1 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2 * bug fix: read-write consistency on streams of element-type
3 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
4 * bug fix: redefiniton of the only method of a generic function with
5 no DEFGENERIC no longer emits a full WARNING. In addition,
6 redefinition of generic functions with no DEFGENERIC to an
7 incompatible lambda list now signals an error. (thanks to Zach
9 * bug fix: DEFGENERIC now works even when there's a function of the
10 same name in an enclosing lexical environment. (thanks to Zach
12 * fixed compiler failure, caused by instrumenting code during
13 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
14 * optimization: added loop analysis and improved register allocation
15 * fixed some bugs revealed by Paul Dietz' test suite:
16 ** POSITION on displaced vectors with non-zero displacement
17 returns the right answer.
18 ** (SIMPLE-STRING) is a valid type specifier for sequence
20 ** *PRINT-LEVEL* handling for slotless structures is pedantically
22 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
23 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
25 ** The justification version of the FORMAT ~< directive treats
26 non-zero minpad parameter correctly.
28 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
29 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
30 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
31 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
32 the supported interface.
33 * new feature: Single-stepping of code compiled with DEBUG 2 or
34 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
35 * new feature: saving cores with foreign code loaded is now
36 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
37 Timothy Moore's work for CMUCL)
38 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
39 parameters to * instead of NIL if no initform is supplied.
40 (reported by Johan Bockgaard)
41 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
42 correctly when FUNCALL appears as the car of the form. Note:
43 despite this FUNCALL forms are not currently subject to
44 compiler-macro expansion. (port of Raymond Toy's fix for the
45 same from CMUCL, reported by Johan Bockgaard)
46 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
47 (thanks for Teemu Kalvas)
48 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
49 case instead of terminating the iteration (thanks for Julian
50 Squires, Sean Champ and Raymond Toy)
51 * bug fix: incorrect expansion of defgeneric that caused a style
52 warning. (thanks for Zach Beane)
53 * bug fix: slot accessor effective method computation works properly
54 for classes with multiple non-standard applicable methods on
55 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
56 * on x86 compiler supports stack allocation of results of LIST and
57 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
58 CMUCL implementation by Gerd Moellmann)
59 * fixed some bugs revealed by Paul Dietz' test suite:
60 ** FORMAT strings with both the ~<~:;~> form of the justification
61 directive and pretty-printing directives cause an error.
63 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
64 * incompatible change: the internal functions
65 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
66 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
67 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
68 instead of the old functions.
69 * new feature: on platforms where "dladdr" is available foreign
70 function names now appear in backtraces. (based on Helmut Eller's
72 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
73 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
75 * documentation: ASDF manual is now distributed with SBCL.
76 * bug fix: SBCL can now load its contributed modules with REQUIRE
77 even if the system-provided entries have been removed from
78 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
80 * bug fix: backtraces involving undefined functions or assembly
81 routines are more informative. (thanks to Brian Downing)
82 * bug fix: mutually referent alien structures now work correctly.
83 (reported by Rick Taube)
84 * bug fix: structures defined by WITH-ALIEN can be referred to
85 within other definitions in the same WITH-ALIEN.
86 * bug fix: division operators (MOD, TRUNCATE and the like) with
87 constant zero divisors and integer dividends no longer generate
89 * bug fix: provide default methods for INPUT-STREAM-P and
90 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
91 * bug fix: improve the detection and resolution of MOP metacycles
92 (where effective method computation depends on the generic function
93 having its effective method computed). (reported by Bruno Haible)
94 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
95 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
96 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
97 * bug fix: pathnames with relative directory components can be
98 represented relative to default pathnames.
99 * optimization: in taking the GCD of bignums, reduce the two bignums
100 to approximately the same size (using Euclid's algorithm) before
101 applying the more sophisticated binary GCD. (thanks to Juho
103 * optimization: COUNT on bitvectors now operates word-at-a-time.
104 * optimization: ASH with a positive, but not necessarily constant,
105 (leftwards) shift, when in a modular context, is compiled to a
107 * fixed some bugs revealed by Paul Dietz' test suite:
108 ** FORMAT variable parameters ("~V<char>") are defaulted properly
109 if the corresponding argument is NIL.
110 ** FORMAT directives accepting a minpad parameter treat negative
112 ** literal commas as character parameters to FORMAT directives are
114 ** literal spaces directly after ~<Newline> directives within a
115 format-logical-block (~:< ~@:>) do not induce :FILL-style
116 conditional newlines.
117 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
118 as specified: it no longer includes conditional newlines).
119 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
121 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
122 :PER-LINE-PREFIX argument does not evaluate to a string.
124 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
125 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
126 the manual for details; for now, package locks can be disabled by
127 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
128 if no major problems are found then it is likely that they will be
129 compiled in unconditionally.
130 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
131 unsupported operators on all platforms. To load a shared library
132 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
133 object file, link it into a shared library outside of SBCL and
134 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
136 * bug workaround: ROOM T and the GC were not working together
137 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
138 reporting facility aren't preserved. That reporting has been
139 disabled, so now until and unless someone figures out how to make it
140 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
141 * minor incompatible change: as threatened around sbcl-0.8.0, the
142 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
143 The symbols are also exported from SB-PCL for backwards
144 compatibility, but more so than before SB-PCL should be treated as
145 an implementation-internal package.
146 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
148 * the SB-SPROF contrib now works on (most) non-x86 architectures.
149 It is known as of this release not to work on the Alpha, however.
150 * fixed bug #167: errors signalled due to illegal syntax in method
151 bodies are now more legible.
152 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
153 designators and can hence be used with TYPEP.
154 * fixed bug #333: CHECK-TYPE now ensures that the type error
155 signalled, if any, has the right object to be accessed by
156 TYPE-ERROR-DATUM. (reported by Tony Martinez)
157 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
158 5.1.2.3 for multiple-value place subforms. (reported by Kalle
160 * fixed bug #334: programmatic addition of slots using specialized
161 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
162 and :ALLOCATION :CLASS slots.
163 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
164 float exponents. (rereported by Peter Seibel)
165 * fixed a bug: #\Space (and other whitespace characters) are no
166 longer considered to be macro characters in standard syntax by
168 * fixed bug: initialization of condition class metaobjects no longer
169 causes an instance of the condition to be created. (reported by
171 * fixed bug: it is now possible to have slots such that
172 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
173 description is neither :INSTANCE nor :CLASS.
174 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
175 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
176 SLOT-VALUE-USING-CLASS), and no longer causes errors with
177 non-standard SLOT-DEFINITION-ALLOCATION values.
178 * fixed bugs: various race conditions handling exiting threads.
179 CL-PPCRE's thread torture test now passes.
180 * fixed arguably-a-bug: GC time proportional to number of threads in
181 system even when most of them are idle
182 * optimization: improved performance of BIT and SBIT on bit-vectors.
183 * .fasl file incompatibility: The fasl file version number has
184 been incremented because of changes associated with package locks.
186 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
187 * minor incompatible change: the system no longer provides
188 optimization hints (conditions which are TYPEP
189 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
190 mode; these hints are still emitted when higher SPEED optimization
192 * new contrib module: a sampling profiler (profiling by statistical
193 sampling, rather than by instrumenting functions) is available as
194 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
195 * the behaviour of the standard function ED is now customizeable by
196 third parties through a hook variable: see ED's documentation
197 string for information on the protocol.
198 * the compiler no longer emits efficiency notes for (FUNCALL X)
199 when the type of X is uncertain under default optimization
201 * fixed bug 276: mutating a binding of a specialized parameter to a
202 method to something that is not TYPEP the specializer is now
204 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
205 exhibit the required behaviour on the x86 platform. (thanks to
206 Peter van Eynde, Eric Marsden and Bruno Haible)
207 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
208 even for difficult arguments. (reported by Peter Graves)
209 * fixed bug 141a: the backquote printer now descends quoted
211 * fixed another bug in backquote printing: no more destructive
212 modification of the form's list structure. (reported by Brian
214 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
215 move between its address being taken and the call to
216 interrupt_thread, fixing a crashing race condition.
217 * the SB-POSIX contrib implementation has been adjusted so that it
218 no longer exhibits ridiculously poor performance when constructing
219 instances corresponding to C structs.
221 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
222 * minor incompatible change: the sb-grovel contrib now treats C
223 structures as alien (in the sense of SB-ALIEN) objects rather than
224 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
225 has implications for memory management of client code
226 (sb-grovel-returned objects must now be manually managed) and for
227 type safety (alien objects now have full types).
228 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
229 used to control emission of compiler diagnostics, rather than the
230 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
231 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
232 quality should be considered deprecated.
233 * install.sh now installs the user manual as well
234 * (not quite a new documentable feature, but worth considering in
235 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
236 beginnings of a semantically meaningful condition hierarchy is
237 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
239 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
240 designator as the defaults argument.
241 * fixed bug: Displaced arrays whose displaced-to array has become
242 too small now cause ARRAY-DIMENSION to signal an error, providing
243 for safer bounds-checking. (reported by Bruno Haible)
244 * fixed bug: DEFCLASS slot definitions with identical :READER and
245 :WRITER names now signal a reasonable error. (reported by Thomas
247 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
249 * fixed bug: Invalid slot specification errors now print correctly.
250 (thanks to Zach Beane)
251 * fixed bug 320: Shared to local slot value transfers in class
252 redefinitions now happen corrently from superclasses as
253 well. (reported by Bruno Haible)
254 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
256 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
257 improper lists correctly. (reported by Bruno Haible)
258 * fixed bug 313: source-transform for <fun-name> was erroneously
259 applied to a call of a value of a variable with name <fun-name>.
260 (reported by Antonio Menezes Leitao)
261 * fixed bug 307: The obsolete instance protocol ensures that
262 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
263 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
264 remove dead unknown-values globs from the middle of the stack.
