1318 on a big-endian kernel, and works sufficiently well to be able to
1319 rebuild itself; it has not been tested in little-endian mode.
1320 * fixed an inconsistency between gencgc.c and purify.c which made
1321 dumping/loading .core files unreliable
1322 * fixed bug 120a: The compiler now deals correctly with IFs where
1323 the consequent is the same as the alternative, instead of
1324 misderiving the return type. (thanks to Alexey Dejneka)
1325 * fixed bug 113: Logical pathnames are now dumpable (the logical
1326 host is resolved at load-time, throwing an error if it is not
1328 * fixed bug 174: FORMAT's error message is slightly clearer when a
1329 non-printing character is used in a format directive.
1330 * fixed several bugs in compiler checking of type declarations, i.e.
1331 violations of the Python "declarations are assertions" principle
1332 (thanks to Alexey Dejneka)
1333 * fixed several bugs in PCL's error checking (thanks to Gerd
1335 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1337 * fixed bug in compilation of functions as first class values
1338 (thanks to Antonio Martinez-Shotton)
1339 * The compiler's handling TYPE-ERRORs which it can prove will
1340 inevitably happen at runtime has been cleaned up and corrected
1341 in several ways. (thanks to Alexey Dejneka)
1342 * improved argument type checking for various basic arithmetic
1343 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1344 lavished on them in the past that they can be compiled in many
1345 ways in different special cases
1346 * fixed bug 181: compiler checks validity of user supplied type
1348 * cleaned up code flushing in optimization: Function calls which
1349 should signal errors for safety purposes (e.g. which ANSI says
1350 should signal errors when their arguments are of incorrect type)
1351 are no longer optimized away.
1352 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1353 * incremented fasl file version number, because changes in the
1354 implementation of sequence functions like COERCE caused
1355 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1356 in old inline expansions) to become undefined. (Actually these
1357 changes were later undone, so we might very well be binary
1358 compatible with 0.7.7 after all, but leaving the version number
1359 incremented seemed like the simplest and most conservative
1362 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1363 * An alpha-quality port to the parisc architecture running Linux,
1364 based on the old CMUCL backend, has been made. This, even more so
1365 than the other backends, should be considered still a work in
1366 progress; known problems include that the Linux kernel in 64-bit
1367 mode does not propagate the correct sigcontext structure to
1368 userspace, and consequently SBCL on a parisc64 kernel will not
1370 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1371 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1372 suppressed.) Also now that the compiler is looking at declarations
1373 in the environment, it checks optimization declarations as well,
1374 and suppresses inlining when (> DEBUG SPEED).
1375 * More fixes have been made to treatment of floating point exception
1376 treatment and other Unix signals. In particular, floating point
1377 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1378 * The detection and handling of control stack exhaustion (infinite
1379 or very deeply nested recursion) has changed. Stack exhaustion
1380 detection is now done by write-protecting pages at the OS level
1381 and applies at all optimization settings; when found, a
1382 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1383 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1384 be caught using IGNORE-ERRORS.
1385 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1386 correctly when transforming recursive LABELS functions to LETs.
1387 (thanks to Alexey Dejneka)
1388 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1389 that are names of constants or global variables.
1390 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1391 alien routines with docstrings.
1392 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1393 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1395 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1396 (thanks to Eric Marsden)
1397 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1398 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1399 * Bug fix: The SPARC backend can now compile functions involving
1400 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1401 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1402 to a functions expecting a PATHNAME-DESIGNATOR.
1403 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1404 lambda lists. (thanks to Alexey Dejneka)
1405 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1406 functions so that EXPORTing the name of the class doesn't cause
1407 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1408 to Antonio Martinez for reporting this)
1409 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1410 constant-folded without failing an assertion. (thanks to Einar
1411 Floystad Dorum for reporting this)
1412 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1413 types are now dealt with more correctly by the compiler.
1414 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1415 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1416 behaviour analogous to RENAME-FILE. This puts its behaviour more
1417 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1419 * The fasl file version number has changed again. (because of the
1420 bug fix involving the names of PCL MAKE-INSTANCE functions)
1422 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1423 * bug fix: Floating point exceptions are treated much more
1424 consistently on the x86/Linux and PPC/Linux platforms.
1425 * Array initialization with :INITIAL-ELEMENT is now much faster for
1426 cases when the compiler cannot open code the array creation, but
1427 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1428 array accesses have also seen a speed increase.
1429 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1430 specified. (This was at the root of some bad interactions between
1431 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1433 * bug fix: Internal error arguments for undefined functions are now
1434 computed correctly on the PPC/Linux platform.
1435 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1436 Raymond Toy's patch for CMU CL)
1437 * Support for the Solaris 9 operating environment has been included
1438 (thanks to Daniel Merritt)
1439 * A very ugly but hopefully complete draft of the missing FFI chapter
1440 of the manual has been created by reformatting the corresponding
1441 CMU CL manual chapter into (currently very ugly and incoherent)
1442 DocBook and bringing it up to date for SBCL behavior. Thus, the
1443 manual is now essentially complete, at least by my extreme
1444 once-and-only-once standards, whereby it's acceptable to refer to
1445 the doc strings of SB-EXT functions as the primary documentation.
1446 * The fasl file version number has changed again, due to cleanup of
1447 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1448 is no longer a static symbol.)
1450 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1451 * SBCL now builds with OpenMCL (version 0.12) as the
1452 cross-compilation host; also, more progress has been made toward
1453 bootstrapping under CLISP.
1454 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1456 * bug 158 fixed: The compiler can now deal with integer loop
1457 increments different from 1; fixing this turned out also to fix
1459 * bug 169 fixed: no more bogus warnings about using lexical bindings
1460 despite the presence of perfectly good SPECIAL declarations (thanks
1461 to David Lichteblau)
1462 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1463 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1464 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1466 * bug 180 fixed: Method combination specifications no longer ignore
1467 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1468 * bug fix: Structure type predicate functions now check their argument
1469 count as they should.
1470 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1471 correctly. (thanks to Pierre Mai)
1472 * minor incompatible change: The --noprogrammer option is deprecated
1473 in favor of the new --disable-debugger option, which is very similar.
1474 (The major difference is that it takes effect at a slightly different
1475 time at startup, causing handling of errors in --sysinit and
1476 --userinit files will be affected differently.) The
1477 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1478 been added to allow this functionality to be controlled from ordinary
1479 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1480 maintainers, who might want to run non-interactive scripts to
1481 build SBCL cores which will later be used interactively.)
1482 * minor incompatible change: The LOAD function no longer, when given
1483 a wild pathname to load, loads all files matching that pathname.
1484 Instead, an error of type FILE-ERROR is signalled.
1486 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1487 * bug 147 fixed: The compiler preserves its block link/count
1488 invariants more correctly now so that it doesn't crash. (thanks
1490 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1492 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1493 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1494 no-value-for-this-character as NIL (as specified by ANSI).
