1 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
3 the manual for details; for now, package locks can be disabled by
4 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
5 if no major problems are found then it is likely that they will be
6 compiled in unconditionally.
7 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
8 unsupported operators on all platforms. To load a shared library
9 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
10 object file, link it into a shared library outside of SBCL and
11 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
13 * minor incompatible change: as threatened around sbcl-0.8.0, the
14 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
15 The symbols are also exported from SB-PCL for backwards
16 compatibility, but more so than before SB-PCL should be treated as
17 an implementation-internal package.
18 * the SB-SPROF contrib now works on (most) non-x86 architectures.
19 It is known as of this release not to work on the Alpha, however.
20 * fixed bug #167: errors signalled due to illegal syntax in method
21 bodies are now more legible.
22 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
23 designators and can hence be used with TYPEP.
24 * fixed bug #333: CHECK-TYPE now ensures that the type error
25 signalled, if any, has the right object to be accessed by
26 TYPE-ERROR-DATUM. (reported by Tony Martinez)
27 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
28 5.1.2.3 for multiple-value place subforms. (reported by Kalle
30 * fixed bug #334: programmatic addition of slots using specialized
31 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
32 and :ALLOCATION :CLASS slots.
33 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
34 float exponents. (rereported by Peter Seibel)
35 * fixed a bug: #\Space (and other whitespace characters) are no
36 longer considered to be macro characters in standard syntax by
38 * fixed bug: initialization of condition class metaobjects no longer
39 causes an instance of the condition to be created. (reported by
41 * fixed bug: it is now possible to have slots such that
42 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
43 description is neither :INSTANCE nor :CLASS.
44 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
45 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
46 SLOT-VALUE-USING-CLASS), and no longer causes errors with
47 non-standard SLOT-DEFINITION-ALLOCATION values.
48 * fixed bugs: various race conditions handling exiting threads.
49 CL-PPCRE's thread torture test now passes.
50 * fixed arguably-a-bug: GC time proportional to number of threads in
51 system even when most of them are idle
52 * optimization: improved performance of BIT and SBIT on bit-vectors.
54 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
55 * minor incompatible change: the system no longer provides
56 optimization hints (conditions which are TYPEP
57 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
58 mode; these hints are still emitted when higher SPEED optimization
60 * new contrib module: a sampling profiler (profiling by statistical
61 sampling, rather than by instrumenting functions) is available as
62 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
63 * the behaviour of the standard function ED is now customizeable by
64 third parties through a hook variable: see ED's documentation
65 string for information on the protocol.
66 * the compiler no longer emits efficiency notes for (FUNCALL X)
67 when the type of X is uncertain under default optimization
69 * fixed bug 276: mutating a binding of a specialized parameter to a
70 method to something that is not TYPEP the specializer is now
72 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
73 exhibit the required behaviour on the x86 platform. (thanks to
74 Peter van Eynde, Eric Marsden and Bruno Haible)
75 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
76 even for difficult arguments. (reported by Peter Graves)
77 * fixed bug 141a: the backquote printer now descends quoted
79 * fixed another bug in backquote printing: no more destructive
80 modification of the form's list structure. (reported by Brian
82 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
83 move between its address being taken and the call to
84 interrupt_thread, fixing a crashing race condition.
85 * the SB-POSIX contrib implementation has been adjusted so that it
86 no longer exhibits ridiculously poor performance when constructing
87 instances corresponding to C structs.
89 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
90 * minor incompatible change: the sb-grovel contrib now treats C
91 structures as alien (in the sense of SB-ALIEN) objects rather than
92 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
93 has implications for memory management of client code
94 (sb-grovel-returned objects must now be manually managed) and for
95 type safety (alien objects now have full types).
96 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
97 used to control emission of compiler diagnostics, rather than the
98 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
99 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
100 quality should be considered deprecated.
101 * install.sh now installs the user manual as well
102 * (not quite a new documentable feature, but worth considering in
103 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
104 beginnings of a semantically meaningful condition hierarchy is
105 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
107 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
108 designator as the defaults argument.
109 * fixed bug: Displaced arrays whose displaced-to array has become
110 too small now cause ARRAY-DIMENSION to signal an error, providing
111 for safer bounds-checking. (reported by Bruno Haible)
112 * fixed bug: DEFCLASS slot definitions with identical :READER and
113 :WRITER names now signal a reasonable error. (reported by Thomas
115 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
117 * fixed bug: Invalid slot specification errors now print correctly.
118 (thanks to Zach Beane)
119 * fixed bug 320: Shared to local slot value transfers in class
120 redefinitions now happen corrently from superclasses as
121 well. (reported by Bruno Haible)
122 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
124 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
125 improper lists correctly. (reported by Bruno Haible)
126 * fixed bug 313: source-transform for <fun-name> was erroneously
127 applied to a call of a value of a variable with name <fun-name>.
128 (reported by Antonio Menezes Leitao)
129 * fixed bug 307: The obsolete instance protocol ensures that
130 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
131 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
132 remove dead unknown-values globs from the middle of the stack.
133 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
134 easier to resume long computations after using *BREAK-ON-SIGNALS*
135 to diagnose and fix failures. (thanks to Nikodemus Siivola)
136 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
137 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
138 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
139 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
140 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
142 * fixed a (fairly theoretical) bug in string printing: if
143 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
144 does not have array-element-type equal to the most general string
146 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
147 function name. (thanks to Nikodemus Siivola)
148 * fixed bug: compile-time format string checker failed on
149 non-closed ~{. (reported by Thomas F Burdick)
150 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
151 don't warn on legal loop code involving more than one aggregate
152 boolean. (thanks to Nikodemus Siivola)
153 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
154 in keyword position to MAKE-PACKAGE was not being reported as
156 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
157 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
159 * fixed bugs 280 and 312: the checking for multiple definitions in a
160 file is less likely to become confused by uses of inline
162 * fixed bug: the #S reader macro performs the keyword coercion
163 specified for slot names. (reported by Kalle Niemitalo)
164 * fixed bug: lambda lists may contain symbols whose names start with
165 & but are not lambda-list-keywords; their occurrence triggers a
167 * fixed bug 321: define-method-combination argument lambda lists do
168 not cause invalid code to be generated when &optional or &aux
169 variables have default values. (reported by Bruno Haible)
170 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
171 have CLOS classes; this appears also to have fixed reported
172 instability in stack exhaustion detection.
173 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
174 MUFFLE-WARNING when no associated restart is present is now
176 * optimization: rearranged the expansion of various defining macros
177 so that each expands into only one top-level form in a
178 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
180 * optimization: used a previously little-used slot in symbols to
181 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
183 * fixed some bugs revealed by Paul Dietz' test suite:
184 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
186 ** Fixed an optimization bug related to inheritance of initforms
187 from local to shared slots.
188 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
189 ** CAST optimizer forgot to flush argument derived type.
190 ** print/read consistency on floats is now orders of magnitude
191 more likely. (thanks also to Bruno Haible for a similar report
193 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
194 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
195 for objects of type REAL. Make it so.
196 ** SXHASH is specified (infelicitously) to respect similarity,
197 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
198 it so. (thanks to Markus Ziegler)
199 ** on the Alpha, the compiler succeeds in compiling functions
200 returning a known number of arguments greater than 63.
201 ** fixed handling of invalid NIL arguments in keyword position
203 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
205 ** provided a readably-printable representation for RANDOM-STATE
207 ** ensured that pathnames, where they have a namestring, always
208 print using #P"..." syntax.
210 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
211 * Support for the forthcoming 2.0 version of the NetBSD kernel
212 running on x86 hardware has been added. (thanks to Perry
213 E. Metzger most immediately, and others for their past work)
214 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
215 current release version; SBCL's OpenBSD support had been broken
216 since about the time of OpenBSD's switch to ELF binary format.)
217 * [placeholder for DX summary]
218 ** user code with &REST lists declared dynamic-extent, under high
219 speed or space and low safety and debug optimization policy.
220 * The manual has been converted to Texinfo format and the debugger
221 chapter from the cmucl manual has been added.
222 * A facility has been added to extract documentation strings from
223 sbcl and store them as Texinfo-formatted snippets for inclusion in
224 the manual (via Texinfo's @include directive)
225 * bug fix: compiler emitted division in optimized DEREF. (thanks for
226 the test case to Dave Roberts)
227 * bug fix: multidimensional simple arrays loaded from FASLs had fill
228 pointers. (reported by Sean Ross)
229 * bug fix: PROFILE output is printed nicely even for large numerical
230 values. (thanks to Zach Beane)
231 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
232 greater than 32 handle EOF correctly.
233 * bug fix: on X86 an immediate argument of the IMUL instruction is
234 correctly printed in disassembly. (thanks to Lutz Euler)
235 * bug fix: class slots in redefined classes preserve their old
236 values. (thanks to Bruno Haible and Nikodemus Siivola)
237 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
238 succeeds. (reported by Marco Baringer)
239 * fixed some bugs revealed by Paul Dietz' test suite:
240 ** READ-SEQUENCE now works on ECHO-STREAMs.
241 ** RATIONALIZE works more according to its specification. (thanks
244 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
245 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
246 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
247 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
248 should matter to you only if you rebind the printer control
249 variables and then find you want different bindings in the
250 debugger than in the ordinary execution of your program.)
251 * The runtime build system has been tweaked to support building
252 (on SPARC/SunOS) using a C compiler which invokes Sun's own
253 assembler and linker. (thanks to Nikodemus Siivola)
254 * Unbound, undefined, undeclared variables now trigger full
255 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
256 more often programmer error than deliberate exploitation of undefined
258 * optimization: the hash algorithm for strings has changed to one
259 that is less vulnerable to spurious collisions. (thanks to Juho
261 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
262 needless bounds checking. (thanks to Juho Snellman)
263 * optimization: implemented multiplication as a modular
264 (UNSIGNED-BYTE 32) operation on the PPC backend.
265 * fixed some bugs revealed by Paul Dietz' test suite:
266 ** ADJUST-ARRAY now copies the datum in a zero rank array if
268 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
269 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
270 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
272 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
274 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
276 ** ECHO-STREAMs no longer attempt to echo the end of file value to
277 their output stream on EOF from read.
278 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
279 have been read to end-of-file.
280 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
282 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
283 description of determination of which consecutive characters
285 ** Printing the "Space" character with escaping on now yields "#\\ ",
286 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
287 ** Reading floating-point numbers with *READ-BASE* set to a number
288 less than 10 works correctly.
289 ** Reading floating-point numbers with *READ-BASE* set to a number
290 more than 10 works correctly.
291 ** Printing with *PRINT-READABLY* targets the standard readtable, not
292 the readtable currently in effect.
294 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
295 * minor incompatible change: parsing of namestrings on a physical
296 (Unix) host has changed; numbers after the final #\. in a
297 namestring are no longer interpreted as a version field. This is
298 intented to be largely invisible to the user, except that the
299 meaning of the namestring "*.*.*" has changed: it now refers to a
300 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
301 should usually be replaced by
302 (make-pathname :name :wild :type :wild :version :wild)
303 with the added benefit that this is more likely to be portable.
304 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
305 option to OPEN now signals an error if the file being opened
306 exists; this may have an impact on existing code.
307 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
308 signals received, on the PowerPC platforms (both Linux and
309 Darwin). (thanks to Pierre Mai for pointing out the location of
311 * several fixes on OS X: The system now builds and runs cleanly on
312 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
313 (thanks to Brian Mastenbrook)
314 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
315 second-resolution: integer multiples of 1/3600 between -24 and 24.
316 (thanks to Vincent Arkesteijn)
317 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
318 type when called with 1 argument; PEEK-CHAR checked type of
319 PEEK-TYPE only after having read first character from a
320 stream. (reported by Peter Graves)
321 * bug fix: the garbage collector now has much better locality
322 behaviour, and in particular no longer treats all memory as being
323 exhausted when many small objects point to each other in a deeply
325 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
326 recognized as being TYPEP their class.
327 * bug fix: the PUSHNEW documentation string has been corrected.
328 (thanks to Vincent Arkesteijn)
329 * bug fix: defaulting of the value for the last of an atypically
330 large number of multiple values being bound was not being
331 performed correctly on the Alpha or PPC platforms
332 * optimization: implemented multiplication as a modular
333 (UNSIGNED-BYTE 32) operation on the x86 backend.
334 * optimization: SEARCH on simple-base-strings can now be open-coded.
335 (see also contrib/compiler-extras.lisp for inspiration for
336 teaching the compiler about the Boyer-Moore algorithm).
337 * value, returned by MAX (and MIN) called with several EQUALP, but
338 not EQL, arguments now does not depend on compiler settings.
339 * fixed some bugs revealed by Paul Dietz' test suite:
340 ** in stack analysis liveness information is propagated from
341 non-local entry points.
342 ** pathwise CAST removing failed when the CAST node did not start
344 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
345 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
347 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
348 host is already defined.
349 ** RENAME-FILE works on streams instead of signalling an internal
351 ** PEEK-CHAR uses the current readtable when determining whether
352 or not a character is whitespace.