265 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
266 easier to resume long computations after using *BREAK-ON-SIGNALS*
267 to diagnose and fix failures. (thanks to Nikodemus Siivola)
268 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
269 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
270 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
271 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
272 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
274 * fixed a (fairly theoretical) bug in string printing: if
275 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
276 does not have array-element-type equal to the most general string
278 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
279 function name. (thanks to Nikodemus Siivola)
280 * fixed bug: compile-time format string checker failed on
281 non-closed ~{. (reported by Thomas F Burdick)
282 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
283 don't warn on legal loop code involving more than one aggregate
284 boolean. (thanks to Nikodemus Siivola)
285 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
286 in keyword position to MAKE-PACKAGE was not being reported as
288 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
289 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
291 * fixed bugs 280 and 312: the checking for multiple definitions in a
292 file is less likely to become confused by uses of inline
294 * fixed bug: the #S reader macro performs the keyword coercion
295 specified for slot names. (reported by Kalle Niemitalo)
296 * fixed bug: lambda lists may contain symbols whose names start with
297 & but are not lambda-list-keywords; their occurrence triggers a
299 * fixed bug 321: define-method-combination argument lambda lists do
300 not cause invalid code to be generated when &optional or &aux
301 variables have default values. (reported by Bruno Haible)
302 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
303 have CLOS classes; this appears also to have fixed reported
304 instability in stack exhaustion detection.
305 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
306 MUFFLE-WARNING when no associated restart is present is now
308 * optimization: rearranged the expansion of various defining macros
309 so that each expands into only one top-level form in a
310 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
312 * optimization: used a previously little-used slot in symbols to
313 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
315 * fixed some bugs revealed by Paul Dietz' test suite:
316 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
318 ** Fixed an optimization bug related to inheritance of initforms
319 from local to shared slots.
320 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
321 ** CAST optimizer forgot to flush argument derived type.
322 ** print/read consistency on floats is now orders of magnitude
323 more likely. (thanks also to Bruno Haible for a similar report
325 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
326 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
327 for objects of type REAL. Make it so.
328 ** SXHASH is specified (infelicitously) to respect similarity,
329 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
330 it so. (thanks to Markus Ziegler)
331 ** on the Alpha, the compiler succeeds in compiling functions
332 returning a known number of arguments greater than 63.
333 ** fixed handling of invalid NIL arguments in keyword position
335 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
337 ** provided a readably-printable representation for RANDOM-STATE
339 ** ensured that pathnames, where they have a namestring, always
340 print using #P"..." syntax.
342 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
343 * Support for the forthcoming 2.0 version of the NetBSD kernel
344 running on x86 hardware has been added. (thanks to Perry
345 E. Metzger most immediately, and others for their past work)
346 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
347 current release version; SBCL's OpenBSD support had been broken
348 since about the time of OpenBSD's switch to ELF binary format.)
349 * [placeholder for DX summary]
350 ** user code with &REST lists declared dynamic-extent, under high
351 speed or space and low safety and debug optimization policy.
352 * The manual has been converted to Texinfo format and the debugger
353 chapter from the cmucl manual has been added.
354 * A facility has been added to extract documentation strings from
355 sbcl and store them as Texinfo-formatted snippets for inclusion in
356 the manual (via Texinfo's @include directive)
357 * bug fix: compiler emitted division in optimized DEREF. (thanks for
358 the test case to Dave Roberts)
359 * bug fix: multidimensional simple arrays loaded from FASLs had fill
360 pointers. (reported by Sean Ross)
361 * bug fix: PROFILE output is printed nicely even for large numerical
362 values. (thanks to Zach Beane)
363 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
364 greater than 32 handle EOF correctly.
365 * bug fix: on X86 an immediate argument of the IMUL instruction is
366 correctly printed in disassembly. (thanks to Lutz Euler)
367 * bug fix: class slots in redefined classes preserve their old
368 values. (thanks to Bruno Haible and Nikodemus Siivola)
369 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
370 succeeds. (reported by Marco Baringer)
371 * fixed some bugs revealed by Paul Dietz' test suite:
372 ** READ-SEQUENCE now works on ECHO-STREAMs.
373 ** RATIONALIZE works more according to its specification. (thanks
376 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
377 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
378 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
379 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
380 should matter to you only if you rebind the printer control
381 variables and then find you want different bindings in the
382 debugger than in the ordinary execution of your program.)
383 * The runtime build system has been tweaked to support building
384 (on SPARC/SunOS) using a C compiler which invokes Sun's own
385 assembler and linker. (thanks to Nikodemus Siivola)
386 * Unbound, undefined, undeclared variables now trigger full
387 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
388 more often programmer error than deliberate exploitation of undefined
390 * optimization: the hash algorithm for strings has changed to one
391 that is less vulnerable to spurious collisions. (thanks to Juho
393 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
394 needless bounds checking. (thanks to Juho Snellman)
395 * optimization: implemented multiplication as a modular
396 (UNSIGNED-BYTE 32) operation on the PPC backend.
397 * fixed some bugs revealed by Paul Dietz' test suite:
398 ** ADJUST-ARRAY now copies the datum in a zero rank array if
400 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
401 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
402 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
404 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
406 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
408 ** ECHO-STREAMs no longer attempt to echo the end of file value to
409 their output stream on EOF from read.
410 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
411 have been read to end-of-file.
412 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
414 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
415 description of determination of which consecutive characters
417 ** Printing the "Space" character with escaping on now yields "#\\ ",
418 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
419 ** Reading floating-point numbers with *READ-BASE* set to a number
420 less than 10 works correctly.
421 ** Reading floating-point numbers with *READ-BASE* set to a number
422 more than 10 works correctly.
423 ** Printing with *PRINT-READABLY* targets the standard readtable, not
424 the readtable currently in effect.
426 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
427 * minor incompatible change: parsing of namestrings on a physical
428 (Unix) host has changed; numbers after the final #\. in a
429 namestring are no longer interpreted as a version field. This is
430 intented to be largely invisible to the user, except that the
431 meaning of the namestring "*.*.*" has changed: it now refers to a
432 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
433 should usually be replaced by
434 (make-pathname :name :wild :type :wild :version :wild)
435 with the added benefit that this is more likely to be portable.
436 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
437 option to OPEN now signals an error if the file being opened
438 exists; this may have an impact on existing code.
439 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
440 signals received, on the PowerPC platforms (both Linux and
441 Darwin). (thanks to Pierre Mai for pointing out the location of
443 * several fixes on OS X: The system now builds and runs cleanly on
444 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
445 (thanks to Brian Mastenbrook)
446 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
447 second-resolution: integer multiples of 1/3600 between -24 and 24.
448 (thanks to Vincent Arkesteijn)
449 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
450 type when called with 1 argument; PEEK-CHAR checked type of
451 PEEK-TYPE only after having read first character from a
452 stream. (reported by Peter Graves)
453 * bug fix: the garbage collector now has much better locality
454 behaviour, and in particular no longer treats all memory as being
455 exhausted when many small objects point to each other in a deeply
457 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
458 recognized as being TYPEP their class.
459 * bug fix: the PUSHNEW documentation string has been corrected.
460 (thanks to Vincent Arkesteijn)
461 * bug fix: defaulting of the value for the last of an atypically
462 large number of multiple values being bound was not being
463 performed correctly on the Alpha or PPC platforms
464 * optimization: implemented multiplication as a modular
465 (UNSIGNED-BYTE 32) operation on the x86 backend.
466 * optimization: SEARCH on simple-base-strings can now be open-coded.
467 (see also contrib/compiler-extras.lisp for inspiration for
468 teaching the compiler about the Boyer-Moore algorithm).
469 * value, returned by MAX (and MIN) called with several EQUALP, but
470 not EQL, arguments now does not depend on compiler settings.
471 * fixed some bugs revealed by Paul Dietz' test suite:
472 ** in stack analysis liveness information is propagated from
473 non-local entry points.
474 ** pathwise CAST removing failed when the CAST node did not start
476 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
477 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
479 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
480 host is already defined.
481 ** RENAME-FILE works on streams instead of signalling an internal
483 ** PEEK-CHAR uses the current readtable when determining whether
484 or not a character is whitespace.
485 ** MERGE-PATHNAMES handles the case when the pathname does not
486 specify a name while the default-pathname specifies a version.
487 ** Pathnames now stand a better chance of respecting print/read
489 ** Attempting to use standardized file system operators with a
490 pathname with invalid :DIRECTORY components signals a
492 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
494 ** DIRECTORY on logical pathnames is more correct.
495 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
496 signal a TYPE-ERROR if their argument is not a stream.
497 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
498 designator argument does not designate a stream.
499 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
500 examining the synonym.
501 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
503 ** OPEN and WITH-OPEN-STREAM allow opening streams with
504 element-type larger than ([UN]SIGNED-BYTE 32).
506 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
507 * When built with the :SB-FUTEX feature, threaded builds now take
508 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
509 for faster/more reliable mutex and condition variable support.
510 * Incompatible change (but one you probably shouldn't have been using
511 anyway): the interface and code for arbitrating between multiple
512 threads in the same user session has been redesigned.
513 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
514 error when called without an explicit environment argument.
515 (thanks to Nikodemus Siivola)
516 * bug fix: buffered :DIRECTION :IO streams are less likely to become
517 confused about their position. (thanks to Adam Warner and Gerd
519 * bug fix: Pretty printing backquoted forms with unquotations in the
520 argument list position of various code constructs such as LAMBDA
521 now works correctly. (reported by Paul Dietz)
522 * bug fix: Pretty printing unquotations no longer loses all
523 stream position information.
524 * optimization: performance of string output streams is now less
525 poor for multiple small sequence writes.
526 * optimization: performance of CSUBTYPEP in the presence of complex
527 expressions involving CONS and NOT many times has been improved.
528 (reported by Paul Dietz)
529 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
531 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
532 streams for the REPL, for the debugger, and so on. (thanks to
534 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
535 * fixed some bugs revealed by Paul Dietz' test suite:
536 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
537 is ignored if it is too small, rather than propagating through
538 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
539 ** extremely complex negations of CONS types were not being
540 sufficiently canonicalized, leading to inconsistencies in
542 ** VALUES tranformer lost derived type.
544 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
545 * fixed a bootstrapping bug: the build process no longer assumes
546 that the various BOOLE-related constants have the same value in
547 host and target lisps. (noted by Paul Dietz' test suite on an
548 SBCL binary built from CLISP)
549 * The system can now be dynamically linked on the MIPS platform,
550 which enables dynamic loading of foreign code from Lisp. (thanks
551 to Ralf Baechle for discussions on the MIPS ABI)
552 * The system now records debugging information for its own source
553 files in a filesystem-position-independent manner, relative to
554 the "SYS" logical host.
