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 * minor incompatible change: as threatened around sbcl-0.8.0, the
8 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
9 The symbols are also exported from SB-PCL for backwards
10 compatibility, but more so than before SB-PCL should be treated as
11 an implementation-internal package.
12 * the SB-SPROF contrib now works on (most) non-x86 architectures.
13 It is known as of this release not to work on the Alpha, however.
14 * fixed bug #167: errors signalled due to illegal syntax in method
15 bodies are now more legible.
16 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
17 designators and can hence be used with TYPEP.
18 * fixed bug #333: CHECK-TYPE now ensures that the type error
19 signalled, if any, has the right object to be accessed by
20 TYPE-ERROR-DATUM. (reported by Tony Martinez)
21 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
22 5.1.2.3 for multiple-value place subforms. (reported by Kalle
24 * fixed bug #334: programmatic addition of slots using specialized
25 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
26 and :ALLOCATION :CLASS slots.
27 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
28 float exponents. (rereported by Peter Seibel)
29 * fixed a bug: #\Space (and other whitespace characters) are no
30 longer considered to be macro characters in standard syntax by
32 * fixed bug: initialization of condition class metaobjects no longer
33 causes an instance of the condition to be created. (reported by
35 * fixed bug: it is now possible to have slots such that
36 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
37 description is neither :INSTANCE nor :CLASS.
38 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
39 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
40 SLOT-VALUE-USING-CLASS), and no longer causes errors with
41 non-standard SLOT-DEFINITION-ALLOCATION values.
43 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
44 * minor incompatible change: the system no longer provides
45 optimization hints (conditions which are TYPEP
46 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
47 mode; these hints are still emitted when higher SPEED optimization
49 * new contrib module: a sampling profiler (profiling by statistical
50 sampling, rather than by instrumenting functions) is available as
51 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
52 * the behaviour of the standard function ED is now customizeable by
53 third parties through a hook variable: see ED's documentation
54 string for information on the protocol.
55 * the compiler no longer emits efficiency notes for (FUNCALL X)
56 when the type of X is uncertain under default optimization
58 * fixed bug 276: mutating a binding of a specialized parameter to a
59 method to something that is not TYPEP the specializer is now
61 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
62 exhibit the required behaviour on the x86 platform. (thanks to
63 Peter van Eynde, Eric Marsden and Bruno Haible)
64 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
65 even for difficult arguments. (reported by Peter Graves)
66 * fixed bug 141a: the backquote printer now descends quoted
68 * fixed another bug in backquote printing: no more destructive
69 modification of the form's list structure. (reported by Brian
71 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
72 move between its address being taken and the call to
73 interrupt_thread, fixing a crashing race condition.
74 * the SB-POSIX contrib implementation has been adjusted so that it
75 no longer exhibits ridiculously poor performance when constructing
76 instances corresponding to C structs.
78 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
79 * minor incompatible change: the sb-grovel contrib now treats C
80 structures as alien (in the sense of SB-ALIEN) objects rather than
81 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
82 has implications for memory management of client code
83 (sb-grovel-returned objects must now be manually managed) and for
84 type safety (alien objects now have full types).
85 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
86 used to control emission of compiler diagnostics, rather than the
87 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
88 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
89 quality should be considered deprecated.
90 * install.sh now installs the user manual as well
91 * (not quite a new documentable feature, but worth considering in
92 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
93 beginnings of a semantically meaningful condition hierarchy is
94 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
96 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
97 designator as the defaults argument.
98 * fixed bug: Displaced arrays whose displaced-to array has become
99 too small now cause ARRAY-DIMENSION to signal an error, providing
100 for safer bounds-checking. (reported by Bruno Haible)
101 * fixed bug: DEFCLASS slot definitions with identical :READER and
102 :WRITER names now signal a reasonable error. (reported by Thomas
104 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
106 * fixed bug: Invalid slot specification errors now print correctly.
107 (thanks to Zach Beane)
108 * fixed bug 320: Shared to local slot value transfers in class
109 redefinitions now happen corrently from superclasses as
110 well. (reported by Bruno Haible)
111 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
113 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
114 improper lists correctly. (reported by Bruno Haible)
115 * fixed bug 313: source-transform for <fun-name> was erroneously
116 applied to a call of a value of a variable with name <fun-name>.
117 (reported by Antonio Menezes Leitao)
118 * fixed bug 307: The obsolete instance protocol ensures that
119 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
120 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
121 remove dead unknown-values globs from the middle of the stack.
122 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
123 easier to resume long computations after using *BREAK-ON-SIGNALS*
124 to diagnose and fix failures. (thanks to Nikodemus Siivola)
125 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
126 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
127 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
128 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
129 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
131 * fixed a (fairly theoretical) bug in string printing: if
132 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
133 does not have array-element-type equal to the most general string
135 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
136 function name. (thanks to Nikodemus Siivola)
137 * fixed bug: compile-time format string checker failed on
138 non-closed ~{. (reported by Thomas F Burdick)
139 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
140 don't warn on legal loop code involving more than one aggregate
141 boolean. (thanks to Nikodemus Siivola)
142 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
143 in keyword position to MAKE-PACKAGE was not being reported as
145 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
146 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
148 * fixed bugs 280 and 312: the checking for multiple definitions in a
149 file is less likely to become confused by uses of inline
151 * fixed bug: the #S reader macro performs the keyword coercion
152 specified for slot names. (reported by Kalle Niemitalo)
153 * fixed bug: lambda lists may contain symbols whose names start with
154 & but are not lambda-list-keywords; their occurrence triggers a
156 * fixed bug 321: define-method-combination argument lambda lists do
157 not cause invalid code to be generated when &optional or &aux
158 variables have default values. (reported by Bruno Haible)
159 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
160 have CLOS classes; this appears also to have fixed reported
161 instability in stack exhaustion detection.
162 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
163 MUFFLE-WARNING when no associated restart is present is now
165 * optimization: rearranged the expansion of various defining macros
166 so that each expands into only one top-level form in a
167 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
169 * optimization: used a previously little-used slot in symbols to
170 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
172 * fixed some bugs revealed by Paul Dietz' test suite:
173 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
175 ** Fixed an optimization bug related to inheritance of initforms
176 from local to shared slots.
177 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
178 ** CAST optimizer forgot to flush argument derived type.
179 ** print/read consistency on floats is now orders of magnitude
180 more likely. (thanks also to Bruno Haible for a similar report
182 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
183 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
184 for objects of type REAL. Make it so.
185 ** SXHASH is specified (infelicitously) to respect similarity,
186 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
187 it so. (thanks to Markus Ziegler)
188 ** on the Alpha, the compiler succeeds in compiling functions
189 returning a known number of arguments greater than 63.
190 ** fixed handling of invalid NIL arguments in keyword position
192 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
194 ** provided a readably-printable representation for RANDOM-STATE
196 ** ensured that pathnames, where they have a namestring, always
197 print using #P"..." syntax.
199 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
200 * Support for the forthcoming 2.0 version of the NetBSD kernel
201 running on x86 hardware has been added. (thanks to Perry
202 E. Metzger most immediately, and others for their past work)
203 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
204 current release version; SBCL's OpenBSD support had been broken
205 since about the time of OpenBSD's switch to ELF binary format.)
206 * [placeholder for DX summary]
207 ** user code with &REST lists declared dynamic-extent, under high
208 speed or space and low safety and debug optimization policy.
209 * The manual has been converted to Texinfo format and the debugger
210 chapter from the cmucl manual has been added.
211 * A facility has been added to extract documentation strings from
212 sbcl and store them as Texinfo-formatted snippets for inclusion in
213 the manual (via Texinfo's @include directive)
214 * bug fix: compiler emitted division in optimized DEREF. (thanks for
215 the test case to Dave Roberts)
216 * bug fix: multidimensional simple arrays loaded from FASLs had fill
217 pointers. (reported by Sean Ross)
218 * bug fix: PROFILE output is printed nicely even for large numerical
219 values. (thanks to Zach Beane)
220 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
221 greater than 32 handle EOF correctly.
222 * bug fix: on X86 an immediate argument of the IMUL instruction is
223 correctly printed in disassembly. (thanks to Lutz Euler)
224 * bug fix: class slots in redefined classes preserve their old
225 values. (thanks to Bruno Haible and Nikodemus Siivola)
226 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
227 succeeds. (reported by Marco Baringer)
228 * fixed some bugs revealed by Paul Dietz' test suite:
229 ** READ-SEQUENCE now works on ECHO-STREAMs.
230 ** RATIONALIZE works more according to its specification. (thanks
233 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
234 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
235 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
236 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
237 should matter to you only if you rebind the printer control
238 variables and then find you want different bindings in the
239 debugger than in the ordinary execution of your program.)
240 * The runtime build system has been tweaked to support building
241 (on SPARC/SunOS) using a C compiler which invokes Sun's own
242 assembler and linker. (thanks to Nikodemus Siivola)
243 * Unbound, undefined, undeclared variables now trigger full
244 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
245 more often programmer error than deliberate exploitation of undefined
247 * optimization: the hash algorithm for strings has changed to one
248 that is less vulnerable to spurious collisions. (thanks to Juho
250 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
251 needless bounds checking. (thanks to Juho Snellman)
252 * optimization: implemented multiplication as a modular
253 (UNSIGNED-BYTE 32) operation on the PPC backend.
254 * fixed some bugs revealed by Paul Dietz' test suite:
255 ** ADJUST-ARRAY now copies the datum in a zero rank array if
257 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
258 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
259 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
261 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
263 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
265 ** ECHO-STREAMs no longer attempt to echo the end of file value to
266 their output stream on EOF from read.
267 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
268 have been read to end-of-file.
269 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
271 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
272 description of determination of which consecutive characters
274 ** Printing the "Space" character with escaping on now yields "#\\ ",
275 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
276 ** Reading floating-point numbers with *READ-BASE* set to a number
277 less than 10 works correctly.
278 ** Reading floating-point numbers with *READ-BASE* set to a number
279 more than 10 works correctly.
280 ** Printing with *PRINT-READABLY* targets the standard readtable, not
281 the readtable currently in effect.
283 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
284 * minor incompatible change: parsing of namestrings on a physical
285 (Unix) host has changed; numbers after the final #\. in a
286 namestring are no longer interpreted as a version field. This is
287 intented to be largely invisible to the user, except that the
288 meaning of the namestring "*.*.*" has changed: it now refers to a
289 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
290 should usually be replaced by
291 (make-pathname :name :wild :type :wild :version :wild)
292 with the added benefit that this is more likely to be portable.
293 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
294 option to OPEN now signals an error if the file being opened
295 exists; this may have an impact on existing code.
296 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
297 signals received, on the PowerPC platforms (both Linux and
298 Darwin). (thanks to Pierre Mai for pointing out the location of
300 * several fixes on OS X: The system now builds and runs cleanly on
301 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
302 (thanks to Brian Mastenbrook)
303 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
304 second-resolution: integer multiples of 1/3600 between -24 and 24.
305 (thanks to Vincent Arkesteijn)
306 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
307 type when called with 1 argument; PEEK-CHAR checked type of
308 PEEK-TYPE only after having read first character from a
309 stream. (reported by Peter Graves)
310 * bug fix: the garbage collector now has much better locality
311 behaviour, and in particular no longer treats all memory as being
312 exhausted when many small objects point to each other in a deeply
314 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
315 recognized as being TYPEP their class.
316 * bug fix: the PUSHNEW documentation string has been corrected.
317 (thanks to Vincent Arkesteijn)
318 * bug fix: defaulting of the value for the last of an atypically
319 large number of multiple values being bound was not being
320 performed correctly on the Alpha or PPC platforms
321 * optimization: implemented multiplication as a modular
322 (UNSIGNED-BYTE 32) operation on the x86 backend.
323 * optimization: SEARCH on simple-base-strings can now be open-coded.
324 (see also contrib/compiler-extras.lisp for inspiration for
325 teaching the compiler about the Boyer-Moore algorithm).
326 * value, returned by MAX (and MIN) called with several EQUALP, but
327 not EQL, arguments now does not depend on compiler settings.
328 * fixed some bugs revealed by Paul Dietz' test suite:
329 ** in stack analysis liveness information is propagated from
330 non-local entry points.
331 ** pathwise CAST removing failed when the CAST node did not start
333 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
334 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
336 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
337 host is already defined.
338 ** RENAME-FILE works on streams instead of signalling an internal
340 ** PEEK-CHAR uses the current readtable when determining whether
341 or not a character is whitespace.
342 ** MERGE-PATHNAMES handles the case when the pathname does not
343 specify a name while the default-pathname specifies a version.
344 ** Pathnames now stand a better chance of respecting print/read
346 ** Attempting to use standardized file system operators with a
347 pathname with invalid :DIRECTORY components signals a
349 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
351 ** DIRECTORY on logical pathnames is more correct.
352 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
353 signal a TYPE-ERROR if their argument is not a stream.
354 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
355 designator argument does not designate a stream.