1495 * HOST-NAMESTRING on physical pathnames now returns a string that is
1496 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1497 (thanks to Christophe Rhodes)
1498 * The Alpha port handles icache flushing more correctly. (thanks to
1500 * More progress has been made toward bootstrapping under CLISP. (thanks
1501 to Christophe Rhodes)
1502 * The fasl file format has changed again, because dynamic loading
1503 on OpenBSD (which has non-ELF object files) motivated some cleanups
1504 in the way that foreign symbols are transformed and passed around.
1505 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1506 is no longer treated as whitespace by the reader, but instead as
1507 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
1508 |A
\x7fB|, instead of A as it used to.
1510 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1511 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1512 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1513 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1514 this as of 0.7.1.45, but was left out of the previous news section
1515 (thanks to Dan Barlow)
1516 * SBCL now runs on the Solaris operating system on SPARC architectures
1517 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1518 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1519 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1520 * SPARC backend cleanups, allowing builds of cores optimized for V8
1521 and V9 SPARCS, and also emission of code targeted to a particular
1522 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1524 * SBCL is closer to bootstrapping under CLISP, thanks to various
1525 fixes by Christophe Rhodes.
1526 * The fasl file format has changed again, to allow the compiler's
1527 INFO database to support symbol macros.
1528 * The user manual (in doc/) is formatted into HTML more nicely.
1529 (thanks to coreythomas)
1530 * The system is smarter about SUBTYPEP relationships, especially
1531 those involving NOT types (including types such as ATOM which are
1532 represented internally using NOT types). Thus SUBTYPEP is less
1533 likely to return (VALUES NIL NIL) in general, and in particular
1534 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1536 * The fasl file format has changed again, because the internal
1537 representation of types now includes a new slot to support the new
1538 SUBTYPEP-of-NOT-types logic.
1539 * (not a change in the main branch of SBCL, but a related prototype
1540 which can hopefully be merged into the main branch of SBCL in the
1541 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1542 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1543 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1544 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1545 dispatch macro character combinations. (thanks to Alexey Dejneka)
1546 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1547 that superficially look like logical namestrings correctly.
1548 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1550 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1551 default supplied-p) for &optional and &key arguments. (thanks to
1554 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1555 * incompatible change: The compiler is now less aggressive about
1556 tail call optimization, doing it only when (> SPACE DEBUG) or
1557 (> SPEED DEBUG). (This is an incompatible change because there are
1558 programs which relied on the old CMU-CL-style behavior to optimize
1559 away their unbounded recursion which will now die of stack overflow.)
1560 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1561 for non-GENCGC systems has been increased to 20M (since that
1562 seems much closer to the likely performance optimum for modern
1563 systems than the old 4M value was)
1564 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1565 and *DEBUG-PRINT-LEVEL*
1566 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1567 of CMU CL's support for SPARC, and various endianness and other
1568 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1569 * new syntactic sugar for the Unix command line: --load foo.bar is now
1570 an alternate notation for --eval '(load "foo.bar")'.
1572 ** The system now detects stack overflow and handles it gracefully,
1573 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1574 optimization settings. (This is a good thing in general, and
1575 its introduction in this version should be particularly timely
1576 for anyone whose code fails because of suppression of tail
1578 ** The system now hunts for the C variable "environ" in a more
1579 devious way, to avoid segfaults when the C library version differs
1580 between compile time and run time. (thanks to Christophe Rhodes)
1581 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1582 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1583 ** The compiler no longer issues bogus style warnings for undefined
1584 classes in the same source file as the DEFCLASSes which defined
1585 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1587 ** fixes in CONDITION class precedence list for undefined function
1588 errors (thanks to Alexei Dejneka)
1589 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1590 correctly. (thanks to Dan Barlow)
1591 ** portability fixes aiming at bootstrapping under CLISP (thanks
1592 to Dave McDonald and Christophe Rhodes)
1593 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1594 ** fixes in type translation and and type inference (thanks to
1596 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1597 ** various fixes in the new ports (thanks to Dan Barlow)
1598 * several changes related to debugging:
1599 ** suppression of tail recursion, as noted above
1600 ** stack overflow detection, as noted above
1601 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1602 is now the default. (For some time encapsulation has been more
1603 reliable than the breakpoint-based :ENCAPSULATE NIL
1604 implementation, at least on X86 systems; and I just noticed that
1605 encapsulation also seems closer to the spirit of the ANSI
1608 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1610 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1611 up properly again. (There was a packaging bug in 0.7.0 which
1612 left their definitions in SB-SYS::LOAD-FOREIGN and
1613 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1614 vital for most things which interface to C-level interfaces,
1615 like extensions working with sockets or databases or
1616 Perl-compatible regexes or whatever, and the need to fix
1617 this bug is the main reason that 0.7.1 was released so
1619 ** DEFGENERIC is now choosier about the methods it redefines, so that
1620 reLOADing a previously-LOADed file containing DEFGENERICs does
1621 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1622 cycle now works as it should. (thanks to Alexey Dejneka)
1623 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1624 fixed by Christophe Rhodes. (He actually submitted this patch
1625 months ago, and I delayed until after 0.7.0.)
1626 ** Bug 111 (internal compiler confusion about runtime checks on
1627 FUNCTION types) was fixed by Alexey Dejneka.
1628 * Some internal cleanups (getting rid of variables which aren't
1629 needed now that the byte interpreter is gone) caused the fasl
1630 file format number to change again.
1632 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1633 * major incompatible change: The default fasl file extension, i.e. the
1634 default extension for files produced by COMPILE-FILE, has changed
1635 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1637 ** There are many changes in the implementation of the compiler.
1638 SBCL is now essentially a compiler-only implementation of ANSI
1639 Common Lisp. EVAL still "interprets" a few special cases, but
1640 almost all the interesting cases are handled by creating
1641 a LAMBDA expression, calling COMPILE on it, then calling
1642 FUNCALL on the result.
1643 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1644 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1645 Since the code is newer, there might still be some new bugs
1646 (though not as many as before Martin Atzmueller's fixes:-). But
1647 the new code is substantially simpler and clearer, and hopefully
1648 any remaining bugs will be simpler, less fundamental, and more
1649 fixable then the bugs in the old code.
1650 ** The revised compiler is still a little unsteady on its feet.
1652 *** The debugging information it produces (particularly the names
1653 of FUNCTION objects) is sometimes much less useful than what
1654 the old compiler produced.
1655 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1656 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1657 within a MACROLET) has been temporarily weakened.
