1 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
3 the manual for details; for now, package locks can be disabled by
4 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
5 if no major problems are found then it is likely that they will be
6 compiled in unconditionally.
7 * minor incompatible change: as threatened around sbcl-0.8.0, the
8 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
9 The symbols are also exported from SB-PCL for backwards
10 compatibility, but more so than before SB-PCL should be treated as
11 an implementation-internal package.
12 * the SB-SPROF contrib now works on (most) non-x86 architectures.
13 It is known as of this release not to work on the Alpha, however.
14 * fixed bug #167: errors signalled due to illegal syntax in method
15 bodies are now more legible.
16 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
17 designators and can hence be used with TYPEP.
18 * fixed bug #333: CHECK-TYPE now ensures that the type error
19 signalled, if any, has the right object to be accessed by
20 TYPE-ERROR-DATUM. (reported by Tony Martinez)
21 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
22 5.1.2.3 for multiple-value place subforms. (reported by Kalle
24 * fixed bug #334: programmatic addition of slots using specialized
25 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
26 and :ALLOCATION :CLASS slots.
27 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
28 float exponents. (rereported by Peter Seibel)
29 * fixed a bug: #\Space (and other whitespace characters) are no
30 longer considered to be macro characters in standard syntax by
32 * fixed bug: initialization of condition class metaobjects no longer
33 causes an instance of the condition to be created. (reported by
35 * fixed bug: it is now possible to have slots such that
36 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
37 description is neither :INSTANCE nor :CLASS.
38 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
39 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
40 SLOT-VALUE-USING-CLASS), and no longer causes errors with
41 non-standard SLOT-DEFINITION-ALLOCATION values.
42 * optimization: improved performance of BIT and SBIT on bit-vectors.
44 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
45 * minor incompatible change: the system no longer provides
46 optimization hints (conditions which are TYPEP
47 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
48 mode; these hints are still emitted when higher SPEED optimization
50 * new contrib module: a sampling profiler (profiling by statistical
51 sampling, rather than by instrumenting functions) is available as
52 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
53 * the behaviour of the standard function ED is now customizeable by
54 third parties through a hook variable: see ED's documentation
55 string for information on the protocol.
56 * the compiler no longer emits efficiency notes for (FUNCALL X)
57 when the type of X is uncertain under default optimization
59 * fixed bug 276: mutating a binding of a specialized parameter to a
60 method to something that is not TYPEP the specializer is now
62 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
63 exhibit the required behaviour on the x86 platform. (thanks to
64 Peter van Eynde, Eric Marsden and Bruno Haible)
65 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
66 even for difficult arguments. (reported by Peter Graves)
67 * fixed bug 141a: the backquote printer now descends quoted
69 * fixed another bug in backquote printing: no more destructive
70 modification of the form's list structure. (reported by Brian
72 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
73 move between its address being taken and the call to
74 interrupt_thread, fixing a crashing race condition.
75 * the SB-POSIX contrib implementation has been adjusted so that it
76 no longer exhibits ridiculously poor performance when constructing
77 instances corresponding to C structs.
79 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
80 * minor incompatible change: the sb-grovel contrib now treats C
81 structures as alien (in the sense of SB-ALIEN) objects rather than
82 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
83 has implications for memory management of client code
84 (sb-grovel-returned objects must now be manually managed) and for
85 type safety (alien objects now have full types).
86 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
87 used to control emission of compiler diagnostics, rather than the
88 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
89 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
90 quality should be considered deprecated.
91 * install.sh now installs the user manual as well
92 * (not quite a new documentable feature, but worth considering in
93 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
94 beginnings of a semantically meaningful condition hierarchy is
95 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
97 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
98 designator as the defaults argument.
99 * fixed bug: Displaced arrays whose displaced-to array has become
100 too small now cause ARRAY-DIMENSION to signal an error, providing
101 for safer bounds-checking. (reported by Bruno Haible)
102 * fixed bug: DEFCLASS slot definitions with identical :READER and
103 :WRITER names now signal a reasonable error. (reported by Thomas
105 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
107 * fixed bug: Invalid slot specification errors now print correctly.
108 (thanks to Zach Beane)
109 * fixed bug 320: Shared to local slot value transfers in class
110 redefinitions now happen corrently from superclasses as
111 well. (reported by Bruno Haible)
112 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
114 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
115 improper lists correctly. (reported by Bruno Haible)
116 * fixed bug 313: source-transform for <fun-name> was erroneously
117 applied to a call of a value of a variable with name <fun-name>.
118 (reported by Antonio Menezes Leitao)
119 * fixed bug 307: The obsolete instance protocol ensures that
120 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
121 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
122 remove dead unknown-values globs from the middle of the stack.
123 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
124 easier to resume long computations after using *BREAK-ON-SIGNALS*
125 to diagnose and fix failures. (thanks to Nikodemus Siivola)
126 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
127 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
128 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
129 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
130 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
132 * fixed a (fairly theoretical) bug in string printing: if
133 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
134 does not have array-element-type equal to the most general string
136 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
137 function name. (thanks to Nikodemus Siivola)
138 * fixed bug: compile-time format string checker failed on
139 non-closed ~{. (reported by Thomas F Burdick)
140 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
141 don't warn on legal loop code involving more than one aggregate
142 boolean. (thanks to Nikodemus Siivola)
143 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
144 in keyword position to MAKE-PACKAGE was not being reported as
146 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
147 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
149 * fixed bugs 280 and 312: the checking for multiple definitions in a
150 file is less likely to become confused by uses of inline
152 * fixed bug: the #S reader macro performs the keyword coercion
153 specified for slot names. (reported by Kalle Niemitalo)
154 * fixed bug: lambda lists may contain symbols whose names start with
155 & but are not lambda-list-keywords; their occurrence triggers a
157 * fixed bug 321: define-method-combination argument lambda lists do
158 not cause invalid code to be generated when &optional or &aux
159 variables have default values. (reported by Bruno Haible)
160 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
161 have CLOS classes; this appears also to have fixed reported
162 instability in stack exhaustion detection.
163 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
164 MUFFLE-WARNING when no associated restart is present is now
166 * optimization: rearranged the expansion of various defining macros
167 so that each expands into only one top-level form in a
168 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
170 * optimization: used a previously little-used slot in symbols to
171 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
173 * fixed some bugs revealed by Paul Dietz' test suite:
174 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
176 ** Fixed an optimization bug related to inheritance of initforms
177 from local to shared slots.
178 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
179 ** CAST optimizer forgot to flush argument derived type.
180 ** print/read consistency on floats is now orders of magnitude
181 more likely. (thanks also to Bruno Haible for a similar report
183 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
184 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
185 for objects of type REAL. Make it so.
186 ** SXHASH is specified (infelicitously) to respect similarity,
187 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
188 it so. (thanks to Markus Ziegler)
189 ** on the Alpha, the compiler succeeds in compiling functions
190 returning a known number of arguments greater than 63.
191 ** fixed handling of invalid NIL arguments in keyword position
193 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
195 ** provided a readably-printable representation for RANDOM-STATE
197 ** ensured that pathnames, where they have a namestring, always
198 print using #P"..." syntax.
200 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
201 * Support for the forthcoming 2.0 version of the NetBSD kernel
202 running on x86 hardware has been added. (thanks to Perry
203 E. Metzger most immediately, and others for their past work)
204 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
205 current release version; SBCL's OpenBSD support had been broken
206 since about the time of OpenBSD's switch to ELF binary format.)
207 * [placeholder for DX summary]
208 ** user code with &REST lists declared dynamic-extent, under high
209 speed or space and low safety and debug optimization policy.
210 * The manual has been converted to Texinfo format and the debugger
211 chapter from the cmucl manual has been added.
212 * A facility has been added to extract documentation strings from
213 sbcl and store them as Texinfo-formatted snippets for inclusion in
214 the manual (via Texinfo's @include directive)
215 * bug fix: compiler emitted division in optimized DEREF. (thanks for
216 the test case to Dave Roberts)
217 * bug fix: multidimensional simple arrays loaded from FASLs had fill
218 pointers. (reported by Sean Ross)
219 * bug fix: PROFILE output is printed nicely even for large numerical
220 values. (thanks to Zach Beane)
221 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
222 greater than 32 handle EOF correctly.
223 * bug fix: on X86 an immediate argument of the IMUL instruction is
224 correctly printed in disassembly. (thanks to Lutz Euler)
225 * bug fix: class slots in redefined classes preserve their old
226 values. (thanks to Bruno Haible and Nikodemus Siivola)
227 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
228 succeeds. (reported by Marco Baringer)
229 * fixed some bugs revealed by Paul Dietz' test suite:
230 ** READ-SEQUENCE now works on ECHO-STREAMs.
231 ** RATIONALIZE works more according to its specification. (thanks
234 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
235 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
236 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
237 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
238 should matter to you only if you rebind the printer control
239 variables and then find you want different bindings in the
240 debugger than in the ordinary execution of your program.)
241 * The runtime build system has been tweaked to support building
242 (on SPARC/SunOS) using a C compiler which invokes Sun's own
243 assembler and linker. (thanks to Nikodemus Siivola)
244 * Unbound, undefined, undeclared variables now trigger full
245 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
246 more often programmer error than deliberate exploitation of undefined
248 * optimization: the hash algorithm for strings has changed to one
249 that is less vulnerable to spurious collisions. (thanks to Juho
251 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
252 needless bounds checking. (thanks to Juho Snellman)
253 * optimization: implemented multiplication as a modular
254 (UNSIGNED-BYTE 32) operation on the PPC backend.
255 * fixed some bugs revealed by Paul Dietz' test suite:
256 ** ADJUST-ARRAY now copies the datum in a zero rank array if
258 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
259 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
260 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
262 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
264 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
266 ** ECHO-STREAMs no longer attempt to echo the end of file value to
267 their output stream on EOF from read.
268 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
269 have been read to end-of-file.
270 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
272 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
273 description of determination of which consecutive characters
275 ** Printing the "Space" character with escaping on now yields "#\\ ",
276 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
277 ** Reading floating-point numbers with *READ-BASE* set to a number
278 less than 10 works correctly.
279 ** Reading floating-point numbers with *READ-BASE* set to a number
280 more than 10 works correctly.
281 ** Printing with *PRINT-READABLY* targets the standard readtable, not
282 the readtable currently in effect.
284 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
285 * minor incompatible change: parsing of namestrings on a physical
286 (Unix) host has changed; numbers after the final #\. in a
287 namestring are no longer interpreted as a version field. This is
288 intented to be largely invisible to the user, except that the
289 meaning of the namestring "*.*.*" has changed: it now refers to a
290 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
291 should usually be replaced by
292 (make-pathname :name :wild :type :wild :version :wild)
293 with the added benefit that this is more likely to be portable.
294 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
295 option to OPEN now signals an error if the file being opened
296 exists; this may have an impact on existing code.
297 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
298 signals received, on the PowerPC platforms (both Linux and
299 Darwin). (thanks to Pierre Mai for pointing out the location of
301 * several fixes on OS X: The system now builds and runs cleanly on
302 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
303 (thanks to Brian Mastenbrook)
304 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
305 second-resolution: integer multiples of 1/3600 between -24 and 24.
306 (thanks to Vincent Arkesteijn)
307 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
308 type when called with 1 argument; PEEK-CHAR checked type of
309 PEEK-TYPE only after having read first character from a
310 stream. (reported by Peter Graves)
311 * bug fix: the garbage collector now has much better locality
312 behaviour, and in particular no longer treats all memory as being
313 exhausted when many small objects point to each other in a deeply
315 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
316 recognized as being TYPEP their class.
317 * bug fix: the PUSHNEW documentation string has been corrected.
318 (thanks to Vincent Arkesteijn)
319 * bug fix: defaulting of the value for the last of an atypically
320 large number of multiple values being bound was not being
321 performed correctly on the Alpha or PPC platforms
322 * optimization: implemented multiplication as a modular
323 (UNSIGNED-BYTE 32) operation on the x86 backend.
324 * optimization: SEARCH on simple-base-strings can now be open-coded.
325 (see also contrib/compiler-extras.lisp for inspiration for
326 teaching the compiler about the Boyer-Moore algorithm).
327 * value, returned by MAX (and MIN) called with several EQUALP, but
328 not EQL, arguments now does not depend on compiler settings.
329 * fixed some bugs revealed by Paul Dietz' test suite:
330 ** in stack analysis liveness information is propagated from
331 non-local entry points.
332 ** pathwise CAST removing failed when the CAST node did not start
334 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
335 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
337 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
338 host is already defined.
339 ** RENAME-FILE works on streams instead of signalling an internal
341 ** PEEK-CHAR uses the current readtable when determining whether
342 or not a character is whitespace.
343 ** MERGE-PATHNAMES handles the case when the pathname does not
344 specify a name while the default-pathname specifies a version.
345 ** Pathnames now stand a better chance of respecting print/read
347 ** Attempting to use standardized file system operators with a
348 pathname with invalid :DIRECTORY components signals a
350 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
352 ** DIRECTORY on logical pathnames is more correct.