356 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
357 examining the synonym.
358 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
360 ** OPEN and WITH-OPEN-STREAM allow opening streams with
361 element-type larger than ([UN]SIGNED-BYTE 32).
363 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
364 * When built with the :SB-FUTEX feature, threaded builds now take
365 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
366 for faster/more reliable mutex and condition variable support.
367 * Incompatible change (but one you probably shouldn't have been using
368 anyway): the interface and code for arbitrating between multiple
369 threads in the same user session has been redesigned.
370 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
371 error when called without an explicit environment argument.
372 (thanks to Nikodemus Siivola)
373 * bug fix: buffered :DIRECTION :IO streams are less likely to become
374 confused about their position. (thanks to Adam Warner and Gerd
376 * bug fix: Pretty printing backquoted forms with unquotations in the
377 argument list position of various code constructs such as LAMBDA
378 now works correctly. (reported by Paul Dietz)
379 * bug fix: Pretty printing unquotations no longer loses all
380 stream position information.
381 * optimization: performance of string output streams is now less
382 poor for multiple small sequence writes.
383 * optimization: performance of CSUBTYPEP in the presence of complex
384 expressions involving CONS and NOT many times has been improved.
385 (reported by Paul Dietz)
386 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
388 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
389 streams for the REPL, for the debugger, and so on. (thanks to
391 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
392 * fixed some bugs revealed by Paul Dietz' test suite:
393 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
394 is ignored if it is too small, rather than propagating through
395 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
396 ** extremely complex negations of CONS types were not being
397 sufficiently canonicalized, leading to inconsistencies in
399 ** VALUES tranformer lost derived type.
401 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
402 * fixed a bootstrapping bug: the build process no longer assumes
403 that the various BOOLE-related constants have the same value in
404 host and target lisps. (noted by Paul Dietz' test suite on an
405 SBCL binary built from CLISP)
406 * The system can now be dynamically linked on the MIPS platform,
407 which enables dynamic loading of foreign code from Lisp. (thanks
408 to Ralf Baechle for discussions on the MIPS ABI)
409 * The system now records debugging information for its own source
410 files in a filesystem-position-independent manner, relative to
411 the "SYS" logical host.
412 * fixed a compiler bug: MV-LET convertion did not check references
413 to the "max args" entry point. (reported by Brian Downing)
414 * tweaked disassembly notes to be less confident about proclaiming
415 some instruction as an LRA. (thanks to Brian Downing)
416 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
417 now each have their own history, command character, and other
418 characteristics. (thanks to David Lichteblau)
419 * fixed some bugs revealed by Paul Dietz' test suite:
420 ** compiler failure in compiling LOGAND expressions including a
422 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
423 shift greater than 32.
424 ** FLUSH-DEST did not mark blocks for type check regeneration.
425 ** HANDLER-CASE failed to accept declarations in handler clauses
426 in some circumstances.
428 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
429 * New code in contrib/sb-introspect (still probably not entirely
430 stable yet) provides some support for smart Lisp development
431 environments like SLIME.
432 * The conditions signalled for errors occurring when loading .fasl
433 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
434 in a way which should help ASDF recover gracefully.
435 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
436 stuff has been cleaned up. If you code contrib/ stuff, this might
437 affect you, and you can look at contrib/README, contrib/STANDARDS,
438 and/or the 0.8.4.27 diff to check.
439 * In full calls the compiler now does not generate checks for declared
440 argument types for all arguments.
441 * various threading fixes
442 ** and some experimental patches which didn't make it into
443 the main tree for this release, but which are shipped in
444 contrib/experimental-thread.patch as a possible fix for some
445 failures (deadlock, spinning...) in GC-intensive multithreaded
447 * fixed PPC build problem (source code incompatibility of different
448 library versions): added offsetof() hackery which attempts to divine
449 where glibc maintainers put uc_mcontext today
450 * fixed bug 282: compiler does not trust type assertions while passing
451 arguments to a full call.
452 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
453 &OPTIONAL VALUES type parameter.
454 * fix bug 214: algorithm for noting rejected templates is now more
455 similar to that of template seletion. (also reported by rydis on
457 * fixed bug 141b: printing backquoted information readably and prettily
458 inserts a space where necessary.
459 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
460 (CHARACTER 10) are now reported as errors, rather than propagated
461 as unknown types. (reported by piso on #lisp)
462 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
463 with values NIL and :ERROR. (thanks to Milan Zamazal)
464 * fixed bug 191c: CLOS now does proper keyword argument checking as
465 described in CLHS 7.6.5 and 7.6.5.1.
466 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
467 longer raise an error; further, using a list as a for-as-arithmetic
468 counter now raises a meaningful error.
469 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
470 now understood by sequence creation functions such as MAKE-SEQUENCE
472 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
473 a STRING-INPUT-STREAM.
474 * compiler enhancement: SIGNUM is now better able to derive the type
476 * type declarations inside WITH-SLOTS are checked. (reported by
478 * fixed some bugs revealed by Paul Dietz' test suite:
479 ** incorrect optimization of TRUNCATE for a positive first
480 argument and negative second.
481 ** compiler failure in let-convertion during flushing dead code.
482 ** compiler failure while deriving type of TRUNCATE on an
483 interval, containing 0.
484 ** ASH of a negative bignum by a negative bignum count now returns
486 ** intersection of CONS types now canonicalizes properly, fixing
487 inconsistencies in SUBTYPEP.
489 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
490 * incompatible change: The --disable-debugger command line
491 option now clobbers the debugger at a more fundamental
492 level, by redefining #'INVOKE-DEBUGGER instead of by
493 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
494 is specified by ANSI to ignore *DEBUGGER-HOOK* and
495 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
496 enter the debugger REPL and then suffer recursive errors
497 because *DEBUG-IO* is also messed up in --disable-debugger mode;
498 while under the new system, BREAK in --disable-debugger mode
499 terminates the system just as an unhandled error would.
500 * fixed compiler performance when processing loops with a step >1;
501 * bug fix: DOCUMENTATION now retrieves generic function
502 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
503 support has been systematized, and now supports the methods
504 specified by ANSI, along with a default method and a method for
505 slot documentation. (reported by Nathan Froyd)
506 * bug fix: effective methods associated with a generic function are
507 no longer cached over a change of that generic function's method
508 combination. (reported by Andreas Fuchs)
509 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
510 names a non-existent directory. (thanks to Andreas Fuchs)
511 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
512 the PPC platform, lead to stack corruption; this has been fixed.
513 (reported by Rainer Joswig)
514 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
515 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
516 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
517 arithmetic in general, is now much more reliable on the PPC
519 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
520 platform now returns the right answer.
521 * optimization: restored some effective method precomputation in
522 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
523 precomputation is now tunable.
524 * optimization: compiler-internal data structure use has been
525 reviewed, and changes have been made that should improve the
526 performance of the compiler by about 20%.
527 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
528 simple-base-strings and simple-bit-vectors is improved.
529 * optimization: the optimization of 32-bit logical and arithmetic
530 functions introduced in version 0.8.3 on the x86 has been
531 implemented on the mips, ppc and sparc platforms; an
532 implementation of the same facility, but for 64-bit arithmetic,
533 has been added for the alpha.
534 * microoptimization: the compiler is better able to make use of the
535 x86 LEA instruction for multiplication by constants.
536 * bug fix: in some situations compiler did not report usage of
537 generic arithmetic in (SPEED 3) policy.
538 * bug 145b fix: compiler used wrong type specifier while converting
539 MEMBER-types to numeric.
540 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
542 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
543 index argument now works properly on the Alpha platform.
544 * bug fix: floating point exception treatment on the Alpha platform
546 * bug fix: FILE-POSITION works much better on string input and
547 output streams. (thanks to Nikodemus Siivola)
548 * bug fix: many threading/garbage collection symptoms sorted.
549 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
550 might be pseudo-atomic.
551 * internal change: Stopping for GC is now done with signals not
552 ptrace. GC is now done in whichever thread wanted it, instead of
554 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
556 * bug fix: result form in DO is not contained in the implicit
558 * incompatible change: ICR structure is changed; the value part of
559 CONTINUATION is now called LVAR; corresponding functions are
560 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
561 * added type deriver for ISQRT (thanks to Robert E. Brown).
562 * bug fix: better support for loading from the command line when an
563 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
565 * fixed some bugs revealed by Paul Dietz' test suite:
566 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
567 ** ROUND and FROUND now give the right answer when given very
568 small float arguments.
569 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
571 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
572 ** (GCD 0 <negative-integer>) returned <negative-integer>.
573 ** LCM should return a non-negative integer.
574 ** PARSE-INTEGER returned the index of a terminator instead of the
575 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
576 ** PARSE-INTEGER returned an incorrect index being applied to a
578 ** LCM with two arguments of 0 returns 0 rather than signalling
580 ** unsigned addition of a 32-bit constant with the high bit set no
581 longer causes an internal compiler error on the x86.
582 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
583 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
584 with negative last argument.
585 ** byte specifiers with zero size and position no longer cause
586 an error during type derivation.
587 ** bignum multiplication on the Alpha platform now returns the
589 * porting: The system now builds on SuSE AMD64, although it still
590 generates a 32-bit binary.
591 * .fasl file incompatibility: The fasl file version number has
592 been incremented (because of the changes to internal compiler
593 data structures referred to above).
595 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
596 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
597 more accurately, on the Darwin kernel running on PowerPC hardware.
598 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
599 * Compiler code deletion notes now signal a condition of type
600 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
601 an associated MUFFLE-WARNING restart.
602 * The compiler now performs limited argument count validation of
603 constant format strings in FORMAT, and where appropriate in ERROR,
604 CERROR and WARN. (thanks to Gerd Moellmann)
605 * New ASDF-INSTALL contrib can be used for automatic download and
606 installation of third-party Lisp code from CCLAN or other sites
608 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
609 which forces another thread to execute a function supplied by the
611 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
612 now accept and act upon their :ELEMENT-TYPE keyword argument.
613 (reported by Martin Atzmueller, Edi Weitz)
614 * bug fix: FILE-POSITION now accepts position designators up to
615 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
616 is the greater. (thanks to Patrik Nordebo)
617 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
618 Kalle Olavi Niemitalo)
619 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
620 printable. (reported by Eric Marsden)
621 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
622 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
623 * x86 bug fix in control stack exhaustion checking: now shows backtrace
624 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
625 (thanks to Stig Sandoe)
626 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
627 UNREAD-CHAR now correctly includes the unread character in the
628 target sequence. (thanks to Gerd Moellmann)
629 * bug fix in threaded builds: the system can now be suspended and
630 resumed by shell job control with minimal disruption.
631 * bug fixes in times and timezones >2038AD
632 * better handling of "where is GNU make?" problem in build scripts
633 (thanks to Nikodemus Siivola)
634 * new optimization: inside a named function any reference to a
635 function with the same name is considered to be a self-reference;
636 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
637 optimization quality.
638 * new optimization on x86: logical functions and + now have
639 optimized (UNSIGNED-BYTE 32) versions, which are automatically
640 used when the result is truncated to 32 bits.
641 * VALUES declaration is partially enabled.
642 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
643 * fixed some bugs revealed by Paul Dietz' test suite:
644 ** The system now obeys the constraint imposed by
645 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
646 types form a lattice under type intersection.
647 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
648 ** ASSOC now ignores NIL elements in an alist.
649 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
650 and (1+ MOST-POSITIVE-FIXNUM) answers.
651 ** The addition of a method with invalid qualifiers to a generic
652 function does not cause an error to be signalled immediately;
653 a warning is signalled, and the error is generated only on
654 calling the generic function.
655 * changed .fasl file version number, in part to add type codes for
656 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
657 obscure ANSI requirements
659 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
660 * fixed bug 148: failure to inline-expand a local function left
661 garbage, confusing the compiler.
662 * fixed bugs 3cd: structure slot readers perform type check if the
663 slot can have an invalid value (i.e. it is either not initialized
664 or can be written with a less specific slot writer).
665 * bug fix: the compiler now traps array references to elements off
666 the end of an array; previously, the bounds checking in some
667 circumstances could go off-by-one.
668 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
670 * type declarations for array element types now obey the description
671 on the CLHS page "Declaration TYPE", as per discussions on
672 sbcl-help around 2003-05-08. This means that a declaration
673 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
674 declaration, all references to BAR will be asserted or assumed
675 (with THE, so dependent on compiler policy) to involve objects of
676 type FOO. Note that no such declaration is implied in
677 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
678 * declared types of functions from the "Conditions"
679 chapter. (reported by Paul Dietz)
680 * bug fix: CERROR accepts a function as its first argument.
681 * bug fix: NTH an NTHCDR accept a bignum as index
682 arguments. (reported by Adam Warner)
683 * optimization: character compare routines now optimize comparing
684 against a constant character. (reported by Gilbert Baumann)
685 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
686 argument now works properly on the MIPS platform.