1658 ** There are new compiler optimizations for various functions:
1659 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1660 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1661 *** the math functions TRUNCATE, FLOOR, and CEILING
1662 *** the function-of-all-trades COERCE
1663 Mostly these should be transparent, but there's one
1664 potentially-annoying problem (bug 117): when the compiler
1665 inline-expands a function and does type analysis on the result,
1666 it can create control paths which have type mismatches, and
1667 when it can't prove that those control paths aren't taken,
1668 it will issue WARNINGs about the type mismatches. This is
1669 a particular problem in practice for the new sequence functions.
1670 It's not clear how this should be fixed, and for now, a
1671 workaround is given in the entry for 117 in the BUGS file.
1672 ** (Because of the interaction between the two previous items --
1673 occasional inlining problems and new inline expansions -- some
1674 of the new sequence function optimizations won't really kick in
1675 completely until debugging information, and then inlining, are
1676 straightened out in some future version.)
1677 * minor incompatible changes:
1678 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1679 DIRECTORY no longer implicitly promotes NIL slots of its
1680 pathname argument to :WILD. In particular, when you ask for the
1681 contents of a directory (which you used to be able to do without
1682 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1683 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1684 ** changes in behavior that ANSI explicitly defines to be
1685 implementation dependent:
1686 *** The new compiler-only implementation still conforms with ANSI,
1687 but acts a little different than before. Besides the obvious
1688 changes in performance tradeoffs (that the cost per form passed
1689 to EVAL has gone up, and the cost per form executed by EVAL
1690 has gone down), the behavior of the system changes a little
1691 because there are no longer any interpreted function objects.
1692 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1693 e.g. doing COMPILE on the output of interactive DEFUN is
1695 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1697 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1698 has changed from (:CL) to NIL.
1699 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1700 e.g. #\Newline and #\Tab, don't have names specified in the
1701 ANSI Common Lisp standard, is now based on their ASCII symbolic
1702 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1703 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1704 are no longer used for output.
1705 ** changes in internal implementation constants:
1706 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1707 4 million. (If your application spends a lot of time GCing and
1708 you have a lot of RAM, you might want to experiment with
1709 increasing it even more.)
1710 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1711 However, almost all old code should still continue to work without
1712 immediate update, as SB-C-CALL is now a (deprecated) nickname
1714 ** Old operator names in the style DEF-FOO are now deprecated in
1715 favor of new corresponding names DEFINE-FOO, for consistency with
1716 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1717 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1718 internal symbols, but a few supported extensions like
1719 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1720 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1721 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1722 etc. as you get deeper into recursive calls to the debugger
1723 command loop, instead of the old "5]", "5]]", "5]]]"
1724 sequence. (I was motivated to do this when squabbles between
1725 ILISP and SBCL left me very deeply nested in the debugger. In the
1726 short term, this change will probably provoke more ILISP/SBCL
1727 squabbles, but hopefully it will be an improvement in the long run.)
1728 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1729 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1730 function, since declaiming return types involving aliens is
1731 (1) annoyingly messy to do by hand and (2) vital to efficient
1732 compilation of code which calls such functions.
1733 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1734 longer reexported by the SB-EXT package. They're solely useful
1735 for alien code, so it seems more logical that you should get
1736 them from the SB-ALIEN package, not in SB-EXT.
1737 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1738 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1739 features. Instead, the code that they used to control is always
1740 built into the system.
1741 * many other bug fixes
1742 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1743 advantage of the new EVAL-WHEN stuff and to clean them up in
1744 general, and they are now more ANSI-compliant in a number of
1745 ways. Martin Atzmueller is responsible for a lot of this.
1746 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1748 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1749 *** correct ERROR type for various file operations
1750 *** some fixes for Lisp streams
1751 *** DEFMETHOD syntax checking
1752 *** changing old weird representation of debug information as
1753 strings (which, among their other deficiencies, don't transform
1754 correctly when you rename packages, and don't change their
1755 print representation when you change things like *PACKAGE*
1756 and *PRINT-LENGTH*) to symbols and lists of symbols
1757 He also made several improvements and fixed several bugs in DESCRIBE.
1758 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1760 *** misbehavior of WRITE-STRING/WRITE-LINE
1761 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1762 and several other LOOP problems as well
1763 *** DIRECTORY when similar filenames are present
1764 *** DEFGENERIC with :METHOD options
1765 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1766 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1767 *** argument ordering in FIND with :TEST option
1768 *** mishandled package designator argument in APROPOS-LIST
1769 *** various problems in the backquote readmacro
1770 *** a bug in APROPOS
1771 *** probably some others that I'm not describing very well here,
1772 since the CVS log documents them by reference to sbcl-devel
1773 messages, and the SourceForge archives aren't working well.:-(
1774 ** Dan Barlow improved the Alpha port (and is making progress on the
1775 PPC port, for those of you who think different).
1776 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
1777 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
1778 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
1779 continues to work on the SPARC port (for those of you in a position
1780 to look down upon our little PC-compatible boxes from a great height).
1781 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
1782 and uses that copy, rather than the current dynamic value, when
1783 it's trying to decide whether to truncate output. Thus e.g.
1784 (let ((*print-lines* 50))
1785 (pprint-logical-block (stream nil)
1787 (let ((*print-lines* 8))
1788 (print (aref possiblybigthings i) stream)))))
1789 should now truncate the logical block only at 50 lines, instead of
1790 often truncating it at 8 lines, as it did before.
1791 * The doc/cmucl/ directory, containing old CMU CL documentation
1792 from the time of the fork, is no longer part of the base system.
1793 SourceForge has shut down its anonymous FTP service, and with it
1794 my original plan for distributing the old CMU CL documentation
1795 there. For now, if you need these files you can download an old
1796 SBCL source release and extract them from it.
1797 * The fasl file version number changed again, for dozens of reasons,
1798 some of which are apparent above.
1800 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
1801 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
1802 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
1803 patch, so that DISASSEMBLE output is much nicer.
1804 * The code in the SB-PROFILE package now seems reasonably stable.
1805 I still haven't decided what the final interface should look like
1806 (I'd like PROFILE to interact cleanly with TRACE, since both
1807 facilities use function encapsulation) but if you have a need
1808 for profiling now, you can probably use it successfully with
1809 the current CMU-CL-style interface.
1810 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
1811 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
1812 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
1813 link now returns the dangling link itself, and for similar
1814 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
1815 link itself. (In these cases the old code signalled an error and
1816 looped endlessly, respectively.) Thus, DIRECTORY now works even
1817 in the presence of dangling and cyclic symbolic links.
1818 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
1819 is now a supported extension again, since the consensus on
1820 sbcl-devel was that it can be useful for ordinary development
1821 work, not just for debugging SBCL itself.
1822 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
1823 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
1824 that currently-defined functions might be redefined later with
1825 different return types.
1826 * Hash tables can be printed readably, as inspired by CMU CL code
1827 of Eric Marsden and SBCL code of Martin Atzmueller.