353 ** MERGE-PATHNAMES handles the case when the pathname does not
354 specify a name while the default-pathname specifies a version.
355 ** Pathnames now stand a better chance of respecting print/read
357 ** Attempting to use standardized file system operators with a
358 pathname with invalid :DIRECTORY components signals a
360 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
362 ** DIRECTORY on logical pathnames is more correct.
363 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
364 signal a TYPE-ERROR if their argument is not a stream.
365 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
366 designator argument does not designate a stream.
367 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
368 examining the synonym.
369 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
371 ** OPEN and WITH-OPEN-STREAM allow opening streams with
372 element-type larger than ([UN]SIGNED-BYTE 32).
374 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
375 * When built with the :SB-FUTEX feature, threaded builds now take
376 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
377 for faster/more reliable mutex and condition variable support.
378 * Incompatible change (but one you probably shouldn't have been using
379 anyway): the interface and code for arbitrating between multiple
380 threads in the same user session has been redesigned.
381 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
382 error when called without an explicit environment argument.
383 (thanks to Nikodemus Siivola)
384 * bug fix: buffered :DIRECTION :IO streams are less likely to become
385 confused about their position. (thanks to Adam Warner and Gerd
387 * bug fix: Pretty printing backquoted forms with unquotations in the
388 argument list position of various code constructs such as LAMBDA
389 now works correctly. (reported by Paul Dietz)
390 * bug fix: Pretty printing unquotations no longer loses all
391 stream position information.
392 * optimization: performance of string output streams is now less
393 poor for multiple small sequence writes.
394 * optimization: performance of CSUBTYPEP in the presence of complex
395 expressions involving CONS and NOT many times has been improved.
396 (reported by Paul Dietz)
397 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
399 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
400 streams for the REPL, for the debugger, and so on. (thanks to
402 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
403 * fixed some bugs revealed by Paul Dietz' test suite:
404 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
405 is ignored if it is too small, rather than propagating through
406 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
407 ** extremely complex negations of CONS types were not being
408 sufficiently canonicalized, leading to inconsistencies in
410 ** VALUES tranformer lost derived type.
412 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
413 * fixed a bootstrapping bug: the build process no longer assumes
414 that the various BOOLE-related constants have the same value in
415 host and target lisps. (noted by Paul Dietz' test suite on an
416 SBCL binary built from CLISP)
417 * The system can now be dynamically linked on the MIPS platform,
418 which enables dynamic loading of foreign code from Lisp. (thanks
419 to Ralf Baechle for discussions on the MIPS ABI)
420 * The system now records debugging information for its own source
421 files in a filesystem-position-independent manner, relative to
422 the "SYS" logical host.
423 * fixed a compiler bug: MV-LET convertion did not check references
424 to the "max args" entry point. (reported by Brian Downing)
425 * tweaked disassembly notes to be less confident about proclaiming
426 some instruction as an LRA. (thanks to Brian Downing)
427 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
428 now each have their own history, command character, and other
429 characteristics. (thanks to David Lichteblau)
430 * fixed some bugs revealed by Paul Dietz' test suite:
431 ** compiler failure in compiling LOGAND expressions including a
433 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
434 shift greater than 32.
435 ** FLUSH-DEST did not mark blocks for type check regeneration.
436 ** HANDLER-CASE failed to accept declarations in handler clauses
437 in some circumstances.
439 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
440 * New code in contrib/sb-introspect (still probably not entirely
441 stable yet) provides some support for smart Lisp development
442 environments like SLIME.
443 * The conditions signalled for errors occurring when loading .fasl
444 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
445 in a way which should help ASDF recover gracefully.
446 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
447 stuff has been cleaned up. If you code contrib/ stuff, this might
448 affect you, and you can look at contrib/README, contrib/STANDARDS,
449 and/or the 0.8.4.27 diff to check.
450 * In full calls the compiler now does not generate checks for declared
451 argument types for all arguments.
452 * various threading fixes
453 ** and some experimental patches which didn't make it into
454 the main tree for this release, but which are shipped in
455 contrib/experimental-thread.patch as a possible fix for some
456 failures (deadlock, spinning...) in GC-intensive multithreaded
458 * fixed PPC build problem (source code incompatibility of different
459 library versions): added offsetof() hackery which attempts to divine
460 where glibc maintainers put uc_mcontext today
461 * fixed bug 282: compiler does not trust type assertions while passing
462 arguments to a full call.
463 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
464 &OPTIONAL VALUES type parameter.
465 * fix bug 214: algorithm for noting rejected templates is now more
466 similar to that of template seletion. (also reported by rydis on
468 * fixed bug 141b: printing backquoted information readably and prettily
469 inserts a space where necessary.
470 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
471 (CHARACTER 10) are now reported as errors, rather than propagated
472 as unknown types. (reported by piso on #lisp)
473 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
474 with values NIL and :ERROR. (thanks to Milan Zamazal)
475 * fixed bug 191c: CLOS now does proper keyword argument checking as
476 described in CLHS 7.6.5 and 7.6.5.1.
477 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
478 longer raise an error; further, using a list as a for-as-arithmetic
479 counter now raises a meaningful error.
480 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
481 now understood by sequence creation functions such as MAKE-SEQUENCE
483 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
484 a STRING-INPUT-STREAM.
485 * compiler enhancement: SIGNUM is now better able to derive the type
487 * type declarations inside WITH-SLOTS are checked. (reported by
489 * fixed some bugs revealed by Paul Dietz' test suite:
490 ** incorrect optimization of TRUNCATE for a positive first
491 argument and negative second.
492 ** compiler failure in let-convertion during flushing dead code.
493 ** compiler failure while deriving type of TRUNCATE on an
494 interval, containing 0.
495 ** ASH of a negative bignum by a negative bignum count now returns
497 ** intersection of CONS types now canonicalizes properly, fixing
498 inconsistencies in SUBTYPEP.
500 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
501 * incompatible change: The --disable-debugger command line
502 option now clobbers the debugger at a more fundamental
503 level, by redefining #'INVOKE-DEBUGGER instead of by
504 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
505 is specified by ANSI to ignore *DEBUGGER-HOOK* and
506 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
507 enter the debugger REPL and then suffer recursive errors
508 because *DEBUG-IO* is also messed up in --disable-debugger mode;
509 while under the new system, BREAK in --disable-debugger mode
510 terminates the system just as an unhandled error would.
511 * fixed compiler performance when processing loops with a step >1;
512 * bug fix: DOCUMENTATION now retrieves generic function
513 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
514 support has been systematized, and now supports the methods
515 specified by ANSI, along with a default method and a method for
516 slot documentation. (reported by Nathan Froyd)
517 * bug fix: effective methods associated with a generic function are
518 no longer cached over a change of that generic function's method
519 combination. (reported by Andreas Fuchs)
520 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
521 names a non-existent directory. (thanks to Andreas Fuchs)
522 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
523 the PPC platform, lead to stack corruption; this has been fixed.
524 (reported by Rainer Joswig)
525 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
526 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
527 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
528 arithmetic in general, is now much more reliable on the PPC
530 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
531 platform now returns the right answer.
532 * optimization: restored some effective method precomputation in
533 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
534 precomputation is now tunable.
535 * optimization: compiler-internal data structure use has been
536 reviewed, and changes have been made that should improve the
537 performance of the compiler by about 20%.
538 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
539 simple-base-strings and simple-bit-vectors is improved.
540 * optimization: the optimization of 32-bit logical and arithmetic
541 functions introduced in version 0.8.3 on the x86 has been
542 implemented on the mips, ppc and sparc platforms; an
543 implementation of the same facility, but for 64-bit arithmetic,
544 has been added for the alpha.
545 * microoptimization: the compiler is better able to make use of the
546 x86 LEA instruction for multiplication by constants.
547 * bug fix: in some situations compiler did not report usage of
548 generic arithmetic in (SPEED 3) policy.
549 * bug 145b fix: compiler used wrong type specifier while converting
550 MEMBER-types to numeric.
551 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
553 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
554 index argument now works properly on the Alpha platform.
555 * bug fix: floating point exception treatment on the Alpha platform
557 * bug fix: FILE-POSITION works much better on string input and
558 output streams. (thanks to Nikodemus Siivola)
559 * bug fix: many threading/garbage collection symptoms sorted.
560 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
561 might be pseudo-atomic.
562 * internal change: Stopping for GC is now done with signals not
563 ptrace. GC is now done in whichever thread wanted it, instead of
565 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
567 * bug fix: result form in DO is not contained in the implicit
569 * incompatible change: ICR structure is changed; the value part of
570 CONTINUATION is now called LVAR; corresponding functions are
571 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
572 * added type deriver for ISQRT (thanks to Robert E. Brown).
573 * bug fix: better support for loading from the command line when an
574 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
576 * fixed some bugs revealed by Paul Dietz' test suite:
577 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
578 ** ROUND and FROUND now give the right answer when given very
579 small float arguments.
580 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
582 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
583 ** (GCD 0 <negative-integer>) returned <negative-integer>.
584 ** LCM should return a non-negative integer.
585 ** PARSE-INTEGER returned the index of a terminator instead of the
586 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
587 ** PARSE-INTEGER returned an incorrect index being applied to a
589 ** LCM with two arguments of 0 returns 0 rather than signalling
591 ** unsigned addition of a 32-bit constant with the high bit set no
592 longer causes an internal compiler error on the x86.
593 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
594 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
595 with negative last argument.
596 ** byte specifiers with zero size and position no longer cause
597 an error during type derivation.
598 ** bignum multiplication on the Alpha platform now returns the
600 * porting: The system now builds on SuSE AMD64, although it still
601 generates a 32-bit binary.
602 * .fasl file incompatibility: The fasl file version number has
603 been incremented (because of the changes to internal compiler
604 data structures referred to above).
606 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
607 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
608 more accurately, on the Darwin kernel running on PowerPC hardware.
609 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
610 * Compiler code deletion notes now signal a condition of type
611 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
612 an associated MUFFLE-WARNING restart.
613 * The compiler now performs limited argument count validation of
614 constant format strings in FORMAT, and where appropriate in ERROR,
615 CERROR and WARN. (thanks to Gerd Moellmann)
616 * New ASDF-INSTALL contrib can be used for automatic download and
617 installation of third-party Lisp code from CCLAN or other sites
619 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
620 which forces another thread to execute a function supplied by the
622 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
623 now accept and act upon their :ELEMENT-TYPE keyword argument.
624 (reported by Martin Atzmueller, Edi Weitz)
625 * bug fix: FILE-POSITION now accepts position designators up to
626 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
627 is the greater. (thanks to Patrik Nordebo)
628 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
629 Kalle Olavi Niemitalo)
630 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
631 printable. (reported by Eric Marsden)
632 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
633 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
634 * x86 bug fix in control stack exhaustion checking: now shows backtrace
635 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
636 (thanks to Stig Sandoe)
637 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
638 UNREAD-CHAR now correctly includes the unread character in the
639 target sequence. (thanks to Gerd Moellmann)
640 * bug fix in threaded builds: the system can now be suspended and
641 resumed by shell job control with minimal disruption.
642 * bug fixes in times and timezones >2038AD
643 * better handling of "where is GNU make?" problem in build scripts
644 (thanks to Nikodemus Siivola)
645 * new optimization: inside a named function any reference to a
646 function with the same name is considered to be a self-reference;
647 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
648 optimization quality.
649 * new optimization on x86: logical functions and + now have
650 optimized (UNSIGNED-BYTE 32) versions, which are automatically
651 used when the result is truncated to 32 bits.
652 * VALUES declaration is partially enabled.
653 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
654 * fixed some bugs revealed by Paul Dietz' test suite:
655 ** The system now obeys the constraint imposed by
656 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
657 types form a lattice under type intersection.
658 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
659 ** ASSOC now ignores NIL elements in an alist.
660 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
661 and (1+ MOST-POSITIVE-FIXNUM) answers.
662 ** The addition of a method with invalid qualifiers to a generic
663 function does not cause an error to be signalled immediately;
664 a warning is signalled, and the error is generated only on
665 calling the generic function.
666 * changed .fasl file version number, in part to add type codes for
667 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
668 obscure ANSI requirements
670 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
671 * fixed bug 148: failure to inline-expand a local function left
672 garbage, confusing the compiler.
673 * fixed bugs 3cd: structure slot readers perform type check if the
674 slot can have an invalid value (i.e. it is either not initialized
675 or can be written with a less specific slot writer).
676 * bug fix: the compiler now traps array references to elements off
677 the end of an array; previously, the bounds checking in some
678 circumstances could go off-by-one.
679 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
681 * type declarations for array element types now obey the description
682 on the CLHS page "Declaration TYPE", as per discussions on
683 sbcl-help around 2003-05-08. This means that a declaration
684 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
685 declaration, all references to BAR will be asserted or assumed
686 (with THE, so dependent on compiler policy) to involve objects of
687 type FOO. Note that no such declaration is implied in
688 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
689 * declared types of functions from the "Conditions"
690 chapter. (reported by Paul Dietz)
691 * bug fix: CERROR accepts a function as its first argument.