687 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
688 * repeated evaluation of the same DEFSTRUCT, a slot of which is
689 declared to have a functional type, does not cause an error
691 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
692 many :DEFAULT-INITARGS. (reported by Istvan Marko)
693 * fixed bug: if last continuation of a deleted block has a
694 destination, this destination should be deleted too. (reported by
696 * fixed a bug in the bootstrap process: the host compiler's values
697 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
698 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
699 test case from Patrik Nordebo)
700 * improved the ability of the disassembler on the PPC platform to
701 provide helpful disassembly notes.
702 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
703 the class in more cases than previously.
704 * bug fix: FILE-POSITION now understands :START and :END for
705 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
706 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
707 8) data. (thanks to David Lichteblau)
708 * bug fix: it is possible to add a method to a generic function
710 * bug fix: reader failed to signal END-OF-FILE inside an
711 object representation. (reported by Nikodemus Siivola)
712 * fixed some bugs revealed by Paul Dietz' test suite:
713 ** LAST and [N]BUTLAST should accept a bignum.
714 ** condition slot accessors are methods.
715 ** (VECTOR NIL) is a subtype of STRING.
717 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
718 * minor incompatible change: some nonsensical specialized lambda
719 lists (used in DEFMETHOD) which were previously ignored now signal
721 * minor incompatible change: the system is now aware of the types of
722 variables in the COMMON-LISP package, and will signal errors for
723 most violations of these type constraints (where previously they
724 were silently accepted).
725 * minor incompatible change: COMPILE-FILE now uses the freedom
726 afforded (ANSI 3.2.2.3) to use derived function types for
727 functions defined in the same file. This also permits the system
728 to warn on static type mismatches and function
729 redefinition. (Currently it does not work with high DEBUG level.)
730 * minor incompatible change: VALUES declaration is disabled.
731 * When issuing notes, the compiler now signals a condition of type
732 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
733 restart for use in user handlers. It is expected that the
734 COMPILER-NOTE condition will eventually become a condition
735 supertype to a hierarchy of note types, which will then be
736 handleable in a similar fashion. However, at the moment, no such
737 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
738 but it's an implementation detail, not a classification for the
740 * Changes in type checking closed the following bugs:
741 ** type checking of unused values (192b, 194d, 203);
742 ** template selection based on unsafe type assertions (192c, 236);
743 ** type checking in branches (194bc).
744 * A short form of VALUES type specifier has ANSI meaning (it has
745 increased the number of situations when SBCL cannot perform type
747 * fixed bug in DEFSTRUCT: once again, naming structure slots with
748 keywords or constants is permissible.
749 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
750 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
751 classes. (thanks to Antonio Martinez)
752 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
753 outputs FRESH-LINE or TERPRI, and no longer converts its stream
754 argument to a pretty-print stream. Instead, it leaves any such
755 operations to DESCRIBE-OBJECT methods.
756 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
758 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
759 lambda lists. (thanks to David Lichteblau)
760 * bug fix: Defining a generic function with a :METHOD-CLASS being a
761 subclass of STANDARD-METHOD no longer causes stack exhaustion.
762 (thanks to Gerd Moellmann)
763 * fixed bug 246: increased compilation speed of long
764 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
766 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
768 * DEFCONSTANT now throws a condition of type
769 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
770 constant to a non-EQL value; CONTINUE and ABORT restarts
771 respectively change and preserve the value.
772 * fixed bug 63: The code walker, part of the implementation of CLOS,
773 is now better at handling symbol macros.
774 * bug fix: There is no longer an internal implementation type named
775 CL:LENGTH. (reported by Raymond Toy)
776 * bug fix: In macro-like defining macros/special operators the
777 implicit block does not enclose the lambda list.
778 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
779 enclosed in parentheses) are not suitable as type specifiers, and
780 their use properly signals an error now.
781 * bug fix: An argument count mismatch for a type specifier in code
782 being compiled no longer causes an unhandled error at compile
783 time, but signals a compile-time warning.
784 * fixed simple vector readable printing
785 * bug fix: DESCRIBE takes more care over whether the class
786 precedence list slot of a class is bound before accessing it.
787 (reported by Markus Krummenacker)
788 * bug fix: FORMATTER can successfully compile pretty-printer format
789 strings which use variants of the ~* directive inside.
790 * bug fix: SEARCH now applies its TEST predicate to the elements of
791 the arguments in the correct order. (thanks to Wolfhard Buss)
792 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
793 in type inference. (thanks to Robert E. Brown)
794 * bug fix: Reading in symbols with an explicit package name of ""
795 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
796 * fixed some bugs revealed by Paul Dietz' test suite:
797 ** NIL is now allowed as a structure slot name.
798 ** Arbitrary numbers, not just REALs, are allowed in certain
799 circumstances in LOOP for-as-arithmetic clauses.
800 ** Multiple class redefinitions before slot access no longer
802 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
803 association between the name and a class.
804 ** Generic functions with non-standard method-combination and over
805 five methods all of which return constants no longer return NIL
806 after the first few invocations. (thanks to Gerd Moellmann)
807 ** CALL-NEXT-METHOD with no arguments now passes the original
808 values of the arguments, even in the presence of assignment.
809 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
811 ** DISASSEMBLE works with closures and funcallable instances.
812 ** ADD-METHOD now returns the generic function, not the new method.
813 ** FIND-METHOD signals an error if the lengths of the specializers
814 is incompatible with the generic function, even if the ERRORP
816 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
817 which its argument is a member.
818 ** DEFCLASS only redefines the class named by its class-name
819 argument if that name is the proper name of the class;
820 otherwise, it creates a new class.
821 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
822 of the UNBOUND-SLOT condition to the name of the slot.
823 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
824 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
825 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
826 SLOT-MAKUNBOUND in the specified fashion.
828 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
829 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
830 cross-compilation host. As a consequence, we can now bootstrap our
831 way up to SBCL starting with a bare gcc toolchain and human-readable
832 source code (first the source to CLISP, then the source to SBCL).
833 * A contributed module containing a partial implementation of the
834 simple-streams interface has been included. (thanks to Rudi
836 * A contributed module implementing the RFC1321 Message Digest
837 Algorithm, known as MD5, has been included.
838 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
839 no longer has any effect, as the code controlled by this feature
840 has been deleted. (As far as we know, no-one has ever built using
841 this feature, and its semantics were confused in any case).
842 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
843 work on conditions (as required by the ANSI specification),
844 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
845 expected behaviour on conditions. Users should note, however,
846 that such behaviour is not required by the ANSI specification,
847 and so use of this behaviour may render their code unportable.
848 * fixed some bugs revealed by Paul Dietz' test suite:
849 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
851 ** &ENVIRONMENT parameter in macro lambda list is bound first.
852 ** SXHASH on condition objects no longer returns NIL.
853 ** :ALLOCATION :CLASS slots are better treated; their values are
854 updated on class redefinition, and initforms inherited from
855 superclasses are applied.
856 ** REMOVE-METHOD returns its generic function argument even when
857 no method was removed.
858 ** SHARED-INITIALIZE now initializes the values of the requested
859 slots, including those with :ALLOCATION :CLASS.
860 ** ALLOCATE-INSTANCE now works on structure classes defined via
861 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
863 ** SLOT-EXISTS-P now works on conditions, as well as structures
865 ** MAKE-LOAD-FORM now has the required methods on
866 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
867 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
868 keyword, and now implements the SLOT-NAMES argument.
869 ** methods with &OPTIONAL arguments no longer allow too many
870 arguments to be passed in the call without error.
871 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
872 option is consistent with the required arguments of the generic
873 function lambda list.
874 * bug fix: REQUIRE accepts a string designator. (Thanks to
876 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
877 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
878 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
879 * bug fix: The compiler now checks for duplicated variables in macro
881 * bug fix: SETQ on globals returns the correct value.
882 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
883 not a COMPILER-ERROR (followed by some other strange error on
884 choosing the CONTINUE restart).
885 * bug fix: make.sh and friends are now more consistent in the way that
886 they look for GNU "make".
888 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
889 * experimental native threads support (on x86 Linux >=2.4 only).
890 This is not compiled in by default: you need to add :SB-THREAD to
891 the target features. See the "Beyond ANSI" chapter of the manual
893 * fix for longstanding nonANSIism: The old distinction between
894 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
895 The return value from CL:FIND-CLASS is now a CLOS class, and
896 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
897 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
898 * An interface to the MetaObject Protocol, as described in Kiczales,
899 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
900 MIT Press, 1991, is available from the SB-MOP package.
901 * incompatible change: the SB-PCL package should now be considered
902 a private implementation detail, and no longer a semi-private MOP
904 * minor incompatible change: due to rearrangement for threads, the
905 control stack and binding stack are now allocated at arbitrary
906 addresses instead of being hardcoded per-port. Users affected by
907 this probably have to be doing advanced things with shared
908 libraries, and will know who they are.
909 * minor incompatible change: Previously, all --eval forms used were
910 processed with READ before any of them were processed with EVAL.
911 Now each --eval form is processed with both READ and EVAL before
912 the next --eval form is processed. (Thus package operations like
913 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
914 work as the user might reasonably expect.)
915 * minor incompatible change: *STANDARD-INPUT* is now only an
916 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
918 * minor incompatible change: Y-OR-N-P is now character-oriented,
919 not line oriented. Also, YES-OR-NO-P now works without errors.
920 (thanks to Antonio Martinez)
921 * sb-aclrepl module improvements: an integrated inspector, added
922 repl features, and a bug fix to :trace command.
923 * Known functions, which cannot be open coded by the backend, are
924 considered to be able to check types of their arguments. (fixing
925 a bug report by Nathan J. Froyd)
926 * fixed a bug in computing method discriminating functions: It is
927 now possible to define methods specialized on classes which have
928 forward-referenced superclasses. (thanks to Gerd Moellmann)
929 * fixed evaluation order in optional entries (reported by Gilbert
931 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
932 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
933 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
934 required-by-AMOP NAME argument, as well as CLASS and
935 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
936 * fixed bug 20: DEFMETHOD can define methods using names that are
937 not the proper names of classes to designate class specializers.
938 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
939 Unix file descriptors, instead of blowing up. (thanks to Antonio
941 * Garbage collection refactoring: user-visible change is that a
942 call to the GC function during WITHOUT-GCING will not do garbage
943 collection until the end of the WITHOUT-GCING. If you were doing
944 this you were probably losing anyway.
945 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
946 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
947 TYPEP the latter but not the former.
948 * The compiler issues a full WARNING for calls to undefined functions
949 with names from the CL package.
950 * MAP-INTO for a vector destination is open coded. (reported by
951 Brian Downing on c.l.l)
952 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
953 documentation string.
954 * fixed some bugs revealed by Paul Dietz' test suite:
955 ** COPY-ALIST now signals an error if its argument is a dotted
957 ** Condition slots are now accessed more correctly in the presence
958 of multiple initargs for a given slot.
959 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
960 exclude restarts of the same name associated with a different
962 ** DEFCLASS of forward-referenced classes with another
963 forward-referenced class in the superclasses list no longer
965 ** Condition slots are now initialized once each, not multiple
966 times. (thanks to Gerd Moellmann)
967 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
969 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
970 ** NSET-EXCLUSIVE-OR does not return extra elements when its
971 arguments contain duplicated elements.
972 ** RESTART-CASE understands local macros.
973 ** RESTART-CASE associates exactly its own restarts with a condition.
974 ** ENDP in safe mode checks its argument to be of type LIST.
975 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
976 in question is unbound.
977 ** Optimized MAKE-INSTANCE functions no longer cause internal
978 assertion failures in the presence of duplicate initargs.
979 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
980 ** Side-effectful :DEFAULT-INITARGS have their side-effects
981 propagated even in the ctor optimized implementation of
983 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
985 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
986 * a better implementation of SXHASH on (simple) bit vectors,
987 measured both in execution speed and in distribution of results
988 over the positive fixnums, has been installed. Likewise, a better
989 implementation of EQUAL for simple bit vectors is now available.
990 * fixed CEILING optimization for a divisor of form 2^k.
991 * fixed bug 240 (emitting extra style warnings "using the lexical
992 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
994 * fixed SXHASH, giving different results for NIL depending on type
995 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
996 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
997 variable. (found by Rolf Wester)
998 * SBCL does not ignore type declarations for special
999 variables. (reported by rif on c.l.l 2003-03-05)
1000 * some bug fixes in contrib/sb-aclrepl/
1001 * fixed some bugs revealed by Paul Dietz' test suite:
1002 ** a bug in the CONS type specifier, whereby the CAR and CDR
1003 types got intertwined, has been fixed;
1004 ** the type system is now able to reason about the interaction
1005 between INTEGER and RATIO types more completely;
1006 ** APPEND, [N]REVERSE and NRECONC check that those their
1007 arguments, which must be proper lists, are really so;
1008 ** An array specialized to be unable to hold elements has been
1009 implemented, as required -- yes, really -- by ANSI;
1010 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1011 on malformed property lists;
1013 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1014 * incompatible packaging change: in line with Unix convention,
1015 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1016 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1017 * REQUIRE and PROVIDE are now optionally capable of doing something
1018 useful. See the documentation string for REQUIRE.