353 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
354 signal a TYPE-ERROR if their argument is not a stream.
355 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
356 designator argument does not designate a stream.
357 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
358 examining the synonym.
359 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
361 ** OPEN and WITH-OPEN-STREAM allow opening streams with
362 element-type larger than ([UN]SIGNED-BYTE 32).
364 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
365 * When built with the :SB-FUTEX feature, threaded builds now take
366 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
367 for faster/more reliable mutex and condition variable support.
368 * Incompatible change (but one you probably shouldn't have been using
369 anyway): the interface and code for arbitrating between multiple
370 threads in the same user session has been redesigned.
371 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
372 error when called without an explicit environment argument.
373 (thanks to Nikodemus Siivola)
374 * bug fix: buffered :DIRECTION :IO streams are less likely to become
375 confused about their position. (thanks to Adam Warner and Gerd
377 * bug fix: Pretty printing backquoted forms with unquotations in the
378 argument list position of various code constructs such as LAMBDA
379 now works correctly. (reported by Paul Dietz)
380 * bug fix: Pretty printing unquotations no longer loses all
381 stream position information.
382 * optimization: performance of string output streams is now less
383 poor for multiple small sequence writes.
384 * optimization: performance of CSUBTYPEP in the presence of complex
385 expressions involving CONS and NOT many times has been improved.
386 (reported by Paul Dietz)
387 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
389 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
390 streams for the REPL, for the debugger, and so on. (thanks to
392 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
393 * fixed some bugs revealed by Paul Dietz' test suite:
394 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
395 is ignored if it is too small, rather than propagating through
396 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
397 ** extremely complex negations of CONS types were not being
398 sufficiently canonicalized, leading to inconsistencies in
400 ** VALUES tranformer lost derived type.
402 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
403 * fixed a bootstrapping bug: the build process no longer assumes
404 that the various BOOLE-related constants have the same value in
405 host and target lisps. (noted by Paul Dietz' test suite on an
406 SBCL binary built from CLISP)
407 * The system can now be dynamically linked on the MIPS platform,
408 which enables dynamic loading of foreign code from Lisp. (thanks
409 to Ralf Baechle for discussions on the MIPS ABI)
410 * The system now records debugging information for its own source
411 files in a filesystem-position-independent manner, relative to
412 the "SYS" logical host.
413 * fixed a compiler bug: MV-LET convertion did not check references
414 to the "max args" entry point. (reported by Brian Downing)
415 * tweaked disassembly notes to be less confident about proclaiming
416 some instruction as an LRA. (thanks to Brian Downing)
417 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
418 now each have their own history, command character, and other
419 characteristics. (thanks to David Lichteblau)
420 * fixed some bugs revealed by Paul Dietz' test suite:
421 ** compiler failure in compiling LOGAND expressions including a
423 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
424 shift greater than 32.
425 ** FLUSH-DEST did not mark blocks for type check regeneration.
426 ** HANDLER-CASE failed to accept declarations in handler clauses
427 in some circumstances.
429 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
430 * New code in contrib/sb-introspect (still probably not entirely
431 stable yet) provides some support for smart Lisp development
432 environments like SLIME.
433 * The conditions signalled for errors occurring when loading .fasl
434 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
435 in a way which should help ASDF recover gracefully.
436 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
437 stuff has been cleaned up. If you code contrib/ stuff, this might
438 affect you, and you can look at contrib/README, contrib/STANDARDS,
439 and/or the 0.8.4.27 diff to check.
440 * In full calls the compiler now does not generate checks for declared
441 argument types for all arguments.
442 * various threading fixes
443 ** and some experimental patches which didn't make it into
444 the main tree for this release, but which are shipped in
445 contrib/experimental-thread.patch as a possible fix for some
446 failures (deadlock, spinning...) in GC-intensive multithreaded
448 * fixed PPC build problem (source code incompatibility of different
449 library versions): added offsetof() hackery which attempts to divine
450 where glibc maintainers put uc_mcontext today
451 * fixed bug 282: compiler does not trust type assertions while passing
452 arguments to a full call.
453 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
454 &OPTIONAL VALUES type parameter.
455 * fix bug 214: algorithm for noting rejected templates is now more
456 similar to that of template seletion. (also reported by rydis on
458 * fixed bug 141b: printing backquoted information readably and prettily
459 inserts a space where necessary.
460 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
461 (CHARACTER 10) are now reported as errors, rather than propagated
462 as unknown types. (reported by piso on #lisp)
463 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
464 with values NIL and :ERROR. (thanks to Milan Zamazal)
465 * fixed bug 191c: CLOS now does proper keyword argument checking as
466 described in CLHS 7.6.5 and 7.6.5.1.
467 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
468 longer raise an error; further, using a list as a for-as-arithmetic
469 counter now raises a meaningful error.
470 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
471 now understood by sequence creation functions such as MAKE-SEQUENCE
473 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
474 a STRING-INPUT-STREAM.
475 * compiler enhancement: SIGNUM is now better able to derive the type
477 * type declarations inside WITH-SLOTS are checked. (reported by
479 * fixed some bugs revealed by Paul Dietz' test suite:
480 ** incorrect optimization of TRUNCATE for a positive first
481 argument and negative second.
482 ** compiler failure in let-convertion during flushing dead code.
483 ** compiler failure while deriving type of TRUNCATE on an
484 interval, containing 0.
485 ** ASH of a negative bignum by a negative bignum count now returns
487 ** intersection of CONS types now canonicalizes properly, fixing
488 inconsistencies in SUBTYPEP.
490 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
491 * incompatible change: The --disable-debugger command line
492 option now clobbers the debugger at a more fundamental
493 level, by redefining #'INVOKE-DEBUGGER instead of by
494 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
495 is specified by ANSI to ignore *DEBUGGER-HOOK* and
496 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
497 enter the debugger REPL and then suffer recursive errors
498 because *DEBUG-IO* is also messed up in --disable-debugger mode;
499 while under the new system, BREAK in --disable-debugger mode
500 terminates the system just as an unhandled error would.
501 * fixed compiler performance when processing loops with a step >1;
502 * bug fix: DOCUMENTATION now retrieves generic function
503 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
504 support has been systematized, and now supports the methods
505 specified by ANSI, along with a default method and a method for
506 slot documentation. (reported by Nathan Froyd)
507 * bug fix: effective methods associated with a generic function are
508 no longer cached over a change of that generic function's method
509 combination. (reported by Andreas Fuchs)
510 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
511 names a non-existent directory. (thanks to Andreas Fuchs)
512 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
513 the PPC platform, lead to stack corruption; this has been fixed.
514 (reported by Rainer Joswig)
515 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
516 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
517 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
518 arithmetic in general, is now much more reliable on the PPC
520 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
521 platform now returns the right answer.
522 * optimization: restored some effective method precomputation in
523 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
524 precomputation is now tunable.
525 * optimization: compiler-internal data structure use has been
526 reviewed, and changes have been made that should improve the
527 performance of the compiler by about 20%.
528 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
529 simple-base-strings and simple-bit-vectors is improved.
530 * optimization: the optimization of 32-bit logical and arithmetic
531 functions introduced in version 0.8.3 on the x86 has been
532 implemented on the mips, ppc and sparc platforms; an
533 implementation of the same facility, but for 64-bit arithmetic,
534 has been added for the alpha.
535 * microoptimization: the compiler is better able to make use of the
536 x86 LEA instruction for multiplication by constants.
537 * bug fix: in some situations compiler did not report usage of
538 generic arithmetic in (SPEED 3) policy.
539 * bug 145b fix: compiler used wrong type specifier while converting
540 MEMBER-types to numeric.
541 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
543 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
544 index argument now works properly on the Alpha platform.
545 * bug fix: floating point exception treatment on the Alpha platform
547 * bug fix: FILE-POSITION works much better on string input and
548 output streams. (thanks to Nikodemus Siivola)
549 * bug fix: many threading/garbage collection symptoms sorted.
550 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
551 might be pseudo-atomic.
552 * internal change: Stopping for GC is now done with signals not
553 ptrace. GC is now done in whichever thread wanted it, instead of
555 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
557 * bug fix: result form in DO is not contained in the implicit
559 * incompatible change: ICR structure is changed; the value part of
560 CONTINUATION is now called LVAR; corresponding functions are
561 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
562 * added type deriver for ISQRT (thanks to Robert E. Brown).
563 * bug fix: better support for loading from the command line when an
564 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
566 * fixed some bugs revealed by Paul Dietz' test suite:
567 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
568 ** ROUND and FROUND now give the right answer when given very
569 small float arguments.
570 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
572 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
573 ** (GCD 0 <negative-integer>) returned <negative-integer>.
574 ** LCM should return a non-negative integer.
575 ** PARSE-INTEGER returned the index of a terminator instead of the
576 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
577 ** PARSE-INTEGER returned an incorrect index being applied to a
579 ** LCM with two arguments of 0 returns 0 rather than signalling
581 ** unsigned addition of a 32-bit constant with the high bit set no
582 longer causes an internal compiler error on the x86.
583 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
584 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
585 with negative last argument.
586 ** byte specifiers with zero size and position no longer cause
587 an error during type derivation.
588 ** bignum multiplication on the Alpha platform now returns the
590 * porting: The system now builds on SuSE AMD64, although it still
591 generates a 32-bit binary.
592 * .fasl file incompatibility: The fasl file version number has
593 been incremented (because of the changes to internal compiler
594 data structures referred to above).
596 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
597 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
598 more accurately, on the Darwin kernel running on PowerPC hardware.
599 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
600 * Compiler code deletion notes now signal a condition of type
601 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
602 an associated MUFFLE-WARNING restart.
603 * The compiler now performs limited argument count validation of
604 constant format strings in FORMAT, and where appropriate in ERROR,
605 CERROR and WARN. (thanks to Gerd Moellmann)
606 * New ASDF-INSTALL contrib can be used for automatic download and
607 installation of third-party Lisp code from CCLAN or other sites
609 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
610 which forces another thread to execute a function supplied by the
612 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
613 now accept and act upon their :ELEMENT-TYPE keyword argument.
614 (reported by Martin Atzmueller, Edi Weitz)
615 * bug fix: FILE-POSITION now accepts position designators up to
616 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
617 is the greater. (thanks to Patrik Nordebo)
618 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
619 Kalle Olavi Niemitalo)
620 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
621 printable. (reported by Eric Marsden)
622 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
623 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
624 * x86 bug fix in control stack exhaustion checking: now shows backtrace
625 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
626 (thanks to Stig Sandoe)
627 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
628 UNREAD-CHAR now correctly includes the unread character in the
629 target sequence. (thanks to Gerd Moellmann)
630 * bug fix in threaded builds: the system can now be suspended and
631 resumed by shell job control with minimal disruption.
632 * bug fixes in times and timezones >2038AD
633 * better handling of "where is GNU make?" problem in build scripts
634 (thanks to Nikodemus Siivola)
635 * new optimization: inside a named function any reference to a
636 function with the same name is considered to be a self-reference;
637 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
638 optimization quality.
639 * new optimization on x86: logical functions and + now have
640 optimized (UNSIGNED-BYTE 32) versions, which are automatically
641 used when the result is truncated to 32 bits.
642 * VALUES declaration is partially enabled.
643 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
644 * fixed some bugs revealed by Paul Dietz' test suite:
645 ** The system now obeys the constraint imposed by
646 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
647 types form a lattice under type intersection.
648 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
649 ** ASSOC now ignores NIL elements in an alist.
650 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
651 and (1+ MOST-POSITIVE-FIXNUM) answers.
652 ** The addition of a method with invalid qualifiers to a generic
653 function does not cause an error to be signalled immediately;
654 a warning is signalled, and the error is generated only on
655 calling the generic function.
656 * changed .fasl file version number, in part to add type codes for
657 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
658 obscure ANSI requirements
660 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
661 * fixed bug 148: failure to inline-expand a local function left
662 garbage, confusing the compiler.
663 * fixed bugs 3cd: structure slot readers perform type check if the
664 slot can have an invalid value (i.e. it is either not initialized
665 or can be written with a less specific slot writer).
666 * bug fix: the compiler now traps array references to elements off
667 the end of an array; previously, the bounds checking in some
668 circumstances could go off-by-one.
669 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
671 * type declarations for array element types now obey the description
672 on the CLHS page "Declaration TYPE", as per discussions on
673 sbcl-help around 2003-05-08. This means that a declaration
674 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
675 declaration, all references to BAR will be asserted or assumed
676 (with THE, so dependent on compiler policy) to involve objects of
677 type FOO. Note that no such declaration is implied in
678 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
679 * declared types of functions from the "Conditions"
680 chapter. (reported by Paul Dietz)
681 * bug fix: CERROR accepts a function as its first argument.
682 * bug fix: NTH an NTHCDR accept a bignum as index
683 arguments. (reported by Adam Warner)
684 * optimization: character compare routines now optimize comparing
685 against a constant character. (reported by Gilbert Baumann)
686 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
687 argument now works properly on the MIPS platform.