692 * bug fix: NTH an NTHCDR accept a bignum as index
693 arguments. (reported by Adam Warner)
694 * optimization: character compare routines now optimize comparing
695 against a constant character. (reported by Gilbert Baumann)
696 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
697 argument now works properly on the MIPS platform.
698 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
699 * repeated evaluation of the same DEFSTRUCT, a slot of which is
700 declared to have a functional type, does not cause an error
702 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
703 many :DEFAULT-INITARGS. (reported by Istvan Marko)
704 * fixed bug: if last continuation of a deleted block has a
705 destination, this destination should be deleted too. (reported by
707 * fixed a bug in the bootstrap process: the host compiler's values
708 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
709 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
710 test case from Patrik Nordebo)
711 * improved the ability of the disassembler on the PPC platform to
712 provide helpful disassembly notes.
713 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
714 the class in more cases than previously.
715 * bug fix: FILE-POSITION now understands :START and :END for
716 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
717 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
718 8) data. (thanks to David Lichteblau)
719 * bug fix: it is possible to add a method to a generic function
721 * bug fix: reader failed to signal END-OF-FILE inside an
722 object representation. (reported by Nikodemus Siivola)
723 * fixed some bugs revealed by Paul Dietz' test suite:
724 ** LAST and [N]BUTLAST should accept a bignum.
725 ** condition slot accessors are methods.
726 ** (VECTOR NIL) is a subtype of STRING.
728 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
729 * minor incompatible change: some nonsensical specialized lambda
730 lists (used in DEFMETHOD) which were previously ignored now signal
732 * minor incompatible change: the system is now aware of the types of
733 variables in the COMMON-LISP package, and will signal errors for
734 most violations of these type constraints (where previously they
735 were silently accepted).
736 * minor incompatible change: COMPILE-FILE now uses the freedom
737 afforded (ANSI 3.2.2.3) to use derived function types for
738 functions defined in the same file. This also permits the system
739 to warn on static type mismatches and function
740 redefinition. (Currently it does not work with high DEBUG level.)
741 * minor incompatible change: VALUES declaration is disabled.
742 * When issuing notes, the compiler now signals a condition of type
743 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
744 restart for use in user handlers. It is expected that the
745 COMPILER-NOTE condition will eventually become a condition
746 supertype to a hierarchy of note types, which will then be
747 handleable in a similar fashion. However, at the moment, no such
748 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
749 but it's an implementation detail, not a classification for the
751 * Changes in type checking closed the following bugs:
752 ** type checking of unused values (192b, 194d, 203);
753 ** template selection based on unsafe type assertions (192c, 236);
754 ** type checking in branches (194bc).
755 * A short form of VALUES type specifier has ANSI meaning (it has
756 increased the number of situations when SBCL cannot perform type
758 * fixed bug in DEFSTRUCT: once again, naming structure slots with
759 keywords or constants is permissible.
760 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
761 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
762 classes. (thanks to Antonio Martinez)
763 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
764 outputs FRESH-LINE or TERPRI, and no longer converts its stream
765 argument to a pretty-print stream. Instead, it leaves any such
766 operations to DESCRIBE-OBJECT methods.
767 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
769 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
770 lambda lists. (thanks to David Lichteblau)
771 * bug fix: Defining a generic function with a :METHOD-CLASS being a
772 subclass of STANDARD-METHOD no longer causes stack exhaustion.
773 (thanks to Gerd Moellmann)
774 * fixed bug 246: increased compilation speed of long
775 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
777 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
779 * DEFCONSTANT now throws a condition of type
780 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
781 constant to a non-EQL value; CONTINUE and ABORT restarts
782 respectively change and preserve the value.
783 * fixed bug 63: The code walker, part of the implementation of CLOS,
784 is now better at handling symbol macros.
785 * bug fix: There is no longer an internal implementation type named
786 CL:LENGTH. (reported by Raymond Toy)
787 * bug fix: In macro-like defining macros/special operators the
788 implicit block does not enclose the lambda list.
789 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
790 enclosed in parentheses) are not suitable as type specifiers, and
791 their use properly signals an error now.
792 * bug fix: An argument count mismatch for a type specifier in code
793 being compiled no longer causes an unhandled error at compile
794 time, but signals a compile-time warning.
795 * fixed simple vector readable printing
796 * bug fix: DESCRIBE takes more care over whether the class
797 precedence list slot of a class is bound before accessing it.
798 (reported by Markus Krummenacker)
799 * bug fix: FORMATTER can successfully compile pretty-printer format
800 strings which use variants of the ~* directive inside.
801 * bug fix: SEARCH now applies its TEST predicate to the elements of
802 the arguments in the correct order. (thanks to Wolfhard Buss)
803 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
804 in type inference. (thanks to Robert E. Brown)
805 * bug fix: Reading in symbols with an explicit package name of ""
806 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
807 * fixed some bugs revealed by Paul Dietz' test suite:
808 ** NIL is now allowed as a structure slot name.
809 ** Arbitrary numbers, not just REALs, are allowed in certain
810 circumstances in LOOP for-as-arithmetic clauses.
811 ** Multiple class redefinitions before slot access no longer
813 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
814 association between the name and a class.
815 ** Generic functions with non-standard method-combination and over
816 five methods all of which return constants no longer return NIL
817 after the first few invocations. (thanks to Gerd Moellmann)
818 ** CALL-NEXT-METHOD with no arguments now passes the original
819 values of the arguments, even in the presence of assignment.
820 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
822 ** DISASSEMBLE works with closures and funcallable instances.
823 ** ADD-METHOD now returns the generic function, not the new method.
824 ** FIND-METHOD signals an error if the lengths of the specializers
825 is incompatible with the generic function, even if the ERRORP
827 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
828 which its argument is a member.
829 ** DEFCLASS only redefines the class named by its class-name
830 argument if that name is the proper name of the class;
831 otherwise, it creates a new class.
832 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
833 of the UNBOUND-SLOT condition to the name of the slot.
834 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
835 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
836 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
837 SLOT-MAKUNBOUND in the specified fashion.
839 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
840 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
841 cross-compilation host. As a consequence, we can now bootstrap our
842 way up to SBCL starting with a bare gcc toolchain and human-readable
843 source code (first the source to CLISP, then the source to SBCL).
844 * A contributed module containing a partial implementation of the
845 simple-streams interface has been included. (thanks to Rudi
847 * A contributed module implementing the RFC1321 Message Digest
848 Algorithm, known as MD5, has been included.
849 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
850 no longer has any effect, as the code controlled by this feature
851 has been deleted. (As far as we know, no-one has ever built using
852 this feature, and its semantics were confused in any case).
853 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
854 work on conditions (as required by the ANSI specification),
855 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
856 expected behaviour on conditions. Users should note, however,
857 that such behaviour is not required by the ANSI specification,
858 and so use of this behaviour may render their code unportable.
859 * fixed some bugs revealed by Paul Dietz' test suite:
860 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
862 ** &ENVIRONMENT parameter in macro lambda list is bound first.
863 ** SXHASH on condition objects no longer returns NIL.
864 ** :ALLOCATION :CLASS slots are better treated; their values are
865 updated on class redefinition, and initforms inherited from
866 superclasses are applied.
867 ** REMOVE-METHOD returns its generic function argument even when
868 no method was removed.
869 ** SHARED-INITIALIZE now initializes the values of the requested
870 slots, including those with :ALLOCATION :CLASS.
871 ** ALLOCATE-INSTANCE now works on structure classes defined via
872 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
874 ** SLOT-EXISTS-P now works on conditions, as well as structures
876 ** MAKE-LOAD-FORM now has the required methods on
877 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
878 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
879 keyword, and now implements the SLOT-NAMES argument.
880 ** methods with &OPTIONAL arguments no longer allow too many
881 arguments to be passed in the call without error.
882 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
883 option is consistent with the required arguments of the generic
884 function lambda list.
885 * bug fix: REQUIRE accepts a string designator. (Thanks to
887 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
888 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
889 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
890 * bug fix: The compiler now checks for duplicated variables in macro
892 * bug fix: SETQ on globals returns the correct value.
893 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
894 not a COMPILER-ERROR (followed by some other strange error on
895 choosing the CONTINUE restart).
896 * bug fix: make.sh and friends are now more consistent in the way that
897 they look for GNU "make".
899 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
900 * experimental native threads support (on x86 Linux >=2.4 only).
901 This is not compiled in by default: you need to add :SB-THREAD to
902 the target features. See the "Beyond ANSI" chapter of the manual
904 * fix for longstanding nonANSIism: The old distinction between
905 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
906 The return value from CL:FIND-CLASS is now a CLOS class, and
907 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
908 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
909 * An interface to the MetaObject Protocol, as described in Kiczales,
910 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
911 MIT Press, 1991, is available from the SB-MOP package.
912 * incompatible change: the SB-PCL package should now be considered
913 a private implementation detail, and no longer a semi-private MOP
915 * minor incompatible change: due to rearrangement for threads, the
916 control stack and binding stack are now allocated at arbitrary
917 addresses instead of being hardcoded per-port. Users affected by
918 this probably have to be doing advanced things with shared
919 libraries, and will know who they are.
920 * minor incompatible change: Previously, all --eval forms used were
921 processed with READ before any of them were processed with EVAL.
922 Now each --eval form is processed with both READ and EVAL before
923 the next --eval form is processed. (Thus package operations like
924 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
925 work as the user might reasonably expect.)
926 * minor incompatible change: *STANDARD-INPUT* is now only an
927 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
929 * minor incompatible change: Y-OR-N-P is now character-oriented,
930 not line oriented. Also, YES-OR-NO-P now works without errors.
931 (thanks to Antonio Martinez)
932 * sb-aclrepl module improvements: an integrated inspector, added
933 repl features, and a bug fix to :trace command.
934 * Known functions, which cannot be open coded by the backend, are
935 considered to be able to check types of their arguments. (fixing
936 a bug report by Nathan J. Froyd)
937 * fixed a bug in computing method discriminating functions: It is
938 now possible to define methods specialized on classes which have
939 forward-referenced superclasses. (thanks to Gerd Moellmann)
940 * fixed evaluation order in optional entries (reported by Gilbert
942 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
943 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
944 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
945 required-by-AMOP NAME argument, as well as CLASS and
946 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
947 * fixed bug 20: DEFMETHOD can define methods using names that are
948 not the proper names of classes to designate class specializers.
949 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
950 Unix file descriptors, instead of blowing up. (thanks to Antonio
952 * Garbage collection refactoring: user-visible change is that a
953 call to the GC function during WITHOUT-GCING will not do garbage
954 collection until the end of the WITHOUT-GCING. If you were doing
955 this you were probably losing anyway.
956 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
957 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
958 TYPEP the latter but not the former.
959 * The compiler issues a full WARNING for calls to undefined functions
960 with names from the CL package.
961 * MAP-INTO for a vector destination is open coded. (reported by
962 Brian Downing on c.l.l)
963 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
964 documentation string.
965 * fixed some bugs revealed by Paul Dietz' test suite:
966 ** COPY-ALIST now signals an error if its argument is a dotted
968 ** Condition slots are now accessed more correctly in the presence
969 of multiple initargs for a given slot.
970 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
971 exclude restarts of the same name associated with a different
973 ** DEFCLASS of forward-referenced classes with another
974 forward-referenced class in the superclasses list no longer
976 ** Condition slots are now initialized once each, not multiple
977 times. (thanks to Gerd Moellmann)
978 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
980 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
981 ** NSET-EXCLUSIVE-OR does not return extra elements when its
982 arguments contain duplicated elements.
983 ** RESTART-CASE understands local macros.
984 ** RESTART-CASE associates exactly its own restarts with a condition.
985 ** ENDP in safe mode checks its argument to be of type LIST.
986 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
987 in question is unbound.
988 ** Optimized MAKE-INSTANCE functions no longer cause internal
989 assertion failures in the presence of duplicate initargs.
990 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
991 ** Side-effectful :DEFAULT-INITARGS have their side-effects
992 propagated even in the ctor optimized implementation of
994 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
996 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
997 * a better implementation of SXHASH on (simple) bit vectors,
998 measured both in execution speed and in distribution of results
999 over the positive fixnums, has been installed. Likewise, a better
1000 implementation of EQUAL for simple bit vectors is now available.
1001 * fixed CEILING optimization for a divisor of form 2^k.
1002 * fixed bug 240 (emitting extra style warnings "using the lexical
1003 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1004 by Antonio Martinez)
1005 * fixed SXHASH, giving different results for NIL depending on type
1006 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1007 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1008 variable. (found by Rolf Wester)
1009 * SBCL does not ignore type declarations for special
1010 variables. (reported by rif on c.l.l 2003-03-05)
1011 * some bug fixes in contrib/sb-aclrepl/
1012 * fixed some bugs revealed by Paul Dietz' test suite:
1013 ** a bug in the CONS type specifier, whereby the CAR and CDR
1014 types got intertwined, has been fixed;
1015 ** the type system is now able to reason about the interaction
1016 between INTEGER and RATIO types more completely;
1017 ** APPEND, [N]REVERSE and NRECONC check that those their
1018 arguments, which must be proper lists, are really so;
1019 ** An array specialized to be unable to hold elements has been
1020 implemented, as required -- yes, really -- by ANSI;
1021 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1022 on malformed property lists;
1024 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1025 * incompatible packaging change: in line with Unix convention,
1026 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1027 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1028 * REQUIRE and PROVIDE are now optionally capable of doing something
1029 useful. See the documentation string for REQUIRE.