1019 * infrastructure for a managed SBCL contrib system: contributed
1020 modules in this release include:
1021 ** the ASDF system definition facility;
1022 ** an interface to the BSD Sockets API;
1023 ** an ACL-like convenience interface to the repl;
1024 (thanks to Kevin Rosenberg)
1025 ** an implementation of ROTATE-BYTE, with efficient implementation
1027 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1028 gives a closer approximation to the right answer than previously.
1029 (thanks to Raymond Toy)
1030 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1031 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1032 optional environment arguments, as required by ANSI.
1033 * fixed bugs in other functions taking environment objects, allowing
1034 calls with an explicit NIL environment argument to be compiled
1036 * fixed bug 228: primary return values from
1037 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1038 COMPILE or FUNCTION.
1039 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1040 :TYPE will no longer signal errors on innocuous objects.
1041 * fixed bug 231b: SETQ is better at respecting type declarations in
1042 the lexical environment.
1043 * fixed a bug in DEFCLASS: classes named by symbols with no or
1044 unprintable packages can now be defined.
1045 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1046 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1047 * fixed compiler failure related to checking types of functions.
1048 (reported by Robert E. Brown)
1049 * the compiler is now much more consistent in its error-checking
1050 treatment of bounding index arguments to sequence functions: in
1051 (SAFETY 3) code, errors will be signalled in almost all cases if
1052 invalid sequence bounding indices are passed to functions defined
1053 by ANSI to operate on sequences.
1054 * fixed a bug in the build procedure: documentation of SBCL-specific
1055 packages is now preserved and available in the final Lisp image.
1056 * lifted FDEFINITION lookup out of loops in the implementation of
1057 many list operations. (thanks to Robert E. Brown)
1058 * fixed a bug in the reader: the #n# reader macro now works for
1059 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1060 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1061 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1062 E. Brown and Tony Martinez respectively)
1063 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1064 SLOT-BOUNDP: in particular, it is now able to optimize them much
1065 better, and is now not vulnerable to having packages renamed.
1066 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1067 always called when a slot is not present in an instance. (thanks
1069 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1070 longer signal an error on unreading a character following EOF on
1071 the previous constituent stream. (thanks to Tony Martinez)
1072 * fixed some bugs revealed by Paul Dietz' test suite:
1073 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1074 not just nonnegative fixnums;
1075 ** the logical bit-array operators such as BIT-AND now accept an
1076 explicit NIL for their "opt-arg" argument (to indicate a
1077 freshly-consed result bit-array);
1078 ** ELT now signals an error on an invalid sequence index in safe
1080 ** the type system is now cleverer about negations of numeric
1081 types, and consequently understands the BIGNUM and RATIO types
1083 ** the type system is now cleverer about the interaction between
1084 INTEGER and RATIO types: while bugs still remain, many more
1085 cases are accurately computed;
1086 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1087 if it is in the last clause;
1088 ** CONSTANTLY now correctly returns a side-effect-free function in
1090 ** DECLARE is no longer treated as a special-operator; in
1091 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1092 * incremented fasl file version number due to the change in the
1093 DEFSTRUCT-SLOT-DESCRIPTION structure.
1095 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1096 * minor incompatible change: code processed by the "interpreter" or
1097 EVAL now has a compilation optimization policy of (DEBUG 2)
1098 (changed from (DEBUG 1)) to improve debuggability of interactive
1099 development, and to allow the use of the debug RETURN command in
1101 * an experimental implementation of the RETURN command for the
1102 debugger has been included. (thanks to Frederik Kuivinen)
1103 * fixed bug 62: constraints were not propagated into a loop.
1104 * fixed bug in embedded calls of SORT (reported and investigated by
1106 * fixed some bugs revealed by Paul Dietz' test suite:
1107 ** printing and reading of arrays with some dimensions having
1108 length 0 (thanks to Gerd Moellmann);
1109 ** BOA constructor with &AUX argument without a default value does
1110 not cause a type error;
1111 ** CONSTANTP now returns true for all self-evaluating objects.
1113 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1114 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1115 accessors that are related by inheritance, as specified in the
1116 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1117 Valtteri Vuorikoski)
1118 * The compiler is now able to inline functions that were defined in
1119 a complex lexical environment (e.g. inside a MACROLET).
1120 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1122 * Support for the upcoming FreeBSD-5.0 release has been included.
1123 (thanks to Dag-Erling Smorgrav)
1124 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1125 effect when it is not in a toplevel context.
1126 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1127 stand a better chance of being correct. (thanks to Gerd
1129 * fixed bug in COERCE, which now signals an error on coercing a
1130 rational to a bounded real type which excludes the expected
1132 * The compiler is now able to derive types more accurately from the
1133 COERCE and COMPILE functions.
1134 * fixed bug 223: functional binding is considered to be constant
1135 only for symbols in the CL package.
1136 * fixed bug 231: SETQ did not check the type of a variable being set
1137 (reported by Robert E. Brown)
1138 * A new optimization for MAKE-INSTANCE has been included, fixing
1139 various bugs (including relating to :ALLOCATION :CLASS slots and
1140 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1142 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1143 ** As required by ANSI, LOOP now disallows anonymous collection
1144 clauses such as COLLECT I in conjunction with aggregate boolean
1145 clauses such as THEREIS (= I 1);
1146 ** LOOP now signals an error when any variable is reused in the
1147 same loop (including the potentially useful construct analogous
1148 to WITH A = 1 WITH A = (1+ A);
1149 ** IT is only a special loop symbol within the first clause of a
1150 conditional loop clause;
1151 ** LOOP with a typed iteration variable over a hashtable now
1152 signals a type error iff it should.
1153 * fixed some other bugs revealed by Paul Dietz' test suite:
1154 ** FILE-STREAM now names the class previously known as FD-STREAM;
1155 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1156 argument) no longer signals an error;
1157 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1158 of the slot symbol, rather than using the current package
1159 ((:CONC-NAME "") continues to intern the slot's name in the
1161 * incremented fasl file version number, because of the incompatible
1162 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1163 of the new implementation of DEFINE-COMPILER-MACRO.
1165 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1166 * Support for building SBCL for MIPS platforms running in
1167 little-endian mode has now been checked in, and basic
1168 functionality on said platforms verified.
1169 * minor incompatible change: PCL now records the pathname of a file
1170 in which methods and the like are defined, rather than its
1172 * minor incompatible change: TRUENAME now considers the truename of
1173 a file naming a directory to be the pathname with :DIRECTORY
1174 component indicating that directory.
1175 * minor incompatible change: a NAMED clause in the extended form of
1176 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1177 reason for the previous behaviour is unclear.
1178 * more systematization and improvement of CLOS and MOP conformance
1179 in PCL (thanks to Gerd Moellman and Pierre Mai):
1180 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1182 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1183 primary methods with no specializers;
1184 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1186 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1187 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1188 FORWARD-REFERENCED-CLASSes; error reporting on
1189 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1191 ** SXHASH on CLOS instances now uses a slot internal to the
1192 instance to return different numbers on distinct instances,
1193 while preserving the same return value through invocations of
1195 ** DEFMETHOD signals errors when methods with longer incongruent
1196 lambda lists are added to generic functions;
1197 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1198 CLASS, as specified in AMOP;
1199 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1200 on the order returned by the primary method for classes of
1201 class STANDARD-CLASS;
1202 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1203 * fixed some bugs shown by Paul Dietz' test suite:
1204 ** DOLIST puts its body in TAGBODY;
1205 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1207 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1208 value producing form;
1209 ** if more variables are given to PROGV than values, extra
1210 variables are bound and made to have no value;
1211 ** NSUBSTITUTE on list arguments gets the right answer with
1213 ** ELT signals an error of type TYPE-ERROR when the index argument
1214 is not a valid sequence index;
1215 ** LOOP signals (at macroexpansion time) an error of type
1216 PROGRAM-ERROR when duplicate variable names are found;
1217 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1218 ** FUNCALL of special-operators now cause an error of type
1220 ** PSETQ now works as required in the presence of side-effecting
1221 symbol-macro places;
1222 ** NCONC accepts any object as its last argument;
1223 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1225 ** loop-for-as-package does not require a package to be explicitely
1227 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1229 * fixed bug 166: compiler preserves "there is a way to go"
1230 invariant when deleting code.
1231 * fixed bug 172: macro lambda lists with required arguments after
1232 &REST arguments now cause an error to be signalled. (thanks to
1234 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1235 bug. (thanks to Matthew Danish)
1236 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1238 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1239 arguments are lexically rebound. (thanks to Gerd Moellmann and
1241 * fixed bug 194: error messages are now more informative when there
1242 is no primary method applicable in a call to a generic
1243 function. (thanks to Gerd Moellmann)
1244 * fixed bug in command line argument checking (thanks to Julian
1246 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1247 (thanks to Matthew Danish)
1248 * incremented fasl file version number, because of the
1249 SXHASH-related changes in the layout of CLOS data structures
1251 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1252 * minor incompatible change: The runtime (the Unix executable named
1253 "sbcl") is now much pickier about the .core files it will load.
1254 Essentially it now requires .core files to descend from the same
1255 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1256 as the runtime does. (The intent is to prevent the crashes which
1257 can occur, and which can even be reported as mysterious failures,
1258 when people patch the sources or change the build parameters
1259 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1260 sbcl and .core files.)
1261 * fixed bug: VALUES-LIST is no longer optimized away.
1262 * fixed bug 142: The FFI conversion of C string values to Lisp
1263 string values no longer conses excessively. (thanks to Nathan
1264 Froyd porting Raymond Toy's fix to CMU CL)
1265 * began to systematize and improve MOP conformance in PCL (thanks to
1266 Nathan Froyd, Gerd Moellman and Pierre Mai):
1267 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1269 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1270 ** FINALIZE-INHERITANCE is now called on class finalization;
1271 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1272 argument precedence order.
1273 * fixed bug 202: The compiler no longer fails on functions whose
1274 derived types contradict their declared type.
1275 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1276 so it can be non-toplevel.
1277 * The fasl file version number has changed (because of the new
1278 implementation of DEFMACRO).
1279 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1280 safe code, that any length requirement by their type-specifier
1281 argument is valid. The exceptions to this are described in bug
1283 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1284 functions now check the types of their inputs as required by ANSI.
1285 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1286 introduced symbol is DECLAREd to be SPECIAL.
1287 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1288 * fixed bug: :COUNT argument to sequence functions may be negative
1289 * fixed bug: body of DO-SYMBOLS may contain declarations
1290 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1291 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1292 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1293 symbol macro only once
1294 * fixed printing of call frame when argument list is unavailable
1295 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1296 * compiler no longer signals WARNING on unknown keyword
1299 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1300 * A beta-quality port to the mips architecture running Linux,
1301 based on the old CMUCL backend, has been made. It has been tested
1302 on a big-endian kernel, and works sufficiently well to be able to
1303 rebuild itself; it has not been tested in little-endian mode.
1304 * fixed an inconsistency between gencgc.c and purify.c which made
1305 dumping/loading .core files unreliable
1306 * fixed bug 120a: The compiler now deals correctly with IFs where
1307 the consequent is the same as the alternative, instead of
1308 misderiving the return type. (thanks to Alexey Dejneka)
1309 * fixed bug 113: Logical pathnames are now dumpable (the logical
1310 host is resolved at load-time, throwing an error if it is not
1312 * fixed bug 174: FORMAT's error message is slightly clearer when a
1313 non-printing character is used in a format directive.
1314 * fixed several bugs in compiler checking of type declarations, i.e.
1315 violations of the Python "declarations are assertions" principle
1316 (thanks to Alexey Dejneka)
1317 * fixed several bugs in PCL's error checking (thanks to Gerd
1319 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1321 * fixed bug in compilation of functions as first class values
1322 (thanks to Antonio Martinez-Shotton)
1323 * The compiler's handling TYPE-ERRORs which it can prove will
1324 inevitably happen at runtime has been cleaned up and corrected
1325 in several ways. (thanks to Alexey Dejneka)
1326 * improved argument type checking for various basic arithmetic
1327 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1328 lavished on them in the past that they can be compiled in many
1329 ways in different special cases
1330 * fixed bug 181: compiler checks validity of user supplied type
1332 * cleaned up code flushing in optimization: Function calls which
1333 should signal errors for safety purposes (e.g. which ANSI says
1334 should signal errors when their arguments are of incorrect type)
1335 are no longer optimized away.