555 * fixed a compiler bug: MV-LET convertion did not check references
556 to the "max args" entry point. (reported by Brian Downing)
557 * tweaked disassembly notes to be less confident about proclaiming
558 some instruction as an LRA. (thanks to Brian Downing)
559 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
560 now each have their own history, command character, and other
561 characteristics. (thanks to David Lichteblau)
562 * fixed some bugs revealed by Paul Dietz' test suite:
563 ** compiler failure in compiling LOGAND expressions including a
565 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
566 shift greater than 32.
567 ** FLUSH-DEST did not mark blocks for type check regeneration.
568 ** HANDLER-CASE failed to accept declarations in handler clauses
569 in some circumstances.
571 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
572 * New code in contrib/sb-introspect (still probably not entirely
573 stable yet) provides some support for smart Lisp development
574 environments like SLIME.
575 * The conditions signalled for errors occurring when loading .fasl
576 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
577 in a way which should help ASDF recover gracefully.
578 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
579 stuff has been cleaned up. If you code contrib/ stuff, this might
580 affect you, and you can look at contrib/README, contrib/STANDARDS,
581 and/or the 0.8.4.27 diff to check.
582 * In full calls the compiler now does not generate checks for declared
583 argument types for all arguments.
584 * various threading fixes
585 ** and some experimental patches which didn't make it into
586 the main tree for this release, but which are shipped in
587 contrib/experimental-thread.patch as a possible fix for some
588 failures (deadlock, spinning...) in GC-intensive multithreaded
590 * fixed PPC build problem (source code incompatibility of different
591 library versions): added offsetof() hackery which attempts to divine
592 where glibc maintainers put uc_mcontext today
593 * fixed bug 282: compiler does not trust type assertions while passing
594 arguments to a full call.
595 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
596 &OPTIONAL VALUES type parameter.
597 * fix bug 214: algorithm for noting rejected templates is now more
598 similar to that of template seletion. (also reported by rydis on
600 * fixed bug 141b: printing backquoted information readably and prettily
601 inserts a space where necessary.
602 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
603 (CHARACTER 10) are now reported as errors, rather than propagated
604 as unknown types. (reported by piso on #lisp)
605 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
606 with values NIL and :ERROR. (thanks to Milan Zamazal)
607 * fixed bug 191c: CLOS now does proper keyword argument checking as
608 described in CLHS 7.6.5 and 7.6.5.1.
609 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
610 longer raise an error; further, using a list as a for-as-arithmetic
611 counter now raises a meaningful error.
612 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
613 now understood by sequence creation functions such as MAKE-SEQUENCE
615 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
616 a STRING-INPUT-STREAM.
617 * compiler enhancement: SIGNUM is now better able to derive the type
619 * type declarations inside WITH-SLOTS are checked. (reported by
621 * fixed some bugs revealed by Paul Dietz' test suite:
622 ** incorrect optimization of TRUNCATE for a positive first
623 argument and negative second.
624 ** compiler failure in let-convertion during flushing dead code.
625 ** compiler failure while deriving type of TRUNCATE on an
626 interval, containing 0.
627 ** ASH of a negative bignum by a negative bignum count now returns
629 ** intersection of CONS types now canonicalizes properly, fixing
630 inconsistencies in SUBTYPEP.
632 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
633 * incompatible change: The --disable-debugger command line
634 option now clobbers the debugger at a more fundamental
635 level, by redefining #'INVOKE-DEBUGGER instead of by
636 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
637 is specified by ANSI to ignore *DEBUGGER-HOOK* and
638 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
639 enter the debugger REPL and then suffer recursive errors
640 because *DEBUG-IO* is also messed up in --disable-debugger mode;
641 while under the new system, BREAK in --disable-debugger mode
642 terminates the system just as an unhandled error would.
643 * fixed compiler performance when processing loops with a step >1;
644 * bug fix: DOCUMENTATION now retrieves generic function
645 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
646 support has been systematized, and now supports the methods
647 specified by ANSI, along with a default method and a method for
648 slot documentation. (reported by Nathan Froyd)
649 * bug fix: effective methods associated with a generic function are
650 no longer cached over a change of that generic function's method
651 combination. (reported by Andreas Fuchs)
652 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
653 names a non-existent directory. (thanks to Andreas Fuchs)
654 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
655 the PPC platform, lead to stack corruption; this has been fixed.
656 (reported by Rainer Joswig)
657 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
658 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
659 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
660 arithmetic in general, is now much more reliable on the PPC
662 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
663 platform now returns the right answer.
664 * optimization: restored some effective method precomputation in
665 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
666 precomputation is now tunable.
667 * optimization: compiler-internal data structure use has been
668 reviewed, and changes have been made that should improve the
669 performance of the compiler by about 20%.
670 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
671 simple-base-strings and simple-bit-vectors is improved.
672 * optimization: the optimization of 32-bit logical and arithmetic
673 functions introduced in version 0.8.3 on the x86 has been
674 implemented on the mips, ppc and sparc platforms; an
675 implementation of the same facility, but for 64-bit arithmetic,
676 has been added for the alpha.
677 * microoptimization: the compiler is better able to make use of the
678 x86 LEA instruction for multiplication by constants.
679 * bug fix: in some situations compiler did not report usage of
680 generic arithmetic in (SPEED 3) policy.
681 * bug 145b fix: compiler used wrong type specifier while converting
682 MEMBER-types to numeric.
683 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
685 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
686 index argument now works properly on the Alpha platform.
687 * bug fix: floating point exception treatment on the Alpha platform
689 * bug fix: FILE-POSITION works much better on string input and
690 output streams. (thanks to Nikodemus Siivola)
691 * bug fix: many threading/garbage collection symptoms sorted.
692 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
693 might be pseudo-atomic.
694 * internal change: Stopping for GC is now done with signals not
695 ptrace. GC is now done in whichever thread wanted it, instead of
697 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
699 * bug fix: result form in DO is not contained in the implicit
701 * incompatible change: ICR structure is changed; the value part of
702 CONTINUATION is now called LVAR; corresponding functions are
703 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
704 * added type deriver for ISQRT (thanks to Robert E. Brown).
705 * bug fix: better support for loading from the command line when an
706 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
708 * fixed some bugs revealed by Paul Dietz' test suite:
709 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
710 ** ROUND and FROUND now give the right answer when given very
711 small float arguments.
712 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
714 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
715 ** (GCD 0 <negative-integer>) returned <negative-integer>.
716 ** LCM should return a non-negative integer.
717 ** PARSE-INTEGER returned the index of a terminator instead of the
718 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
719 ** PARSE-INTEGER returned an incorrect index being applied to a
721 ** LCM with two arguments of 0 returns 0 rather than signalling
723 ** unsigned addition of a 32-bit constant with the high bit set no
724 longer causes an internal compiler error on the x86.
725 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
726 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
727 with negative last argument.
728 ** byte specifiers with zero size and position no longer cause
729 an error during type derivation.
730 ** bignum multiplication on the Alpha platform now returns the
732 * porting: The system now builds on SuSE AMD64, although it still
733 generates a 32-bit binary.
734 * .fasl file incompatibility: The fasl file version number has
735 been incremented (because of the changes to internal compiler
736 data structures referred to above).
738 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
739 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
740 more accurately, on the Darwin kernel running on PowerPC hardware.
741 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
742 * Compiler code deletion notes now signal a condition of type
743 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
744 an associated MUFFLE-WARNING restart.
745 * The compiler now performs limited argument count validation of
746 constant format strings in FORMAT, and where appropriate in ERROR,
747 CERROR and WARN. (thanks to Gerd Moellmann)
748 * New ASDF-INSTALL contrib can be used for automatic download and
749 installation of third-party Lisp code from CCLAN or other sites
751 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
752 which forces another thread to execute a function supplied by the
754 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
755 now accept and act upon their :ELEMENT-TYPE keyword argument.
756 (reported by Martin Atzmueller, Edi Weitz)
757 * bug fix: FILE-POSITION now accepts position designators up to
758 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
759 is the greater. (thanks to Patrik Nordebo)
760 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
761 Kalle Olavi Niemitalo)
762 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
763 printable. (reported by Eric Marsden)
764 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
765 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
766 * x86 bug fix in control stack exhaustion checking: now shows backtrace
767 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
768 (thanks to Stig Sandoe)
769 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
770 UNREAD-CHAR now correctly includes the unread character in the
771 target sequence. (thanks to Gerd Moellmann)
772 * bug fix in threaded builds: the system can now be suspended and
773 resumed by shell job control with minimal disruption.
774 * bug fixes in times and timezones >2038AD
775 * better handling of "where is GNU make?" problem in build scripts
776 (thanks to Nikodemus Siivola)
777 * new optimization: inside a named function any reference to a
778 function with the same name is considered to be a self-reference;
779 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
780 optimization quality.
781 * new optimization on x86: logical functions and + now have
782 optimized (UNSIGNED-BYTE 32) versions, which are automatically
783 used when the result is truncated to 32 bits.
784 * VALUES declaration is partially enabled.
785 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
786 * fixed some bugs revealed by Paul Dietz' test suite:
787 ** The system now obeys the constraint imposed by
788 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
789 types form a lattice under type intersection.
790 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
791 ** ASSOC now ignores NIL elements in an alist.
792 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
793 and (1+ MOST-POSITIVE-FIXNUM) answers.
794 ** The addition of a method with invalid qualifiers to a generic
795 function does not cause an error to be signalled immediately;
796 a warning is signalled, and the error is generated only on
797 calling the generic function.
798 * changed .fasl file version number, in part to add type codes for
799 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
800 obscure ANSI requirements
802 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
803 * fixed bug 148: failure to inline-expand a local function left
804 garbage, confusing the compiler.
805 * fixed bugs 3cd: structure slot readers perform type check if the
806 slot can have an invalid value (i.e. it is either not initialized
807 or can be written with a less specific slot writer).
808 * bug fix: the compiler now traps array references to elements off
809 the end of an array; previously, the bounds checking in some
810 circumstances could go off-by-one.
811 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
813 * type declarations for array element types now obey the description
814 on the CLHS page "Declaration TYPE", as per discussions on
815 sbcl-help around 2003-05-08. This means that a declaration
816 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
817 declaration, all references to BAR will be asserted or assumed
818 (with THE, so dependent on compiler policy) to involve objects of
819 type FOO. Note that no such declaration is implied in
820 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
821 * declared types of functions from the "Conditions"
822 chapter. (reported by Paul Dietz)
823 * bug fix: CERROR accepts a function as its first argument.