1030 * infrastructure for a managed SBCL contrib system: contributed
1031 modules in this release include:
1032 ** the ASDF system definition facility;
1033 ** an interface to the BSD Sockets API;
1034 ** an ACL-like convenience interface to the repl;
1035 (thanks to Kevin Rosenberg)
1036 ** an implementation of ROTATE-BYTE, with efficient implementation
1038 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1039 gives a closer approximation to the right answer than previously.
1040 (thanks to Raymond Toy)
1041 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1042 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1043 optional environment arguments, as required by ANSI.
1044 * fixed bugs in other functions taking environment objects, allowing
1045 calls with an explicit NIL environment argument to be compiled
1047 * fixed bug 228: primary return values from
1048 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1049 COMPILE or FUNCTION.
1050 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1051 :TYPE will no longer signal errors on innocuous objects.
1052 * fixed bug 231b: SETQ is better at respecting type declarations in
1053 the lexical environment.
1054 * fixed a bug in DEFCLASS: classes named by symbols with no or
1055 unprintable packages can now be defined.
1056 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1057 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1058 * fixed compiler failure related to checking types of functions.
1059 (reported by Robert E. Brown)
1060 * the compiler is now much more consistent in its error-checking
1061 treatment of bounding index arguments to sequence functions: in
1062 (SAFETY 3) code, errors will be signalled in almost all cases if
1063 invalid sequence bounding indices are passed to functions defined
1064 by ANSI to operate on sequences.
1065 * fixed a bug in the build procedure: documentation of SBCL-specific
1066 packages is now preserved and available in the final Lisp image.
1067 * lifted FDEFINITION lookup out of loops in the implementation of
1068 many list operations. (thanks to Robert E. Brown)
1069 * fixed a bug in the reader: the #n# reader macro now works for
1070 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1071 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1072 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1073 E. Brown and Tony Martinez respectively)
1074 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1075 SLOT-BOUNDP: in particular, it is now able to optimize them much
1076 better, and is now not vulnerable to having packages renamed.
1077 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1078 always called when a slot is not present in an instance. (thanks
1080 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1081 longer signal an error on unreading a character following EOF on
1082 the previous constituent stream. (thanks to Tony Martinez)
1083 * fixed some bugs revealed by Paul Dietz' test suite:
1084 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1085 not just nonnegative fixnums;
1086 ** the logical bit-array operators such as BIT-AND now accept an
1087 explicit NIL for their "opt-arg" argument (to indicate a
1088 freshly-consed result bit-array);
1089 ** ELT now signals an error on an invalid sequence index in safe
1091 ** the type system is now cleverer about negations of numeric
1092 types, and consequently understands the BIGNUM and RATIO types
1094 ** the type system is now cleverer about the interaction between
1095 INTEGER and RATIO types: while bugs still remain, many more
1096 cases are accurately computed;
1097 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1098 if it is in the last clause;
1099 ** CONSTANTLY now correctly returns a side-effect-free function in
1101 ** DECLARE is no longer treated as a special-operator; in
1102 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1103 * incremented fasl file version number due to the change in the
1104 DEFSTRUCT-SLOT-DESCRIPTION structure.
1106 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1107 * minor incompatible change: code processed by the "interpreter" or
1108 EVAL now has a compilation optimization policy of (DEBUG 2)
1109 (changed from (DEBUG 1)) to improve debuggability of interactive
1110 development, and to allow the use of the debug RETURN command in
1112 * an experimental implementation of the RETURN command for the
1113 debugger has been included. (thanks to Frederik Kuivinen)
1114 * fixed bug 62: constraints were not propagated into a loop.
1115 * fixed bug in embedded calls of SORT (reported and investigated by
1117 * fixed some bugs revealed by Paul Dietz' test suite:
1118 ** printing and reading of arrays with some dimensions having
1119 length 0 (thanks to Gerd Moellmann);
1120 ** BOA constructor with &AUX argument without a default value does
1121 not cause a type error;
1122 ** CONSTANTP now returns true for all self-evaluating objects.
1124 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1125 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1126 accessors that are related by inheritance, as specified in the
1127 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1128 Valtteri Vuorikoski)
1129 * The compiler is now able to inline functions that were defined in
1130 a complex lexical environment (e.g. inside a MACROLET).
1131 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1133 * Support for the upcoming FreeBSD-5.0 release has been included.
1134 (thanks to Dag-Erling Smorgrav)
1135 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1136 effect when it is not in a toplevel context.
1137 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1138 stand a better chance of being correct. (thanks to Gerd
1140 * fixed bug in COERCE, which now signals an error on coercing a
1141 rational to a bounded real type which excludes the expected
1143 * The compiler is now able to derive types more accurately from the
1144 COERCE and COMPILE functions.
1145 * fixed bug 223: functional binding is considered to be constant
1146 only for symbols in the CL package.
1147 * fixed bug 231: SETQ did not check the type of a variable being set
1148 (reported by Robert E. Brown)
1149 * A new optimization for MAKE-INSTANCE has been included, fixing
1150 various bugs (including relating to :ALLOCATION :CLASS slots and
1151 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1153 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1154 ** As required by ANSI, LOOP now disallows anonymous collection
1155 clauses such as COLLECT I in conjunction with aggregate boolean
1156 clauses such as THEREIS (= I 1);
1157 ** LOOP now signals an error when any variable is reused in the
1158 same loop (including the potentially useful construct analogous
1159 to WITH A = 1 WITH A = (1+ A);
1160 ** IT is only a special loop symbol within the first clause of a
1161 conditional loop clause;
1162 ** LOOP with a typed iteration variable over a hashtable now
1163 signals a type error iff it should.
1164 * fixed some other bugs revealed by Paul Dietz' test suite:
1165 ** FILE-STREAM now names the class previously known as FD-STREAM;
1166 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1167 argument) no longer signals an error;
1168 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1169 of the slot symbol, rather than using the current package
1170 ((:CONC-NAME "") continues to intern the slot's name in the
1172 * incremented fasl file version number, because of the incompatible
1173 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1174 of the new implementation of DEFINE-COMPILER-MACRO.
1176 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1177 * Support for building SBCL for MIPS platforms running in
1178 little-endian mode has now been checked in, and basic
1179 functionality on said platforms verified.
1180 * minor incompatible change: PCL now records the pathname of a file
1181 in which methods and the like are defined, rather than its
1183 * minor incompatible change: TRUENAME now considers the truename of
1184 a file naming a directory to be the pathname with :DIRECTORY
1185 component indicating that directory.
1186 * minor incompatible change: a NAMED clause in the extended form of
1187 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1188 reason for the previous behaviour is unclear.
1189 * more systematization and improvement of CLOS and MOP conformance
1190 in PCL (thanks to Gerd Moellman and Pierre Mai):
1191 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1193 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1194 primary methods with no specializers;
1195 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1197 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1198 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1199 FORWARD-REFERENCED-CLASSes; error reporting on
1200 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1202 ** SXHASH on CLOS instances now uses a slot internal to the
1203 instance to return different numbers on distinct instances,
1204 while preserving the same return value through invocations of
1206 ** DEFMETHOD signals errors when methods with longer incongruent
1207 lambda lists are added to generic functions;
1208 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1209 CLASS, as specified in AMOP;
1210 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1211 on the order returned by the primary method for classes of
1212 class STANDARD-CLASS;
1213 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1214 * fixed some bugs shown by Paul Dietz' test suite:
1215 ** DOLIST puts its body in TAGBODY;
1216 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1218 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1219 value producing form;
1220 ** if more variables are given to PROGV than values, extra
1221 variables are bound and made to have no value;
1222 ** NSUBSTITUTE on list arguments gets the right answer with
1224 ** ELT signals an error of type TYPE-ERROR when the index argument
1225 is not a valid sequence index;
1226 ** LOOP signals (at macroexpansion time) an error of type
1227 PROGRAM-ERROR when duplicate variable names are found;
1228 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1229 ** FUNCALL of special-operators now cause an error of type
1231 ** PSETQ now works as required in the presence of side-effecting
1232 symbol-macro places;
1233 ** NCONC accepts any object as its last argument;
1234 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1236 ** loop-for-as-package does not require a package to be explicitely
1238 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1240 * fixed bug 166: compiler preserves "there is a way to go"
1241 invariant when deleting code.
1242 * fixed bug 172: macro lambda lists with required arguments after
1243 &REST arguments now cause an error to be signalled. (thanks to
1245 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1246 bug. (thanks to Matthew Danish)
1247 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1249 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1250 arguments are lexically rebound. (thanks to Gerd Moellmann and
1252 * fixed bug 194: error messages are now more informative when there
1253 is no primary method applicable in a call to a generic
1254 function. (thanks to Gerd Moellmann)
1255 * fixed bug in command line argument checking (thanks to Julian
1257 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1258 (thanks to Matthew Danish)
1259 * incremented fasl file version number, because of the
1260 SXHASH-related changes in the layout of CLOS data structures
1262 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1263 * minor incompatible change: The runtime (the Unix executable named
1264 "sbcl") is now much pickier about the .core files it will load.
1265 Essentially it now requires .core files to descend from the same
1266 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1267 as the runtime does. (The intent is to prevent the crashes which
1268 can occur, and which can even be reported as mysterious failures,
1269 when people patch the sources or change the build parameters
1270 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1271 sbcl and .core files.)
1272 * fixed bug: VALUES-LIST is no longer optimized away.
1273 * fixed bug 142: The FFI conversion of C string values to Lisp
1274 string values no longer conses excessively. (thanks to Nathan
1275 Froyd porting Raymond Toy's fix to CMU CL)
1276 * began to systematize and improve MOP conformance in PCL (thanks to
1277 Nathan Froyd, Gerd Moellman and Pierre Mai):
1278 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1280 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1281 ** FINALIZE-INHERITANCE is now called on class finalization;
1282 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1283 argument precedence order.
1284 * fixed bug 202: The compiler no longer fails on functions whose
1285 derived types contradict their declared type.
1286 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1287 so it can be non-toplevel.
1288 * The fasl file version number has changed (because of the new
1289 implementation of DEFMACRO).
1290 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1291 safe code, that any length requirement by their type-specifier
1292 argument is valid. The exceptions to this are described in bug
1294 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1295 functions now check the types of their inputs as required by ANSI.
1296 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1297 introduced symbol is DECLAREd to be SPECIAL.
1298 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1299 * fixed bug: :COUNT argument to sequence functions may be negative
1300 * fixed bug: body of DO-SYMBOLS may contain declarations
1301 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1302 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1303 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1304 symbol macro only once
1305 * fixed printing of call frame when argument list is unavailable
1306 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1307 * compiler no longer signals WARNING on unknown keyword
1310 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1311 * A beta-quality port to the mips architecture running Linux,
1312 based on the old CMUCL backend, has been made. It has been tested
1313 on a big-endian kernel, and works sufficiently well to be able to
1314 rebuild itself; it has not been tested in little-endian mode.
1315 * fixed an inconsistency between gencgc.c and purify.c which made
1316 dumping/loading .core files unreliable
1317 * fixed bug 120a: The compiler now deals correctly with IFs where
1318 the consequent is the same as the alternative, instead of
1319 misderiving the return type. (thanks to Alexey Dejneka)
1320 * fixed bug 113: Logical pathnames are now dumpable (the logical
1321 host is resolved at load-time, throwing an error if it is not
1323 * fixed bug 174: FORMAT's error message is slightly clearer when a
1324 non-printing character is used in a format directive.
1325 * fixed several bugs in compiler checking of type declarations, i.e.
1326 violations of the Python "declarations are assertions" principle
1327 (thanks to Alexey Dejneka)
1328 * fixed several bugs in PCL's error checking (thanks to Gerd
1330 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1332 * fixed bug in compilation of functions as first class values
1333 (thanks to Antonio Martinez-Shotton)
1334 * The compiler's handling TYPE-ERRORs which it can prove will
1335 inevitably happen at runtime has been cleaned up and corrected
1336 in several ways. (thanks to Alexey Dejneka)
1337 * improved argument type checking for various basic arithmetic
1338 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1339 lavished on them in the past that they can be compiled in many
1340 ways in different special cases
1341 * fixed bug 181: compiler checks validity of user supplied type
1343 * cleaned up code flushing in optimization: Function calls which
1344 should signal errors for safety purposes (e.g. which ANSI says
1345 should signal errors when their arguments are of incorrect type)
1346 are no longer optimized away.