1336 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1337 * incremented fasl file version number, because changes in the
1338 implementation of sequence functions like COERCE caused
1339 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1340 in old inline expansions) to become undefined. (Actually these
1341 changes were later undone, so we might very well be binary
1342 compatible with 0.7.7 after all, but leaving the version number
1343 incremented seemed like the simplest and most conservative
1346 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1347 * An alpha-quality port to the parisc architecture running Linux,
1348 based on the old CMUCL backend, has been made. This, even more so
1349 than the other backends, should be considered still a work in
1350 progress; known problems include that the Linux kernel in 64-bit
1351 mode does not propagate the correct sigcontext structure to
1352 userspace, and consequently SBCL on a parisc64 kernel will not
1354 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1355 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1356 suppressed.) Also now that the compiler is looking at declarations
1357 in the environment, it checks optimization declarations as well,
1358 and suppresses inlining when (> DEBUG SPEED).
1359 * More fixes have been made to treatment of floating point exception
1360 treatment and other Unix signals. In particular, floating point
1361 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1362 * The detection and handling of control stack exhaustion (infinite
1363 or very deeply nested recursion) has changed. Stack exhaustion
1364 detection is now done by write-protecting pages at the OS level
1365 and applies at all optimization settings; when found, a
1366 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1367 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1368 be caught using IGNORE-ERRORS.
1369 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1370 correctly when transforming recursive LABELS functions to LETs.
1371 (thanks to Alexey Dejneka)
1372 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1373 that are names of constants or global variables.
1374 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1375 alien routines with docstrings.
1376 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1377 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1379 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1380 (thanks to Eric Marsden)
1381 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1382 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1383 * Bug fix: The SPARC backend can now compile functions involving
1384 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1385 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1386 to a functions expecting a PATHNAME-DESIGNATOR.
1387 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1388 lambda lists. (thanks to Alexey Dejneka)
1389 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1390 functions so that EXPORTing the name of the class doesn't cause
1391 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1392 to Antonio Martinez for reporting this)
1393 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1394 constant-folded without failing an assertion. (thanks to Einar
1395 Floystad Dorum for reporting this)
1396 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1397 types are now dealt with more correctly by the compiler.
1398 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1399 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1400 behaviour analogous to RENAME-FILE. This puts its behaviour more
1401 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1403 * The fasl file version number has changed again. (because of the
1404 bug fix involving the names of PCL MAKE-INSTANCE functions)
1406 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1407 * bug fix: Floating point exceptions are treated much more
1408 consistently on the x86/Linux and PPC/Linux platforms.
1409 * Array initialization with :INITIAL-ELEMENT is now much faster for
1410 cases when the compiler cannot open code the array creation, but
1411 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1412 array accesses have also seen a speed increase.
1413 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1414 specified. (This was at the root of some bad interactions between
1415 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1417 * bug fix: Internal error arguments for undefined functions are now
1418 computed correctly on the PPC/Linux platform.
1419 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1420 Raymond Toy's patch for CMU CL)
1421 * Support for the Solaris 9 operating environment has been included
1422 (thanks to Daniel Merritt)
1423 * A very ugly but hopefully complete draft of the missing FFI chapter
1424 of the manual has been created by reformatting the corresponding
1425 CMU CL manual chapter into (currently very ugly and incoherent)
1426 DocBook and bringing it up to date for SBCL behavior. Thus, the
1427 manual is now essentially complete, at least by my extreme
1428 once-and-only-once standards, whereby it's acceptable to refer to
1429 the doc strings of SB-EXT functions as the primary documentation.
1430 * The fasl file version number has changed again, due to cleanup of
1431 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1432 is no longer a static symbol.)
1434 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1435 * SBCL now builds with OpenMCL (version 0.12) as the
1436 cross-compilation host; also, more progress has been made toward
1437 bootstrapping under CLISP.
1438 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1440 * bug 158 fixed: The compiler can now deal with integer loop
1441 increments different from 1; fixing this turned out also to fix
1443 * bug 169 fixed: no more bogus warnings about using lexical bindings
1444 despite the presence of perfectly good SPECIAL declarations (thanks
1445 to David Lichteblau)
1446 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1447 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1448 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1450 * bug 180 fixed: Method combination specifications no longer ignore
1451 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1452 * bug fix: Structure type predicate functions now check their argument
1453 count as they should.
1454 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1455 correctly. (thanks to Pierre Mai)
1456 * minor incompatible change: The --noprogrammer option is deprecated
1457 in favor of the new --disable-debugger option, which is very similar.
1458 (The major difference is that it takes effect at a slightly different
1459 time at startup, causing handling of errors in --sysinit and
1460 --userinit files will be affected differently.) The
1461 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1462 been added to allow this functionality to be controlled from ordinary
1463 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1464 maintainers, who might want to run non-interactive scripts to
1465 build SBCL cores which will later be used interactively.)
1466 * minor incompatible change: The LOAD function no longer, when given
1467 a wild pathname to load, loads all files matching that pathname.
1468 Instead, an error of type FILE-ERROR is signalled.
1470 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1471 * bug 147 fixed: The compiler preserves its block link/count
1472 invariants more correctly now so that it doesn't crash. (thanks
1474 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1476 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1477 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1478 no-value-for-this-character as NIL (as specified by ANSI).
1479 * HOST-NAMESTRING on physical pathnames now returns a string that is
1480 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1481 (thanks to Christophe Rhodes)
1482 * The Alpha port handles icache flushing more correctly. (thanks to
1484 * More progress has been made toward bootstrapping under CLISP. (thanks
1485 to Christophe Rhodes)
1486 * The fasl file format has changed again, because dynamic loading
1487 on OpenBSD (which has non-ELF object files) motivated some cleanups
1488 in the way that foreign symbols are transformed and passed around.
1489 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1490 is no longer treated as whitespace by the reader, but instead as
1491 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1492 |A
\7fB|, instead of A as it used to.
1494 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1495 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1496 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1497 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1498 this as of 0.7.1.45, but was left out of the previous news section
1499 (thanks to Dan Barlow)
1500 * SBCL now runs on the Solaris operating system on SPARC architectures
1501 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1502 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1503 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1504 * SPARC backend cleanups, allowing builds of cores optimized for V8
1505 and V9 SPARCS, and also emission of code targeted to a particular
1506 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1508 * SBCL is closer to bootstrapping under CLISP, thanks to various
1509 fixes by Christophe Rhodes.
1510 * The fasl file format has changed again, to allow the compiler's
1511 INFO database to support symbol macros.
1512 * The user manual (in doc/) is formatted into HTML more nicely.
1513 (thanks to coreythomas)
1514 * The system is smarter about SUBTYPEP relationships, especially
1515 those involving NOT types (including types such as ATOM which are
1516 represented internally using NOT types). Thus SUBTYPEP is less
1517 likely to return (VALUES NIL NIL) in general, and in particular
1518 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1520 * The fasl file format has changed again, because the internal
1521 representation of types now includes a new slot to support the new
1522 SUBTYPEP-of-NOT-types logic.
1523 * (not a change in the main branch of SBCL, but a related prototype
1524 which can hopefully be merged into the main branch of SBCL in the
1525 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1526 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1527 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1528 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1529 dispatch macro character combinations. (thanks to Alexey Dejneka)
1530 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1531 that superficially look like logical namestrings correctly.
1532 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1534 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1535 default supplied-p) for &optional and &key arguments. (thanks to
1538 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1539 * incompatible change: The compiler is now less aggressive about
1540 tail call optimization, doing it only when (> SPACE DEBUG) or
1541 (> SPEED DEBUG). (This is an incompatible change because there are
1542 programs which relied on the old CMU-CL-style behavior to optimize
1543 away their unbounded recursion which will now die of stack overflow.)
1544 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1545 for non-GENCGC systems has been increased to 20M (since that
1546 seems much closer to the likely performance optimum for modern
1547 systems than the old 4M value was)
1548 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1549 and *DEBUG-PRINT-LEVEL*
1550 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1551 of CMU CL's support for SPARC, and various endianness and other
1552 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1553 * new syntactic sugar for the Unix command line: --load foo.bar is now
1554 an alternate notation for --eval '(load "foo.bar")'.
1556 ** The system now detects stack overflow and handles it gracefully,
1557 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1558 optimization settings. (This is a good thing in general, and
1559 its introduction in this version should be particularly timely
1560 for anyone whose code fails because of suppression of tail
1562 ** The system now hunts for the C variable "environ" in a more
1563 devious way, to avoid segfaults when the C library version differs
1564 between compile time and run time. (thanks to Christophe Rhodes)
1565 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1566 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1567 ** The compiler no longer issues bogus style warnings for undefined
1568 classes in the same source file as the DEFCLASSes which defined
1569 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1571 ** fixes in CONDITION class precedence list for undefined function
1572 errors (thanks to Alexei Dejneka)
1573 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1574 correctly. (thanks to Dan Barlow)
1575 ** portability fixes aiming at bootstrapping under CLISP (thanks
1576 to Dave McDonald and Christophe Rhodes)
1577 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1578 ** fixes in type translation and and type inference (thanks to
1580 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1581 ** various fixes in the new ports (thanks to Dan Barlow)
1582 * several changes related to debugging:
1583 ** suppression of tail recursion, as noted above
1584 ** stack overflow detection, as noted above
1585 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1586 is now the default. (For some time encapsulation has been more
1587 reliable than the breakpoint-based :ENCAPSULATE NIL
1588 implementation, at least on X86 systems; and I just noticed that
1589 encapsulation also seems closer to the spirit of the ANSI
1592 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1594 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1595 up properly again. (There was a packaging bug in 0.7.0 which
1596 left their definitions in SB-SYS::LOAD-FOREIGN and
1597 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1598 vital for most things which interface to C-level interfaces,
1599 like extensions working with sockets or databases or
1600 Perl-compatible regexes or whatever, and the need to fix
1601 this bug is the main reason that 0.7.1 was released so
1603 ** DEFGENERIC is now choosier about the methods it redefines, so that
1604 reLOADing a previously-LOADed file containing DEFGENERICs does
1605 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1606 cycle now works as it should. (thanks to Alexey Dejneka)
1607 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1608 fixed by Christophe Rhodes. (He actually submitted this patch
1609 months ago, and I delayed until after 0.7.0.)
1610 ** Bug 111 (internal compiler confusion about runtime checks on
1611 FUNCTION types) was fixed by Alexey Dejneka.
1612 * Some internal cleanups (getting rid of variables which aren't
1613 needed now that the byte interpreter is gone) caused the fasl
1614 file format number to change again.
1616 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1617 * major incompatible change: The default fasl file extension, i.e. the
1618 default extension for files produced by COMPILE-FILE, has changed
1619 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1621 ** There are many changes in the implementation of the compiler.
1622 SBCL is now essentially a compiler-only implementation of ANSI
1623 Common Lisp. EVAL still "interprets" a few special cases, but
1624 almost all the interesting cases are handled by creating
1625 a LAMBDA expression, calling COMPILE on it, then calling
1626 FUNCALL on the result.
1627 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1628 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1629 Since the code is newer, there might still be some new bugs
1630 (though not as many as before Martin Atzmueller's fixes:-). But
1631 the new code is substantially simpler and clearer, and hopefully
1632 any remaining bugs will be simpler, less fundamental, and more
1633 fixable then the bugs in the old code.
1634 ** The revised compiler is still a little unsteady on its feet.
1636 *** The debugging information it produces (particularly the names
1637 of FUNCTION objects) is sometimes much less useful than what
1638 the old compiler produced.
1639 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1640 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1641 within a MACROLET) has been temporarily weakened.
1642 ** There are new compiler optimizations for various functions:
1643 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1644 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1645 *** the math functions TRUNCATE, FLOOR, and CEILING
1646 *** the function-of-all-trades COERCE
1647 Mostly these should be transparent, but there's one
1648 potentially-annoying problem (bug 117): when the compiler
1649 inline-expands a function and does type analysis on the result,
1650 it can create control paths which have type mismatches, and
1651 when it can't prove that those control paths aren't taken,
1652 it will issue WARNINGs about the type mismatches. This is
1653 a particular problem in practice for the new sequence functions.
1654 It's not clear how this should be fixed, and for now, a
1655 workaround is given in the entry for 117 in the BUGS file.
1656 ** (Because of the interaction between the two previous items --
1657 occasional inlining problems and new inline expansions -- some
1658 of the new sequence function optimizations won't really kick in
1659 completely until debugging information, and then inlining, are
1660 straightened out in some future version.)
1661 * minor incompatible changes:
1662 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1663 DIRECTORY no longer implicitly promotes NIL slots of its
1664 pathname argument to :WILD. In particular, when you ask for the
1665 contents of a directory (which you used to be able to do without
1666 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1667 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1668 ** changes in behavior that ANSI explicitly defines to be
1669 implementation dependent:
1670 *** The new compiler-only implementation still conforms with ANSI,
1671 but acts a little different than before. Besides the obvious
1672 changes in performance tradeoffs (that the cost per form passed
1673 to EVAL has gone up, and the cost per form executed by EVAL
1674 has gone down), the behavior of the system changes a little
1675 because there are no longer any interpreted function objects.
1676 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1677 e.g. doing COMPILE on the output of interactive DEFUN is
1679 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1681 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1682 has changed from (:CL) to NIL.