1828 * better error handling in CLOS method combination, thanks to
1829 Martin Atzmueller porting Pierre Mai's CMU CL patches
1830 * more overflow fixes for >16Mbyte I/O buffers
1831 * A bug in READ has been fixed, so that now a single Ctrl-D
1832 character suffices to cause end-of-file on character streams.
1833 In particular, now you only need one Ctrl-D at the command
1834 line (not two) to exit SBCL.
1835 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
1837 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
1838 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
1839 does the right thing.
1840 * The implementation of some type tests, especially for CONDITION
1841 types, is now tidier and maybe faster, due to CMU CL code
1842 originally by Douglas Crosher, ported by Martin Atzmueller.
1843 * Some math functions have been fixed, and there are new
1844 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
1845 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
1846 * (There are also some new optimizers in contrib/*-extras.lisp. Those
1847 aren't built into sbcl-0.6.13, but are a sneak preview of what's
1848 likely to be built into sbcl-0.7.0.)
1849 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
1850 CL, ported to SBCL by Martin Atzmueller)
1851 * DESCRIBE now gives more information in some cases. (Pierre Mai's
1852 patch to CMU CL, ported to SBCL by Martin Atzmueller)
1853 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
1854 * There's a new slam.sh hack to shorten the edit/compile/debug
1855 cycle for low-level changes to SBCL itself, and a new
1856 :SB-AFTER-XC-CORE target feature to control the generation of
1857 the after-xc.core file needed by slam.sh.
1858 * minor incompatible change: The ENTRY-POINTS &KEY argument to
1859 COMPILE-FILE is no longer supported, so that now every function
1860 gets an entry point, so that block compilation looks a little
1861 more like the plain vanilla ANSI section 3.2.2.3 scheme.
1862 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
1863 returns the number of bytes consed since the system started,
1864 rather than the number consed since the first time the function
1865 was called. (The new definition parallels ANSI functions like
1866 CL:GET-INTERNAL-RUN-TIME.)
1867 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
1868 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
1869 supported. Now DIRECTORY always does the abstract Common-Lisp-y
1870 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
1871 * Fasl file version numbers are now independent of the target CPU,
1872 since historically most system changes which required version
1873 number changes have affected all CPUs equally. Similarly,
1874 the byte fasl file version is now equal to the ordinary
1877 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
1878 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
1879 is no longer recognized. I apologize for this, because it was
1880 listed in SB-EXT as a supported extension, but I found that
1881 its existing behavior was poorly specified, as well as incorrectly
1882 specified, and it looked like too much of a mess to straighten it
1883 out. I have enough on my hands trying to get ANSI stuff to work..
1884 * many patches ported from CMU CL by Martin Atzmueller, with
1885 half a dozen bug fixes in pretty-printing and the debugger, and
1886 half a dozen others elsewhere
1887 * fixed bug 13: Floating point infinities are now supported again.
1888 They might still be a little bit flaky, but thanks to bug reports
1889 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
1890 as flaky as they were.
1891 * The --noprogrammer command line option is now supported. (Its
1892 behavior is slightly different in detail from what the old man
1893 page claimed it would do, but it's still appropriate under the
1894 same circumstances that the man page talks about.)
1895 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
1896 are now supported, and enabled by default. Thus, the compiler can
1897 handle many floating point and complex operations much less
1898 inefficiently. (Thus e.g. you can implement a complex FFT
1900 * The compiler now detects type mismatches between DECLAIM FTYPE
1901 and DEFUN better, and implements CHECK-TYPE more correctly, and
1902 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
1903 thanks to patches from Martin Atzmueller.
1904 * various fixes to make the cross-compiler more portable to
1905 ANSI-conforming-but-different cross-compilation hosts (notably
1906 Lispworks for Windows, following bug reports from Arthur Lemmens)
1907 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
1908 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
1909 fixed thanks to Pierre Mai's CMU CL patches.
1910 * fixes to keep the system from overflowing internal counters when
1911 it tries to use i/o buffers larger than 16M bytes
1912 * fixed bug 45a: Various internal functions required to support
1913 complex special functions have been merged from CMU CL sources.
1914 (When I was first setting up SBCL, I misunderstood a compile-time
1915 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
1916 * improved support for type intersection and union, fixing bug 12
1917 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
1918 more obscure bugs as well
1919 * some steps toward byte-compiling non-performance-critical
1920 parts of the system, courtesy of patches from Martin Atzmueller
1921 * Christophe Rhodes has made some debian packages of sbcl at
1922 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
1923 From his sbcl-devel e-mail of 2001-04-08 they're not completely
1924 stable, but are nonetheless usable. When he's ready, I'd be happy
1925 to add them to the SourceForge "File Releases" section. (And if
1926 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
1927 * new fasl file format version number (because of changes in
1928 internal representation of (OR ..) types to accommodate the new
1929 support for (AND ..) types, among other things)
1931 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
1932 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
1934 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
1936 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
1937 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
1938 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
1939 * DESCRIBE now works on structure objects again.
1940 * Most function call argument type mismatches are now handled as
1941 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
1942 know whether the function will be redefined before the call is
1943 executed. (The compiler could flag local calls with full WARNINGs,
1944 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
1945 it doesn't keep track of enough information to know whether calls
1946 are local in this sense.)
1947 * Compiler output is now more verbose, with messages truncated
1948 later than before. (There should be some supported way for users
1949 to override the default verbosity, but I haven't decided how to
1950 provide it yet, so this behavior is still controlled by the internal
1951 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
1952 src/compiler/ir1util.lisp.)
1953 * Fasl file format version numbers have increased again, because
1954 support for the Gray streams extension changes the layout of the
1955 system's STREAM objects.
1956 * The Gray subclassable streams extension now works, thanks to a
1957 patch from Martin Atzmueller.
1958 * The full LOAD-FOREIGN extension (not just the primitive
1959 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
1960 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
1961 but like most other programs, it defaults to copying the Unix
1962 environment from the original process instead of starting the
1963 new process in an empty environment.
1964 * Extensions which manipulate the Unix environment now support
1965 an :ENVIRONMENT keyword option which doesn't smash case or
1966 do other bad things. The CMU-CL-style :ENV option is retained
1967 for porting convenience.
1968 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
1969 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
1971 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
1973 * A patch from Martin Atzmueller seems to have solved the SIGINT
1974 problem, and as far as we know, signal-handling now works cleanly.
1975 (If you find any new bugs, please report them!)
1976 * The system no longer defaults Lisp source file names to types
1977 ".l", ".cl", or ".lsp", but only to ".lisp".
1978 * The compiler no longer uses special default file extensions for
1979 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
1980 seems to expect a single default extension for all compiled code,
1981 and there's no compelling reason to try to stretch the standard
1982 to allow two different extensions.) Instead, byte-compiled files
1983 default to the same extension as native-compiled files.