824 * bug fix: NTH an NTHCDR accept a bignum as index
825 arguments. (reported by Adam Warner)
826 * optimization: character compare routines now optimize comparing
827 against a constant character. (reported by Gilbert Baumann)
828 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
829 argument now works properly on the MIPS platform.
830 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
831 * repeated evaluation of the same DEFSTRUCT, a slot of which is
832 declared to have a functional type, does not cause an error
834 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
835 many :DEFAULT-INITARGS. (reported by Istvan Marko)
836 * fixed bug: if last continuation of a deleted block has a
837 destination, this destination should be deleted too. (reported by
839 * fixed a bug in the bootstrap process: the host compiler's values
840 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
841 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
842 test case from Patrik Nordebo)
843 * improved the ability of the disassembler on the PPC platform to
844 provide helpful disassembly notes.
845 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
846 the class in more cases than previously.
847 * bug fix: FILE-POSITION now understands :START and :END for
848 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
849 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
850 8) data. (thanks to David Lichteblau)
851 * bug fix: it is possible to add a method to a generic function
853 * bug fix: reader failed to signal END-OF-FILE inside an
854 object representation. (reported by Nikodemus Siivola)
855 * fixed some bugs revealed by Paul Dietz' test suite:
856 ** LAST and [N]BUTLAST should accept a bignum.
857 ** condition slot accessors are methods.
858 ** (VECTOR NIL) is a subtype of STRING.
860 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
861 * minor incompatible change: some nonsensical specialized lambda
862 lists (used in DEFMETHOD) which were previously ignored now signal
864 * minor incompatible change: the system is now aware of the types of
865 variables in the COMMON-LISP package, and will signal errors for
866 most violations of these type constraints (where previously they
867 were silently accepted).
868 * minor incompatible change: COMPILE-FILE now uses the freedom
869 afforded (ANSI 3.2.2.3) to use derived function types for
870 functions defined in the same file. This also permits the system
871 to warn on static type mismatches and function
872 redefinition. (Currently it does not work with high DEBUG level.)
873 * minor incompatible change: VALUES declaration is disabled.
874 * When issuing notes, the compiler now signals a condition of type
875 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
876 restart for use in user handlers. It is expected that the
877 COMPILER-NOTE condition will eventually become a condition
878 supertype to a hierarchy of note types, which will then be
879 handleable in a similar fashion. However, at the moment, no such
880 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
881 but it's an implementation detail, not a classification for the
883 * Changes in type checking closed the following bugs:
884 ** type checking of unused values (192b, 194d, 203);
885 ** template selection based on unsafe type assertions (192c, 236);
886 ** type checking in branches (194bc).
887 * A short form of VALUES type specifier has ANSI meaning (it has
888 increased the number of situations when SBCL cannot perform type
890 * fixed bug in DEFSTRUCT: once again, naming structure slots with
891 keywords or constants is permissible.
892 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
893 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
894 classes. (thanks to Antonio Martinez)
895 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
896 outputs FRESH-LINE or TERPRI, and no longer converts its stream
897 argument to a pretty-print stream. Instead, it leaves any such
898 operations to DESCRIBE-OBJECT methods.
899 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
901 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
902 lambda lists. (thanks to David Lichteblau)
903 * bug fix: Defining a generic function with a :METHOD-CLASS being a
904 subclass of STANDARD-METHOD no longer causes stack exhaustion.
905 (thanks to Gerd Moellmann)
906 * fixed bug 246: increased compilation speed of long
907 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
909 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
911 * DEFCONSTANT now throws a condition of type
912 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
913 constant to a non-EQL value; CONTINUE and ABORT restarts
914 respectively change and preserve the value.
915 * fixed bug 63: The code walker, part of the implementation of CLOS,
916 is now better at handling symbol macros.
917 * bug fix: There is no longer an internal implementation type named
918 CL:LENGTH. (reported by Raymond Toy)
919 * bug fix: In macro-like defining macros/special operators the
920 implicit block does not enclose the lambda list.
921 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
922 enclosed in parentheses) are not suitable as type specifiers, and
923 their use properly signals an error now.
924 * bug fix: An argument count mismatch for a type specifier in code
925 being compiled no longer causes an unhandled error at compile
926 time, but signals a compile-time warning.
927 * fixed simple vector readable printing
928 * bug fix: DESCRIBE takes more care over whether the class
929 precedence list slot of a class is bound before accessing it.
930 (reported by Markus Krummenacker)
931 * bug fix: FORMATTER can successfully compile pretty-printer format
932 strings which use variants of the ~* directive inside.
933 * bug fix: SEARCH now applies its TEST predicate to the elements of
934 the arguments in the correct order. (thanks to Wolfhard Buss)
935 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
936 in type inference. (thanks to Robert E. Brown)
937 * bug fix: Reading in symbols with an explicit package name of ""
938 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
939 * fixed some bugs revealed by Paul Dietz' test suite:
940 ** NIL is now allowed as a structure slot name.
941 ** Arbitrary numbers, not just REALs, are allowed in certain
942 circumstances in LOOP for-as-arithmetic clauses.
943 ** Multiple class redefinitions before slot access no longer
945 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
946 association between the name and a class.
947 ** Generic functions with non-standard method-combination and over
948 five methods all of which return constants no longer return NIL
949 after the first few invocations. (thanks to Gerd Moellmann)
950 ** CALL-NEXT-METHOD with no arguments now passes the original
951 values of the arguments, even in the presence of assignment.
952 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
954 ** DISASSEMBLE works with closures and funcallable instances.
955 ** ADD-METHOD now returns the generic function, not the new method.
956 ** FIND-METHOD signals an error if the lengths of the specializers
957 is incompatible with the generic function, even if the ERRORP
959 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
960 which its argument is a member.
961 ** DEFCLASS only redefines the class named by its class-name
962 argument if that name is the proper name of the class;
963 otherwise, it creates a new class.
964 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
965 of the UNBOUND-SLOT condition to the name of the slot.
966 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
967 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
968 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
969 SLOT-MAKUNBOUND in the specified fashion.
971 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
972 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
973 cross-compilation host. As a consequence, we can now bootstrap our
974 way up to SBCL starting with a bare gcc toolchain and human-readable
975 source code (first the source to CLISP, then the source to SBCL).
976 * A contributed module containing a partial implementation of the
977 simple-streams interface has been included. (thanks to Rudi
979 * A contributed module implementing the RFC1321 Message Digest
980 Algorithm, known as MD5, has been included.
981 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
982 no longer has any effect, as the code controlled by this feature
983 has been deleted. (As far as we know, no-one has ever built using
984 this feature, and its semantics were confused in any case).
985 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
986 work on conditions (as required by the ANSI specification),
987 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
988 expected behaviour on conditions. Users should note, however,
989 that such behaviour is not required by the ANSI specification,
990 and so use of this behaviour may render their code unportable.
991 * fixed some bugs revealed by Paul Dietz' test suite:
992 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
994 ** &ENVIRONMENT parameter in macro lambda list is bound first.
995 ** SXHASH on condition objects no longer returns NIL.
996 ** :ALLOCATION :CLASS slots are better treated; their values are
997 updated on class redefinition, and initforms inherited from
998 superclasses are applied.
999 ** REMOVE-METHOD returns its generic function argument even when
1000 no method was removed.
1001 ** SHARED-INITIALIZE now initializes the values of the requested
1002 slots, including those with :ALLOCATION :CLASS.
1003 ** ALLOCATE-INSTANCE now works on structure classes defined via
1004 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1006 ** SLOT-EXISTS-P now works on conditions, as well as structures
1008 ** MAKE-LOAD-FORM now has the required methods on
1009 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1010 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1011 keyword, and now implements the SLOT-NAMES argument.
1012 ** methods with &OPTIONAL arguments no longer allow too many
1013 arguments to be passed in the call without error.
1014 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1015 option is consistent with the required arguments of the generic
1016 function lambda list.
1017 * bug fix: REQUIRE accepts a string designator. (Thanks to
1019 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1020 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1021 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1022 * bug fix: The compiler now checks for duplicated variables in macro
1024 * bug fix: SETQ on globals returns the correct value.
1025 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1026 not a COMPILER-ERROR (followed by some other strange error on
1027 choosing the CONTINUE restart).
1028 * bug fix: make.sh and friends are now more consistent in the way that
1029 they look for GNU "make".
1031 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1032 * experimental native threads support (on x86 Linux >=2.4 only).
1033 This is not compiled in by default: you need to add :SB-THREAD to
1034 the target features. See the "Beyond ANSI" chapter of the manual
1036 * fix for longstanding nonANSIism: The old distinction between
1037 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1038 The return value from CL:FIND-CLASS is now a CLOS class, and
1039 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1040 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1041 * An interface to the MetaObject Protocol, as described in Kiczales,
1042 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1043 MIT Press, 1991, is available from the SB-MOP package.
1044 * incompatible change: the SB-PCL package should now be considered
1045 a private implementation detail, and no longer a semi-private MOP
1047 * minor incompatible change: due to rearrangement for threads, the
1048 control stack and binding stack are now allocated at arbitrary
1049 addresses instead of being hardcoded per-port. Users affected by
1050 this probably have to be doing advanced things with shared
1051 libraries, and will know who they are.
1052 * minor incompatible change: Previously, all --eval forms used were
1053 processed with READ before any of them were processed with EVAL.
1054 Now each --eval form is processed with both READ and EVAL before
1055 the next --eval form is processed. (Thus package operations like
1056 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1057 work as the user might reasonably expect.)
1058 * minor incompatible change: *STANDARD-INPUT* is now only an
1059 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1061 * minor incompatible change: Y-OR-N-P is now character-oriented,
1062 not line oriented. Also, YES-OR-NO-P now works without errors.
1063 (thanks to Antonio Martinez)
1064 * sb-aclrepl module improvements: an integrated inspector, added
1065 repl features, and a bug fix to :trace command.
1066 * Known functions, which cannot be open coded by the backend, are
1067 considered to be able to check types of their arguments. (fixing
1068 a bug report by Nathan J. Froyd)
1069 * fixed a bug in computing method discriminating functions: It is
1070 now possible to define methods specialized on classes which have
1071 forward-referenced superclasses. (thanks to Gerd Moellmann)
1072 * fixed evaluation order in optional entries (reported by Gilbert
1074 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1075 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1076 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1077 required-by-AMOP NAME argument, as well as CLASS and
1078 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1079 * fixed bug 20: DEFMETHOD can define methods using names that are
1080 not the proper names of classes to designate class specializers.