1683 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1684 e.g. #\Newline and #\Tab, don't have names specified in the
1685 ANSI Common Lisp standard, is now based on their ASCII symbolic
1686 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1687 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1688 are no longer used for output.
1689 ** changes in internal implementation constants:
1690 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1691 4 million. (If your application spends a lot of time GCing and
1692 you have a lot of RAM, you might want to experiment with
1693 increasing it even more.)
1694 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1695 However, almost all old code should still continue to work without
1696 immediate update, as SB-C-CALL is now a (deprecated) nickname
1698 ** Old operator names in the style DEF-FOO are now deprecated in
1699 favor of new corresponding names DEFINE-FOO, for consistency with
1700 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1701 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1702 internal symbols, but a few supported extensions like
1703 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1704 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1705 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1706 etc. as you get deeper into recursive calls to the debugger
1707 command loop, instead of the old "5]", "5]]", "5]]]"
1708 sequence. (I was motivated to do this when squabbles between
1709 ILISP and SBCL left me very deeply nested in the debugger. In the
1710 short term, this change will probably provoke more ILISP/SBCL
1711 squabbles, but hopefully it will be an improvement in the long run.)
1712 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1713 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1714 function, since declaiming return types involving aliens is
1715 (1) annoyingly messy to do by hand and (2) vital to efficient
1716 compilation of code which calls such functions.
1717 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1718 longer reexported by the SB-EXT package. They're solely useful
1719 for alien code, so it seems more logical that you should get
1720 them from the SB-ALIEN package, not in SB-EXT.
1721 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1722 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1723 features. Instead, the code that they used to control is always
1724 built into the system.
1725 * many other bug fixes
1726 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1727 advantage of the new EVAL-WHEN stuff and to clean them up in
1728 general, and they are now more ANSI-compliant in a number of
1729 ways. Martin Atzmueller is responsible for a lot of this.
1730 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1732 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1733 *** correct ERROR type for various file operations
1734 *** some fixes for Lisp streams
1735 *** DEFMETHOD syntax checking
1736 *** changing old weird representation of debug information as
1737 strings (which, among their other deficiencies, don't transform
1738 correctly when you rename packages, and don't change their
1739 print representation when you change things like *PACKAGE*
1740 and *PRINT-LENGTH*) to symbols and lists of symbols
1741 He also made several improvements and fixed several bugs in DESCRIBE.
1742 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1744 *** misbehavior of WRITE-STRING/WRITE-LINE
1745 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1746 and several other LOOP problems as well
1747 *** DIRECTORY when similar filenames are present
1748 *** DEFGENERIC with :METHOD options
1749 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1750 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1751 *** argument ordering in FIND with :TEST option
1752 *** mishandled package designator argument in APROPOS-LIST
1753 *** various problems in the backquote readmacro
1754 *** a bug in APROPOS
1755 *** probably some others that I'm not describing very well here,
1756 since the CVS log documents them by reference to sbcl-devel
1757 messages, and the SourceForge archives aren't working well.:-(
1758 ** Dan Barlow improved the Alpha port (and is making progress on the
1759 PPC port, for those of you who think different).
1760 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
1761 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
1762 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
1763 continues to work on the SPARC port (for those of you in a position
1764 to look down upon our little PC-compatible boxes from a great height).
1765 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
1766 and uses that copy, rather than the current dynamic value, when
1767 it's trying to decide whether to truncate output. Thus e.g.
1768 (let ((*print-lines* 50))
1769 (pprint-logical-block (stream nil)
1771 (let ((*print-lines* 8))
1772 (print (aref possiblybigthings i) stream)))))
1773 should now truncate the logical block only at 50 lines, instead of
1774 often truncating it at 8 lines, as it did before.
1775 * The doc/cmucl/ directory, containing old CMU CL documentation
1776 from the time of the fork, is no longer part of the base system.
1777 SourceForge has shut down its anonymous FTP service, and with it
1778 my original plan for distributing the old CMU CL documentation
1779 there. For now, if you need these files you can download an old
1780 SBCL source release and extract them from it.
1781 * The fasl file version number changed again, for dozens of reasons,
1782 some of which are apparent above.
1784 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
1785 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
1786 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
1787 patch, so that DISASSEMBLE output is much nicer.
1788 * The code in the SB-PROFILE package now seems reasonably stable.
1789 I still haven't decided what the final interface should look like
1790 (I'd like PROFILE to interact cleanly with TRACE, since both
1791 facilities use function encapsulation) but if you have a need
1792 for profiling now, you can probably use it successfully with
1793 the current CMU-CL-style interface.
1794 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
1795 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
1796 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
1797 link now returns the dangling link itself, and for similar
1798 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
1799 link itself. (In these cases the old code signalled an error and
1800 looped endlessly, respectively.) Thus, DIRECTORY now works even
1801 in the presence of dangling and cyclic symbolic links.
1802 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
1803 is now a supported extension again, since the consensus on
1804 sbcl-devel was that it can be useful for ordinary development
1805 work, not just for debugging SBCL itself.
1806 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
1807 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
1808 that currently-defined functions might be redefined later with
1809 different return types.
1810 * Hash tables can be printed readably, as inspired by CMU CL code
1811 of Eric Marsden and SBCL code of Martin Atzmueller.
1812 * better error handling in CLOS method combination, thanks to
1813 Martin Atzmueller porting Pierre Mai's CMU CL patches
1814 * more overflow fixes for >16Mbyte I/O buffers
1815 * A bug in READ has been fixed, so that now a single Ctrl-D
1816 character suffices to cause end-of-file on character streams.
1817 In particular, now you only need one Ctrl-D at the command
1818 line (not two) to exit SBCL.
1819 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
1821 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
1822 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
1823 does the right thing.
1824 * The implementation of some type tests, especially for CONDITION
1825 types, is now tidier and maybe faster, due to CMU CL code
1826 originally by Douglas Crosher, ported by Martin Atzmueller.
1827 * Some math functions have been fixed, and there are new
1828 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
1829 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
1830 * (There are also some new optimizers in contrib/*-extras.lisp. Those
1831 aren't built into sbcl-0.6.13, but are a sneak preview of what's
1832 likely to be built into sbcl-0.7.0.)
1833 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
1834 CL, ported to SBCL by Martin Atzmueller)
1835 * DESCRIBE now gives more information in some cases. (Pierre Mai's
1836 patch to CMU CL, ported to SBCL by Martin Atzmueller)
1837 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
1838 * There's a new slam.sh hack to shorten the edit/compile/debug
1839 cycle for low-level changes to SBCL itself, and a new
1840 :SB-AFTER-XC-CORE target feature to control the generation of
1841 the after-xc.core file needed by slam.sh.
1842 * minor incompatible change: The ENTRY-POINTS &KEY argument to
1843 COMPILE-FILE is no longer supported, so that now every function
1844 gets an entry point, so that block compilation looks a little
1845 more like the plain vanilla ANSI section 3.2.2.3 scheme.
1846 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
1847 returns the number of bytes consed since the system started,
1848 rather than the number consed since the first time the function
1849 was called. (The new definition parallels ANSI functions like
1850 CL:GET-INTERNAL-RUN-TIME.)
1851 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
1852 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
1853 supported. Now DIRECTORY always does the abstract Common-Lisp-y
1854 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
1855 * Fasl file version numbers are now independent of the target CPU,
1856 since historically most system changes which required version
1857 number changes have affected all CPUs equally. Similarly,
1858 the byte fasl file version is now equal to the ordinary
1861 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
1862 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
1863 is no longer recognized. I apologize for this, because it was
1864 listed in SB-EXT as a supported extension, but I found that
1865 its existing behavior was poorly specified, as well as incorrectly
1866 specified, and it looked like too much of a mess to straighten it
1867 out. I have enough on my hands trying to get ANSI stuff to work..
1868 * many patches ported from CMU CL by Martin Atzmueller, with
1869 half a dozen bug fixes in pretty-printing and the debugger, and
1870 half a dozen others elsewhere
1871 * fixed bug 13: Floating point infinities are now supported again.
1872 They might still be a little bit flaky, but thanks to bug reports
1873 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
1874 as flaky as they were.
1875 * The --noprogrammer command line option is now supported. (Its
1876 behavior is slightly different in detail from what the old man
1877 page claimed it would do, but it's still appropriate under the
1878 same circumstances that the man page talks about.)
1879 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
1880 are now supported, and enabled by default. Thus, the compiler can
1881 handle many floating point and complex operations much less
1882 inefficiently. (Thus e.g. you can implement a complex FFT
1884 * The compiler now detects type mismatches between DECLAIM FTYPE
1885 and DEFUN better, and implements CHECK-TYPE more correctly, and
1886 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
1887 thanks to patches from Martin Atzmueller.
1888 * various fixes to make the cross-compiler more portable to
1889 ANSI-conforming-but-different cross-compilation hosts (notably
1890 Lispworks for Windows, following bug reports from Arthur Lemmens)
1891 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
1892 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
1893 fixed thanks to Pierre Mai's CMU CL patches.
1894 * fixes to keep the system from overflowing internal counters when
1895 it tries to use i/o buffers larger than 16M bytes
1896 * fixed bug 45a: Various internal functions required to support
1897 complex special functions have been merged from CMU CL sources.
1898 (When I was first setting up SBCL, I misunderstood a compile-time
1899 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
1900 * improved support for type intersection and union, fixing bug 12
1901 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
1902 more obscure bugs as well
1903 * some steps toward byte-compiling non-performance-critical
1904 parts of the system, courtesy of patches from Martin Atzmueller
1905 * Christophe Rhodes has made some debian packages of sbcl at
1906 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
1907 From his sbcl-devel e-mail of 2001-04-08 they're not completely
1908 stable, but are nonetheless usable. When he's ready, I'd be happy
1909 to add them to the SourceForge "File Releases" section. (And if
1910 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
1911 * new fasl file format version number (because of changes in
1912 internal representation of (OR ..) types to accommodate the new
1913 support for (AND ..) types, among other things)
1915 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
1916 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
1918 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
1920 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
1921 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
1922 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
1923 * DESCRIBE now works on structure objects again.
1924 * Most function call argument type mismatches are now handled as
1925 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
1926 know whether the function will be redefined before the call is
1927 executed. (The compiler could flag local calls with full WARNINGs,
1928 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
1929 it doesn't keep track of enough information to know whether calls
1930 are local in this sense.)
1931 * Compiler output is now more verbose, with messages truncated
1932 later than before. (There should be some supported way for users
1933 to override the default verbosity, but I haven't decided how to
1934 provide it yet, so this behavior is still controlled by the internal
1935 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
1936 src/compiler/ir1util.lisp.)
1937 * Fasl file format version numbers have increased again, because
1938 support for the Gray streams extension changes the layout of the
1939 system's STREAM objects.
1940 * The Gray subclassable streams extension now works, thanks to a
1941 patch from Martin Atzmueller.
1942 * The full LOAD-FOREIGN extension (not just the primitive
1943 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
1944 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
1945 but like most other programs, it defaults to copying the Unix
1946 environment from the original process instead of starting the
1947 new process in an empty environment.
1948 * Extensions which manipulate the Unix environment now support
1949 an :ENVIRONMENT keyword option which doesn't smash case or
1950 do other bad things. The CMU-CL-style :ENV option is retained
1951 for porting convenience.
1952 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
1953 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
1955 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
1957 * A patch from Martin Atzmueller seems to have solved the SIGINT
1958 problem, and as far as we know, signal-handling now works cleanly.
1959 (If you find any new bugs, please report them!)
1960 * The system no longer defaults Lisp source file names to types
1961 ".l", ".cl", or ".lsp", but only to ".lisp".
1962 * The compiler no longer uses special default file extensions for
1963 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
1964 seems to expect a single default extension for all compiled code,
1965 and there's no compelling reason to try to stretch the standard
1966 to allow two different extensions.) Instead, byte-compiled files
1967 default to the same extension as native-compiled files.
1968 * Fasl file format version numbers have increased again, because
1969 a rearrangement of internal implementation packages made some
1970 dumped symbols in old fasl files unreadable in new cores.
1971 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
1972 many fewer weird special cases.
1973 * Bug #17 (differing COMPILE-FILE behavior between logical and
1974 physical pathnames) has been fixed, and some related misbehavior too,
1975 thanks to a patch from Martin Atzmueller.
1976 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
1977 by Tim Moore, ported to SBCL by Martin Atzmueller.
1978 * Martin Atzmueller fixed several filesystem-related problems,
1979 including bug #36, in part by porting CMU CL patches, which were
1980 written in part by Paul Werkowski.
1981 * More compiler warnings in src/runtime/ are gone, thanks to
1982 more patches from Martin Atzmueller.
1983 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
1986 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
1988 * DESCRIBE now works on CONDITION objects.