688 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
689 * repeated evaluation of the same DEFSTRUCT, a slot of which is
690 declared to have a functional type, does not cause an error
692 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
693 many :DEFAULT-INITARGS. (reported by Istvan Marko)
694 * fixed bug: if last continuation of a deleted block has a
695 destination, this destination should be deleted too. (reported by
697 * fixed a bug in the bootstrap process: the host compiler's values
698 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
699 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
700 test case from Patrik Nordebo)
701 * improved the ability of the disassembler on the PPC platform to
702 provide helpful disassembly notes.
703 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
704 the class in more cases than previously.
705 * bug fix: FILE-POSITION now understands :START and :END for
706 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
707 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
708 8) data. (thanks to David Lichteblau)
709 * bug fix: it is possible to add a method to a generic function
711 * bug fix: reader failed to signal END-OF-FILE inside an
712 object representation. (reported by Nikodemus Siivola)
713 * fixed some bugs revealed by Paul Dietz' test suite:
714 ** LAST and [N]BUTLAST should accept a bignum.
715 ** condition slot accessors are methods.
716 ** (VECTOR NIL) is a subtype of STRING.
718 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
719 * minor incompatible change: some nonsensical specialized lambda
720 lists (used in DEFMETHOD) which were previously ignored now signal
722 * minor incompatible change: the system is now aware of the types of
723 variables in the COMMON-LISP package, and will signal errors for
724 most violations of these type constraints (where previously they
725 were silently accepted).
726 * minor incompatible change: COMPILE-FILE now uses the freedom
727 afforded (ANSI 3.2.2.3) to use derived function types for
728 functions defined in the same file. This also permits the system
729 to warn on static type mismatches and function
730 redefinition. (Currently it does not work with high DEBUG level.)
731 * minor incompatible change: VALUES declaration is disabled.
732 * When issuing notes, the compiler now signals a condition of type
733 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
734 restart for use in user handlers. It is expected that the
735 COMPILER-NOTE condition will eventually become a condition
736 supertype to a hierarchy of note types, which will then be
737 handleable in a similar fashion. However, at the moment, no such
738 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
739 but it's an implementation detail, not a classification for the
741 * Changes in type checking closed the following bugs:
742 ** type checking of unused values (192b, 194d, 203);
743 ** template selection based on unsafe type assertions (192c, 236);
744 ** type checking in branches (194bc).
745 * A short form of VALUES type specifier has ANSI meaning (it has
746 increased the number of situations when SBCL cannot perform type
748 * fixed bug in DEFSTRUCT: once again, naming structure slots with
749 keywords or constants is permissible.
750 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
751 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
752 classes. (thanks to Antonio Martinez)
753 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
754 outputs FRESH-LINE or TERPRI, and no longer converts its stream
755 argument to a pretty-print stream. Instead, it leaves any such
756 operations to DESCRIBE-OBJECT methods.
757 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
759 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
760 lambda lists. (thanks to David Lichteblau)
761 * bug fix: Defining a generic function with a :METHOD-CLASS being a
762 subclass of STANDARD-METHOD no longer causes stack exhaustion.
763 (thanks to Gerd Moellmann)
764 * fixed bug 246: increased compilation speed of long
765 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
767 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
769 * DEFCONSTANT now throws a condition of type
770 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
771 constant to a non-EQL value; CONTINUE and ABORT restarts
772 respectively change and preserve the value.
773 * fixed bug 63: The code walker, part of the implementation of CLOS,
774 is now better at handling symbol macros.
775 * bug fix: There is no longer an internal implementation type named
776 CL:LENGTH. (reported by Raymond Toy)
777 * bug fix: In macro-like defining macros/special operators the
778 implicit block does not enclose the lambda list.
779 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
780 enclosed in parentheses) are not suitable as type specifiers, and
781 their use properly signals an error now.
782 * bug fix: An argument count mismatch for a type specifier in code
783 being compiled no longer causes an unhandled error at compile
784 time, but signals a compile-time warning.
785 * fixed simple vector readable printing
786 * bug fix: DESCRIBE takes more care over whether the class
787 precedence list slot of a class is bound before accessing it.
788 (reported by Markus Krummenacker)
789 * bug fix: FORMATTER can successfully compile pretty-printer format
790 strings which use variants of the ~* directive inside.
791 * bug fix: SEARCH now applies its TEST predicate to the elements of
792 the arguments in the correct order. (thanks to Wolfhard Buss)
793 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
794 in type inference. (thanks to Robert E. Brown)
795 * bug fix: Reading in symbols with an explicit package name of ""
796 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
797 * fixed some bugs revealed by Paul Dietz' test suite:
798 ** NIL is now allowed as a structure slot name.
799 ** Arbitrary numbers, not just REALs, are allowed in certain
800 circumstances in LOOP for-as-arithmetic clauses.
801 ** Multiple class redefinitions before slot access no longer
803 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
804 association between the name and a class.
805 ** Generic functions with non-standard method-combination and over
806 five methods all of which return constants no longer return NIL
807 after the first few invocations. (thanks to Gerd Moellmann)
808 ** CALL-NEXT-METHOD with no arguments now passes the original
809 values of the arguments, even in the presence of assignment.
810 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
812 ** DISASSEMBLE works with closures and funcallable instances.
813 ** ADD-METHOD now returns the generic function, not the new method.
814 ** FIND-METHOD signals an error if the lengths of the specializers
815 is incompatible with the generic function, even if the ERRORP
817 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
818 which its argument is a member.
819 ** DEFCLASS only redefines the class named by its class-name
820 argument if that name is the proper name of the class;
821 otherwise, it creates a new class.
822 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
823 of the UNBOUND-SLOT condition to the name of the slot.
824 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
825 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
826 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
827 SLOT-MAKUNBOUND in the specified fashion.
829 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
830 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
831 cross-compilation host. As a consequence, we can now bootstrap our
832 way up to SBCL starting with a bare gcc toolchain and human-readable
833 source code (first the source to CLISP, then the source to SBCL).
834 * A contributed module containing a partial implementation of the
835 simple-streams interface has been included. (thanks to Rudi
837 * A contributed module implementing the RFC1321 Message Digest
838 Algorithm, known as MD5, has been included.
839 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
840 no longer has any effect, as the code controlled by this feature
841 has been deleted. (As far as we know, no-one has ever built using
842 this feature, and its semantics were confused in any case).
843 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
844 work on conditions (as required by the ANSI specification),
845 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
846 expected behaviour on conditions. Users should note, however,
847 that such behaviour is not required by the ANSI specification,
848 and so use of this behaviour may render their code unportable.
849 * fixed some bugs revealed by Paul Dietz' test suite:
850 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
852 ** &ENVIRONMENT parameter in macro lambda list is bound first.
853 ** SXHASH on condition objects no longer returns NIL.
854 ** :ALLOCATION :CLASS slots are better treated; their values are
855 updated on class redefinition, and initforms inherited from
856 superclasses are applied.
857 ** REMOVE-METHOD returns its generic function argument even when
858 no method was removed.
859 ** SHARED-INITIALIZE now initializes the values of the requested
860 slots, including those with :ALLOCATION :CLASS.
861 ** ALLOCATE-INSTANCE now works on structure classes defined via
862 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
864 ** SLOT-EXISTS-P now works on conditions, as well as structures
866 ** MAKE-LOAD-FORM now has the required methods on
867 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
868 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
869 keyword, and now implements the SLOT-NAMES argument.
870 ** methods with &OPTIONAL arguments no longer allow too many
871 arguments to be passed in the call without error.
872 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
873 option is consistent with the required arguments of the generic
874 function lambda list.
875 * bug fix: REQUIRE accepts a string designator. (Thanks to
877 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
878 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
879 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
880 * bug fix: The compiler now checks for duplicated variables in macro
882 * bug fix: SETQ on globals returns the correct value.
883 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
884 not a COMPILER-ERROR (followed by some other strange error on
885 choosing the CONTINUE restart).
886 * bug fix: make.sh and friends are now more consistent in the way that
887 they look for GNU "make".
889 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
890 * experimental native threads support (on x86 Linux >=2.4 only).
891 This is not compiled in by default: you need to add :SB-THREAD to
892 the target features. See the "Beyond ANSI" chapter of the manual
894 * fix for longstanding nonANSIism: The old distinction between
895 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
896 The return value from CL:FIND-CLASS is now a CLOS class, and
897 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
898 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
899 * An interface to the MetaObject Protocol, as described in Kiczales,
900 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
901 MIT Press, 1991, is available from the SB-MOP package.
902 * incompatible change: the SB-PCL package should now be considered
903 a private implementation detail, and no longer a semi-private MOP
905 * minor incompatible change: due to rearrangement for threads, the
906 control stack and binding stack are now allocated at arbitrary
907 addresses instead of being hardcoded per-port. Users affected by
908 this probably have to be doing advanced things with shared
909 libraries, and will know who they are.
910 * minor incompatible change: Previously, all --eval forms used were
911 processed with READ before any of them were processed with EVAL.
912 Now each --eval form is processed with both READ and EVAL before
913 the next --eval form is processed. (Thus package operations like
914 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
915 work as the user might reasonably expect.)
916 * minor incompatible change: *STANDARD-INPUT* is now only an
917 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
919 * minor incompatible change: Y-OR-N-P is now character-oriented,
920 not line oriented. Also, YES-OR-NO-P now works without errors.
921 (thanks to Antonio Martinez)
922 * sb-aclrepl module improvements: an integrated inspector, added
923 repl features, and a bug fix to :trace command.
924 * Known functions, which cannot be open coded by the backend, are
925 considered to be able to check types of their arguments. (fixing
926 a bug report by Nathan J. Froyd)
927 * fixed a bug in computing method discriminating functions: It is
928 now possible to define methods specialized on classes which have
929 forward-referenced superclasses. (thanks to Gerd Moellmann)
930 * fixed evaluation order in optional entries (reported by Gilbert
932 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
933 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
934 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
935 required-by-AMOP NAME argument, as well as CLASS and
936 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
937 * fixed bug 20: DEFMETHOD can define methods using names that are
938 not the proper names of classes to designate class specializers.
939 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
940 Unix file descriptors, instead of blowing up. (thanks to Antonio
942 * Garbage collection refactoring: user-visible change is that a
943 call to the GC function during WITHOUT-GCING will not do garbage
944 collection until the end of the WITHOUT-GCING. If you were doing
945 this you were probably losing anyway.
946 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
947 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
948 TYPEP the latter but not the former.
949 * The compiler issues a full WARNING for calls to undefined functions
950 with names from the CL package.
951 * MAP-INTO for a vector destination is open coded. (reported by
952 Brian Downing on c.l.l)
953 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
954 documentation string.
955 * fixed some bugs revealed by Paul Dietz' test suite:
956 ** COPY-ALIST now signals an error if its argument is a dotted
958 ** Condition slots are now accessed more correctly in the presence
959 of multiple initargs for a given slot.
960 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
961 exclude restarts of the same name associated with a different
963 ** DEFCLASS of forward-referenced classes with another
964 forward-referenced class in the superclasses list no longer
966 ** Condition slots are now initialized once each, not multiple
967 times. (thanks to Gerd Moellmann)
968 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
970 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
971 ** NSET-EXCLUSIVE-OR does not return extra elements when its
972 arguments contain duplicated elements.
973 ** RESTART-CASE understands local macros.
974 ** RESTART-CASE associates exactly its own restarts with a condition.
975 ** ENDP in safe mode checks its argument to be of type LIST.
976 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
977 in question is unbound.
978 ** Optimized MAKE-INSTANCE functions no longer cause internal
979 assertion failures in the presence of duplicate initargs.
980 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
981 ** Side-effectful :DEFAULT-INITARGS have their side-effects
982 propagated even in the ctor optimized implementation of
984 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
986 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
987 * a better implementation of SXHASH on (simple) bit vectors,
988 measured both in execution speed and in distribution of results
989 over the positive fixnums, has been installed. Likewise, a better
990 implementation of EQUAL for simple bit vectors is now available.
991 * fixed CEILING optimization for a divisor of form 2^k.
992 * fixed bug 240 (emitting extra style warnings "using the lexical
993 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
995 * fixed SXHASH, giving different results for NIL depending on type
996 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
997 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
998 variable. (found by Rolf Wester)
999 * SBCL does not ignore type declarations for special
1000 variables. (reported by rif on c.l.l 2003-03-05)
1001 * some bug fixes in contrib/sb-aclrepl/
1002 * fixed some bugs revealed by Paul Dietz' test suite:
1003 ** a bug in the CONS type specifier, whereby the CAR and CDR
1004 types got intertwined, has been fixed;
1005 ** the type system is now able to reason about the interaction
1006 between INTEGER and RATIO types more completely;
1007 ** APPEND, [N]REVERSE and NRECONC check that those their
1008 arguments, which must be proper lists, are really so;
1009 ** An array specialized to be unable to hold elements has been
1010 implemented, as required -- yes, really -- by ANSI;
1011 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1012 on malformed property lists;
1014 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1015 * incompatible packaging change: in line with Unix convention,
1016 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1017 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1018 * REQUIRE and PROVIDE are now optionally capable of doing something
1019 useful. See the documentation string for REQUIRE.