1081 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1082 Unix file descriptors, instead of blowing up. (thanks to Antonio
1084 * Garbage collection refactoring: user-visible change is that a
1085 call to the GC function during WITHOUT-GCING will not do garbage
1086 collection until the end of the WITHOUT-GCING. If you were doing
1087 this you were probably losing anyway.
1088 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1089 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1090 TYPEP the latter but not the former.
1091 * The compiler issues a full WARNING for calls to undefined functions
1092 with names from the CL package.
1093 * MAP-INTO for a vector destination is open coded. (reported by
1094 Brian Downing on c.l.l)
1095 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1096 documentation string.
1097 * fixed some bugs revealed by Paul Dietz' test suite:
1098 ** COPY-ALIST now signals an error if its argument is a dotted
1100 ** Condition slots are now accessed more correctly in the presence
1101 of multiple initargs for a given slot.
1102 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1103 exclude restarts of the same name associated with a different
1105 ** DEFCLASS of forward-referenced classes with another
1106 forward-referenced class in the superclasses list no longer
1108 ** Condition slots are now initialized once each, not multiple
1109 times. (thanks to Gerd Moellmann)
1110 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1112 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1113 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1114 arguments contain duplicated elements.
1115 ** RESTART-CASE understands local macros.
1116 ** RESTART-CASE associates exactly its own restarts with a condition.
1117 ** ENDP in safe mode checks its argument to be of type LIST.
1118 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1119 in question is unbound.
1120 ** Optimized MAKE-INSTANCE functions no longer cause internal
1121 assertion failures in the presence of duplicate initargs.
1122 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1123 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1124 propagated even in the ctor optimized implementation of
1126 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1128 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1129 * a better implementation of SXHASH on (simple) bit vectors,
1130 measured both in execution speed and in distribution of results
1131 over the positive fixnums, has been installed. Likewise, a better
1132 implementation of EQUAL for simple bit vectors is now available.
1133 * fixed CEILING optimization for a divisor of form 2^k.
1134 * fixed bug 240 (emitting extra style warnings "using the lexical
1135 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1136 by Antonio Martinez)
1137 * fixed SXHASH, giving different results for NIL depending on type
1138 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1139 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1140 variable. (found by Rolf Wester)
1141 * SBCL does not ignore type declarations for special
1142 variables. (reported by rif on c.l.l 2003-03-05)
1143 * some bug fixes in contrib/sb-aclrepl/
1144 * fixed some bugs revealed by Paul Dietz' test suite:
1145 ** a bug in the CONS type specifier, whereby the CAR and CDR
1146 types got intertwined, has been fixed;
1147 ** the type system is now able to reason about the interaction
1148 between INTEGER and RATIO types more completely;
1149 ** APPEND, [N]REVERSE and NRECONC check that those their
1150 arguments, which must be proper lists, are really so;
1151 ** An array specialized to be unable to hold elements has been
1152 implemented, as required -- yes, really -- by ANSI;
1153 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1154 on malformed property lists;
1156 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1157 * incompatible packaging change: in line with Unix convention,
1158 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1159 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1160 * REQUIRE and PROVIDE are now optionally capable of doing something
1161 useful. See the documentation string for REQUIRE.
1162 * infrastructure for a managed SBCL contrib system: contributed
1163 modules in this release include:
1164 ** the ASDF system definition facility;
1165 ** an interface to the BSD Sockets API;
1166 ** an ACL-like convenience interface to the repl;
1167 (thanks to Kevin Rosenberg)
1168 ** an implementation of ROTATE-BYTE, with efficient implementation
1170 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1171 gives a closer approximation to the right answer than previously.
1172 (thanks to Raymond Toy)
1173 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1174 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1175 optional environment arguments, as required by ANSI.
1176 * fixed bugs in other functions taking environment objects, allowing
1177 calls with an explicit NIL environment argument to be compiled
1179 * fixed bug 228: primary return values from
1180 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1181 COMPILE or FUNCTION.
1182 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1183 :TYPE will no longer signal errors on innocuous objects.
1184 * fixed bug 231b: SETQ is better at respecting type declarations in
1185 the lexical environment.
1186 * fixed a bug in DEFCLASS: classes named by symbols with no or
1187 unprintable packages can now be defined.
1188 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1189 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1190 * fixed compiler failure related to checking types of functions.
1191 (reported by Robert E. Brown)
1192 * the compiler is now much more consistent in its error-checking
1193 treatment of bounding index arguments to sequence functions: in
1194 (SAFETY 3) code, errors will be signalled in almost all cases if
1195 invalid sequence bounding indices are passed to functions defined
1196 by ANSI to operate on sequences.
1197 * fixed a bug in the build procedure: documentation of SBCL-specific
1198 packages is now preserved and available in the final Lisp image.
1199 * lifted FDEFINITION lookup out of loops in the implementation of
1200 many list operations. (thanks to Robert E. Brown)
1201 * fixed a bug in the reader: the #n# reader macro now works for
1202 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1203 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1204 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1205 E. Brown and Tony Martinez respectively)
1206 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1207 SLOT-BOUNDP: in particular, it is now able to optimize them much
1208 better, and is now not vulnerable to having packages renamed.
1209 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1210 always called when a slot is not present in an instance. (thanks
1212 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1213 longer signal an error on unreading a character following EOF on
1214 the previous constituent stream. (thanks to Tony Martinez)
1215 * fixed some bugs revealed by Paul Dietz' test suite:
1216 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1217 not just nonnegative fixnums;
1218 ** the logical bit-array operators such as BIT-AND now accept an
1219 explicit NIL for their "opt-arg" argument (to indicate a
1220 freshly-consed result bit-array);
1221 ** ELT now signals an error on an invalid sequence index in safe
1223 ** the type system is now cleverer about negations of numeric
1224 types, and consequently understands the BIGNUM and RATIO types
1226 ** the type system is now cleverer about the interaction between
1227 INTEGER and RATIO types: while bugs still remain, many more
1228 cases are accurately computed;
1229 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1230 if it is in the last clause;
1231 ** CONSTANTLY now correctly returns a side-effect-free function in
1233 ** DECLARE is no longer treated as a special-operator; in
1234 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1235 * incremented fasl file version number due to the change in the
1236 DEFSTRUCT-SLOT-DESCRIPTION structure.
1238 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1239 * minor incompatible change: code processed by the "interpreter" or
1240 EVAL now has a compilation optimization policy of (DEBUG 2)
1241 (changed from (DEBUG 1)) to improve debuggability of interactive
1242 development, and to allow the use of the debug RETURN command in
1244 * an experimental implementation of the RETURN command for the
1245 debugger has been included. (thanks to Frederik Kuivinen)
1246 * fixed bug 62: constraints were not propagated into a loop.
1247 * fixed bug in embedded calls of SORT (reported and investigated by
1249 * fixed some bugs revealed by Paul Dietz' test suite:
1250 ** printing and reading of arrays with some dimensions having
1251 length 0 (thanks to Gerd Moellmann);
1252 ** BOA constructor with &AUX argument without a default value does
1253 not cause a type error;
1254 ** CONSTANTP now returns true for all self-evaluating objects.
1256 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1257 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1258 accessors that are related by inheritance, as specified in the
1259 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1260 Valtteri Vuorikoski)
1261 * The compiler is now able to inline functions that were defined in
1262 a complex lexical environment (e.g. inside a MACROLET).
1263 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1265 * Support for the upcoming FreeBSD-5.0 release has been included.
1266 (thanks to Dag-Erling Smorgrav)
1267 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1268 effect when it is not in a toplevel context.
1269 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1270 stand a better chance of being correct. (thanks to Gerd
1272 * fixed bug in COERCE, which now signals an error on coercing a
1273 rational to a bounded real type which excludes the expected
1275 * The compiler is now able to derive types more accurately from the
1276 COERCE and COMPILE functions.
1277 * fixed bug 223: functional binding is considered to be constant
1278 only for symbols in the CL package.
1279 * fixed bug 231: SETQ did not check the type of a variable being set
1280 (reported by Robert E. Brown)
1281 * A new optimization for MAKE-INSTANCE has been included, fixing
1282 various bugs (including relating to :ALLOCATION :CLASS slots and
1283 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1285 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1286 ** As required by ANSI, LOOP now disallows anonymous collection
1287 clauses such as COLLECT I in conjunction with aggregate boolean
1288 clauses such as THEREIS (= I 1);
1289 ** LOOP now signals an error when any variable is reused in the
1290 same loop (including the potentially useful construct analogous
1291 to WITH A = 1 WITH A = (1+ A);
1292 ** IT is only a special loop symbol within the first clause of a
1293 conditional loop clause;
1294 ** LOOP with a typed iteration variable over a hashtable now
1295 signals a type error iff it should.
1296 * fixed some other bugs revealed by Paul Dietz' test suite:
1297 ** FILE-STREAM now names the class previously known as FD-STREAM;
1298 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1299 argument) no longer signals an error;
1300 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1301 of the slot symbol, rather than using the current package
1302 ((:CONC-NAME "") continues to intern the slot's name in the
1304 * incremented fasl file version number, because of the incompatible
1305 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1306 of the new implementation of DEFINE-COMPILER-MACRO.
1308 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1309 * Support for building SBCL for MIPS platforms running in
1310 little-endian mode has now been checked in, and basic
1311 functionality on said platforms verified.
1312 * minor incompatible change: PCL now records the pathname of a file
1313 in which methods and the like are defined, rather than its
1315 * minor incompatible change: TRUENAME now considers the truename of
1316 a file naming a directory to be the pathname with :DIRECTORY
1317 component indicating that directory.
1318 * minor incompatible change: a NAMED clause in the extended form of
1319 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1320 reason for the previous behaviour is unclear.