1989 * The debugger now handles errors which arise when trying to print
1990 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
1992 * The build system now uses an additional file, customize-target-features.lisp,
1993 to allow local modifications to the target *FEATURES* list. (The point of
1994 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
1995 debugging features enabled, without having to worry about propagating it
1996 into everyone's system when I do a "cvs update".) When no
1997 customize-target-features.lisp file exists, the target *FEATURES* list
1998 should be constructed the same way as before.
1999 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2000 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2001 warn about behavior which is undefined under the ANSI spec. Note: This
2002 is specified by ANSI, but it's not very popular with programmers.
2003 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2004 macro in the SBCL sources for an example of a workaround which you
2005 might use to make portable ANSI-standard code which does what you want.
2006 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2007 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2008 when it's used as a non-toplevel form. (This is required in order
2009 to implement the DEFCONSTANT-EQX macro.)
2010 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2011 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2012 funny things, due to bugs in the implementation of EVAL-WHEN
2013 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2014 non-toplevel DEFCONSTANTs are uncommon.)
2015 * The core file version number and fasl file version number have been
2016 incremented, because the old noncompliant DEFCONSTANT behavior involved
2017 calling functions which no longer exist, and because I also took the
2018 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2019 * fixed bug 1 (error handling before read-eval-print loop starts), and
2020 redid debugger restarts and related debugger commands somewhat while
2022 ** The QUIT debugger command is gone, since it did something
2023 rather different than the SB-EXT:QUIT command, and since it never
2024 worked properly outside the main toplevel read/eval/print loop.
2025 Invoking the new TOPLEVEL restart provides the same functionality.
2026 ** The GO debugger command is also gone, since you can just invoke
2027 the CONTINUE restart directly instead.
2028 ** The TOP debugger command is also gone, since it's redundant with the
2029 FRAME 0 command, and since it interfered with abbreviations for the
2031 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2033 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2034 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2035 some porting work by Martin Atzmueller)
2036 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2037 with a boatload of patches which he ported to SBCL. Now that those
2039 ** The system tries to make sure that its low-priority messages
2040 are prefixed by semicolons, to help people who like to use
2041 syntax highlighting in their ILISP buffer. (This patch
2042 was originally due to Raymond Toy.)
2043 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2044 patches originally written by Raymond Toy.
2045 ** The compiler understands coercion between single-value and
2046 multiple-VALUES type expressions better, getting rid of some very
2047 weird behavior, thanks to patches originally by Robert MacLachlan
2048 and Douglas Crosher.
2049 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2050 lambda lists, thanks to a patch originally by Pierre Mai.
2051 ** The system no longer bogusly warns about "abbreviated type
2053 ** The compiler gets less confused by inlining and RETURN-FROM,
2054 thanks to some patches originally by Tim Moore.
2055 ** The system no longer hangs when dumping circular lists to fasl
2056 files, thanks to a patch originally from Douglas Crosher.
2057 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2058 undefined function error.
2059 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2060 in structure slot initforms) for now, documented workaround instead:-|
2061 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2062 * fixed bug 5: added stubs for various Gray stream functions called
2063 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2064 installed, at least appropriate type errors are generated
2065 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2066 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2067 consistently in DEFMETHOD forms.
2068 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2069 it doesn't seem to affect SBCL after all
2070 * The C runtime system now builds with better optimization and many
2071 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2073 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2075 * The system is now under CVS at SourceForge (instead of the
2076 CVS repository on my home machine).
2077 * The new signal handling code has been tweaked to treat register
2078 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2079 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2080 downstream, e.g. in debug-int.lisp, has implicit dependencies
2081 on the unsignedness of integer representation of machine words,
2082 and that caused the system to bomb out with infinite regress
2083 when trying to recover from type errors involving signed values,
2084 e.g. (BUTLAST '(1 2 3) -1).)
2085 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2086 (This was one of the bugs Peter Van Eynde reported back in July.)
2087 * The system now uses code inspired by Colin Walters' O(N)
2088 implementation of MAP (from the cmucl-imp@cons.org mailing
2089 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2090 inline the MAP operation, and there is more than one
2091 sequence argument to the MAP call (so that it can't just
2092 do ETYPECASE once and for all based on the type of the
2093 single sequence argument). (The old non-inline implementation
2094 of the general M-argument sequence-of-length-N case required
2095 O(M*N*N) time when any of the sequence arguments were LISTs.)
2096 * The QUIT :UNIX-CODE keyword argument has been renamed to
2097 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2098 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2100 * Raymond Wiker's patches to port dynamic loading from Linux to
2101 FreeBSD have been added.
2102 * The BUGS file is now more nearly up to date, thanks in large part
2103 to Martin Atzmueller's review of it.
2104 * The debugger now flushes standard output streams before it begins
2105 its output ("debugger invoked" and so forth).
2106 * The core version number and fasl file version number have both
2107 been incremented, because of incompatible changes in the layout
2109 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2110 used to not be called for a saved Lisp image.)
2111 * Martin Atzmueller's version of a patch to fix a compiler crash,
2112 as posted on sbcl-devel 13 September 2000, has been installed.
2113 * Instead of installing Martin Atzmueller's patch for the
2114 compiler transform for SUBSEQ, I deleted the compiler transform,
2115 and transforms for some similar consing operations.
2116 * A bug in signal handling which kept TRACE from working on OpenBSD
2118 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2119 away by constant folding
2120 * The system now defines its address space constants in one place
2121 (in the Lisp sources), and propagates them automatically elsewhere
2122 (through GENESIS and the sbcl.h file). Therefore, patching the
2123 address map is less unnecessarily tedious and error-prone. The
2124 Lisp names of address space constants have also been systematized.
2125 * CVS tags like dollar-Header-dollar have been removed from
2126 the sources, because they have never saved me trouble and
2127 they've been source of trouble working with patches and other
2128 diff-related operations.
2129 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2130 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2132 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2134 * The system has been ported to OpenBSD.
2135 * The system now compiles with a simple "sh make.sh" on the systems
2136 that it's supported on. I.e., now you no longer need to tweak
2137 text in the target-features.lisp-expr and symlinks in src/runtime/
2138 by hand, the make.sh takes care of it for you.
2139 * The system is no longer so grossly inefficient when compiling code
2140 involving vectors implemented as general (not simple) vectors (VECTOR T),
2141 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2142 which dares to use the various sequence functions on non-simple
2143 vectors, takes less of a performance hit.
2144 * There is now a primitive type predicate VECTOR-T-P
2145 to test for the (VECTOR T) type, so that e.g.
2146 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2147 can now be compiled with some semblance of efficiency. (The old code
2148 turned the type declaration into a full call to %TYPEP at runtime!)
2149 * AREF on (VECTOR T) is still not fast, since it's still compiled
2150 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2151 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2152 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2153 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2154 without making full calls to SUBTYPEP at runtime.
2155 (Some analogous efficiency issues for non-simple vectors specialized to
2156 element types other than T, or for non-simple multidimensional arrays,
2157 have not been addressed. They could almost certainly be handled the
2158 same way if anyone is motivated to do so.)
2159 * The changes in array handling break binary compatibility, so
2160 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2161 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2162 instead of (VALUES T T).
2163 * By following the instructions that Dan Barlow posted to sbcl-devel
2164 on 2 July 2000, I was able to enable primitive dynamic object
2165 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2166 functionality is not implemented (since it calls ld to resolve
2167 library references automatically, requiring RUN-PROGRAM for its
2168 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2169 try to resolve library references) is now supported.
2170 * The system now flushes the standard output streams when it terminates,
2171 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2172 them at several other probably-convenient times, e.g. in each pass of
2173 the toplevel read-eval-print loop, and after evaluating a form given
2174 as an "--eval" command-line option. (These changes were motivated by a
2175 discussion of stream flushing issues on cmucl-imp in August 2000.)
2176 * The source transform for TYPEP of array types no longer assumes
2177 that an array whose element type is a not-yet-defined type
2178 is implemented as an array of T, but instead punts, so that the
2179 type will be interpreted at runtime.
2180 * There is now some support for cross-compiling in make.sh: each of
2181 the phases of make.sh has its own script. (This should be transparent
2182 to people doing ordinary, non-cross-compile builds.)
2183 * Since my laptop doesn't have hundreds of megabytes of memory like
2184 my desktop machine, I became more motivated to do some items on
2185 my to-do list in order to reduce the size of the system a little:
2186 ** Arrange for various needed-only-at-cold-init things to be
2187 uninterned after cold init. To support this, those things have
2188 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2189 symbols with such names are now uninterned after cold init).
2190 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2191 fasl files for cold load.
2192 ** Remove the old compiler structure pooling code (which used to
2193 be conditional on the target feature :SB-ALLOC) completely.
2194 ** Redo the representation of some data in cold init to be more compact.
2195 (I also looked into supporting byte compiled code at bootstrap time,
2196 which would probably reduce the size of the system a lot, but that
2197 looked too complicated, so I punted for now.)
2198 * The maximum signal nesting depth in the src/runtime/ support code has
2199 been reduced from 4096 to 256. (I don't know any reason for the very
2200 large old value. If the new smaller value turns out to break something,
2201 I'll probably just bump it back up.)
2202 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2204 * Many, many bugs reported by Peter Van Eynde have been added to
2205 the BUGS list; some have even been fixed.
2206 * While enabling dynamic object file loading, I tried to make the
2207 code easier to understand, renaming various functions and variables
2208 with less ambiguous names, and changing some function calling
2209 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2210 * While trying to figure out how to do the OpenBSD port, I tried to
2211 clean up some of the code in src/runtime/. In particular, I dropped
2212 support for non-POSIX signal handling, added various comments,
2213 tweaked the code to reduce the number of compilation warnings, and
2214 renamed some files to increase consistency.
2215 * To support the new automatic configuration functionality in make.sh,
2216 the source file target-features.lisp-expr has been replaced with the
2217 source file base-target-features.lisp-expr and the machine-generated
2218 file local-target-features.lisp-expr.
2219 * fixed a stupid quoting error in make.sh so that using CMU CL
2220 "lisp -batch" as cross-compilation host works again
2222 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2224 * DESCRIBE no longer tries to call itself recursively to describe
2225 bound/fbound values, so that it no longer fails on symbols which are
2226 bound to themselves (like keywords, T, and NIL).
2227 * DESCRIBE now works on generic functions.
2228 * The printer now prints less-screwed-up representations of closures
2229 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2230 * A private symbol is used instead of the :EMPTY keyword previously
2231 used to mark empty slots in hash tables. Thus
2232 (DEFVAR *HT* (MAKE-HASH-TABLE))
2233 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2234 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2235 now does what ANSI says that it should. (You can still get
2236 similar noncompliant behavior if bang on the hash table
2237 implementation with all the symbols you get back from
2238 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2239 This breaks binary compatibility, since tests for equality to
2240 :EMPTY are wired into things like the macroexpansion of
2241 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2243 * There's now a minimal placeholder implementation for CL:STEP,
2244 as required by ANSI.
2245 * An obscure bug in the interaction of the normal compiler, the byte
2246 compiler, inlining, and structure predicates has been patched
2247 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2248 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2250 * Missing ordinary arguments in a macro call are now detected even
2251 when the macro lambda list contains &KEY or &REST.
2252 * The debugger no longer complains about encountering the top of the
2253 stack when you type "FRAME 0" to explicitly instruct it to go to
2254 the top of the stack. And it now prints the frame you request even
2255 if it's the current frame (instead of saying "You are here.").
2256 * As specified by ANSI, the system now always prints keywords
2257 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2259 * The default initial SIZE of HASH-TABLEs is now smaller.
2260 * Type information from CLOS class dispatch is now propagated
2261 into DEFMETHOD bodies, so that e.g.
2262 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2264 is now basically equivalent to
2265 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2266 (DECLARE (TYPE SINGLE-FLOAT X))
2268 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2269 operation, without having to do run-time type dispatch.
2270 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2271 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2272 * Since I don't seem to be making big file reorganizations very often
2273 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2274 is overflowing my ability to conveniently back them up), I've finally
2275 checked the system into CVS. (The CVS repository is on my home system,
2276 not at SourceForge -- putting it on SourceForge might come later.)
2277 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2278 high-level GC-NOTIFY-FOO functions send their output. (There's
2279 still very little control of where low-level verbose GC functions
2280 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2281 less than it used to -- the GC-NOTIFY-FOO functions are now under
2282 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2283 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2284 in only one place in the source code, and propagates it automatically
2285 everywhere that it's needed. Thus e.g. when I bump the version from
2286 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2287 * The C source files now include boilerplate legalese and documentation
2288 at the head of each file (just as the Lisp source files already did).
2289 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2290 to his page will be replaced with a link to his new CLiki service.
2292 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2294 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2295 * The build process now looks for GNU make under the default name "gmake",
2296 instead of "make" as it used to. If GNU make is not available as "gmake"
2297 on your system, you can change this default behavior by setting the
2298 GNUMAKE environment variable.
2299 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2300 can build without error under CMU CL.
2302 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2304 * There is now a partial SBCL user manual (with some new text and some
2305 text cribbed from the CMU CL manual).