1984 * Fasl file format version numbers have increased again, because
1985 a rearrangement of internal implementation packages made some
1986 dumped symbols in old fasl files unreadable in new cores.
1987 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
1988 many fewer weird special cases.
1989 * Bug #17 (differing COMPILE-FILE behavior between logical and
1990 physical pathnames) has been fixed, and some related misbehavior too,
1991 thanks to a patch from Martin Atzmueller.
1992 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
1993 by Tim Moore, ported to SBCL by Martin Atzmueller.
1994 * Martin Atzmueller fixed several filesystem-related problems,
1995 including bug #36, in part by porting CMU CL patches, which were
1996 written in part by Paul Werkowski.
1997 * More compiler warnings in src/runtime/ are gone, thanks to
1998 more patches from Martin Atzmueller.
1999 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2002 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2004 * DESCRIBE now works on CONDITION objects.
2005 * The debugger now handles errors which arise when trying to print
2006 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2008 * The build system now uses an additional file, customize-target-features.lisp,
2009 to allow local modifications to the target *FEATURES* list. (The point of
2010 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2011 debugging features enabled, without having to worry about propagating it
2012 into everyone's system when I do a "cvs update".) When no
2013 customize-target-features.lisp file exists, the target *FEATURES* list
2014 should be constructed the same way as before.
2015 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2016 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2017 warn about behavior which is undefined under the ANSI spec. Note: This
2018 is specified by ANSI, but it's not very popular with programmers.
2019 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2020 macro in the SBCL sources for an example of a workaround which you
2021 might use to make portable ANSI-standard code which does what you want.
2022 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2023 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2024 when it's used as a non-toplevel form. (This is required in order
2025 to implement the DEFCONSTANT-EQX macro.)
2026 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2027 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2028 funny things, due to bugs in the implementation of EVAL-WHEN
2029 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2030 non-toplevel DEFCONSTANTs are uncommon.)
2031 * The core file version number and fasl file version number have been
2032 incremented, because the old noncompliant DEFCONSTANT behavior involved
2033 calling functions which no longer exist, and because I also took the
2034 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2035 * fixed bug 1 (error handling before read-eval-print loop starts), and
2036 redid debugger restarts and related debugger commands somewhat while
2038 ** The QUIT debugger command is gone, since it did something
2039 rather different than the SB-EXT:QUIT command, and since it never
2040 worked properly outside the main toplevel read/eval/print loop.
2041 Invoking the new TOPLEVEL restart provides the same functionality.
2042 ** The GO debugger command is also gone, since you can just invoke
2043 the CONTINUE restart directly instead.
2044 ** The TOP debugger command is also gone, since it's redundant with the
2045 FRAME 0 command, and since it interfered with abbreviations for the
2047 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2049 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2050 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2051 some porting work by Martin Atzmueller)
2052 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2053 with a boatload of patches which he ported to SBCL. Now that those
2055 ** The system tries to make sure that its low-priority messages
2056 are prefixed by semicolons, to help people who like to use
2057 syntax highlighting in their ILISP buffer. (This patch
2058 was originally due to Raymond Toy.)
2059 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2060 patches originally written by Raymond Toy.
2061 ** The compiler understands coercion between single-value and
2062 multiple-VALUES type expressions better, getting rid of some very
2063 weird behavior, thanks to patches originally by Robert MacLachlan
2064 and Douglas Crosher.
2065 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2066 lambda lists, thanks to a patch originally by Pierre Mai.
2067 ** The system no longer bogusly warns about "abbreviated type
2069 ** The compiler gets less confused by inlining and RETURN-FROM,
2070 thanks to some patches originally by Tim Moore.
2071 ** The system no longer hangs when dumping circular lists to fasl
2072 files, thanks to a patch originally from Douglas Crosher.
2073 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2074 undefined function error.
2075 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2076 in structure slot initforms) for now, documented workaround instead:-|
2077 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2078 * fixed bug 5: added stubs for various Gray stream functions called
2079 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2080 installed, at least appropriate type errors are generated
2081 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2082 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2083 consistently in DEFMETHOD forms.
2084 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2085 it doesn't seem to affect SBCL after all
2086 * The C runtime system now builds with better optimization and many
2087 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2089 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2091 * The system is now under CVS at SourceForge (instead of the
2092 CVS repository on my home machine).
2093 * The new signal handling code has been tweaked to treat register
2094 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2095 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2096 downstream, e.g. in debug-int.lisp, has implicit dependencies
2097 on the unsignedness of integer representation of machine words,
2098 and that caused the system to bomb out with infinite regress
2099 when trying to recover from type errors involving signed values,
2100 e.g. (BUTLAST '(1 2 3) -1).)
2101 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2102 (This was one of the bugs Peter Van Eynde reported back in July.)
2103 * The system now uses code inspired by Colin Walters' O(N)
2104 implementation of MAP (from the cmucl-imp@cons.org mailing
2105 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2106 inline the MAP operation, and there is more than one
2107 sequence argument to the MAP call (so that it can't just
2108 do ETYPECASE once and for all based on the type of the
2109 single sequence argument). (The old non-inline implementation
2110 of the general M-argument sequence-of-length-N case required
2111 O(M*N*N) time when any of the sequence arguments were LISTs.)
2112 * The QUIT :UNIX-CODE keyword argument has been renamed to
2113 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2114 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2116 * Raymond Wiker's patches to port dynamic loading from Linux to
2117 FreeBSD have been added.
2118 * The BUGS file is now more nearly up to date, thanks in large part
2119 to Martin Atzmueller's review of it.
2120 * The debugger now flushes standard output streams before it begins
2121 its output ("debugger invoked" and so forth).
2122 * The core version number and fasl file version number have both
2123 been incremented, because of incompatible changes in the layout
2125 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2126 used to not be called for a saved Lisp image.)
2127 * Martin Atzmueller's version of a patch to fix a compiler crash,
2128 as posted on sbcl-devel 13 September 2000, has been installed.
2129 * Instead of installing Martin Atzmueller's patch for the
2130 compiler transform for SUBSEQ, I deleted the compiler transform,
2131 and transforms for some similar consing operations.
2132 * A bug in signal handling which kept TRACE from working on OpenBSD
2134 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2135 away by constant folding
2136 * The system now defines its address space constants in one place
2137 (in the Lisp sources), and propagates them automatically elsewhere
2138 (through GENESIS and the sbcl.h file). Therefore, patching the
2139 address map is less unnecessarily tedious and error-prone. The
2140 Lisp names of address space constants have also been systematized.
2141 * CVS tags like dollar-Header-dollar have been removed from
2142 the sources, because they have never saved me trouble and
2143 they've been source of trouble working with patches and other
2144 diff-related operations.
2145 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2146 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2148 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2150 * The system has been ported to OpenBSD.