1020 * infrastructure for a managed SBCL contrib system: contributed
1021 modules in this release include:
1022 ** the ASDF system definition facility;
1023 ** an interface to the BSD Sockets API;
1024 ** an ACL-like convenience interface to the repl;
1025 (thanks to Kevin Rosenberg)
1026 ** an implementation of ROTATE-BYTE, with efficient implementation
1028 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1029 gives a closer approximation to the right answer than previously.
1030 (thanks to Raymond Toy)
1031 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1032 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1033 optional environment arguments, as required by ANSI.
1034 * fixed bugs in other functions taking environment objects, allowing
1035 calls with an explicit NIL environment argument to be compiled
1037 * fixed bug 228: primary return values from
1038 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1039 COMPILE or FUNCTION.
1040 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1041 :TYPE will no longer signal errors on innocuous objects.
1042 * fixed bug 231b: SETQ is better at respecting type declarations in
1043 the lexical environment.
1044 * fixed a bug in DEFCLASS: classes named by symbols with no or
1045 unprintable packages can now be defined.
1046 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1047 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1048 * fixed compiler failure related to checking types of functions.
1049 (reported by Robert E. Brown)
1050 * the compiler is now much more consistent in its error-checking
1051 treatment of bounding index arguments to sequence functions: in
1052 (SAFETY 3) code, errors will be signalled in almost all cases if
1053 invalid sequence bounding indices are passed to functions defined
1054 by ANSI to operate on sequences.
1055 * fixed a bug in the build procedure: documentation of SBCL-specific
1056 packages is now preserved and available in the final Lisp image.
1057 * lifted FDEFINITION lookup out of loops in the implementation of
1058 many list operations. (thanks to Robert E. Brown)
1059 * fixed a bug in the reader: the #n# reader macro now works for
1060 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1061 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1062 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1063 E. Brown and Tony Martinez respectively)
1064 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1065 SLOT-BOUNDP: in particular, it is now able to optimize them much
1066 better, and is now not vulnerable to having packages renamed.
1067 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1068 always called when a slot is not present in an instance. (thanks
1070 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1071 longer signal an error on unreading a character following EOF on
1072 the previous constituent stream. (thanks to Tony Martinez)
1073 * fixed some bugs revealed by Paul Dietz' test suite:
1074 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1075 not just nonnegative fixnums;
1076 ** the logical bit-array operators such as BIT-AND now accept an
1077 explicit NIL for their "opt-arg" argument (to indicate a
1078 freshly-consed result bit-array);
1079 ** ELT now signals an error on an invalid sequence index in safe
1081 ** the type system is now cleverer about negations of numeric
1082 types, and consequently understands the BIGNUM and RATIO types
1084 ** the type system is now cleverer about the interaction between
1085 INTEGER and RATIO types: while bugs still remain, many more
1086 cases are accurately computed;
1087 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1088 if it is in the last clause;
1089 ** CONSTANTLY now correctly returns a side-effect-free function in
1091 ** DECLARE is no longer treated as a special-operator; in
1092 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1093 * incremented fasl file version number due to the change in the
1094 DEFSTRUCT-SLOT-DESCRIPTION structure.
1096 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1097 * minor incompatible change: code processed by the "interpreter" or
1098 EVAL now has a compilation optimization policy of (DEBUG 2)
1099 (changed from (DEBUG 1)) to improve debuggability of interactive
1100 development, and to allow the use of the debug RETURN command in
1102 * an experimental implementation of the RETURN command for the
1103 debugger has been included. (thanks to Frederik Kuivinen)
1104 * fixed bug 62: constraints were not propagated into a loop.
1105 * fixed bug in embedded calls of SORT (reported and investigated by
1107 * fixed some bugs revealed by Paul Dietz' test suite:
1108 ** printing and reading of arrays with some dimensions having
1109 length 0 (thanks to Gerd Moellmann);
1110 ** BOA constructor with &AUX argument without a default value does
1111 not cause a type error;
1112 ** CONSTANTP now returns true for all self-evaluating objects.
1114 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1115 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1116 accessors that are related by inheritance, as specified in the
1117 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1118 Valtteri Vuorikoski)
1119 * The compiler is now able to inline functions that were defined in
1120 a complex lexical environment (e.g. inside a MACROLET).
1121 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1123 * Support for the upcoming FreeBSD-5.0 release has been included.
1124 (thanks to Dag-Erling Smorgrav)
1125 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1126 effect when it is not in a toplevel context.
1127 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1128 stand a better chance of being correct. (thanks to Gerd
1130 * fixed bug in COERCE, which now signals an error on coercing a
1131 rational to a bounded real type which excludes the expected
1133 * The compiler is now able to derive types more accurately from the
1134 COERCE and COMPILE functions.
1135 * fixed bug 223: functional binding is considered to be constant
1136 only for symbols in the CL package.
1137 * fixed bug 231: SETQ did not check the type of a variable being set
1138 (reported by Robert E. Brown)
1139 * A new optimization for MAKE-INSTANCE has been included, fixing
1140 various bugs (including relating to :ALLOCATION :CLASS slots and
1141 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1143 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1144 ** As required by ANSI, LOOP now disallows anonymous collection
1145 clauses such as COLLECT I in conjunction with aggregate boolean
1146 clauses such as THEREIS (= I 1);
1147 ** LOOP now signals an error when any variable is reused in the
1148 same loop (including the potentially useful construct analogous
1149 to WITH A = 1 WITH A = (1+ A);
1150 ** IT is only a special loop symbol within the first clause of a
1151 conditional loop clause;
1152 ** LOOP with a typed iteration variable over a hashtable now
1153 signals a type error iff it should.
1154 * fixed some other bugs revealed by Paul Dietz' test suite:
1155 ** FILE-STREAM now names the class previously known as FD-STREAM;
1156 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1157 argument) no longer signals an error;
1158 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1159 of the slot symbol, rather than using the current package
1160 ((:CONC-NAME "") continues to intern the slot's name in the
1162 * incremented fasl file version number, because of the incompatible
1163 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1164 of the new implementation of DEFINE-COMPILER-MACRO.
1166 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1167 * Support for building SBCL for MIPS platforms running in
1168 little-endian mode has now been checked in, and basic
1169 functionality on said platforms verified.
1170 * minor incompatible change: PCL now records the pathname of a file
1171 in which methods and the like are defined, rather than its
1173 * minor incompatible change: TRUENAME now considers the truename of
1174 a file naming a directory to be the pathname with :DIRECTORY
1175 component indicating that directory.
1176 * minor incompatible change: a NAMED clause in the extended form of
1177 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1178 reason for the previous behaviour is unclear.
1179 * more systematization and improvement of CLOS and MOP conformance
1180 in PCL (thanks to Gerd Moellman and Pierre Mai):
1181 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1183 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1184 primary methods with no specializers;
1185 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1187 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1188 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1189 FORWARD-REFERENCED-CLASSes; error reporting on
1190 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1192 ** SXHASH on CLOS instances now uses a slot internal to the
1193 instance to return different numbers on distinct instances,
1194 while preserving the same return value through invocations of
1196 ** DEFMETHOD signals errors when methods with longer incongruent
1197 lambda lists are added to generic functions;
1198 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1199 CLASS, as specified in AMOP;
1200 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1201 on the order returned by the primary method for classes of
1202 class STANDARD-CLASS;
1203 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1204 * fixed some bugs shown by Paul Dietz' test suite:
1205 ** DOLIST puts its body in TAGBODY;
1206 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1208 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1209 value producing form;
1210 ** if more variables are given to PROGV than values, extra
1211 variables are bound and made to have no value;
1212 ** NSUBSTITUTE on list arguments gets the right answer with
1214 ** ELT signals an error of type TYPE-ERROR when the index argument
1215 is not a valid sequence index;
1216 ** LOOP signals (at macroexpansion time) an error of type
1217 PROGRAM-ERROR when duplicate variable names are found;
1218 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1219 ** FUNCALL of special-operators now cause an error of type
1221 ** PSETQ now works as required in the presence of side-effecting
1222 symbol-macro places;
1223 ** NCONC accepts any object as its last argument;
1224 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1226 ** loop-for-as-package does not require a package to be explicitely
1228 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1230 * fixed bug 166: compiler preserves "there is a way to go"
1231 invariant when deleting code.
1232 * fixed bug 172: macro lambda lists with required arguments after
1233 &REST arguments now cause an error to be signalled. (thanks to
1235 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1236 bug. (thanks to Matthew Danish)
1237 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1239 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1240 arguments are lexically rebound. (thanks to Gerd Moellmann and
1242 * fixed bug 194: error messages are now more informative when there
1243 is no primary method applicable in a call to a generic
1244 function. (thanks to Gerd Moellmann)
1245 * fixed bug in command line argument checking (thanks to Julian
1247 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1248 (thanks to Matthew Danish)
1249 * incremented fasl file version number, because of the
1250 SXHASH-related changes in the layout of CLOS data structures
1252 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1253 * minor incompatible change: The runtime (the Unix executable named
1254 "sbcl") is now much pickier about the .core files it will load.
1255 Essentially it now requires .core files to descend from the same
1256 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1257 as the runtime does. (The intent is to prevent the crashes which
1258 can occur, and which can even be reported as mysterious failures,
1259 when people patch the sources or change the build parameters
1260 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1261 sbcl and .core files.)
1262 * fixed bug: VALUES-LIST is no longer optimized away.
1263 * fixed bug 142: The FFI conversion of C string values to Lisp
1264 string values no longer conses excessively. (thanks to Nathan
1265 Froyd porting Raymond Toy's fix to CMU CL)
1266 * began to systematize and improve MOP conformance in PCL (thanks to
1267 Nathan Froyd, Gerd Moellman and Pierre Mai):
1268 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1270 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1271 ** FINALIZE-INHERITANCE is now called on class finalization;
1272 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1273 argument precedence order.
1274 * fixed bug 202: The compiler no longer fails on functions whose
1275 derived types contradict their declared type.
1276 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1277 so it can be non-toplevel.
1278 * The fasl file version number has changed (because of the new
1279 implementation of DEFMACRO).
1280 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1281 safe code, that any length requirement by their type-specifier
1282 argument is valid. The exceptions to this are described in bug
1284 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1285 functions now check the types of their inputs as required by ANSI.
1286 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1287 introduced symbol is DECLAREd to be SPECIAL.
1288 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1289 * fixed bug: :COUNT argument to sequence functions may be negative
1290 * fixed bug: body of DO-SYMBOLS may contain declarations
1291 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1292 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1293 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1294 symbol macro only once
1295 * fixed printing of call frame when argument list is unavailable
1296 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1297 * compiler no longer signals WARNING on unknown keyword
1300 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1301 * A beta-quality port to the mips architecture running Linux,
1302 based on the old CMUCL backend, has been made. It has been tested
1303 on a big-endian kernel, and works sufficiently well to be able to
1304 rebuild itself; it has not been tested in little-endian mode.
1305 * fixed an inconsistency between gencgc.c and purify.c which made
1306 dumping/loading .core files unreliable
1307 * fixed bug 120a: The compiler now deals correctly with IFs where
1308 the consequent is the same as the alternative, instead of
1309 misderiving the return type. (thanks to Alexey Dejneka)
1310 * fixed bug 113: Logical pathnames are now dumpable (the logical
1311 host is resolved at load-time, throwing an error if it is not
1313 * fixed bug 174: FORMAT's error message is slightly clearer when a
1314 non-printing character is used in a format directive.
1315 * fixed several bugs in compiler checking of type declarations, i.e.
1316 violations of the Python "declarations are assertions" principle
1317 (thanks to Alexey Dejneka)
1318 * fixed several bugs in PCL's error checking (thanks to Gerd
1320 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1322 * fixed bug in compilation of functions as first class values
1323 (thanks to Antonio Martinez-Shotton)
1324 * The compiler's handling TYPE-ERRORs which it can prove will
1325 inevitably happen at runtime has been cleaned up and corrected
1326 in several ways. (thanks to Alexey Dejneka)
1327 * improved argument type checking for various basic arithmetic
1328 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1329 lavished on them in the past that they can be compiled in many
1330 ways in different special cases
1331 * fixed bug 181: compiler checks validity of user supplied type
1333 * cleaned up code flushing in optimization: Function calls which
1334 should signal errors for safety purposes (e.g. which ANSI says
1335 should signal errors when their arguments are of incorrect type)
1336 are no longer optimized away.