2306 * The beginnings of a profiler have been added (starting with the
2307 CMU CL profiler and simplifying and cleaning up). Eventually the
2308 main interface should be through the TRACE macro, but for now,
2309 it's still accessed through vaguely CMU-CL-style functions and macros
2310 exported from the package SB-PROFILE.
2311 * Some problems left over from porting CMU CL to the new
2312 cross-compilation bootstrap process have been cleaned up:
2313 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2314 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2315 part when getting rid of the compiler *BACKEND* stuff.)
2316 ** The value of *NULL-TYPE* was screwed up, because it was
2317 being initialized before the type system knew the final
2318 definition of the 'NULL type. This screwed up several key
2319 optimizations in the compiler, causing inefficiency in all sorts
2320 of places. (I found it because I wanted to understand why
2321 GET-INTERNAL-RUN-TIME was consing.)
2322 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2323 PROCLAIM FTYPE information. Unfortunately this broke binary
2324 compatibility again, since now the forms output by DEFGENERIC
2325 to refer to functions which didn't exist in 0.6.3.
2326 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2327 can use the new (as of 0.6.3) transform for SOME into MAP into
2329 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2330 values to (INTEGER 0 1000000), so that the time code will no longer
2331 occasionally get blown up by Linux returning 1000000 microseconds
2332 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2333 its output conform to the ANSI spec. (Alas, this makes its output
2334 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2335 * A full call to MAP NIL with a single sequence argument no longer conses.
2336 * fixes to problems pointed out by Martin Atzmueller:
2337 * The manual page no longer talks about multiprocessing as though
2338 it were currently supported.
2339 * The ILISP support patches have been removed from the distribution,
2340 because as of version 5.10.1, ILISP now supports SBCL without us
2341 having to maintain patches.
2342 * added a modified version of Raymond Toy's recent CMU CL patch for
2343 EQUALP comparison of HASH-TABLE
2345 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2347 * The system still can't cross-compile itself with
2348 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2349 entails), but at least it can compile more of itself that way
2350 than it used to be able to, and various buglets which were uncovered
2351 by trying to cross-compile itself that way have now been fixed.
2352 * This release breaks binary compatibility again. This time
2353 at least I've incremented the FASL file format version to 2, so that the
2354 problem can be detected reliably instead of just causing weird errors.
2355 * various new style warnings:
2356 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2357 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2358 ** using the lexical binding of a variable named in the *FOO* style
2359 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2360 as specified by ANSI.
2361 * *RANDOM-STATE* is no longer automatically initialized from
2362 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2363 default behavior of the system is to repeat its behavior every time
2364 it's run. If you'd like to change this behavior, you can always
2365 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2366 old convention there was no comparably easy way to get the system to
2367 repeat its behavior every time it was run.
2368 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2369 FTYPE declarations has been removed, in favor of their ANSI
2370 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2371 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2372 the types of their sequence arguments can be determined at compile time.
2373 This is done through a new open code expansion for MAP which eliminates
2374 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2375 argument types can be determined at compile time.
2376 * The optimizer now transforms COERCE into an identity operation when it
2377 can prove that the coerced object is already of the correct type. (This
2378 can be a win for machine generated code, including the output of other
2379 optimization transforms, such as the MAP transform above.)
2380 * Credit information has been moved from source file headers into CREDITS.
2381 * Source file headers have been made more standard.
2382 * The CASE macro now compiles without complaining even when it has
2385 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2387 * (Note that the way that the PCL macroexpansions were rewritten
2388 to accommodate the change in DEFGENERIC below breaks binary
2389 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2390 not run under sbcl-0.6.2. Once we get out of alpha releases,
2391 i.e. hit release 1.0.0, we'll probably try to maintain binary
2392 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2393 and sbcl-1.4.4. Until then, however, it might be fairly common
2394 for maintenance releases to break binary compatibility.)
2395 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2397 * The reporting of unhandled signals has been changed to print some
2398 explanatory text as well as the report form. (Previously only
2399 the report form was printed.)
2400 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2401 it defines, so that the compiler no longer issues undefined function
2402 warnings for compiled-but-not-yet-loaded generic functions.
2403 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2404 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2405 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2406 section of the ANSI spec are supported.
2407 * The "" nickname for the "KEYWORD" package has been removed.
2408 The reader still handles symbol tokens which begin with a package marker
2409 as keywords, but it doesn't expose its mechanism for doing so in the
2410 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2411 * The system now issues STYLE-WARNINGs for contradictory TYPE
2412 proclamations. (Warnings for contradictory FTYPE proclamations would
2413 be nice too, but those can't be done usefully unless the type system
2414 is made smarter about FUNCTION types.)
2415 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2416 systematized, so that "*target-*.lisp is supposed to exist only on the
2417 target and imply that there's a related file which exists on the
2418 host, and *host-*.lisp is supposed to exist only on the host and imply
2419 that there's a related file which exists on the target. This involves a
2420 lot of renaming. Hopefully the acute confusion caused by the renaming
2421 will be justified by the reduction in chronic confusion..
2422 ** runtime-type.lisp -> early-target-type.lisp
2423 ** target-type.lisp -> late-target-type.lisp
2424 ** early-host-format.lisp -> early-format.lisp
2425 ** late-host-format.lisp -> late-format.lisp
2426 ** host-error.lisp -> misc-error.lisp
2427 ** early-error.lisp -> early-target-error.lisp
2428 ** late-error.lisp -> late-target-error.lisp
2429 ** host-defboot.lisp -> early-defboot.lisp
2430 ** code/misc.lisp -> code/target-misc.lisp
2431 ** code/host-misc.lisp -> code/misc.lisp
2432 ** code/numbers.lisp -> code/target-numbers.lisp
2433 ** code/early-numbers.lisp -> numbers.lisp
2434 ** early-host-type.lisp -> early-type.lisp
2435 ** late-host-type.lisp -> late-type.lisp
2436 ** host-typep.lisp -> typep.lisp
2437 ** load.lisp -> target-load.lisp
2438 ** host-load.lisp -> load.lisp
2439 ** host-disassem.lisp -> disassem.lisp
2440 ** host-insts.lisp -> insts.lisp
2441 ** byte-comp.lisp -> target-byte-comp.lisp
2442 ** host-byte-comp.lisp -> byte-comp.lisp
2443 ** host-signal.lisp -> signal.lisp
2444 ** host-defstruct.lisp -> defstruct.lisp
2445 ** late-target-type.lisp -> deftypes-for-target.lisp
2446 Furthermore, several other previously target-only files foo.lisp (e.g.
2447 hash-table.lisp and random.lisp) have been split into a target-and-host
2448 foo.lisp file and a target-only target-foo.lisp file, with their key type
2449 definitions in the target-and-host part, so that the cross-compiler will
2450 know more about target types.
2451 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2452 gone away. In their place are various *BACKEND-FOO* variables
2453 corresponding to the slots of the old structure.
2454 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2455 into the target SBCL has been fixed.
2456 * The chill.lisp system for loading cold code into a running SBCL
2458 * Support for the CMU CL "scavenger hook" extension has been removed.
2459 (It was undocumented and unused in the CMU CL sources that SBCL was
2460 derived from, and stale in sbcl-0.6.1.)
2461 * Various errors in the cross-compiler type system were detected
2462 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2463 (enabling various consistency checks). Many of them were fixed,
2464 but some hard problems remain, so the compiler is back to
2465 running without *TYPE-SYSTEM-INITIALIZED* for now.
2466 * As part of the cross-compiler type system cleanup, I implemented
2467 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2468 * I have started adding UNCROSS calls throughout the type system
2469 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2470 on cross-compiler input files will be able to go away, and various
2472 * CONSTANTP now returns true for quoted forms (as explicitly required
2475 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2477 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2478 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2479 (MAKE-LIST -1). In the longer term, it ought to become true
2480 that declarations are assertions even at SAFETY 1. For now, it's not
2481 quite true even at SAFETY 3, but it's at least more nearly true..
2482 (Note that this change seems to increases the size of the system by
2483 O(5%) and to decrease the speed of the compiler by 20% or more.)
2484 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2485 for the problem of printing dozens of lines of distracting information
2486 about low-level system machinery as part of the top stack frame
2487 on entry to the debugger when an undefined function was called.
2488 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2489 is not reset to COMMON-LISP-USER.
2490 * Compilation of stuff related to dyncount.lisp has been made conditional
2491 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2492 smaller. The various dyncount-related symbols have been moved into
2493 a new "SB-DYNCOUNT" package.
2494 * tty-inspect.lisp has been renamed to inspect.lisp.
2495 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2496 feature has gone away. (When we eventually port to other flavors of
2497 libc and/or Unix, we'll try to make the differences between flavors
2498 invisible at the user level.)
2499 * Various other *FEATURES* tags, and/or their associated conditionals,
2500 have been removed if obsolescent, or given better documentation, or
2501 sometimes given more-mnemonic names.
2503 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2505 * tidied up "make.sh" script
2506 * tidied up system directory structure
2507 * better "clean.sh" behavior
2508 * added doc/FOR-CMUCL-DEVELOPERS
2509 * many many small tweaks to output format, e.g. removing possibly-confusing
2510 trailing #\. character in DESCRIBE-INSTANCE
2511 * (EQUALP #\A 'A) no longer signals an error.
2512 * new hashing code, including EQUALP hashing
2513 * tidied up Lisp initialization and toplevel
2514 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2515 * command line argument processing
2516 * added POSIX-GETENV function to deal with Unix-ish environment variables
2517 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2518 terminating SBCL on EOF
2519 * non-verbose GC by default
2520 * There is no more "sbcl" shell script; the sbcl file is now the C
2521 runtime executable (just like CMU CL).
2522 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2524 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2525 symbol and package information as Lisp native symbol and package objects
2526 instead of strings naming symbols and strings naming packages. This way,
2527 whenever packages are renamed (as in warm init), debug information is
2528 transformed along with everything else.
2529 * tweaked the optimization policy declarations which control the building
2530 of SBCL itself. Now, among other things, the system no longer saves
2531 source location debugging information. (This helps two problems at once
2532 by reducing SBCL size and by keeping SBCL from trying to look for its
2533 sources -- which may not exist -- when reporting errors.)
2534 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2535 debugging and testing purposes
2536 * cleaned up printing, making the printer call PRINT-OBJECT for
2537 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2538 methods, giving nearly-ANSI behavior
2539 * converted almost all special variables to use *FOO* naming convention
2540 * deleted PARSE-TIME functionality, since it can be done portably
2541 * moved some files out of cold init into warm init
2542 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2544 * regularized formatting of source files
2545 * added an install.sh script
2546 * fixed ridiculous memory usage of cross-compiler by making
2547 compiler/alloc.lisp not try to do pooling unless it can hook
2548 itself into the GC of the cross-compilation host. Now the system
2549 builds nicely on my old laptop.
2550 * added :SB-ALLOC in target-features.lisp-expr
2551 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2552 was not implemented)
2553 * re-did condition handling and note reporting in the compiler. Notes
2554 are no longer handled by signalling conditions. Style warnings
2555 and warnings are handled more correctly and reported in such a way
2556 that it's easy to find one or the other in your output (so that you
2557 can e.g. figure out which of many problems caused COMPILE-FILE to
2559 * changed the severity of several compiler warnings from full WARNING
2560 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2561 compiler note reporting so that it doesn't use the condition system
2562 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2563 COMPILE-FILE command)
2564 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2565 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2566 no longer supported, since their implementation was deeply intertwingled
2567 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2568 * removed bogus "support" for compiler macros named (SETF FOO), and
2569 removed the compiler macro for SETF INFO (but only after making a fool
2570 of myself on the cmucl-imp mailing list by posting a bogus patch for
2571 DEFINE-COMPILER-MACRO..)
2572 * Compiled files containing forms which have side effects on the Lisp
2573 reader (such as DEFPACKAGE forms) are now handled more correctly.
2574 (Compiler queuing of top level lambdas has been suppressed by setting
2575 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2576 * deleted various currently-unused source files, e.g. gengc.lisp. They
2577 may be added back at some point e.g. when porting to other architectures,
2578 but until they are it's distracting to distribute them and to try to
2580 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2581 there were so many of them they're just distractions, and UNCROSS is
2582 known to be able to handle the current sources
2583 * moved PROFILE functionality into TRACE, so that it will be clear
2584 how the wrapping and unwrapping of functions when you profile them
2585 interacts with the wrapping and unwrapping of functions when you
2586 trace them. (Actually, the functionality isn't there yet, but at least
2587 the interface specification is there. Hopefully, the functionality will
2588 arrive with some future maintenance release.)
2589 * removed host-oops.lisp
2590 * changed signature of QUIT function to allow UNIX-CODE argument
2591 * fixed READ-SEQUENCE bug
2592 * tweaked verbose GC output so that it looks more like the progress
2593 output that ANSI specifies for functions like LOAD
2594 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2595 * added <http://sbcl.sourceforge.com> to the banner information printed by