1321 * more systematization and improvement of CLOS and MOP conformance
1322 in PCL (thanks to Gerd Moellman and Pierre Mai):
1323 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1325 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1326 primary methods with no specializers;
1327 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1329 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1330 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1331 FORWARD-REFERENCED-CLASSes; error reporting on
1332 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1334 ** SXHASH on CLOS instances now uses a slot internal to the
1335 instance to return different numbers on distinct instances,
1336 while preserving the same return value through invocations of
1338 ** DEFMETHOD signals errors when methods with longer incongruent
1339 lambda lists are added to generic functions;
1340 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1341 CLASS, as specified in AMOP;
1342 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1343 on the order returned by the primary method for classes of
1344 class STANDARD-CLASS;
1345 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1346 * fixed some bugs shown by Paul Dietz' test suite:
1347 ** DOLIST puts its body in TAGBODY;
1348 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1350 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1351 value producing form;
1352 ** if more variables are given to PROGV than values, extra
1353 variables are bound and made to have no value;
1354 ** NSUBSTITUTE on list arguments gets the right answer with
1356 ** ELT signals an error of type TYPE-ERROR when the index argument
1357 is not a valid sequence index;
1358 ** LOOP signals (at macroexpansion time) an error of type
1359 PROGRAM-ERROR when duplicate variable names are found;
1360 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1361 ** FUNCALL of special-operators now cause an error of type
1363 ** PSETQ now works as required in the presence of side-effecting
1364 symbol-macro places;
1365 ** NCONC accepts any object as its last argument;
1366 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1368 ** loop-for-as-package does not require a package to be explicitely
1370 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1372 * fixed bug 166: compiler preserves "there is a way to go"
1373 invariant when deleting code.
1374 * fixed bug 172: macro lambda lists with required arguments after
1375 &REST arguments now cause an error to be signalled. (thanks to
1377 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1378 bug. (thanks to Matthew Danish)
1379 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1381 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1382 arguments are lexically rebound. (thanks to Gerd Moellmann and
1384 * fixed bug 194: error messages are now more informative when there
1385 is no primary method applicable in a call to a generic
1386 function. (thanks to Gerd Moellmann)
1387 * fixed bug in command line argument checking (thanks to Julian
1389 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1390 (thanks to Matthew Danish)
1391 * incremented fasl file version number, because of the
1392 SXHASH-related changes in the layout of CLOS data structures
1394 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1395 * minor incompatible change: The runtime (the Unix executable named
1396 "sbcl") is now much pickier about the .core files it will load.
1397 Essentially it now requires .core files to descend from the same
1398 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1399 as the runtime does. (The intent is to prevent the crashes which
1400 can occur, and which can even be reported as mysterious failures,
1401 when people patch the sources or change the build parameters
1402 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1403 sbcl and .core files.)
1404 * fixed bug: VALUES-LIST is no longer optimized away.
1405 * fixed bug 142: The FFI conversion of C string values to Lisp
1406 string values no longer conses excessively. (thanks to Nathan
1407 Froyd porting Raymond Toy's fix to CMU CL)
1408 * began to systematize and improve MOP conformance in PCL (thanks to
1409 Nathan Froyd, Gerd Moellman and Pierre Mai):
1410 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1412 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1413 ** FINALIZE-INHERITANCE is now called on class finalization;
1414 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1415 argument precedence order.
1416 * fixed bug 202: The compiler no longer fails on functions whose
1417 derived types contradict their declared type.
1418 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1419 so it can be non-toplevel.
1420 * The fasl file version number has changed (because of the new
1421 implementation of DEFMACRO).
1422 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1423 safe code, that any length requirement by their type-specifier
1424 argument is valid. The exceptions to this are described in bug
1426 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1427 functions now check the types of their inputs as required by ANSI.
1428 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1429 introduced symbol is DECLAREd to be SPECIAL.
1430 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1431 * fixed bug: :COUNT argument to sequence functions may be negative
1432 * fixed bug: body of DO-SYMBOLS may contain declarations
1433 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1434 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1435 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1436 symbol macro only once
1437 * fixed printing of call frame when argument list is unavailable
1438 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1439 * compiler no longer signals WARNING on unknown keyword
1442 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1443 * A beta-quality port to the mips architecture running Linux,
1444 based on the old CMUCL backend, has been made. It has been tested
1445 on a big-endian kernel, and works sufficiently well to be able to
1446 rebuild itself; it has not been tested in little-endian mode.
1447 * fixed an inconsistency between gencgc.c and purify.c which made
1448 dumping/loading .core files unreliable
1449 * fixed bug 120a: The compiler now deals correctly with IFs where
1450 the consequent is the same as the alternative, instead of
1451 misderiving the return type. (thanks to Alexey Dejneka)
1452 * fixed bug 113: Logical pathnames are now dumpable (the logical
1453 host is resolved at load-time, throwing an error if it is not
1455 * fixed bug 174: FORMAT's error message is slightly clearer when a
1456 non-printing character is used in a format directive.
1457 * fixed several bugs in compiler checking of type declarations, i.e.
1458 violations of the Python "declarations are assertions" principle
1459 (thanks to Alexey Dejneka)
1460 * fixed several bugs in PCL's error checking (thanks to Gerd
1462 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1464 * fixed bug in compilation of functions as first class values
1465 (thanks to Antonio Martinez-Shotton)
1466 * The compiler's handling TYPE-ERRORs which it can prove will
1467 inevitably happen at runtime has been cleaned up and corrected
1468 in several ways. (thanks to Alexey Dejneka)
1469 * improved argument type checking for various basic arithmetic
1470 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1471 lavished on them in the past that they can be compiled in many
1472 ways in different special cases
1473 * fixed bug 181: compiler checks validity of user supplied type
1475 * cleaned up code flushing in optimization: Function calls which
1476 should signal errors for safety purposes (e.g. which ANSI says
1477 should signal errors when their arguments are of incorrect type)
1478 are no longer optimized away.
1479 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1480 * incremented fasl file version number, because changes in the
1481 implementation of sequence functions like COERCE caused
1482 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1483 in old inline expansions) to become undefined. (Actually these
1484 changes were later undone, so we might very well be binary
1485 compatible with 0.7.7 after all, but leaving the version number
1486 incremented seemed like the simplest and most conservative
1489 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1490 * An alpha-quality port to the parisc architecture running Linux,
1491 based on the old CMUCL backend, has been made. This, even more so
1492 than the other backends, should be considered still a work in
1493 progress; known problems include that the Linux kernel in 64-bit
1494 mode does not propagate the correct sigcontext structure to
1495 userspace, and consequently SBCL on a parisc64 kernel will not
1497 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1498 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1499 suppressed.) Also now that the compiler is looking at declarations
1500 in the environment, it checks optimization declarations as well,
1501 and suppresses inlining when (> DEBUG SPEED).
1502 * More fixes have been made to treatment of floating point exception
1503 treatment and other Unix signals. In particular, floating point
1504 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1505 * The detection and handling of control stack exhaustion (infinite
1506 or very deeply nested recursion) has changed. Stack exhaustion
1507 detection is now done by write-protecting pages at the OS level
1508 and applies at all optimization settings; when found, a
1509 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1510 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1511 be caught using IGNORE-ERRORS.
1512 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1513 correctly when transforming recursive LABELS functions to LETs.
1514 (thanks to Alexey Dejneka)
1515 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1516 that are names of constants or global variables.
1517 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1518 alien routines with docstrings.
1519 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1520 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1522 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1523 (thanks to Eric Marsden)
1524 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1525 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1526 * Bug fix: The SPARC backend can now compile functions involving
1527 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1528 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1529 to a functions expecting a PATHNAME-DESIGNATOR.
1530 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1531 lambda lists. (thanks to Alexey Dejneka)
1532 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1533 functions so that EXPORTing the name of the class doesn't cause
1534 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1535 to Antonio Martinez for reporting this)
1536 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1537 constant-folded without failing an assertion. (thanks to Einar
1538 Floystad Dorum for reporting this)
1539 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1540 types are now dealt with more correctly by the compiler.
1541 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1542 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1543 behaviour analogous to RENAME-FILE. This puts its behaviour more
1544 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1546 * The fasl file version number has changed again. (because of the
1547 bug fix involving the names of PCL MAKE-INSTANCE functions)
1549 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1550 * bug fix: Floating point exceptions are treated much more
1551 consistently on the x86/Linux and PPC/Linux platforms.
1552 * Array initialization with :INITIAL-ELEMENT is now much faster for
1553 cases when the compiler cannot open code the array creation, but
1554 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1555 array accesses have also seen a speed increase.
1556 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1557 specified. (This was at the root of some bad interactions between
1558 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1560 * bug fix: Internal error arguments for undefined functions are now
1561 computed correctly on the PPC/Linux platform.
1562 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1563 Raymond Toy's patch for CMU CL)
1564 * Support for the Solaris 9 operating environment has been included
1565 (thanks to Daniel Merritt)
1566 * A very ugly but hopefully complete draft of the missing FFI chapter
1567 of the manual has been created by reformatting the corresponding
1568 CMU CL manual chapter into (currently very ugly and incoherent)
1569 DocBook and bringing it up to date for SBCL behavior. Thus, the
1570 manual is now essentially complete, at least by my extreme
1571 once-and-only-once standards, whereby it's acceptable to refer to
1572 the doc strings of SB-EXT functions as the primary documentation.
1573 * The fasl file version number has changed again, due to cleanup of
1574 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1575 is no longer a static symbol.)
1577 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1578 * SBCL now builds with OpenMCL (version 0.12) as the
1579 cross-compilation host; also, more progress has been made toward
1580 bootstrapping under CLISP.
1581 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1583 * bug 158 fixed: The compiler can now deal with integer loop
1584 increments different from 1; fixing this turned out also to fix
1586 * bug 169 fixed: no more bogus warnings about using lexical bindings
1587 despite the presence of perfectly good SPECIAL declarations (thanks
1588 to David Lichteblau)
1589 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1590 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1591 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1593 * bug 180 fixed: Method combination specifications no longer ignore
1594 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1595 * bug fix: Structure type predicate functions now check their argument
1596 count as they should.
1597 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1598 correctly. (thanks to Pierre Mai)
1599 * minor incompatible change: The --noprogrammer option is deprecated
1600 in favor of the new --disable-debugger option, which is very similar.
1601 (The major difference is that it takes effect at a slightly different
1602 time at startup, causing handling of errors in --sysinit and
1603 --userinit files will be affected differently.) The
1604 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1605 been added to allow this functionality to be controlled from ordinary
1606 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1607 maintainers, who might want to run non-interactive scripts to
1608 build SBCL cores which will later be used interactively.)
1609 * minor incompatible change: The LOAD function no longer, when given
1610 a wild pathname to load, loads all files matching that pathname.
1611 Instead, an error of type FILE-ERROR is signalled.
1613 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1614 * bug 147 fixed: The compiler preserves its block link/count
1615 invariants more correctly now so that it doesn't crash. (thanks
1617 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1619 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1620 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1621 no-value-for-this-character as NIL (as specified by ANSI).