1347 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1348 * incremented fasl file version number, because changes in the
1349 implementation of sequence functions like COERCE caused
1350 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1351 in old inline expansions) to become undefined. (Actually these
1352 changes were later undone, so we might very well be binary
1353 compatible with 0.7.7 after all, but leaving the version number
1354 incremented seemed like the simplest and most conservative
1357 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1358 * An alpha-quality port to the parisc architecture running Linux,
1359 based on the old CMUCL backend, has been made. This, even more so
1360 than the other backends, should be considered still a work in
1361 progress; known problems include that the Linux kernel in 64-bit
1362 mode does not propagate the correct sigcontext structure to
1363 userspace, and consequently SBCL on a parisc64 kernel will not
1365 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1366 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1367 suppressed.) Also now that the compiler is looking at declarations
1368 in the environment, it checks optimization declarations as well,
1369 and suppresses inlining when (> DEBUG SPEED).
1370 * More fixes have been made to treatment of floating point exception
1371 treatment and other Unix signals. In particular, floating point
1372 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1373 * The detection and handling of control stack exhaustion (infinite
1374 or very deeply nested recursion) has changed. Stack exhaustion
1375 detection is now done by write-protecting pages at the OS level
1376 and applies at all optimization settings; when found, a
1377 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1378 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1379 be caught using IGNORE-ERRORS.
1380 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1381 correctly when transforming recursive LABELS functions to LETs.
1382 (thanks to Alexey Dejneka)
1383 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1384 that are names of constants or global variables.
1385 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1386 alien routines with docstrings.
1387 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1388 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1390 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1391 (thanks to Eric Marsden)
1392 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1393 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1394 * Bug fix: The SPARC backend can now compile functions involving
1395 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1396 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1397 to a functions expecting a PATHNAME-DESIGNATOR.
1398 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1399 lambda lists. (thanks to Alexey Dejneka)
1400 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1401 functions so that EXPORTing the name of the class doesn't cause
1402 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1403 to Antonio Martinez for reporting this)
1404 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1405 constant-folded without failing an assertion. (thanks to Einar
1406 Floystad Dorum for reporting this)
1407 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1408 types are now dealt with more correctly by the compiler.
1409 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1410 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1411 behaviour analogous to RENAME-FILE. This puts its behaviour more
1412 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1414 * The fasl file version number has changed again. (because of the
1415 bug fix involving the names of PCL MAKE-INSTANCE functions)
1417 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1418 * bug fix: Floating point exceptions are treated much more
1419 consistently on the x86/Linux and PPC/Linux platforms.
1420 * Array initialization with :INITIAL-ELEMENT is now much faster for
1421 cases when the compiler cannot open code the array creation, but
1422 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1423 array accesses have also seen a speed increase.
1424 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1425 specified. (This was at the root of some bad interactions between
1426 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1428 * bug fix: Internal error arguments for undefined functions are now
1429 computed correctly on the PPC/Linux platform.
1430 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1431 Raymond Toy's patch for CMU CL)
1432 * Support for the Solaris 9 operating environment has been included
1433 (thanks to Daniel Merritt)
1434 * A very ugly but hopefully complete draft of the missing FFI chapter
1435 of the manual has been created by reformatting the corresponding
1436 CMU CL manual chapter into (currently very ugly and incoherent)
1437 DocBook and bringing it up to date for SBCL behavior. Thus, the
1438 manual is now essentially complete, at least by my extreme
1439 once-and-only-once standards, whereby it's acceptable to refer to
1440 the doc strings of SB-EXT functions as the primary documentation.
1441 * The fasl file version number has changed again, due to cleanup of
1442 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1443 is no longer a static symbol.)
1445 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1446 * SBCL now builds with OpenMCL (version 0.12) as the
1447 cross-compilation host; also, more progress has been made toward
1448 bootstrapping under CLISP.
1449 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1451 * bug 158 fixed: The compiler can now deal with integer loop
1452 increments different from 1; fixing this turned out also to fix
1454 * bug 169 fixed: no more bogus warnings about using lexical bindings
1455 despite the presence of perfectly good SPECIAL declarations (thanks
1456 to David Lichteblau)
1457 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1458 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1459 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1461 * bug 180 fixed: Method combination specifications no longer ignore
1462 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1463 * bug fix: Structure type predicate functions now check their argument
1464 count as they should.
1465 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1466 correctly. (thanks to Pierre Mai)
1467 * minor incompatible change: The --noprogrammer option is deprecated
1468 in favor of the new --disable-debugger option, which is very similar.
1469 (The major difference is that it takes effect at a slightly different
1470 time at startup, causing handling of errors in --sysinit and
1471 --userinit files will be affected differently.) The
1472 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1473 been added to allow this functionality to be controlled from ordinary
1474 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1475 maintainers, who might want to run non-interactive scripts to
1476 build SBCL cores which will later be used interactively.)
1477 * minor incompatible change: The LOAD function no longer, when given
1478 a wild pathname to load, loads all files matching that pathname.
1479 Instead, an error of type FILE-ERROR is signalled.
1481 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1482 * bug 147 fixed: The compiler preserves its block link/count
1483 invariants more correctly now so that it doesn't crash. (thanks
1485 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1487 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1488 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1489 no-value-for-this-character as NIL (as specified by ANSI).
1490 * HOST-NAMESTRING on physical pathnames now returns a string that is
1491 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1492 (thanks to Christophe Rhodes)
1493 * The Alpha port handles icache flushing more correctly. (thanks to
1495 * More progress has been made toward bootstrapping under CLISP. (thanks
1496 to Christophe Rhodes)
1497 * The fasl file format has changed again, because dynamic loading
1498 on OpenBSD (which has non-ELF object files) motivated some cleanups
1499 in the way that foreign symbols are transformed and passed around.
1500 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1501 is no longer treated as whitespace by the reader, but instead as
1502 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1503 |A
\7fB|, instead of A as it used to.
1505 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1506 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1507 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1508 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1509 this as of 0.7.1.45, but was left out of the previous news section
1510 (thanks to Dan Barlow)
1511 * SBCL now runs on the Solaris operating system on SPARC architectures
1512 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1513 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1514 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1515 * SPARC backend cleanups, allowing builds of cores optimized for V8
1516 and V9 SPARCS, and also emission of code targeted to a particular
1517 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1519 * SBCL is closer to bootstrapping under CLISP, thanks to various
1520 fixes by Christophe Rhodes.
1521 * The fasl file format has changed again, to allow the compiler's
1522 INFO database to support symbol macros.
1523 * The user manual (in doc/) is formatted into HTML more nicely.
1524 (thanks to coreythomas)
1525 * The system is smarter about SUBTYPEP relationships, especially
1526 those involving NOT types (including types such as ATOM which are
1527 represented internally using NOT types). Thus SUBTYPEP is less
1528 likely to return (VALUES NIL NIL) in general, and in particular
1529 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1531 * The fasl file format has changed again, because the internal
1532 representation of types now includes a new slot to support the new
1533 SUBTYPEP-of-NOT-types logic.
1534 * (not a change in the main branch of SBCL, but a related prototype
1535 which can hopefully be merged into the main branch of SBCL in the
1536 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1537 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1538 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1539 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1540 dispatch macro character combinations. (thanks to Alexey Dejneka)
1541 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1542 that superficially look like logical namestrings correctly.
1543 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1545 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1546 default supplied-p) for &optional and &key arguments. (thanks to
1549 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1550 * incompatible change: The compiler is now less aggressive about
1551 tail call optimization, doing it only when (> SPACE DEBUG) or
1552 (> SPEED DEBUG). (This is an incompatible change because there are
1553 programs which relied on the old CMU-CL-style behavior to optimize
1554 away their unbounded recursion which will now die of stack overflow.)
1555 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1556 for non-GENCGC systems has been increased to 20M (since that
1557 seems much closer to the likely performance optimum for modern
1558 systems than the old 4M value was)
1559 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1560 and *DEBUG-PRINT-LEVEL*
1561 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1562 of CMU CL's support for SPARC, and various endianness and other
1563 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1564 * new syntactic sugar for the Unix command line: --load foo.bar is now
1565 an alternate notation for --eval '(load "foo.bar")'.
1567 ** The system now detects stack overflow and handles it gracefully,
1568 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1569 optimization settings. (This is a good thing in general, and
1570 its introduction in this version should be particularly timely
1571 for anyone whose code fails because of suppression of tail
1573 ** The system now hunts for the C variable "environ" in a more
1574 devious way, to avoid segfaults when the C library version differs
1575 between compile time and run time. (thanks to Christophe Rhodes)
1576 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1577 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1578 ** The compiler no longer issues bogus style warnings for undefined
1579 classes in the same source file as the DEFCLASSes which defined
1580 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1582 ** fixes in CONDITION class precedence list for undefined function
1583 errors (thanks to Alexei Dejneka)
1584 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1585 correctly. (thanks to Dan Barlow)
1586 ** portability fixes aiming at bootstrapping under CLISP (thanks
1587 to Dave McDonald and Christophe Rhodes)
1588 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1589 ** fixes in type translation and and type inference (thanks to
1591 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1592 ** various fixes in the new ports (thanks to Dan Barlow)
1593 * several changes related to debugging:
1594 ** suppression of tail recursion, as noted above
1595 ** stack overflow detection, as noted above
1596 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1597 is now the default. (For some time encapsulation has been more
1598 reliable than the breakpoint-based :ENCAPSULATE NIL
1599 implementation, at least on X86 systems; and I just noticed that
1600 encapsulation also seems closer to the spirit of the ANSI
1603 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1605 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1606 up properly again. (There was a packaging bug in 0.7.0 which
1607 left their definitions in SB-SYS::LOAD-FOREIGN and
1608 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1609 vital for most things which interface to C-level interfaces,
1610 like extensions working with sockets or databases or
1611 Perl-compatible regexes or whatever, and the need to fix
1612 this bug is the main reason that 0.7.1 was released so
1614 ** DEFGENERIC is now choosier about the methods it redefines, so that
1615 reLOADing a previously-LOADed file containing DEFGENERICs does
1616 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1617 cycle now works as it should. (thanks to Alexey Dejneka)
1618 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1619 fixed by Christophe Rhodes. (He actually submitted this patch
1620 months ago, and I delayed until after 0.7.0.)
1621 ** Bug 111 (internal compiler confusion about runtime checks on
1622 FUNCTION types) was fixed by Alexey Dejneka.
1623 * Some internal cleanups (getting rid of variables which aren't
1624 needed now that the byte interpreter is gone) caused the fasl
1625 file format number to change again.
1627 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1628 * major incompatible change: The default fasl file extension, i.e. the
1629 default extension for files produced by COMPILE-FILE, has changed
1630 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1632 ** There are many changes in the implementation of the compiler.
1633 SBCL is now essentially a compiler-only implementation of ANSI
1634 Common Lisp. EVAL still "interprets" a few special cases, but
1635 almost all the interesting cases are handled by creating
1636 a LAMBDA expression, calling COMPILE on it, then calling
1637 FUNCALL on the result.
1638 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1639 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1640 Since the code is newer, there might still be some new bugs
1641 (though not as many as before Martin Atzmueller's fixes:-). But
1642 the new code is substantially simpler and clearer, and hopefully
1643 any remaining bugs will be simpler, less fundamental, and more
1644 fixable then the bugs in the old code.
1645 ** The revised compiler is still a little unsteady on its feet.
1647 *** The debugging information it produces (particularly the names
1648 of FUNCTION objects) is sometimes much less useful than what
1649 the old compiler produced.
1650 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1651 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1652 within a MACROLET) has been temporarily weakened.
1653 ** There are new compiler optimizations for various functions:
1654 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1655 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1656 *** the math functions TRUNCATE, FLOOR, and CEILING
1657 *** the function-of-all-trades COERCE
1658 Mostly these should be transparent, but there's one
1659 potentially-annoying problem (bug 117): when the compiler
1660 inline-expands a function and does type analysis on the result,
1661 it can create control paths which have type mismatches, and
1662 when it can't prove that those control paths aren't taken,
1663 it will issue WARNINGs about the type mismatches. This is
1664 a particular problem in practice for the new sequence functions.
1665 It's not clear how this should be fixed, and for now, a
1666 workaround is given in the entry for 117 in the BUGS file.
1667 ** (Because of the interaction between the two previous items --
1668 occasional inlining problems and new inline expansions -- some
1669 of the new sequence function optimizations won't really kick in
1670 completely until debugging information, and then inlining, are
1671 straightened out in some future version.)
1672 * minor incompatible changes:
1673 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1674 DIRECTORY no longer implicitly promotes NIL slots of its
1675 pathname argument to :WILD. In particular, when you ask for the
1676 contents of a directory (which you used to be able to do without
1677 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1678 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1679 ** changes in behavior that ANSI explicitly defines to be
1680 implementation dependent:
1681 *** The new compiler-only implementation still conforms with ANSI,
1682 but acts a little different than before. Besides the obvious
1683 changes in performance tradeoffs (that the cost per form passed
1684 to EVAL has gone up, and the cost per form executed by EVAL
1685 has gone down), the behavior of the system changes a little
1686 because there are no longer any interpreted function objects.