1337 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1338 * incremented fasl file version number, because changes in the
1339 implementation of sequence functions like COERCE caused
1340 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1341 in old inline expansions) to become undefined. (Actually these
1342 changes were later undone, so we might very well be binary
1343 compatible with 0.7.7 after all, but leaving the version number
1344 incremented seemed like the simplest and most conservative
1347 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1348 * An alpha-quality port to the parisc architecture running Linux,
1349 based on the old CMUCL backend, has been made. This, even more so
1350 than the other backends, should be considered still a work in
1351 progress; known problems include that the Linux kernel in 64-bit
1352 mode does not propagate the correct sigcontext structure to
1353 userspace, and consequently SBCL on a parisc64 kernel will not
1355 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1356 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1357 suppressed.) Also now that the compiler is looking at declarations
1358 in the environment, it checks optimization declarations as well,
1359 and suppresses inlining when (> DEBUG SPEED).
1360 * More fixes have been made to treatment of floating point exception
1361 treatment and other Unix signals. In particular, floating point
1362 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1363 * The detection and handling of control stack exhaustion (infinite
1364 or very deeply nested recursion) has changed. Stack exhaustion
1365 detection is now done by write-protecting pages at the OS level
1366 and applies at all optimization settings; when found, a
1367 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1368 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1369 be caught using IGNORE-ERRORS.
1370 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1371 correctly when transforming recursive LABELS functions to LETs.
1372 (thanks to Alexey Dejneka)
1373 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1374 that are names of constants or global variables.
1375 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1376 alien routines with docstrings.
1377 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1378 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1380 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1381 (thanks to Eric Marsden)
1382 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1383 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1384 * Bug fix: The SPARC backend can now compile functions involving
1385 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1386 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1387 to a functions expecting a PATHNAME-DESIGNATOR.
1388 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1389 lambda lists. (thanks to Alexey Dejneka)
1390 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1391 functions so that EXPORTing the name of the class doesn't cause
1392 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1393 to Antonio Martinez for reporting this)
1394 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1395 constant-folded without failing an assertion. (thanks to Einar
1396 Floystad Dorum for reporting this)
1397 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1398 types are now dealt with more correctly by the compiler.
1399 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1400 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1401 behaviour analogous to RENAME-FILE. This puts its behaviour more
1402 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1404 * The fasl file version number has changed again. (because of the
1405 bug fix involving the names of PCL MAKE-INSTANCE functions)
1407 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1408 * bug fix: Floating point exceptions are treated much more
1409 consistently on the x86/Linux and PPC/Linux platforms.
1410 * Array initialization with :INITIAL-ELEMENT is now much faster for
1411 cases when the compiler cannot open code the array creation, but
1412 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1413 array accesses have also seen a speed increase.
1414 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1415 specified. (This was at the root of some bad interactions between
1416 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1418 * bug fix: Internal error arguments for undefined functions are now
1419 computed correctly on the PPC/Linux platform.
1420 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1421 Raymond Toy's patch for CMU CL)
1422 * Support for the Solaris 9 operating environment has been included
1423 (thanks to Daniel Merritt)
1424 * A very ugly but hopefully complete draft of the missing FFI chapter
1425 of the manual has been created by reformatting the corresponding
1426 CMU CL manual chapter into (currently very ugly and incoherent)
1427 DocBook and bringing it up to date for SBCL behavior. Thus, the
1428 manual is now essentially complete, at least by my extreme
1429 once-and-only-once standards, whereby it's acceptable to refer to
1430 the doc strings of SB-EXT functions as the primary documentation.
1431 * The fasl file version number has changed again, due to cleanup of
1432 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1433 is no longer a static symbol.)
1435 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1436 * SBCL now builds with OpenMCL (version 0.12) as the
1437 cross-compilation host; also, more progress has been made toward
1438 bootstrapping under CLISP.
1439 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1441 * bug 158 fixed: The compiler can now deal with integer loop
1442 increments different from 1; fixing this turned out also to fix
1444 * bug 169 fixed: no more bogus warnings about using lexical bindings
1445 despite the presence of perfectly good SPECIAL declarations (thanks
1446 to David Lichteblau)
1447 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1448 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1449 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1451 * bug 180 fixed: Method combination specifications no longer ignore
1452 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1453 * bug fix: Structure type predicate functions now check their argument
1454 count as they should.
1455 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1456 correctly. (thanks to Pierre Mai)
1457 * minor incompatible change: The --noprogrammer option is deprecated
1458 in favor of the new --disable-debugger option, which is very similar.
1459 (The major difference is that it takes effect at a slightly different
1460 time at startup, causing handling of errors in --sysinit and
1461 --userinit files will be affected differently.) The
1462 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1463 been added to allow this functionality to be controlled from ordinary
1464 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1465 maintainers, who might want to run non-interactive scripts to
1466 build SBCL cores which will later be used interactively.)
1467 * minor incompatible change: The LOAD function no longer, when given
1468 a wild pathname to load, loads all files matching that pathname.
1469 Instead, an error of type FILE-ERROR is signalled.
1471 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1472 * bug 147 fixed: The compiler preserves its block link/count
1473 invariants more correctly now so that it doesn't crash. (thanks
1475 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1477 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1478 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1479 no-value-for-this-character as NIL (as specified by ANSI).
1480 * HOST-NAMESTRING on physical pathnames now returns a string that is
1481 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1482 (thanks to Christophe Rhodes)
1483 * The Alpha port handles icache flushing more correctly. (thanks to
1485 * More progress has been made toward bootstrapping under CLISP. (thanks
1486 to Christophe Rhodes)
1487 * The fasl file format has changed again, because dynamic loading
1488 on OpenBSD (which has non-ELF object files) motivated some cleanups
1489 in the way that foreign symbols are transformed and passed around.
1490 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1491 is no longer treated as whitespace by the reader, but instead as
1492 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1493 |A
\7fB|, instead of A as it used to.
1495 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1496 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1497 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1498 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1499 this as of 0.7.1.45, but was left out of the previous news section
1500 (thanks to Dan Barlow)
1501 * SBCL now runs on the Solaris operating system on SPARC architectures
1502 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1503 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1504 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1505 * SPARC backend cleanups, allowing builds of cores optimized for V8
1506 and V9 SPARCS, and also emission of code targeted to a particular
1507 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1509 * SBCL is closer to bootstrapping under CLISP, thanks to various
1510 fixes by Christophe Rhodes.
1511 * The fasl file format has changed again, to allow the compiler's
1512 INFO database to support symbol macros.
1513 * The user manual (in doc/) is formatted into HTML more nicely.
1514 (thanks to coreythomas)
1515 * The system is smarter about SUBTYPEP relationships, especially
1516 those involving NOT types (including types such as ATOM which are
1517 represented internally using NOT types). Thus SUBTYPEP is less
1518 likely to return (VALUES NIL NIL) in general, and in particular
1519 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1521 * The fasl file format has changed again, because the internal
1522 representation of types now includes a new slot to support the new
1523 SUBTYPEP-of-NOT-types logic.
1524 * (not a change in the main branch of SBCL, but a related prototype
1525 which can hopefully be merged into the main branch of SBCL in the
1526 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1527 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1528 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1529 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1530 dispatch macro character combinations. (thanks to Alexey Dejneka)
1531 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1532 that superficially look like logical namestrings correctly.
1533 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1535 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1536 default supplied-p) for &optional and &key arguments. (thanks to
1539 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1540 * incompatible change: The compiler is now less aggressive about
1541 tail call optimization, doing it only when (> SPACE DEBUG) or
1542 (> SPEED DEBUG). (This is an incompatible change because there are
1543 programs which relied on the old CMU-CL-style behavior to optimize
1544 away their unbounded recursion which will now die of stack overflow.)
1545 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1546 for non-GENCGC systems has been increased to 20M (since that
1547 seems much closer to the likely performance optimum for modern
1548 systems than the old 4M value was)
1549 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1550 and *DEBUG-PRINT-LEVEL*
1551 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1552 of CMU CL's support for SPARC, and various endianness and other
1553 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1554 * new syntactic sugar for the Unix command line: --load foo.bar is now
1555 an alternate notation for --eval '(load "foo.bar")'.
1557 ** The system now detects stack overflow and handles it gracefully,
1558 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1559 optimization settings. (This is a good thing in general, and
1560 its introduction in this version should be particularly timely
1561 for anyone whose code fails because of suppression of tail
1563 ** The system now hunts for the C variable "environ" in a more
1564 devious way, to avoid segfaults when the C library version differs
1565 between compile time and run time. (thanks to Christophe Rhodes)
1566 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1567 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1568 ** The compiler no longer issues bogus style warnings for undefined
1569 classes in the same source file as the DEFCLASSes which defined
1570 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1572 ** fixes in CONDITION class precedence list for undefined function
1573 errors (thanks to Alexei Dejneka)
1574 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1575 correctly. (thanks to Dan Barlow)
1576 ** portability fixes aiming at bootstrapping under CLISP (thanks
1577 to Dave McDonald and Christophe Rhodes)
1578 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1579 ** fixes in type translation and and type inference (thanks to
1581 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1582 ** various fixes in the new ports (thanks to Dan Barlow)
1583 * several changes related to debugging:
1584 ** suppression of tail recursion, as noted above
1585 ** stack overflow detection, as noted above
1586 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1587 is now the default. (For some time encapsulation has been more
1588 reliable than the breakpoint-based :ENCAPSULATE NIL
1589 implementation, at least on X86 systems; and I just noticed that
1590 encapsulation also seems closer to the spirit of the ANSI
1593 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1595 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1596 up properly again. (There was a packaging bug in 0.7.0 which
1597 left their definitions in SB-SYS::LOAD-FOREIGN and
1598 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1599 vital for most things which interface to C-level interfaces,
1600 like extensions working with sockets or databases or
1601 Perl-compatible regexes or whatever, and the need to fix
1602 this bug is the main reason that 0.7.1 was released so
1604 ** DEFGENERIC is now choosier about the methods it redefines, so that
1605 reLOADing a previously-LOADed file containing DEFGENERICs does
1606 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1607 cycle now works as it should. (thanks to Alexey Dejneka)
1608 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1609 fixed by Christophe Rhodes. (He actually submitted this patch
1610 months ago, and I delayed until after 0.7.0.)
1611 ** Bug 111 (internal compiler confusion about runtime checks on
1612 FUNCTION types) was fixed by Alexey Dejneka.
1613 * Some internal cleanups (getting rid of variables which aren't
1614 needed now that the byte interpreter is gone) caused the fasl
1615 file format number to change again.
1617 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1618 * major incompatible change: The default fasl file extension, i.e. the
1619 default extension for files produced by COMPILE-FILE, has changed
1620 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1622 ** There are many changes in the implementation of the compiler.
1623 SBCL is now essentially a compiler-only implementation of ANSI
1624 Common Lisp. EVAL still "interprets" a few special cases, but
1625 almost all the interesting cases are handled by creating
1626 a LAMBDA expression, calling COMPILE on it, then calling
1627 FUNCALL on the result.
1628 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1629 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1630 Since the code is newer, there might still be some new bugs
1631 (though not as many as before Martin Atzmueller's fixes:-). But
1632 the new code is substantially simpler and clearer, and hopefully
1633 any remaining bugs will be simpler, less fundamental, and more
1634 fixable then the bugs in the old code.
1635 ** The revised compiler is still a little unsteady on its feet.
1637 *** The debugging information it produces (particularly the names
1638 of FUNCTION objects) is sometimes much less useful than what
1639 the old compiler produced.
1640 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1641 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1642 within a MACROLET) has been temporarily weakened.
1643 ** There are new compiler optimizations for various functions:
1644 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1645 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1646 *** the math functions TRUNCATE, FLOOR, and CEILING
1647 *** the function-of-all-trades COERCE
1648 Mostly these should be transparent, but there's one
1649 potentially-annoying problem (bug 117): when the compiler
1650 inline-expands a function and does type analysis on the result,
1651 it can create control paths which have type mismatches, and
1652 when it can't prove that those control paths aren't taken,
1653 it will issue WARNINGs about the type mismatches. This is
1654 a particular problem in practice for the new sequence functions.
1655 It's not clear how this should be fixed, and for now, a
1656 workaround is given in the entry for 117 in the BUGS file.
1657 ** (Because of the interaction between the two previous items --
1658 occasional inlining problems and new inline expansions -- some
1659 of the new sequence function optimizations won't really kick in
1660 completely until debugging information, and then inlining, are
1661 straightened out in some future version.)
1662 * minor incompatible changes:
1663 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1664 DIRECTORY no longer implicitly promotes NIL slots of its
1665 pathname argument to :WILD. In particular, when you ask for the
1666 contents of a directory (which you used to be able to do without
1667 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1668 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1669 ** changes in behavior that ANSI explicitly defines to be
1670 implementation dependent:
1671 *** The new compiler-only implementation still conforms with ANSI,
1672 but acts a little different than before. Besides the obvious
1673 changes in performance tradeoffs (that the cost per form passed
1674 to EVAL has gone up, and the cost per form executed by EVAL
1675 has gone down), the behavior of the system changes a little
1676 because there are no longer any interpreted function objects.