2151 * The system now compiles with a simple "sh make.sh" on the systems
2152 that it's supported on. I.e., now you no longer need to tweak
2153 text in the target-features.lisp-expr and symlinks in src/runtime/
2154 by hand, the make.sh takes care of it for you.
2155 * The system is no longer so grossly inefficient when compiling code
2156 involving vectors implemented as general (not simple) vectors (VECTOR T),
2157 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2158 which dares to use the various sequence functions on non-simple
2159 vectors, takes less of a performance hit.
2160 * There is now a primitive type predicate VECTOR-T-P
2161 to test for the (VECTOR T) type, so that e.g.
2162 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2163 can now be compiled with some semblance of efficiency. (The old code
2164 turned the type declaration into a full call to %TYPEP at runtime!)
2165 * AREF on (VECTOR T) is still not fast, since it's still compiled
2166 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2167 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2168 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2169 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2170 without making full calls to SUBTYPEP at runtime.
2171 (Some analogous efficiency issues for non-simple vectors specialized to
2172 element types other than T, or for non-simple multidimensional arrays,
2173 have not been addressed. They could almost certainly be handled the
2174 same way if anyone is motivated to do so.)
2175 * The changes in array handling break binary compatibility, so
2176 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2177 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2178 instead of (VALUES T T).
2179 * By following the instructions that Dan Barlow posted to sbcl-devel
2180 on 2 July 2000, I was able to enable primitive dynamic object
2181 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2182 functionality is not implemented (since it calls ld to resolve
2183 library references automatically, requiring RUN-PROGRAM for its
2184 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2185 try to resolve library references) is now supported.
2186 * The system now flushes the standard output streams when it terminates,
2187 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2188 them at several other probably-convenient times, e.g. in each pass of
2189 the toplevel read-eval-print loop, and after evaluating a form given
2190 as an "--eval" command-line option. (These changes were motivated by a
2191 discussion of stream flushing issues on cmucl-imp in August 2000.)
2192 * The source transform for TYPEP of array types no longer assumes
2193 that an array whose element type is a not-yet-defined type
2194 is implemented as an array of T, but instead punts, so that the
2195 type will be interpreted at runtime.
2196 * There is now some support for cross-compiling in make.sh: each of
2197 the phases of make.sh has its own script. (This should be transparent
2198 to people doing ordinary, non-cross-compile builds.)
2199 * Since my laptop doesn't have hundreds of megabytes of memory like
2200 my desktop machine, I became more motivated to do some items on
2201 my to-do list in order to reduce the size of the system a little:
2202 ** Arrange for various needed-only-at-cold-init things to be
2203 uninterned after cold init. To support this, those things have
2204 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2205 symbols with such names are now uninterned after cold init).
2206 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2207 fasl files for cold load.
2208 ** Remove the old compiler structure pooling code (which used to
2209 be conditional on the target feature :SB-ALLOC) completely.
2210 ** Redo the representation of some data in cold init to be more compact.
2211 (I also looked into supporting byte compiled code at bootstrap time,
2212 which would probably reduce the size of the system a lot, but that
2213 looked too complicated, so I punted for now.)
2214 * The maximum signal nesting depth in the src/runtime/ support code has
2215 been reduced from 4096 to 256. (I don't know any reason for the very
2216 large old value. If the new smaller value turns out to break something,
2217 I'll probably just bump it back up.)
2218 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2220 * Many, many bugs reported by Peter Van Eynde have been added to
2221 the BUGS list; some have even been fixed.
2222 * While enabling dynamic object file loading, I tried to make the
2223 code easier to understand, renaming various functions and variables
2224 with less ambiguous names, and changing some function calling
2225 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2226 * While trying to figure out how to do the OpenBSD port, I tried to
2227 clean up some of the code in src/runtime/. In particular, I dropped
2228 support for non-POSIX signal handling, added various comments,
2229 tweaked the code to reduce the number of compilation warnings, and
2230 renamed some files to increase consistency.
2231 * To support the new automatic configuration functionality in make.sh,
2232 the source file target-features.lisp-expr has been replaced with the
2233 source file base-target-features.lisp-expr and the machine-generated
2234 file local-target-features.lisp-expr.
2235 * fixed a stupid quoting error in make.sh so that using CMU CL
2236 "lisp -batch" as cross-compilation host works again
2238 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2240 * DESCRIBE no longer tries to call itself recursively to describe
2241 bound/fbound values, so that it no longer fails on symbols which are
2242 bound to themselves (like keywords, T, and NIL).
2243 * DESCRIBE now works on generic functions.
2244 * The printer now prints less-screwed-up representations of closures
2245 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2246 * A private symbol is used instead of the :EMPTY keyword previously
2247 used to mark empty slots in hash tables. Thus
2248 (DEFVAR *HT* (MAKE-HASH-TABLE))
2249 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2250 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2251 now does what ANSI says that it should. (You can still get
2252 similar noncompliant behavior if bang on the hash table
2253 implementation with all the symbols you get back from
2254 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2255 This breaks binary compatibility, since tests for equality to
2256 :EMPTY are wired into things like the macroexpansion of
2257 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2259 * There's now a minimal placeholder implementation for CL:STEP,
2260 as required by ANSI.
2261 * An obscure bug in the interaction of the normal compiler, the byte
2262 compiler, inlining, and structure predicates has been patched
2263 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2264 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2266 * Missing ordinary arguments in a macro call are now detected even
2267 when the macro lambda list contains &KEY or &REST.
2268 * The debugger no longer complains about encountering the top of the
2269 stack when you type "FRAME 0" to explicitly instruct it to go to
2270 the top of the stack. And it now prints the frame you request even
2271 if it's the current frame (instead of saying "You are here.").
2272 * As specified by ANSI, the system now always prints keywords
2273 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2275 * The default initial SIZE of HASH-TABLEs is now smaller.
2276 * Type information from CLOS class dispatch is now propagated
2277 into DEFMETHOD bodies, so that e.g.
2278 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2280 is now basically equivalent to
2281 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2282 (DECLARE (TYPE SINGLE-FLOAT X))
2284 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2285 operation, without having to do run-time type dispatch.
2286 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2287 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2288 * Since I don't seem to be making big file reorganizations very often
2289 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2290 is overflowing my ability to conveniently back them up), I've finally
2291 checked the system into CVS. (The CVS repository is on my home system,
2292 not at SourceForge -- putting it on SourceForge might come later.)
2293 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2294 high-level GC-NOTIFY-FOO functions send their output. (There's
2295 still very little control of where low-level verbose GC functions
2296 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2297 less than it used to -- the GC-NOTIFY-FOO functions are now under
2298 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2299 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2300 in only one place in the source code, and propagates it automatically
2301 everywhere that it's needed. Thus e.g. when I bump the version from
2302 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2303 * The C source files now include boilerplate legalese and documentation
2304 at the head of each file (just as the Lisp source files already did).