1687 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1688 e.g. doing COMPILE on the output of interactive DEFUN is
1690 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1692 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1693 has changed from (:CL) to NIL.
1694 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1695 e.g. #\Newline and #\Tab, don't have names specified in the
1696 ANSI Common Lisp standard, is now based on their ASCII symbolic
1697 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1698 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1699 are no longer used for output.
1700 ** changes in internal implementation constants:
1701 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1702 4 million. (If your application spends a lot of time GCing and
1703 you have a lot of RAM, you might want to experiment with
1704 increasing it even more.)
1705 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1706 However, almost all old code should still continue to work without
1707 immediate update, as SB-C-CALL is now a (deprecated) nickname
1709 ** Old operator names in the style DEF-FOO are now deprecated in
1710 favor of new corresponding names DEFINE-FOO, for consistency with
1711 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1712 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1713 internal symbols, but a few supported extensions like
1714 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1715 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1716 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1717 etc. as you get deeper into recursive calls to the debugger
1718 command loop, instead of the old "5]", "5]]", "5]]]"
1719 sequence. (I was motivated to do this when squabbles between
1720 ILISP and SBCL left me very deeply nested in the debugger. In the
1721 short term, this change will probably provoke more ILISP/SBCL
1722 squabbles, but hopefully it will be an improvement in the long run.)
1723 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1724 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1725 function, since declaiming return types involving aliens is
1726 (1) annoyingly messy to do by hand and (2) vital to efficient
1727 compilation of code which calls such functions.
1728 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1729 longer reexported by the SB-EXT package. They're solely useful
1730 for alien code, so it seems more logical that you should get
1731 them from the SB-ALIEN package, not in SB-EXT.
1732 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1733 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1734 features. Instead, the code that they used to control is always
1735 built into the system.
1736 * many other bug fixes
1737 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1738 advantage of the new EVAL-WHEN stuff and to clean them up in
1739 general, and they are now more ANSI-compliant in a number of
1740 ways. Martin Atzmueller is responsible for a lot of this.
1741 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1743 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1744 *** correct ERROR type for various file operations
1745 *** some fixes for Lisp streams
1746 *** DEFMETHOD syntax checking
1747 *** changing old weird representation of debug information as
1748 strings (which, among their other deficiencies, don't transform
1749 correctly when you rename packages, and don't change their
1750 print representation when you change things like *PACKAGE*
1751 and *PRINT-LENGTH*) to symbols and lists of symbols
1752 He also made several improvements and fixed several bugs in DESCRIBE.
1753 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1755 *** misbehavior of WRITE-STRING/WRITE-LINE
1756 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1757 and several other LOOP problems as well
1758 *** DIRECTORY when similar filenames are present
1759 *** DEFGENERIC with :METHOD options
1760 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1761 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1762 *** argument ordering in FIND with :TEST option
1763 *** mishandled package designator argument in APROPOS-LIST
1764 *** various problems in the backquote readmacro
1765 *** a bug in APROPOS
1766 *** probably some others that I'm not describing very well here,
1767 since the CVS log documents them by reference to sbcl-devel
1768 messages, and the SourceForge archives aren't working well.:-(
1769 ** Dan Barlow improved the Alpha port (and is making progress on the
1770 PPC port, for those of you who think different).
1771 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
1772 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
1773 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
1774 continues to work on the SPARC port (for those of you in a position
1775 to look down upon our little PC-compatible boxes from a great height).
1776 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
1777 and uses that copy, rather than the current dynamic value, when
1778 it's trying to decide whether to truncate output. Thus e.g.
1779 (let ((*print-lines* 50))
1780 (pprint-logical-block (stream nil)
1782 (let ((*print-lines* 8))
1783 (print (aref possiblybigthings i) stream)))))
1784 should now truncate the logical block only at 50 lines, instead of
1785 often truncating it at 8 lines, as it did before.
1786 * The doc/cmucl/ directory, containing old CMU CL documentation
1787 from the time of the fork, is no longer part of the base system.
1788 SourceForge has shut down its anonymous FTP service, and with it
1789 my original plan for distributing the old CMU CL documentation
1790 there. For now, if you need these files you can download an old
1791 SBCL source release and extract them from it.
1792 * The fasl file version number changed again, for dozens of reasons,
1793 some of which are apparent above.
1795 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
1796 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
1797 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
1798 patch, so that DISASSEMBLE output is much nicer.
1799 * The code in the SB-PROFILE package now seems reasonably stable.
1800 I still haven't decided what the final interface should look like
1801 (I'd like PROFILE to interact cleanly with TRACE, since both
1802 facilities use function encapsulation) but if you have a need
1803 for profiling now, you can probably use it successfully with
1804 the current CMU-CL-style interface.
1805 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
1806 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
1807 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
1808 link now returns the dangling link itself, and for similar
1809 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
1810 link itself. (In these cases the old code signalled an error and
1811 looped endlessly, respectively.) Thus, DIRECTORY now works even
1812 in the presence of dangling and cyclic symbolic links.
1813 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
1814 is now a supported extension again, since the consensus on
1815 sbcl-devel was that it can be useful for ordinary development
1816 work, not just for debugging SBCL itself.
1817 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
1818 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
1819 that currently-defined functions might be redefined later with
1820 different return types.
1821 * Hash tables can be printed readably, as inspired by CMU CL code
1822 of Eric Marsden and SBCL code of Martin Atzmueller.
1823 * better error handling in CLOS method combination, thanks to
1824 Martin Atzmueller porting Pierre Mai's CMU CL patches
1825 * more overflow fixes for >16Mbyte I/O buffers
1826 * A bug in READ has been fixed, so that now a single Ctrl-D
1827 character suffices to cause end-of-file on character streams.
1828 In particular, now you only need one Ctrl-D at the command
1829 line (not two) to exit SBCL.
1830 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
1832 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
1833 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
1834 does the right thing.
1835 * The implementation of some type tests, especially for CONDITION
1836 types, is now tidier and maybe faster, due to CMU CL code
1837 originally by Douglas Crosher, ported by Martin Atzmueller.
1838 * Some math functions have been fixed, and there are new
1839 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
1840 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
1841 * (There are also some new optimizers in contrib/*-extras.lisp. Those
1842 aren't built into sbcl-0.6.13, but are a sneak preview of what's
1843 likely to be built into sbcl-0.7.0.)
1844 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
1845 CL, ported to SBCL by Martin Atzmueller)
1846 * DESCRIBE now gives more information in some cases. (Pierre Mai's
1847 patch to CMU CL, ported to SBCL by Martin Atzmueller)
1848 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
1849 * There's a new slam.sh hack to shorten the edit/compile/debug
1850 cycle for low-level changes to SBCL itself, and a new
1851 :SB-AFTER-XC-CORE target feature to control the generation of
1852 the after-xc.core file needed by slam.sh.
1853 * minor incompatible change: The ENTRY-POINTS &KEY argument to
1854 COMPILE-FILE is no longer supported, so that now every function
1855 gets an entry point, so that block compilation looks a little
1856 more like the plain vanilla ANSI section 3.2.2.3 scheme.
1857 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
1858 returns the number of bytes consed since the system started,
1859 rather than the number consed since the first time the function
1860 was called. (The new definition parallels ANSI functions like
1861 CL:GET-INTERNAL-RUN-TIME.)
1862 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
1863 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
1864 supported. Now DIRECTORY always does the abstract Common-Lisp-y
1865 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
1866 * Fasl file version numbers are now independent of the target CPU,
1867 since historically most system changes which required version
1868 number changes have affected all CPUs equally. Similarly,
1869 the byte fasl file version is now equal to the ordinary
1872 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
1873 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
1874 is no longer recognized. I apologize for this, because it was
1875 listed in SB-EXT as a supported extension, but I found that
1876 its existing behavior was poorly specified, as well as incorrectly
1877 specified, and it looked like too much of a mess to straighten it
1878 out. I have enough on my hands trying to get ANSI stuff to work..
1879 * many patches ported from CMU CL by Martin Atzmueller, with
1880 half a dozen bug fixes in pretty-printing and the debugger, and
1881 half a dozen others elsewhere
1882 * fixed bug 13: Floating point infinities are now supported again.
1883 They might still be a little bit flaky, but thanks to bug reports
1884 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
1885 as flaky as they were.
1886 * The --noprogrammer command line option is now supported. (Its
1887 behavior is slightly different in detail from what the old man
1888 page claimed it would do, but it's still appropriate under the
1889 same circumstances that the man page talks about.)
1890 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
1891 are now supported, and enabled by default. Thus, the compiler can
1892 handle many floating point and complex operations much less
1893 inefficiently. (Thus e.g. you can implement a complex FFT
1895 * The compiler now detects type mismatches between DECLAIM FTYPE
1896 and DEFUN better, and implements CHECK-TYPE more correctly, and
1897 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
1898 thanks to patches from Martin Atzmueller.
1899 * various fixes to make the cross-compiler more portable to
1900 ANSI-conforming-but-different cross-compilation hosts (notably
1901 Lispworks for Windows, following bug reports from Arthur Lemmens)
1902 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
1903 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
1904 fixed thanks to Pierre Mai's CMU CL patches.
1905 * fixes to keep the system from overflowing internal counters when
1906 it tries to use i/o buffers larger than 16M bytes
1907 * fixed bug 45a: Various internal functions required to support
1908 complex special functions have been merged from CMU CL sources.
1909 (When I was first setting up SBCL, I misunderstood a compile-time
1910 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
1911 * improved support for type intersection and union, fixing bug 12
1912 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
1913 more obscure bugs as well
1914 * some steps toward byte-compiling non-performance-critical
1915 parts of the system, courtesy of patches from Martin Atzmueller
1916 * Christophe Rhodes has made some debian packages of sbcl at
1917 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
1918 From his sbcl-devel e-mail of 2001-04-08 they're not completely
1919 stable, but are nonetheless usable. When he's ready, I'd be happy
1920 to add them to the SourceForge "File Releases" section. (And if
1921 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
1922 * new fasl file format version number (because of changes in
1923 internal representation of (OR ..) types to accommodate the new
1924 support for (AND ..) types, among other things)
1926 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
1927 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
1929 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
1931 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
1932 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
1933 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
1934 * DESCRIBE now works on structure objects again.
1935 * Most function call argument type mismatches are now handled as
1936 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
1937 know whether the function will be redefined before the call is
1938 executed. (The compiler could flag local calls with full WARNINGs,
1939 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
1940 it doesn't keep track of enough information to know whether calls
1941 are local in this sense.)
1942 * Compiler output is now more verbose, with messages truncated
1943 later than before. (There should be some supported way for users
1944 to override the default verbosity, but I haven't decided how to
1945 provide it yet, so this behavior is still controlled by the internal
1946 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
1947 src/compiler/ir1util.lisp.)
1948 * Fasl file format version numbers have increased again, because
1949 support for the Gray streams extension changes the layout of the
1950 system's STREAM objects.
1951 * The Gray subclassable streams extension now works, thanks to a
1952 patch from Martin Atzmueller.
1953 * The full LOAD-FOREIGN extension (not just the primitive
1954 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
1955 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
1956 but like most other programs, it defaults to copying the Unix
1957 environment from the original process instead of starting the
1958 new process in an empty environment.
1959 * Extensions which manipulate the Unix environment now support
1960 an :ENVIRONMENT keyword option which doesn't smash case or
1961 do other bad things. The CMU-CL-style :ENV option is retained
1962 for porting convenience.
1963 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
1964 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
1966 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
1968 * A patch from Martin Atzmueller seems to have solved the SIGINT
1969 problem, and as far as we know, signal-handling now works cleanly.
1970 (If you find any new bugs, please report them!)
1971 * The system no longer defaults Lisp source file names to types
1972 ".l", ".cl", or ".lsp", but only to ".lisp".
1973 * The compiler no longer uses special default file extensions for
1974 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
1975 seems to expect a single default extension for all compiled code,
1976 and there's no compelling reason to try to stretch the standard
1977 to allow two different extensions.) Instead, byte-compiled files
1978 default to the same extension as native-compiled files.
1979 * Fasl file format version numbers have increased again, because
1980 a rearrangement of internal implementation packages made some
1981 dumped symbols in old fasl files unreadable in new cores.
1982 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
1983 many fewer weird special cases.
1984 * Bug #17 (differing COMPILE-FILE behavior between logical and
1985 physical pathnames) has been fixed, and some related misbehavior too,
1986 thanks to a patch from Martin Atzmueller.
1987 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
1988 by Tim Moore, ported to SBCL by Martin Atzmueller.
1989 * Martin Atzmueller fixed several filesystem-related problems,
1990 including bug #36, in part by porting CMU CL patches, which were
1991 written in part by Paul Werkowski.
1992 * More compiler warnings in src/runtime/ are gone, thanks to
1993 more patches from Martin Atzmueller.
1994 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
1997 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
1999 * DESCRIBE now works on CONDITION objects.