1677 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1678 e.g. doing COMPILE on the output of interactive DEFUN is
1680 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1682 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1683 has changed from (:CL) to NIL.
1684 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1685 e.g. #\Newline and #\Tab, don't have names specified in the
1686 ANSI Common Lisp standard, is now based on their ASCII symbolic
1687 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1688 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1689 are no longer used for output.
1690 ** changes in internal implementation constants:
1691 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1692 4 million. (If your application spends a lot of time GCing and
1693 you have a lot of RAM, you might want to experiment with
1694 increasing it even more.)
1695 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1696 However, almost all old code should still continue to work without
1697 immediate update, as SB-C-CALL is now a (deprecated) nickname
1699 ** Old operator names in the style DEF-FOO are now deprecated in
1700 favor of new corresponding names DEFINE-FOO, for consistency with
1701 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1702 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1703 internal symbols, but a few supported extensions like
1704 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1705 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1706 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1707 etc. as you get deeper into recursive calls to the debugger
1708 command loop, instead of the old "5]", "5]]", "5]]]"
1709 sequence. (I was motivated to do this when squabbles between
1710 ILISP and SBCL left me very deeply nested in the debugger. In the
1711 short term, this change will probably provoke more ILISP/SBCL
1712 squabbles, but hopefully it will be an improvement in the long run.)
1713 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1714 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1715 function, since declaiming return types involving aliens is
1716 (1) annoyingly messy to do by hand and (2) vital to efficient
1717 compilation of code which calls such functions.
1718 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1719 longer reexported by the SB-EXT package. They're solely useful
1720 for alien code, so it seems more logical that you should get
1721 them from the SB-ALIEN package, not in SB-EXT.
1722 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1723 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1724 features. Instead, the code that they used to control is always
1725 built into the system.
1726 * many other bug fixes
1727 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1728 advantage of the new EVAL-WHEN stuff and to clean them up in
1729 general, and they are now more ANSI-compliant in a number of
1730 ways. Martin Atzmueller is responsible for a lot of this.
1731 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1733 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1734 *** correct ERROR type for various file operations
1735 *** some fixes for Lisp streams
1736 *** DEFMETHOD syntax checking
1737 *** changing old weird representation of debug information as
1738 strings (which, among their other deficiencies, don't transform
1739 correctly when you rename packages, and don't change their
1740 print representation when you change things like *PACKAGE*
1741 and *PRINT-LENGTH*) to symbols and lists of symbols
1742 He also made several improvements and fixed several bugs in DESCRIBE.
1743 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1745 *** misbehavior of WRITE-STRING/WRITE-LINE
1746 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1747 and several other LOOP problems as well
1748 *** DIRECTORY when similar filenames are present
1749 *** DEFGENERIC with :METHOD options
1750 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1751 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1752 *** argument ordering in FIND with :TEST option
1753 *** mishandled package designator argument in APROPOS-LIST
1754 *** various problems in the backquote readmacro
1755 *** a bug in APROPOS
1756 *** probably some others that I'm not describing very well here,
1757 since the CVS log documents them by reference to sbcl-devel
1758 messages, and the SourceForge archives aren't working well.:-(
1759 ** Dan Barlow improved the Alpha port (and is making progress on the
1760 PPC port, for those of you who think different).
1761 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
1762 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
1763 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
1764 continues to work on the SPARC port (for those of you in a position
1765 to look down upon our little PC-compatible boxes from a great height).
1766 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
1767 and uses that copy, rather than the current dynamic value, when
1768 it's trying to decide whether to truncate output. Thus e.g.
1769 (let ((*print-lines* 50))
1770 (pprint-logical-block (stream nil)
1772 (let ((*print-lines* 8))
1773 (print (aref possiblybigthings i) stream)))))
1774 should now truncate the logical block only at 50 lines, instead of
1775 often truncating it at 8 lines, as it did before.
1776 * The doc/cmucl/ directory, containing old CMU CL documentation
1777 from the time of the fork, is no longer part of the base system.
1778 SourceForge has shut down its anonymous FTP service, and with it
1779 my original plan for distributing the old CMU CL documentation
1780 there. For now, if you need these files you can download an old
1781 SBCL source release and extract them from it.
1782 * The fasl file version number changed again, for dozens of reasons,
1783 some of which are apparent above.
1785 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
1786 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
1787 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
1788 patch, so that DISASSEMBLE output is much nicer.
1789 * The code in the SB-PROFILE package now seems reasonably stable.
1790 I still haven't decided what the final interface should look like
1791 (I'd like PROFILE to interact cleanly with TRACE, since both
1792 facilities use function encapsulation) but if you have a need
1793 for profiling now, you can probably use it successfully with
1794 the current CMU-CL-style interface.
1795 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
1796 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
1797 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
1798 link now returns the dangling link itself, and for similar
1799 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
1800 link itself. (In these cases the old code signalled an error and
1801 looped endlessly, respectively.) Thus, DIRECTORY now works even
1802 in the presence of dangling and cyclic symbolic links.
1803 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
1804 is now a supported extension again, since the consensus on
1805 sbcl-devel was that it can be useful for ordinary development
1806 work, not just for debugging SBCL itself.
1807 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
1808 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
1809 that currently-defined functions might be redefined later with
1810 different return types.
1811 * Hash tables can be printed readably, as inspired by CMU CL code
1812 of Eric Marsden and SBCL code of Martin Atzmueller.
1813 * better error handling in CLOS method combination, thanks to
1814 Martin Atzmueller porting Pierre Mai's CMU CL patches
1815 * more overflow fixes for >16Mbyte I/O buffers
1816 * A bug in READ has been fixed, so that now a single Ctrl-D
1817 character suffices to cause end-of-file on character streams.
1818 In particular, now you only need one Ctrl-D at the command
1819 line (not two) to exit SBCL.
1820 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
1822 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
1823 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
1824 does the right thing.
1825 * The implementation of some type tests, especially for CONDITION
1826 types, is now tidier and maybe faster, due to CMU CL code
1827 originally by Douglas Crosher, ported by Martin Atzmueller.
1828 * Some math functions have been fixed, and there are new
1829 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
1830 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
1831 * (There are also some new optimizers in contrib/*-extras.lisp. Those
1832 aren't built into sbcl-0.6.13, but are a sneak preview of what's
1833 likely to be built into sbcl-0.7.0.)
1834 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
1835 CL, ported to SBCL by Martin Atzmueller)
1836 * DESCRIBE now gives more information in some cases. (Pierre Mai's
1837 patch to CMU CL, ported to SBCL by Martin Atzmueller)
1838 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
1839 * There's a new slam.sh hack to shorten the edit/compile/debug
1840 cycle for low-level changes to SBCL itself, and a new
1841 :SB-AFTER-XC-CORE target feature to control the generation of
1842 the after-xc.core file needed by slam.sh.
1843 * minor incompatible change: The ENTRY-POINTS &KEY argument to
1844 COMPILE-FILE is no longer supported, so that now every function
1845 gets an entry point, so that block compilation looks a little
1846 more like the plain vanilla ANSI section 3.2.2.3 scheme.
1847 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
1848 returns the number of bytes consed since the system started,
1849 rather than the number consed since the first time the function
1850 was called. (The new definition parallels ANSI functions like
1851 CL:GET-INTERNAL-RUN-TIME.)
1852 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
1853 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
1854 supported. Now DIRECTORY always does the abstract Common-Lisp-y
1855 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
1856 * Fasl file version numbers are now independent of the target CPU,
1857 since historically most system changes which required version
1858 number changes have affected all CPUs equally. Similarly,
1859 the byte fasl file version is now equal to the ordinary
1862 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
1863 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
1864 is no longer recognized. I apologize for this, because it was
1865 listed in SB-EXT as a supported extension, but I found that
1866 its existing behavior was poorly specified, as well as incorrectly
1867 specified, and it looked like too much of a mess to straighten it
1868 out. I have enough on my hands trying to get ANSI stuff to work..
1869 * many patches ported from CMU CL by Martin Atzmueller, with
1870 half a dozen bug fixes in pretty-printing and the debugger, and
1871 half a dozen others elsewhere
1872 * fixed bug 13: Floating point infinities are now supported again.
1873 They might still be a little bit flaky, but thanks to bug reports
1874 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
1875 as flaky as they were.
1876 * The --noprogrammer command line option is now supported. (Its
1877 behavior is slightly different in detail from what the old man
1878 page claimed it would do, but it's still appropriate under the
1879 same circumstances that the man page talks about.)
1880 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
1881 are now supported, and enabled by default. Thus, the compiler can
1882 handle many floating point and complex operations much less
1883 inefficiently. (Thus e.g. you can implement a complex FFT
1885 * The compiler now detects type mismatches between DECLAIM FTYPE
1886 and DEFUN better, and implements CHECK-TYPE more correctly, and
1887 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
1888 thanks to patches from Martin Atzmueller.
1889 * various fixes to make the cross-compiler more portable to
1890 ANSI-conforming-but-different cross-compilation hosts (notably
1891 Lispworks for Windows, following bug reports from Arthur Lemmens)
1892 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
1893 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
1894 fixed thanks to Pierre Mai's CMU CL patches.
1895 * fixes to keep the system from overflowing internal counters when
1896 it tries to use i/o buffers larger than 16M bytes
1897 * fixed bug 45a: Various internal functions required to support
1898 complex special functions have been merged from CMU CL sources.
1899 (When I was first setting up SBCL, I misunderstood a compile-time
1900 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
1901 * improved support for type intersection and union, fixing bug 12
1902 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
1903 more obscure bugs as well
1904 * some steps toward byte-compiling non-performance-critical
1905 parts of the system, courtesy of patches from Martin Atzmueller
1906 * Christophe Rhodes has made some debian packages of sbcl at
1907 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
1908 From his sbcl-devel e-mail of 2001-04-08 they're not completely
1909 stable, but are nonetheless usable. When he's ready, I'd be happy
1910 to add them to the SourceForge "File Releases" section. (And if
1911 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
1912 * new fasl file format version number (because of changes in
1913 internal representation of (OR ..) types to accommodate the new
1914 support for (AND ..) types, among other things)
1916 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
1917 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
1919 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
1921 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
1922 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
1923 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
1924 * DESCRIBE now works on structure objects again.
1925 * Most function call argument type mismatches are now handled as
1926 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
1927 know whether the function will be redefined before the call is
1928 executed. (The compiler could flag local calls with full WARNINGs,
1929 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
1930 it doesn't keep track of enough information to know whether calls
1931 are local in this sense.)
1932 * Compiler output is now more verbose, with messages truncated
1933 later than before. (There should be some supported way for users
1934 to override the default verbosity, but I haven't decided how to
1935 provide it yet, so this behavior is still controlled by the internal
1936 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
1937 src/compiler/ir1util.lisp.)
1938 * Fasl file format version numbers have increased again, because
1939 support for the Gray streams extension changes the layout of the
1940 system's STREAM objects.
1941 * The Gray subclassable streams extension now works, thanks to a
1942 patch from Martin Atzmueller.
1943 * The full LOAD-FOREIGN extension (not just the primitive
1944 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
1945 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
1946 but like most other programs, it defaults to copying the Unix
1947 environment from the original process instead of starting the
1948 new process in an empty environment.
1949 * Extensions which manipulate the Unix environment now support
1950 an :ENVIRONMENT keyword option which doesn't smash case or
1951 do other bad things. The CMU-CL-style :ENV option is retained
1952 for porting convenience.
1953 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
1954 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
1956 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
1958 * A patch from Martin Atzmueller seems to have solved the SIGINT
1959 problem, and as far as we know, signal-handling now works cleanly.
1960 (If you find any new bugs, please report them!)
1961 * The system no longer defaults Lisp source file names to types
1962 ".l", ".cl", or ".lsp", but only to ".lisp".
1963 * The compiler no longer uses special default file extensions for
1964 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
1965 seems to expect a single default extension for all compiled code,
1966 and there's no compelling reason to try to stretch the standard
1967 to allow two different extensions.) Instead, byte-compiled files
1968 default to the same extension as native-compiled files.
1969 * Fasl file format version numbers have increased again, because
1970 a rearrangement of internal implementation packages made some
1971 dumped symbols in old fasl files unreadable in new cores.
1972 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
1973 many fewer weird special cases.
1974 * Bug #17 (differing COMPILE-FILE behavior between logical and
1975 physical pathnames) has been fixed, and some related misbehavior too,
1976 thanks to a patch from Martin Atzmueller.
1977 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
1978 by Tim Moore, ported to SBCL by Martin Atzmueller.
1979 * Martin Atzmueller fixed several filesystem-related problems,
1980 including bug #36, in part by porting CMU CL patches, which were
1981 written in part by Paul Werkowski.
1982 * More compiler warnings in src/runtime/ are gone, thanks to
1983 more patches from Martin Atzmueller.
1984 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
1987 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
1989 * DESCRIBE now works on CONDITION objects.