1622 * HOST-NAMESTRING on physical pathnames now returns a string that is
1623 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1624 (thanks to Christophe Rhodes)
1625 * The Alpha port handles icache flushing more correctly. (thanks to
1627 * More progress has been made toward bootstrapping under CLISP. (thanks
1628 to Christophe Rhodes)
1629 * The fasl file format has changed again, because dynamic loading
1630 on OpenBSD (which has non-ELF object files) motivated some cleanups
1631 in the way that foreign symbols are transformed and passed around.
1632 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1633 is no longer treated as whitespace by the reader, but instead as
1634 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1635 |A
\7fB|, instead of A as it used to.
1637 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1638 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1639 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1640 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1641 this as of 0.7.1.45, but was left out of the previous news section
1642 (thanks to Dan Barlow)
1643 * SBCL now runs on the Solaris operating system on SPARC architectures
1644 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1645 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1646 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1647 * SPARC backend cleanups, allowing builds of cores optimized for V8
1648 and V9 SPARCS, and also emission of code targeted to a particular
1649 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1651 * SBCL is closer to bootstrapping under CLISP, thanks to various
1652 fixes by Christophe Rhodes.
1653 * The fasl file format has changed again, to allow the compiler's
1654 INFO database to support symbol macros.
1655 * The user manual (in doc/) is formatted into HTML more nicely.
1656 (thanks to coreythomas)
1657 * The system is smarter about SUBTYPEP relationships, especially
1658 those involving NOT types (including types such as ATOM which are
1659 represented internally using NOT types). Thus SUBTYPEP is less
1660 likely to return (VALUES NIL NIL) in general, and in particular
1661 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1663 * The fasl file format has changed again, because the internal
1664 representation of types now includes a new slot to support the new
1665 SUBTYPEP-of-NOT-types logic.
1666 * (not a change in the main branch of SBCL, but a related prototype
1667 which can hopefully be merged into the main branch of SBCL in the
1668 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1669 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1670 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1671 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1672 dispatch macro character combinations. (thanks to Alexey Dejneka)
1673 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1674 that superficially look like logical namestrings correctly.
1675 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1677 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1678 default supplied-p) for &optional and &key arguments. (thanks to
1681 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1682 * incompatible change: The compiler is now less aggressive about
1683 tail call optimization, doing it only when (> SPACE DEBUG) or
1684 (> SPEED DEBUG). (This is an incompatible change because there are
1685 programs which relied on the old CMU-CL-style behavior to optimize
1686 away their unbounded recursion which will now die of stack overflow.)
1687 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1688 for non-GENCGC systems has been increased to 20M (since that
1689 seems much closer to the likely performance optimum for modern
1690 systems than the old 4M value was)
1691 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1692 and *DEBUG-PRINT-LEVEL*
1693 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1694 of CMU CL's support for SPARC, and various endianness and other
1695 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1696 * new syntactic sugar for the Unix command line: --load foo.bar is now
1697 an alternate notation for --eval '(load "foo.bar")'.
1699 ** The system now detects stack overflow and handles it gracefully,
1700 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1701 optimization settings. (This is a good thing in general, and
1702 its introduction in this version should be particularly timely
1703 for anyone whose code fails because of suppression of tail
1705 ** The system now hunts for the C variable "environ" in a more
1706 devious way, to avoid segfaults when the C library version differs
1707 between compile time and run time. (thanks to Christophe Rhodes)
1708 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1709 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1710 ** The compiler no longer issues bogus style warnings for undefined
1711 classes in the same source file as the DEFCLASSes which defined
1712 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1714 ** fixes in CONDITION class precedence list for undefined function
1715 errors (thanks to Alexei Dejneka)
1716 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1717 correctly. (thanks to Dan Barlow)
1718 ** portability fixes aiming at bootstrapping under CLISP (thanks
1719 to Dave McDonald and Christophe Rhodes)
1720 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1721 ** fixes in type translation and and type inference (thanks to
1723 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1724 ** various fixes in the new ports (thanks to Dan Barlow)
1725 * several changes related to debugging:
1726 ** suppression of tail recursion, as noted above
1727 ** stack overflow detection, as noted above
1728 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1729 is now the default. (For some time encapsulation has been more
1730 reliable than the breakpoint-based :ENCAPSULATE NIL
1731 implementation, at least on X86 systems; and I just noticed that
1732 encapsulation also seems closer to the spirit of the ANSI
1735 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1737 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1738 up properly again. (There was a packaging bug in 0.7.0 which
1739 left their definitions in SB-SYS::LOAD-FOREIGN and
1740 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1741 vital for most things which interface to C-level interfaces,
1742 like extensions working with sockets or databases or
1743 Perl-compatible regexes or whatever, and the need to fix
1744 this bug is the main reason that 0.7.1 was released so
1746 ** DEFGENERIC is now choosier about the methods it redefines, so that
1747 reLOADing a previously-LOADed file containing DEFGENERICs does
1748 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1749 cycle now works as it should. (thanks to Alexey Dejneka)
1750 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1751 fixed by Christophe Rhodes. (He actually submitted this patch
1752 months ago, and I delayed until after 0.7.0.)
1753 ** Bug 111 (internal compiler confusion about runtime checks on
1754 FUNCTION types) was fixed by Alexey Dejneka.
1755 * Some internal cleanups (getting rid of variables which aren't
1756 needed now that the byte interpreter is gone) caused the fasl
1757 file format number to change again.
1759 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1760 * major incompatible change: The default fasl file extension, i.e. the
1761 default extension for files produced by COMPILE-FILE, has changed
1762 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1764 ** There are many changes in the implementation of the compiler.
1765 SBCL is now essentially a compiler-only implementation of ANSI
1766 Common Lisp. EVAL still "interprets" a few special cases, but
1767 almost all the interesting cases are handled by creating
1768 a LAMBDA expression, calling COMPILE on it, then calling
1769 FUNCALL on the result.
1770 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1771 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1772 Since the code is newer, there might still be some new bugs
1773 (though not as many as before Martin Atzmueller's fixes:-). But
1774 the new code is substantially simpler and clearer, and hopefully
1775 any remaining bugs will be simpler, less fundamental, and more
1776 fixable then the bugs in the old code.
1777 ** The revised compiler is still a little unsteady on its feet.
1779 *** The debugging information it produces (particularly the names
1780 of FUNCTION objects) is sometimes much less useful than what
1781 the old compiler produced.
1782 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1783 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1784 within a MACROLET) has been temporarily weakened.
1785 ** There are new compiler optimizations for various functions:
1786 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1787 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1788 *** the math functions TRUNCATE, FLOOR, and CEILING
1789 *** the function-of-all-trades COERCE
1790 Mostly these should be transparent, but there's one
1791 potentially-annoying problem (bug 117): when the compiler
1792 inline-expands a function and does type analysis on the result,
1793 it can create control paths which have type mismatches, and
1794 when it can't prove that those control paths aren't taken,
1795 it will issue WARNINGs about the type mismatches. This is
1796 a particular problem in practice for the new sequence functions.
1797 It's not clear how this should be fixed, and for now, a
1798 workaround is given in the entry for 117 in the BUGS file.
1799 ** (Because of the interaction between the two previous items --
1800 occasional inlining problems and new inline expansions -- some
1801 of the new sequence function optimizations won't really kick in
1802 completely until debugging information, and then inlining, are
1803 straightened out in some future version.)
1804 * minor incompatible changes:
1805 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1806 DIRECTORY no longer implicitly promotes NIL slots of its
1807 pathname argument to :WILD. In particular, when you ask for the
1808 contents of a directory (which you used to be able to do without
1809 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1810 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1811 ** changes in behavior that ANSI explicitly defines to be
1812 implementation dependent:
1813 *** The new compiler-only implementation still conforms with ANSI,
1814 but acts a little different than before. Besides the obvious
1815 changes in performance tradeoffs (that the cost per form passed
1816 to EVAL has gone up, and the cost per form executed by EVAL
1817 has gone down), the behavior of the system changes a little
1818 because there are no longer any interpreted function objects.
1819 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1820 e.g. doing COMPILE on the output of interactive DEFUN is
1822 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1824 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1825 has changed from (:CL) to NIL.
1826 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1827 e.g. #\Newline and #\Tab, don't have names specified in the
1828 ANSI Common Lisp standard, is now based on their ASCII symbolic
1829 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1830 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1831 are no longer used for output.
1832 ** changes in internal implementation constants:
1833 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1834 4 million. (If your application spends a lot of time GCing and
1835 you have a lot of RAM, you might want to experiment with
1836 increasing it even more.)
1837 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1838 However, almost all old code should still continue to work without
1839 immediate update, as SB-C-CALL is now a (deprecated) nickname
1841 ** Old operator names in the style DEF-FOO are now deprecated in
1842 favor of new corresponding names DEFINE-FOO, for consistency with
1843 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1844 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1845 internal symbols, but a few supported extensions like
1846 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1847 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1848 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1849 etc. as you get deeper into recursive calls to the debugger
1850 command loop, instead of the old "5]", "5]]", "5]]]"
1851 sequence. (I was motivated to do this when squabbles between
1852 ILISP and SBCL left me very deeply nested in the debugger. In the
1853 short term, this change will probably provoke more ILISP/SBCL
1854 squabbles, but hopefully it will be an improvement in the long run.)
1855 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1856 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1857 function, since declaiming return types involving aliens is
1858 (1) annoyingly messy to do by hand and (2) vital to efficient
1859 compilation of code which calls such functions.
1860 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1861 longer reexported by the SB-EXT package. They're solely useful
1862 for alien code, so it seems more logical that you should get
1863 them from the SB-ALIEN package, not in SB-EXT.
1864 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1865 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1866 features. Instead, the code that they used to control is always
1867 built into the system.
1868 * many other bug fixes
1869 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1870 advantage of the new EVAL-WHEN stuff and to clean them up in
1871 general, and they are now more ANSI-compliant in a number of
1872 ways. Martin Atzmueller is responsible for a lot of this.