2000 * The debugger now handles errors which arise when trying to print
2001 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2003 * The build system now uses an additional file, customize-target-features.lisp,
2004 to allow local modifications to the target *FEATURES* list. (The point of
2005 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2006 debugging features enabled, without having to worry about propagating it
2007 into everyone's system when I do a "cvs update".) When no
2008 customize-target-features.lisp file exists, the target *FEATURES* list
2009 should be constructed the same way as before.
2010 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2011 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2012 warn about behavior which is undefined under the ANSI spec. Note: This
2013 is specified by ANSI, but it's not very popular with programmers.
2014 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2015 macro in the SBCL sources for an example of a workaround which you
2016 might use to make portable ANSI-standard code which does what you want.
2017 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2018 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2019 when it's used as a non-toplevel form. (This is required in order
2020 to implement the DEFCONSTANT-EQX macro.)
2021 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2022 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2023 funny things, due to bugs in the implementation of EVAL-WHEN
2024 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2025 non-toplevel DEFCONSTANTs are uncommon.)
2026 * The core file version number and fasl file version number have been
2027 incremented, because the old noncompliant DEFCONSTANT behavior involved
2028 calling functions which no longer exist, and because I also took the
2029 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2030 * fixed bug 1 (error handling before read-eval-print loop starts), and
2031 redid debugger restarts and related debugger commands somewhat while
2033 ** The QUIT debugger command is gone, since it did something
2034 rather different than the SB-EXT:QUIT command, and since it never
2035 worked properly outside the main toplevel read/eval/print loop.
2036 Invoking the new TOPLEVEL restart provides the same functionality.
2037 ** The GO debugger command is also gone, since you can just invoke
2038 the CONTINUE restart directly instead.
2039 ** The TOP debugger command is also gone, since it's redundant with the
2040 FRAME 0 command, and since it interfered with abbreviations for the
2042 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2044 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2045 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2046 some porting work by Martin Atzmueller)
2047 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2048 with a boatload of patches which he ported to SBCL. Now that those
2050 ** The system tries to make sure that its low-priority messages
2051 are prefixed by semicolons, to help people who like to use
2052 syntax highlighting in their ILISP buffer. (This patch
2053 was originally due to Raymond Toy.)
2054 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2055 patches originally written by Raymond Toy.
2056 ** The compiler understands coercion between single-value and
2057 multiple-VALUES type expressions better, getting rid of some very
2058 weird behavior, thanks to patches originally by Robert MacLachlan
2059 and Douglas Crosher.
2060 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2061 lambda lists, thanks to a patch originally by Pierre Mai.
2062 ** The system no longer bogusly warns about "abbreviated type
2064 ** The compiler gets less confused by inlining and RETURN-FROM,
2065 thanks to some patches originally by Tim Moore.
2066 ** The system no longer hangs when dumping circular lists to fasl
2067 files, thanks to a patch originally from Douglas Crosher.
2068 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2069 undefined function error.
2070 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2071 in structure slot initforms) for now, documented workaround instead:-|
2072 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2073 * fixed bug 5: added stubs for various Gray stream functions called
2074 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2075 installed, at least appropriate type errors are generated
2076 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2077 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2078 consistently in DEFMETHOD forms.
2079 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2080 it doesn't seem to affect SBCL after all
2081 * The C runtime system now builds with better optimization and many
2082 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2084 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2086 * The system is now under CVS at SourceForge (instead of the
2087 CVS repository on my home machine).
2088 * The new signal handling code has been tweaked to treat register
2089 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2090 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2091 downstream, e.g. in debug-int.lisp, has implicit dependencies
2092 on the unsignedness of integer representation of machine words,
2093 and that caused the system to bomb out with infinite regress
2094 when trying to recover from type errors involving signed values,
2095 e.g. (BUTLAST '(1 2 3) -1).)
2096 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2097 (This was one of the bugs Peter Van Eynde reported back in July.)
2098 * The system now uses code inspired by Colin Walters' O(N)
2099 implementation of MAP (from the cmucl-imp@cons.org mailing
2100 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2101 inline the MAP operation, and there is more than one
2102 sequence argument to the MAP call (so that it can't just
2103 do ETYPECASE once and for all based on the type of the
2104 single sequence argument). (The old non-inline implementation
2105 of the general M-argument sequence-of-length-N case required
2106 O(M*N*N) time when any of the sequence arguments were LISTs.)
2107 * The QUIT :UNIX-CODE keyword argument has been renamed to
2108 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2109 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2111 * Raymond Wiker's patches to port dynamic loading from Linux to
2112 FreeBSD have been added.
2113 * The BUGS file is now more nearly up to date, thanks in large part
2114 to Martin Atzmueller's review of it.
2115 * The debugger now flushes standard output streams before it begins
2116 its output ("debugger invoked" and so forth).
2117 * The core version number and fasl file version number have both
2118 been incremented, because of incompatible changes in the layout
2120 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2121 used to not be called for a saved Lisp image.)
2122 * Martin Atzmueller's version of a patch to fix a compiler crash,
2123 as posted on sbcl-devel 13 September 2000, has been installed.
2124 * Instead of installing Martin Atzmueller's patch for the
2125 compiler transform for SUBSEQ, I deleted the compiler transform,
2126 and transforms for some similar consing operations.
2127 * A bug in signal handling which kept TRACE from working on OpenBSD
2129 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2130 away by constant folding
2131 * The system now defines its address space constants in one place
2132 (in the Lisp sources), and propagates them automatically elsewhere
2133 (through GENESIS and the sbcl.h file). Therefore, patching the
2134 address map is less unnecessarily tedious and error-prone. The
2135 Lisp names of address space constants have also been systematized.
2136 * CVS tags like dollar-Header-dollar have been removed from
2137 the sources, because they have never saved me trouble and
2138 they've been source of trouble working with patches and other
2139 diff-related operations.
2140 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2141 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2143 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2145 * The system has been ported to OpenBSD.
2146 * The system now compiles with a simple "sh make.sh" on the systems
2147 that it's supported on. I.e., now you no longer need to tweak
2148 text in the target-features.lisp-expr and symlinks in src/runtime/
2149 by hand, the make.sh takes care of it for you.
2150 * The system is no longer so grossly inefficient when compiling code
2151 involving vectors implemented as general (not simple) vectors (VECTOR T),
2152 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2153 which dares to use the various sequence functions on non-simple
2154 vectors, takes less of a performance hit.
2155 * There is now a primitive type predicate VECTOR-T-P
2156 to test for the (VECTOR T) type, so that e.g.
2157 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2158 can now be compiled with some semblance of efficiency. (The old code
2159 turned the type declaration into a full call to %TYPEP at runtime!)
2160 * AREF on (VECTOR T) is still not fast, since it's still compiled
2161 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2162 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2163 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2164 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2165 without making full calls to SUBTYPEP at runtime.
2166 (Some analogous efficiency issues for non-simple vectors specialized to
2167 element types other than T, or for non-simple multidimensional arrays,
2168 have not been addressed. They could almost certainly be handled the
2169 same way if anyone is motivated to do so.)
2170 * The changes in array handling break binary compatibility, so
2171 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2172 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2173 instead of (VALUES T T).
2174 * By following the instructions that Dan Barlow posted to sbcl-devel
2175 on 2 July 2000, I was able to enable primitive dynamic object
2176 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2177 functionality is not implemented (since it calls ld to resolve
2178 library references automatically, requiring RUN-PROGRAM for its
2179 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2180 try to resolve library references) is now supported.
2181 * The system now flushes the standard output streams when it terminates,
2182 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2183 them at several other probably-convenient times, e.g. in each pass of
2184 the toplevel read-eval-print loop, and after evaluating a form given
2185 as an "--eval" command-line option. (These changes were motivated by a
2186 discussion of stream flushing issues on cmucl-imp in August 2000.)
2187 * The source transform for TYPEP of array types no longer assumes
2188 that an array whose element type is a not-yet-defined type
2189 is implemented as an array of T, but instead punts, so that the
2190 type will be interpreted at runtime.
2191 * There is now some support for cross-compiling in make.sh: each of
2192 the phases of make.sh has its own script. (This should be transparent
2193 to people doing ordinary, non-cross-compile builds.)
2194 * Since my laptop doesn't have hundreds of megabytes of memory like
2195 my desktop machine, I became more motivated to do some items on
2196 my to-do list in order to reduce the size of the system a little:
2197 ** Arrange for various needed-only-at-cold-init things to be
2198 uninterned after cold init. To support this, those things have
2199 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2200 symbols with such names are now uninterned after cold init).
2201 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2202 fasl files for cold load.
2203 ** Remove the old compiler structure pooling code (which used to
2204 be conditional on the target feature :SB-ALLOC) completely.
2205 ** Redo the representation of some data in cold init to be more compact.
2206 (I also looked into supporting byte compiled code at bootstrap time,
2207 which would probably reduce the size of the system a lot, but that
2208 looked too complicated, so I punted for now.)
2209 * The maximum signal nesting depth in the src/runtime/ support code has
2210 been reduced from 4096 to 256. (I don't know any reason for the very
2211 large old value. If the new smaller value turns out to break something,
2212 I'll probably just bump it back up.)
2213 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2215 * Many, many bugs reported by Peter Van Eynde have been added to
2216 the BUGS list; some have even been fixed.
2217 * While enabling dynamic object file loading, I tried to make the
2218 code easier to understand, renaming various functions and variables
2219 with less ambiguous names, and changing some function calling
2220 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2221 * While trying to figure out how to do the OpenBSD port, I tried to
2222 clean up some of the code in src/runtime/. In particular, I dropped
2223 support for non-POSIX signal handling, added various comments,
2224 tweaked the code to reduce the number of compilation warnings, and
2225 renamed some files to increase consistency.
2226 * To support the new automatic configuration functionality in make.sh,
2227 the source file target-features.lisp-expr has been replaced with the
2228 source file base-target-features.lisp-expr and the machine-generated
2229 file local-target-features.lisp-expr.
2230 * fixed a stupid quoting error in make.sh so that using CMU CL
2231 "lisp -batch" as cross-compilation host works again
2233 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2235 * DESCRIBE no longer tries to call itself recursively to describe
2236 bound/fbound values, so that it no longer fails on symbols which are
2237 bound to themselves (like keywords, T, and NIL).
2238 * DESCRIBE now works on generic functions.
2239 * The printer now prints less-screwed-up representations of closures
2240 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2241 * A private symbol is used instead of the :EMPTY keyword previously
2242 used to mark empty slots in hash tables. Thus
2243 (DEFVAR *HT* (MAKE-HASH-TABLE))
2244 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2245 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2246 now does what ANSI says that it should. (You can still get
2247 similar noncompliant behavior if bang on the hash table
2248 implementation with all the symbols you get back from
2249 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2250 This breaks binary compatibility, since tests for equality to
2251 :EMPTY are wired into things like the macroexpansion of
2252 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2254 * There's now a minimal placeholder implementation for CL:STEP,
2255 as required by ANSI.
2256 * An obscure bug in the interaction of the normal compiler, the byte
2257 compiler, inlining, and structure predicates has been patched
2258 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2259 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2261 * Missing ordinary arguments in a macro call are now detected even
2262 when the macro lambda list contains &KEY or &REST.
2263 * The debugger no longer complains about encountering the top of the
2264 stack when you type "FRAME 0" to explicitly instruct it to go to
2265 the top of the stack. And it now prints the frame you request even
2266 if it's the current frame (instead of saying "You are here.").
2267 * As specified by ANSI, the system now always prints keywords
2268 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2270 * The default initial SIZE of HASH-TABLEs is now smaller.
2271 * Type information from CLOS class dispatch is now propagated
2272 into DEFMETHOD bodies, so that e.g.
2273 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2275 is now basically equivalent to
2276 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2277 (DECLARE (TYPE SINGLE-FLOAT X))
2279 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2280 operation, without having to do run-time type dispatch.
2281 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2282 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2283 * Since I don't seem to be making big file reorganizations very often
2284 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2285 is overflowing my ability to conveniently back them up), I've finally
2286 checked the system into CVS. (The CVS repository is on my home system,
2287 not at SourceForge -- putting it on SourceForge might come later.)
2288 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2289 high-level GC-NOTIFY-FOO functions send their output. (There's
2290 still very little control of where low-level verbose GC functions
2291 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2292 less than it used to -- the GC-NOTIFY-FOO functions are now under
2293 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2294 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2295 in only one place in the source code, and propagates it automatically
2296 everywhere that it's needed. Thus e.g. when I bump the version from
2297 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2298 * The C source files now include boilerplate legalese and documentation
2299 at the head of each file (just as the Lisp source files already did).
2300 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2301 to his page will be replaced with a link to his new CLiki service.
2303 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2305 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2306 * The build process now looks for GNU make under the default name "gmake",
2307 instead of "make" as it used to. If GNU make is not available as "gmake"
2308 on your system, you can change this default behavior by setting the
2309 GNUMAKE environment variable.
2310 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2311 can build without error under CMU CL.