1990 * The debugger now handles errors which arise when trying to print
1991 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
1993 * The build system now uses an additional file, customize-target-features.lisp,
1994 to allow local modifications to the target *FEATURES* list. (The point of
1995 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
1996 debugging features enabled, without having to worry about propagating it
1997 into everyone's system when I do a "cvs update".) When no
1998 customize-target-features.lisp file exists, the target *FEATURES* list
1999 should be constructed the same way as before.
2000 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2001 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2002 warn about behavior which is undefined under the ANSI spec. Note: This
2003 is specified by ANSI, but it's not very popular with programmers.
2004 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2005 macro in the SBCL sources for an example of a workaround which you
2006 might use to make portable ANSI-standard code which does what you want.
2007 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2008 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2009 when it's used as a non-toplevel form. (This is required in order
2010 to implement the DEFCONSTANT-EQX macro.)
2011 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2012 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2013 funny things, due to bugs in the implementation of EVAL-WHEN
2014 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2015 non-toplevel DEFCONSTANTs are uncommon.)
2016 * The core file version number and fasl file version number have been
2017 incremented, because the old noncompliant DEFCONSTANT behavior involved
2018 calling functions which no longer exist, and because I also took the
2019 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2020 * fixed bug 1 (error handling before read-eval-print loop starts), and
2021 redid debugger restarts and related debugger commands somewhat while
2023 ** The QUIT debugger command is gone, since it did something
2024 rather different than the SB-EXT:QUIT command, and since it never
2025 worked properly outside the main toplevel read/eval/print loop.
2026 Invoking the new TOPLEVEL restart provides the same functionality.
2027 ** The GO debugger command is also gone, since you can just invoke
2028 the CONTINUE restart directly instead.
2029 ** The TOP debugger command is also gone, since it's redundant with the
2030 FRAME 0 command, and since it interfered with abbreviations for the
2032 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2034 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2035 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2036 some porting work by Martin Atzmueller)
2037 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2038 with a boatload of patches which he ported to SBCL. Now that those
2040 ** The system tries to make sure that its low-priority messages
2041 are prefixed by semicolons, to help people who like to use
2042 syntax highlighting in their ILISP buffer. (This patch
2043 was originally due to Raymond Toy.)
2044 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2045 patches originally written by Raymond Toy.
2046 ** The compiler understands coercion between single-value and
2047 multiple-VALUES type expressions better, getting rid of some very
2048 weird behavior, thanks to patches originally by Robert MacLachlan
2049 and Douglas Crosher.
2050 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2051 lambda lists, thanks to a patch originally by Pierre Mai.
2052 ** The system no longer bogusly warns about "abbreviated type
2054 ** The compiler gets less confused by inlining and RETURN-FROM,
2055 thanks to some patches originally by Tim Moore.
2056 ** The system no longer hangs when dumping circular lists to fasl
2057 files, thanks to a patch originally from Douglas Crosher.
2058 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2059 undefined function error.
2060 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2061 in structure slot initforms) for now, documented workaround instead:-|
2062 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2063 * fixed bug 5: added stubs for various Gray stream functions called
2064 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2065 installed, at least appropriate type errors are generated
2066 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2067 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2068 consistently in DEFMETHOD forms.
2069 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2070 it doesn't seem to affect SBCL after all
2071 * The C runtime system now builds with better optimization and many
2072 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2074 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2076 * The system is now under CVS at SourceForge (instead of the
2077 CVS repository on my home machine).
2078 * The new signal handling code has been tweaked to treat register
2079 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2080 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2081 downstream, e.g. in debug-int.lisp, has implicit dependencies
2082 on the unsignedness of integer representation of machine words,
2083 and that caused the system to bomb out with infinite regress
2084 when trying to recover from type errors involving signed values,
2085 e.g. (BUTLAST '(1 2 3) -1).)
2086 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2087 (This was one of the bugs Peter Van Eynde reported back in July.)
2088 * The system now uses code inspired by Colin Walters' O(N)
2089 implementation of MAP (from the cmucl-imp@cons.org mailing
2090 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2091 inline the MAP operation, and there is more than one
2092 sequence argument to the MAP call (so that it can't just
2093 do ETYPECASE once and for all based on the type of the
2094 single sequence argument). (The old non-inline implementation
2095 of the general M-argument sequence-of-length-N case required
2096 O(M*N*N) time when any of the sequence arguments were LISTs.)
2097 * The QUIT :UNIX-CODE keyword argument has been renamed to
2098 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2099 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2101 * Raymond Wiker's patches to port dynamic loading from Linux to
2102 FreeBSD have been added.
2103 * The BUGS file is now more nearly up to date, thanks in large part
2104 to Martin Atzmueller's review of it.
2105 * The debugger now flushes standard output streams before it begins
2106 its output ("debugger invoked" and so forth).
2107 * The core version number and fasl file version number have both
2108 been incremented, because of incompatible changes in the layout
2110 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2111 used to not be called for a saved Lisp image.)
2112 * Martin Atzmueller's version of a patch to fix a compiler crash,
2113 as posted on sbcl-devel 13 September 2000, has been installed.
2114 * Instead of installing Martin Atzmueller's patch for the
2115 compiler transform for SUBSEQ, I deleted the compiler transform,
2116 and transforms for some similar consing operations.
2117 * A bug in signal handling which kept TRACE from working on OpenBSD
2119 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2120 away by constant folding
2121 * The system now defines its address space constants in one place
2122 (in the Lisp sources), and propagates them automatically elsewhere
2123 (through GENESIS and the sbcl.h file). Therefore, patching the
2124 address map is less unnecessarily tedious and error-prone. The
2125 Lisp names of address space constants have also been systematized.
2126 * CVS tags like dollar-Header-dollar have been removed from
2127 the sources, because they have never saved me trouble and
2128 they've been source of trouble working with patches and other
2129 diff-related operations.
2130 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2131 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2133 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2135 * The system has been ported to OpenBSD.
2136 * The system now compiles with a simple "sh make.sh" on the systems
2137 that it's supported on. I.e., now you no longer need to tweak
2138 text in the target-features.lisp-expr and symlinks in src/runtime/
2139 by hand, the make.sh takes care of it for you.
2140 * The system is no longer so grossly inefficient when compiling code
2141 involving vectors implemented as general (not simple) vectors (VECTOR T),
2142 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2143 which dares to use the various sequence functions on non-simple
2144 vectors, takes less of a performance hit.
2145 * There is now a primitive type predicate VECTOR-T-P
2146 to test for the (VECTOR T) type, so that e.g.
2147 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2148 can now be compiled with some semblance of efficiency. (The old code
2149 turned the type declaration into a full call to %TYPEP at runtime!)
2150 * AREF on (VECTOR T) is still not fast, since it's still compiled
2151 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2152 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2153 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2154 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2155 without making full calls to SUBTYPEP at runtime.
2156 (Some analogous efficiency issues for non-simple vectors specialized to
2157 element types other than T, or for non-simple multidimensional arrays,
2158 have not been addressed. They could almost certainly be handled the
2159 same way if anyone is motivated to do so.)
2160 * The changes in array handling break binary compatibility, so
2161 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2162 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2163 instead of (VALUES T T).
2164 * By following the instructions that Dan Barlow posted to sbcl-devel
2165 on 2 July 2000, I was able to enable primitive dynamic object
2166 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2167 functionality is not implemented (since it calls ld to resolve
2168 library references automatically, requiring RUN-PROGRAM for its
2169 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2170 try to resolve library references) is now supported.
2171 * The system now flushes the standard output streams when it terminates,
2172 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2173 them at several other probably-convenient times, e.g. in each pass of
2174 the toplevel read-eval-print loop, and after evaluating a form given
2175 as an "--eval" command-line option. (These changes were motivated by a
2176 discussion of stream flushing issues on cmucl-imp in August 2000.)
2177 * The source transform for TYPEP of array types no longer assumes
2178 that an array whose element type is a not-yet-defined type
2179 is implemented as an array of T, but instead punts, so that the
2180 type will be interpreted at runtime.
2181 * There is now some support for cross-compiling in make.sh: each of
2182 the phases of make.sh has its own script. (This should be transparent
2183 to people doing ordinary, non-cross-compile builds.)
2184 * Since my laptop doesn't have hundreds of megabytes of memory like
2185 my desktop machine, I became more motivated to do some items on
2186 my to-do list in order to reduce the size of the system a little:
2187 ** Arrange for various needed-only-at-cold-init things to be
2188 uninterned after cold init. To support this, those things have
2189 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2190 symbols with such names are now uninterned after cold init).
2191 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2192 fasl files for cold load.
2193 ** Remove the old compiler structure pooling code (which used to
2194 be conditional on the target feature :SB-ALLOC) completely.
2195 ** Redo the representation of some data in cold init to be more compact.
2196 (I also looked into supporting byte compiled code at bootstrap time,
2197 which would probably reduce the size of the system a lot, but that
2198 looked too complicated, so I punted for now.)
2199 * The maximum signal nesting depth in the src/runtime/ support code has
2200 been reduced from 4096 to 256. (I don't know any reason for the very
2201 large old value. If the new smaller value turns out to break something,
2202 I'll probably just bump it back up.)
2203 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2205 * Many, many bugs reported by Peter Van Eynde have been added to
2206 the BUGS list; some have even been fixed.
2207 * While enabling dynamic object file loading, I tried to make the
2208 code easier to understand, renaming various functions and variables
2209 with less ambiguous names, and changing some function calling
2210 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2211 * While trying to figure out how to do the OpenBSD port, I tried to
2212 clean up some of the code in src/runtime/. In particular, I dropped
2213 support for non-POSIX signal handling, added various comments,
2214 tweaked the code to reduce the number of compilation warnings, and
2215 renamed some files to increase consistency.
2216 * To support the new automatic configuration functionality in make.sh,
2217 the source file target-features.lisp-expr has been replaced with the
2218 source file base-target-features.lisp-expr and the machine-generated
2219 file local-target-features.lisp-expr.
2220 * fixed a stupid quoting error in make.sh so that using CMU CL
2221 "lisp -batch" as cross-compilation host works again
2223 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2225 * DESCRIBE no longer tries to call itself recursively to describe
2226 bound/fbound values, so that it no longer fails on symbols which are
2227 bound to themselves (like keywords, T, and NIL).
2228 * DESCRIBE now works on generic functions.
2229 * The printer now prints less-screwed-up representations of closures
2230 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2231 * A private symbol is used instead of the :EMPTY keyword previously
2232 used to mark empty slots in hash tables. Thus
2233 (DEFVAR *HT* (MAKE-HASH-TABLE))
2234 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2235 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2236 now does what ANSI says that it should. (You can still get
2237 similar noncompliant behavior if bang on the hash table
2238 implementation with all the symbols you get back from
2239 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2240 This breaks binary compatibility, since tests for equality to
2241 :EMPTY are wired into things like the macroexpansion of
2242 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2244 * There's now a minimal placeholder implementation for CL:STEP,
2245 as required by ANSI.
2246 * An obscure bug in the interaction of the normal compiler, the byte
2247 compiler, inlining, and structure predicates has been patched
2248 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2249 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2251 * Missing ordinary arguments in a macro call are now detected even
2252 when the macro lambda list contains &KEY or &REST.
2253 * The debugger no longer complains about encountering the top of the
2254 stack when you type "FRAME 0" to explicitly instruct it to go to
2255 the top of the stack. And it now prints the frame you request even
2256 if it's the current frame (instead of saying "You are here.").
2257 * As specified by ANSI, the system now always prints keywords
2258 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2260 * The default initial SIZE of HASH-TABLEs is now smaller.
2261 * Type information from CLOS class dispatch is now propagated
2262 into DEFMETHOD bodies, so that e.g.
2263 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2265 is now basically equivalent to
2266 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2267 (DECLARE (TYPE SINGLE-FLOAT X))
2269 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2270 operation, without having to do run-time type dispatch.
2271 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2272 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2273 * Since I don't seem to be making big file reorganizations very often
2274 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2275 is overflowing my ability to conveniently back them up), I've finally
2276 checked the system into CVS. (The CVS repository is on my home system,
2277 not at SourceForge -- putting it on SourceForge might come later.)
2278 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2279 high-level GC-NOTIFY-FOO functions send their output. (There's
2280 still very little control of where low-level verbose GC functions
2281 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2282 less than it used to -- the GC-NOTIFY-FOO functions are now under
2283 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2284 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2285 in only one place in the source code, and propagates it automatically
2286 everywhere that it's needed. Thus e.g. when I bump the version from
2287 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2288 * The C source files now include boilerplate legalese and documentation
2289 at the head of each file (just as the Lisp source files already did).
2290 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2291 to his page will be replaced with a link to his new CLiki service.
2293 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2295 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2296 * The build process now looks for GNU make under the default name "gmake",
2297 instead of "make" as it used to. If GNU make is not available as "gmake"
2298 on your system, you can change this default behavior by setting the
2299 GNUMAKE environment variable.
2300 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2301 can build without error under CMU CL.
2303 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2305 * There is now a partial SBCL user manual (with some new text and some
2306 text cribbed from the CMU CL manual).