2305 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2306 to his page will be replaced with a link to his new CLiki service.
2308 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2310 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2311 * The build process now looks for GNU make under the default name "gmake",
2312 instead of "make" as it used to. If GNU make is not available as "gmake"
2313 on your system, you can change this default behavior by setting the
2314 GNUMAKE environment variable.
2315 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2316 can build without error under CMU CL.
2318 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2320 * There is now a partial SBCL user manual (with some new text and some
2321 text cribbed from the CMU CL manual).
2322 * The beginnings of a profiler have been added (starting with the
2323 CMU CL profiler and simplifying and cleaning up). Eventually the
2324 main interface should be through the TRACE macro, but for now,
2325 it's still accessed through vaguely CMU-CL-style functions and macros
2326 exported from the package SB-PROFILE.
2327 * Some problems left over from porting CMU CL to the new
2328 cross-compilation bootstrap process have been cleaned up:
2329 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2330 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2331 part when getting rid of the compiler *BACKEND* stuff.)
2332 ** The value of *NULL-TYPE* was screwed up, because it was
2333 being initialized before the type system knew the final
2334 definition of the 'NULL type. This screwed up several key
2335 optimizations in the compiler, causing inefficiency in all sorts
2336 of places. (I found it because I wanted to understand why
2337 GET-INTERNAL-RUN-TIME was consing.)
2338 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2339 PROCLAIM FTYPE information. Unfortunately this broke binary
2340 compatibility again, since now the forms output by DEFGENERIC
2341 to refer to functions which didn't exist in 0.6.3.
2342 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2343 can use the new (as of 0.6.3) transform for SOME into MAP into
2345 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2346 values to (INTEGER 0 1000000), so that the time code will no longer
2347 occasionally get blown up by Linux returning 1000000 microseconds
2348 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2349 its output conform to the ANSI spec. (Alas, this makes its output
2350 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2351 * A full call to MAP NIL with a single sequence argument no longer conses.
2352 * fixes to problems pointed out by Martin Atzmueller:
2353 * The manual page no longer talks about multiprocessing as though
2354 it were currently supported.
2355 * The ILISP support patches have been removed from the distribution,
2356 because as of version 5.10.1, ILISP now supports SBCL without us
2357 having to maintain patches.
2358 * added a modified version of Raymond Toy's recent CMU CL patch for
2359 EQUALP comparison of HASH-TABLE
2361 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2363 * The system still can't cross-compile itself with
2364 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2365 entails), but at least it can compile more of itself that way
2366 than it used to be able to, and various buglets which were uncovered
2367 by trying to cross-compile itself that way have now been fixed.
2368 * This release breaks binary compatibility again. This time
2369 at least I've incremented the FASL file format version to 2, so that the
2370 problem can be detected reliably instead of just causing weird errors.
2371 * various new style warnings:
2372 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2373 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2374 ** using the lexical binding of a variable named in the *FOO* style
2375 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2376 as specified by ANSI.
2377 * *RANDOM-STATE* is no longer automatically initialized from
2378 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2379 default behavior of the system is to repeat its behavior every time
2380 it's run. If you'd like to change this behavior, you can always
2381 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2382 old convention there was no comparably easy way to get the system to
2383 repeat its behavior every time it was run.
2384 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2385 FTYPE declarations has been removed, in favor of their ANSI
2386 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2387 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2388 the types of their sequence arguments can be determined at compile time.
2389 This is done through a new open code expansion for MAP which eliminates
2390 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2391 argument types can be determined at compile time.
2392 * The optimizer now transforms COERCE into an identity operation when it
2393 can prove that the coerced object is already of the correct type. (This
2394 can be a win for machine generated code, including the output of other
2395 optimization transforms, such as the MAP transform above.)
2396 * Credit information has been moved from source file headers into CREDITS.
2397 * Source file headers have been made more standard.
2398 * The CASE macro now compiles without complaining even when it has
2401 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2403 * (Note that the way that the PCL macroexpansions were rewritten
2404 to accommodate the change in DEFGENERIC below breaks binary
2405 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2406 not run under sbcl-0.6.2. Once we get out of alpha releases,
2407 i.e. hit release 1.0.0, we'll probably try to maintain binary
2408 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2409 and sbcl-1.4.4. Until then, however, it might be fairly common
2410 for maintenance releases to break binary compatibility.)
2411 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2413 * The reporting of unhandled signals has been changed to print some
2414 explanatory text as well as the report form. (Previously only
2415 the report form was printed.)
2416 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2417 it defines, so that the compiler no longer issues undefined function
2418 warnings for compiled-but-not-yet-loaded generic functions.
2419 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2420 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2421 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2422 section of the ANSI spec are supported.
2423 * The "" nickname for the "KEYWORD" package has been removed.
2424 The reader still handles symbol tokens which begin with a package marker
2425 as keywords, but it doesn't expose its mechanism for doing so in the
2426 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2427 * The system now issues STYLE-WARNINGs for contradictory TYPE
2428 proclamations. (Warnings for contradictory FTYPE proclamations would
2429 be nice too, but those can't be done usefully unless the type system
2430 is made smarter about FUNCTION types.)
2431 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2432 systematized, so that "*target-*.lisp is supposed to exist only on the
2433 target and imply that there's a related file which exists on the
2434 host, and *host-*.lisp is supposed to exist only on the host and imply
2435 that there's a related file which exists on the target. This involves a
2436 lot of renaming. Hopefully the acute confusion caused by the renaming
2437 will be justified by the reduction in chronic confusion..
2438 ** runtime-type.lisp -> early-target-type.lisp
2439 ** target-type.lisp -> late-target-type.lisp
2440 ** early-host-format.lisp -> early-format.lisp
2441 ** late-host-format.lisp -> late-format.lisp
2442 ** host-error.lisp -> misc-error.lisp
2443 ** early-error.lisp -> early-target-error.lisp
2444 ** late-error.lisp -> late-target-error.lisp
2445 ** host-defboot.lisp -> early-defboot.lisp
2446 ** code/misc.lisp -> code/target-misc.lisp
2447 ** code/host-misc.lisp -> code/misc.lisp
2448 ** code/numbers.lisp -> code/target-numbers.lisp
2449 ** code/early-numbers.lisp -> numbers.lisp
2450 ** early-host-type.lisp -> early-type.lisp
2451 ** late-host-type.lisp -> late-type.lisp
2452 ** host-typep.lisp -> typep.lisp
2453 ** load.lisp -> target-load.lisp
2454 ** host-load.lisp -> load.lisp
2455 ** host-disassem.lisp -> disassem.lisp
2456 ** host-insts.lisp -> insts.lisp
2457 ** byte-comp.lisp -> target-byte-comp.lisp
2458 ** host-byte-comp.lisp -> byte-comp.lisp
2459 ** host-signal.lisp -> signal.lisp
2460 ** host-defstruct.lisp -> defstruct.lisp
2461 ** late-target-type.lisp -> deftypes-for-target.lisp
2462 Furthermore, several other previously target-only files foo.lisp (e.g.
2463 hash-table.lisp and random.lisp) have been split into a target-and-host
2464 foo.lisp file and a target-only target-foo.lisp file, with their key type
2465 definitions in the target-and-host part, so that the cross-compiler will
2466 know more about target types.