2313 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2315 * There is now a partial SBCL user manual (with some new text and some
2316 text cribbed from the CMU CL manual).
2317 * The beginnings of a profiler have been added (starting with the
2318 CMU CL profiler and simplifying and cleaning up). Eventually the
2319 main interface should be through the TRACE macro, but for now,
2320 it's still accessed through vaguely CMU-CL-style functions and macros
2321 exported from the package SB-PROFILE.
2322 * Some problems left over from porting CMU CL to the new
2323 cross-compilation bootstrap process have been cleaned up:
2324 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2325 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2326 part when getting rid of the compiler *BACKEND* stuff.)
2327 ** The value of *NULL-TYPE* was screwed up, because it was
2328 being initialized before the type system knew the final
2329 definition of the 'NULL type. This screwed up several key
2330 optimizations in the compiler, causing inefficiency in all sorts
2331 of places. (I found it because I wanted to understand why
2332 GET-INTERNAL-RUN-TIME was consing.)
2333 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2334 PROCLAIM FTYPE information. Unfortunately this broke binary
2335 compatibility again, since now the forms output by DEFGENERIC
2336 to refer to functions which didn't exist in 0.6.3.
2337 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2338 can use the new (as of 0.6.3) transform for SOME into MAP into
2340 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2341 values to (INTEGER 0 1000000), so that the time code will no longer
2342 occasionally get blown up by Linux returning 1000000 microseconds
2343 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2344 its output conform to the ANSI spec. (Alas, this makes its output
2345 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2346 * A full call to MAP NIL with a single sequence argument no longer conses.
2347 * fixes to problems pointed out by Martin Atzmueller:
2348 * The manual page no longer talks about multiprocessing as though
2349 it were currently supported.
2350 * The ILISP support patches have been removed from the distribution,
2351 because as of version 5.10.1, ILISP now supports SBCL without us
2352 having to maintain patches.
2353 * added a modified version of Raymond Toy's recent CMU CL patch for
2354 EQUALP comparison of HASH-TABLE
2356 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2358 * The system still can't cross-compile itself with
2359 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2360 entails), but at least it can compile more of itself that way
2361 than it used to be able to, and various buglets which were uncovered
2362 by trying to cross-compile itself that way have now been fixed.
2363 * This release breaks binary compatibility again. This time
2364 at least I've incremented the FASL file format version to 2, so that the
2365 problem can be detected reliably instead of just causing weird errors.
2366 * various new style warnings:
2367 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2368 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2369 ** using the lexical binding of a variable named in the *FOO* style
2370 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2371 as specified by ANSI.
2372 * *RANDOM-STATE* is no longer automatically initialized from
2373 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2374 default behavior of the system is to repeat its behavior every time
2375 it's run. If you'd like to change this behavior, you can always
2376 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2377 old convention there was no comparably easy way to get the system to
2378 repeat its behavior every time it was run.
2379 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2380 FTYPE declarations has been removed, in favor of their ANSI
2381 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2382 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2383 the types of their sequence arguments can be determined at compile time.
2384 This is done through a new open code expansion for MAP which eliminates
2385 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2386 argument types can be determined at compile time.
2387 * The optimizer now transforms COERCE into an identity operation when it
2388 can prove that the coerced object is already of the correct type. (This
2389 can be a win for machine generated code, including the output of other
2390 optimization transforms, such as the MAP transform above.)
2391 * Credit information has been moved from source file headers into CREDITS.
2392 * Source file headers have been made more standard.
2393 * The CASE macro now compiles without complaining even when it has
2396 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2398 * (Note that the way that the PCL macroexpansions were rewritten
2399 to accommodate the change in DEFGENERIC below breaks binary
2400 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2401 not run under sbcl-0.6.2. Once we get out of alpha releases,
2402 i.e. hit release 1.0.0, we'll probably try to maintain binary
2403 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2404 and sbcl-1.4.4. Until then, however, it might be fairly common
2405 for maintenance releases to break binary compatibility.)
2406 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2408 * The reporting of unhandled signals has been changed to print some
2409 explanatory text as well as the report form. (Previously only
2410 the report form was printed.)
2411 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2412 it defines, so that the compiler no longer issues undefined function
2413 warnings for compiled-but-not-yet-loaded generic functions.
2414 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2415 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2416 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2417 section of the ANSI spec are supported.
2418 * The "" nickname for the "KEYWORD" package has been removed.
2419 The reader still handles symbol tokens which begin with a package marker
2420 as keywords, but it doesn't expose its mechanism for doing so in the
2421 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2422 * The system now issues STYLE-WARNINGs for contradictory TYPE
2423 proclamations. (Warnings for contradictory FTYPE proclamations would
2424 be nice too, but those can't be done usefully unless the type system
2425 is made smarter about FUNCTION types.)
2426 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2427 systematized, so that "*target-*.lisp is supposed to exist only on the
2428 target and imply that there's a related file which exists on the
2429 host, and *host-*.lisp is supposed to exist only on the host and imply
2430 that there's a related file which exists on the target. This involves a
2431 lot of renaming. Hopefully the acute confusion caused by the renaming
2432 will be justified by the reduction in chronic confusion..
2433 ** runtime-type.lisp -> early-target-type.lisp
2434 ** target-type.lisp -> late-target-type.lisp
2435 ** early-host-format.lisp -> early-format.lisp
2436 ** late-host-format.lisp -> late-format.lisp
2437 ** host-error.lisp -> misc-error.lisp
2438 ** early-error.lisp -> early-target-error.lisp
2439 ** late-error.lisp -> late-target-error.lisp
2440 ** host-defboot.lisp -> early-defboot.lisp
2441 ** code/misc.lisp -> code/target-misc.lisp
2442 ** code/host-misc.lisp -> code/misc.lisp
2443 ** code/numbers.lisp -> code/target-numbers.lisp
2444 ** code/early-numbers.lisp -> numbers.lisp
2445 ** early-host-type.lisp -> early-type.lisp
2446 ** late-host-type.lisp -> late-type.lisp
2447 ** host-typep.lisp -> typep.lisp
2448 ** load.lisp -> target-load.lisp
2449 ** host-load.lisp -> load.lisp
2450 ** host-disassem.lisp -> disassem.lisp
2451 ** host-insts.lisp -> insts.lisp
2452 ** byte-comp.lisp -> target-byte-comp.lisp
2453 ** host-byte-comp.lisp -> byte-comp.lisp
2454 ** host-signal.lisp -> signal.lisp
2455 ** host-defstruct.lisp -> defstruct.lisp
2456 ** late-target-type.lisp -> deftypes-for-target.lisp
2457 Furthermore, several other previously target-only files foo.lisp (e.g.
2458 hash-table.lisp and random.lisp) have been split into a target-and-host
2459 foo.lisp file and a target-only target-foo.lisp file, with their key type
2460 definitions in the target-and-host part, so that the cross-compiler will
2461 know more about target types.
2462 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2463 gone away. In their place are various *BACKEND-FOO* variables
2464 corresponding to the slots of the old structure.
2465 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2466 into the target SBCL has been fixed.
2467 * The chill.lisp system for loading cold code into a running SBCL
2469 * Support for the CMU CL "scavenger hook" extension has been removed.
2470 (It was undocumented and unused in the CMU CL sources that SBCL was
2471 derived from, and stale in sbcl-0.6.1.)
2472 * Various errors in the cross-compiler type system were detected
2473 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2474 (enabling various consistency checks). Many of them were fixed,
2475 but some hard problems remain, so the compiler is back to
2476 running without *TYPE-SYSTEM-INITIALIZED* for now.
2477 * As part of the cross-compiler type system cleanup, I implemented
2478 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2479 * I have started adding UNCROSS calls throughout the type system
2480 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2481 on cross-compiler input files will be able to go away, and various
2483 * CONSTANTP now returns true for quoted forms (as explicitly required
2486 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2488 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2489 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2490 (MAKE-LIST -1). In the longer term, it ought to become true
2491 that declarations are assertions even at SAFETY 1. For now, it's not
2492 quite true even at SAFETY 3, but it's at least more nearly true..
2493 (Note that this change seems to increases the size of the system by
2494 O(5%) and to decrease the speed of the compiler by 20% or more.)
2495 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2496 for the problem of printing dozens of lines of distracting information
2497 about low-level system machinery as part of the top stack frame
2498 on entry to the debugger when an undefined function was called.
2499 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2500 is not reset to COMMON-LISP-USER.
2501 * Compilation of stuff related to dyncount.lisp has been made conditional
2502 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2503 smaller. The various dyncount-related symbols have been moved into
2504 a new "SB-DYNCOUNT" package.
2505 * tty-inspect.lisp has been renamed to inspect.lisp.
2506 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2507 feature has gone away. (When we eventually port to other flavors of
2508 libc and/or Unix, we'll try to make the differences between flavors
2509 invisible at the user level.)
2510 * Various other *FEATURES* tags, and/or their associated conditionals,
2511 have been removed if obsolescent, or given better documentation, or
2512 sometimes given more-mnemonic names.
2514 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2516 * tidied up "make.sh" script
2517 * tidied up system directory structure
2518 * better "clean.sh" behavior
2519 * added doc/FOR-CMUCL-DEVELOPERS
2520 * many many small tweaks to output format, e.g. removing possibly-confusing
2521 trailing #\. character in DESCRIBE-INSTANCE
2522 * (EQUALP #\A 'A) no longer signals an error.
2523 * new hashing code, including EQUALP hashing
2524 * tidied up Lisp initialization and toplevel
2525 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2526 * command line argument processing
2527 * added POSIX-GETENV function to deal with Unix-ish environment variables
2528 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2529 terminating SBCL on EOF
2530 * non-verbose GC by default
2531 * There is no more "sbcl" shell script; the sbcl file is now the C
2532 runtime executable (just like CMU CL).
2533 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2535 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2536 symbol and package information as Lisp native symbol and package objects
2537 instead of strings naming symbols and strings naming packages. This way,
2538 whenever packages are renamed (as in warm init), debug information is
2539 transformed along with everything else.
2540 * tweaked the optimization policy declarations which control the building
2541 of SBCL itself. Now, among other things, the system no longer saves
2542 source location debugging information. (This helps two problems at once
2543 by reducing SBCL size and by keeping SBCL from trying to look for its
2544 sources -- which may not exist -- when reporting errors.)
2545 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2546 debugging and testing purposes
2547 * cleaned up printing, making the printer call PRINT-OBJECT for
2548 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2549 methods, giving nearly-ANSI behavior
2550 * converted almost all special variables to use *FOO* naming convention
2551 * deleted PARSE-TIME functionality, since it can be done portably
2552 * moved some files out of cold init into warm init
2553 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2555 * regularized formatting of source files
2556 * added an install.sh script
2557 * fixed ridiculous memory usage of cross-compiler by making
2558 compiler/alloc.lisp not try to do pooling unless it can hook
2559 itself into the GC of the cross-compilation host. Now the system
2560 builds nicely on my old laptop.
2561 * added :SB-ALLOC in target-features.lisp-expr
2562 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2563 was not implemented)
2564 * re-did condition handling and note reporting in the compiler. Notes
2565 are no longer handled by signalling conditions. Style warnings
2566 and warnings are handled more correctly and reported in such a way
2567 that it's easy to find one or the other in your output (so that you
2568 can e.g. figure out which of many problems caused COMPILE-FILE to
2570 * changed the severity of several compiler warnings from full WARNING
2571 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2572 compiler note reporting so that it doesn't use the condition system
2573 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2574 COMPILE-FILE command)
2575 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2576 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2577 no longer supported, since their implementation was deeply intertwingled
2578 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2579 * removed bogus "support" for compiler macros named (SETF FOO), and
2580 removed the compiler macro for SETF INFO (but only after making a fool
2581 of myself on the cmucl-imp mailing list by posting a bogus patch for
2582 DEFINE-COMPILER-MACRO..)
2583 * Compiled files containing forms which have side effects on the Lisp
2584 reader (such as DEFPACKAGE forms) are now handled more correctly.
2585 (Compiler queuing of top level lambdas has been suppressed by setting
2586 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2587 * deleted various currently-unused source files, e.g. gengc.lisp. They
2588 may be added back at some point e.g. when porting to other architectures,
2589 but until they are it's distracting to distribute them and to try to
2591 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2592 there were so many of them they're just distractions, and UNCROSS is
2593 known to be able to handle the current sources
2594 * moved PROFILE functionality into TRACE, so that it will be clear
2595 how the wrapping and unwrapping of functions when you profile them
2596 interacts with the wrapping and unwrapping of functions when you
2597 trace them. (Actually, the functionality isn't there yet, but at least
2598 the interface specification is there. Hopefully, the functionality will
2599 arrive with some future maintenance release.)
2600 * removed host-oops.lisp
2601 * changed signature of QUIT function to allow UNIX-CODE argument
2602 * fixed READ-SEQUENCE bug
2603 * tweaked verbose GC output so that it looks more like the progress
2604 output that ANSI specifies for functions like LOAD
2605 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2606 * added <http://sbcl.sourceforge.com> to the banner information printed by