2307 * The beginnings of a profiler have been added (starting with the
2308 CMU CL profiler and simplifying and cleaning up). Eventually the
2309 main interface should be through the TRACE macro, but for now,
2310 it's still accessed through vaguely CMU-CL-style functions and macros
2311 exported from the package SB-PROFILE.
2312 * Some problems left over from porting CMU CL to the new
2313 cross-compilation bootstrap process have been cleaned up:
2314 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2315 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2316 part when getting rid of the compiler *BACKEND* stuff.)
2317 ** The value of *NULL-TYPE* was screwed up, because it was
2318 being initialized before the type system knew the final
2319 definition of the 'NULL type. This screwed up several key
2320 optimizations in the compiler, causing inefficiency in all sorts
2321 of places. (I found it because I wanted to understand why
2322 GET-INTERNAL-RUN-TIME was consing.)
2323 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2324 PROCLAIM FTYPE information. Unfortunately this broke binary
2325 compatibility again, since now the forms output by DEFGENERIC
2326 to refer to functions which didn't exist in 0.6.3.
2327 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2328 can use the new (as of 0.6.3) transform for SOME into MAP into
2330 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2331 values to (INTEGER 0 1000000), so that the time code will no longer
2332 occasionally get blown up by Linux returning 1000000 microseconds
2333 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2334 its output conform to the ANSI spec. (Alas, this makes its output
2335 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2336 * A full call to MAP NIL with a single sequence argument no longer conses.
2337 * fixes to problems pointed out by Martin Atzmueller:
2338 * The manual page no longer talks about multiprocessing as though
2339 it were currently supported.
2340 * The ILISP support patches have been removed from the distribution,
2341 because as of version 5.10.1, ILISP now supports SBCL without us
2342 having to maintain patches.
2343 * added a modified version of Raymond Toy's recent CMU CL patch for
2344 EQUALP comparison of HASH-TABLE
2346 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2348 * The system still can't cross-compile itself with
2349 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2350 entails), but at least it can compile more of itself that way
2351 than it used to be able to, and various buglets which were uncovered
2352 by trying to cross-compile itself that way have now been fixed.
2353 * This release breaks binary compatibility again. This time
2354 at least I've incremented the FASL file format version to 2, so that the
2355 problem can be detected reliably instead of just causing weird errors.
2356 * various new style warnings:
2357 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2358 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2359 ** using the lexical binding of a variable named in the *FOO* style
2360 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2361 as specified by ANSI.
2362 * *RANDOM-STATE* is no longer automatically initialized from
2363 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2364 default behavior of the system is to repeat its behavior every time
2365 it's run. If you'd like to change this behavior, you can always
2366 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2367 old convention there was no comparably easy way to get the system to
2368 repeat its behavior every time it was run.
2369 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2370 FTYPE declarations has been removed, in favor of their ANSI
2371 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2372 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2373 the types of their sequence arguments can be determined at compile time.
2374 This is done through a new open code expansion for MAP which eliminates
2375 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2376 argument types can be determined at compile time.
2377 * The optimizer now transforms COERCE into an identity operation when it
2378 can prove that the coerced object is already of the correct type. (This
2379 can be a win for machine generated code, including the output of other
2380 optimization transforms, such as the MAP transform above.)
2381 * Credit information has been moved from source file headers into CREDITS.
2382 * Source file headers have been made more standard.
2383 * The CASE macro now compiles without complaining even when it has
2386 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2388 * (Note that the way that the PCL macroexpansions were rewritten
2389 to accommodate the change in DEFGENERIC below breaks binary
2390 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2391 not run under sbcl-0.6.2. Once we get out of alpha releases,
2392 i.e. hit release 1.0.0, we'll probably try to maintain binary
2393 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2394 and sbcl-1.4.4. Until then, however, it might be fairly common
2395 for maintenance releases to break binary compatibility.)
2396 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2398 * The reporting of unhandled signals has been changed to print some
2399 explanatory text as well as the report form. (Previously only
2400 the report form was printed.)
2401 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2402 it defines, so that the compiler no longer issues undefined function
2403 warnings for compiled-but-not-yet-loaded generic functions.
2404 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2405 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2406 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2407 section of the ANSI spec are supported.
2408 * The "" nickname for the "KEYWORD" package has been removed.
2409 The reader still handles symbol tokens which begin with a package marker
2410 as keywords, but it doesn't expose its mechanism for doing so in the
2411 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2412 * The system now issues STYLE-WARNINGs for contradictory TYPE
2413 proclamations. (Warnings for contradictory FTYPE proclamations would
2414 be nice too, but those can't be done usefully unless the type system
2415 is made smarter about FUNCTION types.)
2416 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2417 systematized, so that "*target-*.lisp is supposed to exist only on the
2418 target and imply that there's a related file which exists on the
2419 host, and *host-*.lisp is supposed to exist only on the host and imply
2420 that there's a related file which exists on the target. This involves a
2421 lot of renaming. Hopefully the acute confusion caused by the renaming
2422 will be justified by the reduction in chronic confusion..
2423 ** runtime-type.lisp -> early-target-type.lisp
2424 ** target-type.lisp -> late-target-type.lisp
2425 ** early-host-format.lisp -> early-format.lisp
2426 ** late-host-format.lisp -> late-format.lisp
2427 ** host-error.lisp -> misc-error.lisp
2428 ** early-error.lisp -> early-target-error.lisp
2429 ** late-error.lisp -> late-target-error.lisp
2430 ** host-defboot.lisp -> early-defboot.lisp
2431 ** code/misc.lisp -> code/target-misc.lisp
2432 ** code/host-misc.lisp -> code/misc.lisp
2433 ** code/numbers.lisp -> code/target-numbers.lisp
2434 ** code/early-numbers.lisp -> numbers.lisp
2435 ** early-host-type.lisp -> early-type.lisp
2436 ** late-host-type.lisp -> late-type.lisp
2437 ** host-typep.lisp -> typep.lisp
2438 ** load.lisp -> target-load.lisp
2439 ** host-load.lisp -> load.lisp
2440 ** host-disassem.lisp -> disassem.lisp
2441 ** host-insts.lisp -> insts.lisp
2442 ** byte-comp.lisp -> target-byte-comp.lisp
2443 ** host-byte-comp.lisp -> byte-comp.lisp
2444 ** host-signal.lisp -> signal.lisp
2445 ** host-defstruct.lisp -> defstruct.lisp
2446 ** late-target-type.lisp -> deftypes-for-target.lisp
2447 Furthermore, several other previously target-only files foo.lisp (e.g.
2448 hash-table.lisp and random.lisp) have been split into a target-and-host
2449 foo.lisp file and a target-only target-foo.lisp file, with their key type
2450 definitions in the target-and-host part, so that the cross-compiler will
2451 know more about target types.
2452 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2453 gone away. In their place are various *BACKEND-FOO* variables
2454 corresponding to the slots of the old structure.
2455 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2456 into the target SBCL has been fixed.
2457 * The chill.lisp system for loading cold code into a running SBCL
2459 * Support for the CMU CL "scavenger hook" extension has been removed.
2460 (It was undocumented and unused in the CMU CL sources that SBCL was
2461 derived from, and stale in sbcl-0.6.1.)
2462 * Various errors in the cross-compiler type system were detected
2463 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2464 (enabling various consistency checks). Many of them were fixed,
2465 but some hard problems remain, so the compiler is back to
2466 running without *TYPE-SYSTEM-INITIALIZED* for now.
2467 * As part of the cross-compiler type system cleanup, I implemented
2468 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2469 * I have started adding UNCROSS calls throughout the type system
2470 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2471 on cross-compiler input files will be able to go away, and various
2473 * CONSTANTP now returns true for quoted forms (as explicitly required
2476 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2478 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2479 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2480 (MAKE-LIST -1). In the longer term, it ought to become true
2481 that declarations are assertions even at SAFETY 1. For now, it's not
2482 quite true even at SAFETY 3, but it's at least more nearly true..
2483 (Note that this change seems to increases the size of the system by
2484 O(5%) and to decrease the speed of the compiler by 20% or more.)
2485 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2486 for the problem of printing dozens of lines of distracting information
2487 about low-level system machinery as part of the top stack frame
2488 on entry to the debugger when an undefined function was called.
2489 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2490 is not reset to COMMON-LISP-USER.
2491 * Compilation of stuff related to dyncount.lisp has been made conditional
2492 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2493 smaller. The various dyncount-related symbols have been moved into
2494 a new "SB-DYNCOUNT" package.
2495 * tty-inspect.lisp has been renamed to inspect.lisp.
2496 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2497 feature has gone away. (When we eventually port to other flavors of
2498 libc and/or Unix, we'll try to make the differences between flavors
2499 invisible at the user level.)
2500 * Various other *FEATURES* tags, and/or their associated conditionals,
2501 have been removed if obsolescent, or given better documentation, or
2502 sometimes given more-mnemonic names.
2504 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2506 * tidied up "make.sh" script
2507 * tidied up system directory structure
2508 * better "clean.sh" behavior
2509 * added doc/FOR-CMUCL-DEVELOPERS
2510 * many many small tweaks to output format, e.g. removing possibly-confusing
2511 trailing #\. character in DESCRIBE-INSTANCE
2512 * (EQUALP #\A 'A) no longer signals an error.
2513 * new hashing code, including EQUALP hashing
2514 * tidied up Lisp initialization and toplevel
2515 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2516 * command line argument processing
2517 * added POSIX-GETENV function to deal with Unix-ish environment variables
2518 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2519 terminating SBCL on EOF
2520 * non-verbose GC by default
2521 * There is no more "sbcl" shell script; the sbcl file is now the C
2522 runtime executable (just like CMU CL).
2523 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2525 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2526 symbol and package information as Lisp native symbol and package objects
2527 instead of strings naming symbols and strings naming packages. This way,
2528 whenever packages are renamed (as in warm init), debug information is
2529 transformed along with everything else.
2530 * tweaked the optimization policy declarations which control the building
2531 of SBCL itself. Now, among other things, the system no longer saves
2532 source location debugging information. (This helps two problems at once
2533 by reducing SBCL size and by keeping SBCL from trying to look for its
2534 sources -- which may not exist -- when reporting errors.)
2535 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2536 debugging and testing purposes
2537 * cleaned up printing, making the printer call PRINT-OBJECT for
2538 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2539 methods, giving nearly-ANSI behavior
2540 * converted almost all special variables to use *FOO* naming convention
2541 * deleted PARSE-TIME functionality, since it can be done portably
2542 * moved some files out of cold init into warm init
2543 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2545 * regularized formatting of source files
2546 * added an install.sh script
2547 * fixed ridiculous memory usage of cross-compiler by making
2548 compiler/alloc.lisp not try to do pooling unless it can hook
2549 itself into the GC of the cross-compilation host. Now the system
2550 builds nicely on my old laptop.
2551 * added :SB-ALLOC in target-features.lisp-expr
2552 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2553 was not implemented)
2554 * re-did condition handling and note reporting in the compiler. Notes
2555 are no longer handled by signalling conditions. Style warnings
2556 and warnings are handled more correctly and reported in such a way
2557 that it's easy to find one or the other in your output (so that you
2558 can e.g. figure out which of many problems caused COMPILE-FILE to
2560 * changed the severity of several compiler warnings from full WARNING
2561 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2562 compiler note reporting so that it doesn't use the condition system
2563 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2564 COMPILE-FILE command)
2565 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2566 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2567 no longer supported, since their implementation was deeply intertwingled
2568 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2569 * removed bogus "support" for compiler macros named (SETF FOO), and
2570 removed the compiler macro for SETF INFO (but only after making a fool
2571 of myself on the cmucl-imp mailing list by posting a bogus patch for
2572 DEFINE-COMPILER-MACRO..)
2573 * Compiled files containing forms which have side effects on the Lisp
2574 reader (such as DEFPACKAGE forms) are now handled more correctly.
2575 (Compiler queuing of top level lambdas has been suppressed by setting
2576 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2577 * deleted various currently-unused source files, e.g. gengc.lisp. They
2578 may be added back at some point e.g. when porting to other architectures,
2579 but until they are it's distracting to distribute them and to try to
2581 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2582 there were so many of them they're just distractions, and UNCROSS is
2583 known to be able to handle the current sources
2584 * moved PROFILE functionality into TRACE, so that it will be clear
2585 how the wrapping and unwrapping of functions when you profile them
2586 interacts with the wrapping and unwrapping of functions when you
2587 trace them. (Actually, the functionality isn't there yet, but at least
2588 the interface specification is there. Hopefully, the functionality will
2589 arrive with some future maintenance release.)
2590 * removed host-oops.lisp
2591 * changed signature of QUIT function to allow UNIX-CODE argument
2592 * fixed READ-SEQUENCE bug
2593 * tweaked verbose GC output so that it looks more like the progress
2594 output that ANSI specifies for functions like LOAD
2595 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2596 * added <http://sbcl.sourceforge.com> to the banner information printed by