2467 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2468 gone away. In their place are various *BACKEND-FOO* variables
2469 corresponding to the slots of the old structure.
2470 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2471 into the target SBCL has been fixed.
2472 * The chill.lisp system for loading cold code into a running SBCL
2474 * Support for the CMU CL "scavenger hook" extension has been removed.
2475 (It was undocumented and unused in the CMU CL sources that SBCL was
2476 derived from, and stale in sbcl-0.6.1.)
2477 * Various errors in the cross-compiler type system were detected
2478 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2479 (enabling various consistency checks). Many of them were fixed,
2480 but some hard problems remain, so the compiler is back to
2481 running without *TYPE-SYSTEM-INITIALIZED* for now.
2482 * As part of the cross-compiler type system cleanup, I implemented
2483 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2484 * I have started adding UNCROSS calls throughout the type system
2485 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2486 on cross-compiler input files will be able to go away, and various
2488 * CONSTANTP now returns true for quoted forms (as explicitly required
2491 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2493 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2494 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2495 (MAKE-LIST -1). In the longer term, it ought to become true
2496 that declarations are assertions even at SAFETY 1. For now, it's not
2497 quite true even at SAFETY 3, but it's at least more nearly true..
2498 (Note that this change seems to increases the size of the system by
2499 O(5%) and to decrease the speed of the compiler by 20% or more.)
2500 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2501 for the problem of printing dozens of lines of distracting information
2502 about low-level system machinery as part of the top stack frame
2503 on entry to the debugger when an undefined function was called.
2504 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2505 is not reset to COMMON-LISP-USER.
2506 * Compilation of stuff related to dyncount.lisp has been made conditional
2507 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2508 smaller. The various dyncount-related symbols have been moved into
2509 a new "SB-DYNCOUNT" package.
2510 * tty-inspect.lisp has been renamed to inspect.lisp.
2511 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2512 feature has gone away. (When we eventually port to other flavors of
2513 libc and/or Unix, we'll try to make the differences between flavors
2514 invisible at the user level.)
2515 * Various other *FEATURES* tags, and/or their associated conditionals,
2516 have been removed if obsolescent, or given better documentation, or
2517 sometimes given more-mnemonic names.
2519 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2521 * tidied up "make.sh" script
2522 * tidied up system directory structure
2523 * better "clean.sh" behavior
2524 * added doc/FOR-CMUCL-DEVELOPERS
2525 * many many small tweaks to output format, e.g. removing possibly-confusing
2526 trailing #\. character in DESCRIBE-INSTANCE
2527 * (EQUALP #\A 'A) no longer signals an error.
2528 * new hashing code, including EQUALP hashing
2529 * tidied up Lisp initialization and toplevel
2530 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2531 * command line argument processing
2532 * added POSIX-GETENV function to deal with Unix-ish environment variables
2533 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2534 terminating SBCL on EOF
2535 * non-verbose GC by default
2536 * There is no more "sbcl" shell script; the sbcl file is now the C
2537 runtime executable (just like CMU CL).
2538 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2540 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2541 symbol and package information as Lisp native symbol and package objects
2542 instead of strings naming symbols and strings naming packages. This way,
2543 whenever packages are renamed (as in warm init), debug information is
2544 transformed along with everything else.
2545 * tweaked the optimization policy declarations which control the building
2546 of SBCL itself. Now, among other things, the system no longer saves
2547 source location debugging information. (This helps two problems at once
2548 by reducing SBCL size and by keeping SBCL from trying to look for its
2549 sources -- which may not exist -- when reporting errors.)
2550 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2551 debugging and testing purposes
2552 * cleaned up printing, making the printer call PRINT-OBJECT for
2553 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2554 methods, giving nearly-ANSI behavior
2555 * converted almost all special variables to use *FOO* naming convention
2556 * deleted PARSE-TIME functionality, since it can be done portably
2557 * moved some files out of cold init into warm init
2558 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2560 * regularized formatting of source files
2561 * added an install.sh script
2562 * fixed ridiculous memory usage of cross-compiler by making
2563 compiler/alloc.lisp not try to do pooling unless it can hook
2564 itself into the GC of the cross-compilation host. Now the system
2565 builds nicely on my old laptop.
2566 * added :SB-ALLOC in target-features.lisp-expr
2567 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2568 was not implemented)
2569 * re-did condition handling and note reporting in the compiler. Notes
2570 are no longer handled by signalling conditions. Style warnings
2571 and warnings are handled more correctly and reported in such a way
2572 that it's easy to find one or the other in your output (so that you
2573 can e.g. figure out which of many problems caused COMPILE-FILE to
2575 * changed the severity of several compiler warnings from full WARNING
2576 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2577 compiler note reporting so that it doesn't use the condition system
2578 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2579 COMPILE-FILE command)
2580 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2581 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2582 no longer supported, since their implementation was deeply intertwingled
2583 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2584 * removed bogus "support" for compiler macros named (SETF FOO), and
2585 removed the compiler macro for SETF INFO (but only after making a fool
2586 of myself on the cmucl-imp mailing list by posting a bogus patch for
2587 DEFINE-COMPILER-MACRO..)
2588 * Compiled files containing forms which have side effects on the Lisp
2589 reader (such as DEFPACKAGE forms) are now handled more correctly.
2590 (Compiler queuing of top level lambdas has been suppressed by setting
2591 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2592 * deleted various currently-unused source files, e.g. gengc.lisp. They
2593 may be added back at some point e.g. when porting to other architectures,
2594 but until they are it's distracting to distribute them and to try to
2596 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2597 there were so many of them they're just distractions, and UNCROSS is
2598 known to be able to handle the current sources
2599 * moved PROFILE functionality into TRACE, so that it will be clear
2600 how the wrapping and unwrapping of functions when you profile them
2601 interacts with the wrapping and unwrapping of functions when you
2602 trace them. (Actually, the functionality isn't there yet, but at least
2603 the interface specification is there. Hopefully, the functionality will
2604 arrive with some future maintenance release.)
2605 * removed host-oops.lisp
2606 * changed signature of QUIT function to allow UNIX-CODE argument
2607 * fixed READ-SEQUENCE bug
2608 * tweaked verbose GC output so that it looks more like the progress
2609 output that ANSI specifies for functions like LOAD
2610 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2611 * added <http://sbcl.sourceforge.com> to the banner information printed by