1873 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1875 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1876 *** correct ERROR type for various file operations
1877 *** some fixes for Lisp streams
1878 *** DEFMETHOD syntax checking
1879 *** changing old weird representation of debug information as
1880 strings (which, among their other deficiencies, don't transform
1881 correctly when you rename packages, and don't change their
1882 print representation when you change things like *PACKAGE*
1883 and *PRINT-LENGTH*) to symbols and lists of symbols
1884 He also made several improvements and fixed several bugs in DESCRIBE.
1885 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1887 *** misbehavior of WRITE-STRING/WRITE-LINE
1888 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1889 and several other LOOP problems as well
1890 *** DIRECTORY when similar filenames are present
1891 *** DEFGENERIC with :METHOD options
1892 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1893 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1894 *** argument ordering in FIND with :TEST option
1895 *** mishandled package designator argument in APROPOS-LIST
1896 *** various problems in the backquote readmacro
1897 *** a bug in APROPOS
1898 *** probably some others that I'm not describing very well here,
1899 since the CVS log documents them by reference to sbcl-devel
1900 messages, and the SourceForge archives aren't working well.:-(
1901 ** Dan Barlow improved the Alpha port (and is making progress on the
1902 PPC port, for those of you who think different).
1903 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
1904 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
1905 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
1906 continues to work on the SPARC port (for those of you in a position
1907 to look down upon our little PC-compatible boxes from a great height).
1908 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
1909 and uses that copy, rather than the current dynamic value, when
1910 it's trying to decide whether to truncate output. Thus e.g.
1911 (let ((*print-lines* 50))
1912 (pprint-logical-block (stream nil)
1914 (let ((*print-lines* 8))
1915 (print (aref possiblybigthings i) stream)))))
1916 should now truncate the logical block only at 50 lines, instead of
1917 often truncating it at 8 lines, as it did before.
1918 * The doc/cmucl/ directory, containing old CMU CL documentation
1919 from the time of the fork, is no longer part of the base system.
1920 SourceForge has shut down its anonymous FTP service, and with it
1921 my original plan for distributing the old CMU CL documentation
1922 there. For now, if you need these files you can download an old
1923 SBCL source release and extract them from it.
1924 * The fasl file version number changed again, for dozens of reasons,
1925 some of which are apparent above.
1927 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
1928 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
1929 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
1930 patch, so that DISASSEMBLE output is much nicer.
1931 * The code in the SB-PROFILE package now seems reasonably stable.
1932 I still haven't decided what the final interface should look like
1933 (I'd like PROFILE to interact cleanly with TRACE, since both
1934 facilities use function encapsulation) but if you have a need
1935 for profiling now, you can probably use it successfully with
1936 the current CMU-CL-style interface.
1937 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
1938 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
1939 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
1940 link now returns the dangling link itself, and for similar
1941 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
1942 link itself. (In these cases the old code signalled an error and
1943 looped endlessly, respectively.) Thus, DIRECTORY now works even
1944 in the presence of dangling and cyclic symbolic links.
1945 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
1946 is now a supported extension again, since the consensus on
1947 sbcl-devel was that it can be useful for ordinary development
1948 work, not just for debugging SBCL itself.
1949 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
1950 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
1951 that currently-defined functions might be redefined later with
1952 different return types.
1953 * Hash tables can be printed readably, as inspired by CMU CL code
1954 of Eric Marsden and SBCL code of Martin Atzmueller.
1955 * better error handling in CLOS method combination, thanks to
1956 Martin Atzmueller porting Pierre Mai's CMU CL patches
1957 * more overflow fixes for >16Mbyte I/O buffers
1958 * A bug in READ has been fixed, so that now a single Ctrl-D
1959 character suffices to cause end-of-file on character streams.
1960 In particular, now you only need one Ctrl-D at the command
1961 line (not two) to exit SBCL.
1962 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
1964 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
1965 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
1966 does the right thing.
1967 * The implementation of some type tests, especially for CONDITION
1968 types, is now tidier and maybe faster, due to CMU CL code
1969 originally by Douglas Crosher, ported by Martin Atzmueller.
1970 * Some math functions have been fixed, and there are new
1971 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
1972 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
1973 * (There are also some new optimizers in contrib/*-extras.lisp. Those
1974 aren't built into sbcl-0.6.13, but are a sneak preview of what's
1975 likely to be built into sbcl-0.7.0.)
1976 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
1977 CL, ported to SBCL by Martin Atzmueller)
1978 * DESCRIBE now gives more information in some cases. (Pierre Mai's
1979 patch to CMU CL, ported to SBCL by Martin Atzmueller)
1980 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
1981 * There's a new slam.sh hack to shorten the edit/compile/debug
1982 cycle for low-level changes to SBCL itself, and a new
1983 :SB-AFTER-XC-CORE target feature to control the generation of
1984 the after-xc.core file needed by slam.sh.
1985 * minor incompatible change: The ENTRY-POINTS &KEY argument to
1986 COMPILE-FILE is no longer supported, so that now every function
1987 gets an entry point, so that block compilation looks a little
1988 more like the plain vanilla ANSI section 3.2.2.3 scheme.
1989 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
1990 returns the number of bytes consed since the system started,
1991 rather than the number consed since the first time the function
1992 was called. (The new definition parallels ANSI functions like
1993 CL:GET-INTERNAL-RUN-TIME.)
1994 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
1995 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
1996 supported. Now DIRECTORY always does the abstract Common-Lisp-y
1997 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
1998 * Fasl file version numbers are now independent of the target CPU,
1999 since historically most system changes which required version
2000 number changes have affected all CPUs equally. Similarly,
2001 the byte fasl file version is now equal to the ordinary
2004 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2005 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2006 is no longer recognized. I apologize for this, because it was
2007 listed in SB-EXT as a supported extension, but I found that
2008 its existing behavior was poorly specified, as well as incorrectly
2009 specified, and it looked like too much of a mess to straighten it
2010 out. I have enough on my hands trying to get ANSI stuff to work..
2011 * many patches ported from CMU CL by Martin Atzmueller, with
2012 half a dozen bug fixes in pretty-printing and the debugger, and
2013 half a dozen others elsewhere
2014 * fixed bug 13: Floating point infinities are now supported again.
2015 They might still be a little bit flaky, but thanks to bug reports
2016 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2017 as flaky as they were.
2018 * The --noprogrammer command line option is now supported. (Its
2019 behavior is slightly different in detail from what the old man
2020 page claimed it would do, but it's still appropriate under the
2021 same circumstances that the man page talks about.)
2022 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2023 are now supported, and enabled by default. Thus, the compiler can
2024 handle many floating point and complex operations much less
2025 inefficiently. (Thus e.g. you can implement a complex FFT
2027 * The compiler now detects type mismatches between DECLAIM FTYPE
2028 and DEFUN better, and implements CHECK-TYPE more correctly, and
2029 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2030 thanks to patches from Martin Atzmueller.
2031 * various fixes to make the cross-compiler more portable to
2032 ANSI-conforming-but-different cross-compilation hosts (notably
2033 Lispworks for Windows, following bug reports from Arthur Lemmens)
2034 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2035 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2036 fixed thanks to Pierre Mai's CMU CL patches.
2037 * fixes to keep the system from overflowing internal counters when
2038 it tries to use i/o buffers larger than 16M bytes
2039 * fixed bug 45a: Various internal functions required to support
2040 complex special functions have been merged from CMU CL sources.
2041 (When I was first setting up SBCL, I misunderstood a compile-time
2042 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2043 * improved support for type intersection and union, fixing bug 12
2044 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2045 more obscure bugs as well
2046 * some steps toward byte-compiling non-performance-critical
2047 parts of the system, courtesy of patches from Martin Atzmueller
2048 * Christophe Rhodes has made some debian packages of sbcl at
2049 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2050 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2051 stable, but are nonetheless usable. When he's ready, I'd be happy
2052 to add them to the SourceForge "File Releases" section. (And if
2053 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2054 * new fasl file format version number (because of changes in
2055 internal representation of (OR ..) types to accommodate the new
2056 support for (AND ..) types, among other things)
2058 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2059 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2061 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2063 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2064 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2065 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2066 * DESCRIBE now works on structure objects again.
2067 * Most function call argument type mismatches are now handled as
2068 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2069 know whether the function will be redefined before the call is
2070 executed. (The compiler could flag local calls with full WARNINGs,
2071 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2072 it doesn't keep track of enough information to know whether calls
2073 are local in this sense.)
2074 * Compiler output is now more verbose, with messages truncated
2075 later than before. (There should be some supported way for users
2076 to override the default verbosity, but I haven't decided how to
2077 provide it yet, so this behavior is still controlled by the internal
2078 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2079 src/compiler/ir1util.lisp.)
2080 * Fasl file format version numbers have increased again, because
2081 support for the Gray streams extension changes the layout of the
2082 system's STREAM objects.
2083 * The Gray subclassable streams extension now works, thanks to a
2084 patch from Martin Atzmueller.
2085 * The full LOAD-FOREIGN extension (not just the primitive
2086 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2087 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2088 but like most other programs, it defaults to copying the Unix
2089 environment from the original process instead of starting the
2090 new process in an empty environment.
2091 * Extensions which manipulate the Unix environment now support
2092 an :ENVIRONMENT keyword option which doesn't smash case or
2093 do other bad things. The CMU-CL-style :ENV option is retained
2094 for porting convenience.
2095 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2096 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2098 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2100 * A patch from Martin Atzmueller seems to have solved the SIGINT
2101 problem, and as far as we know, signal-handling now works cleanly.
2102 (If you find any new bugs, please report them!)
2103 * The system no longer defaults Lisp source file names to types
2104 ".l", ".cl", or ".lsp", but only to ".lisp".
2105 * The compiler no longer uses special default file extensions for
2106 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2107 seems to expect a single default extension for all compiled code,
2108 and there's no compelling reason to try to stretch the standard
2109 to allow two different extensions.) Instead, byte-compiled files
2110 default to the same extension as native-compiled files.
2111 * Fasl file format version numbers have increased again, because
2112 a rearrangement of internal implementation packages made some
2113 dumped symbols in old fasl files unreadable in new cores.
2114 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2115 many fewer weird special cases.
2116 * Bug #17 (differing COMPILE-FILE behavior between logical and
2117 physical pathnames) has been fixed, and some related misbehavior too,
2118 thanks to a patch from Martin Atzmueller.
2119 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2120 by Tim Moore, ported to SBCL by Martin Atzmueller.
2121 * Martin Atzmueller fixed several filesystem-related problems,
2122 including bug #36, in part by porting CMU CL patches, which were
2123 written in part by Paul Werkowski.
2124 * More compiler warnings in src/runtime/ are gone, thanks to
2125 more patches from Martin Atzmueller.
2126 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2129 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2131 * DESCRIBE now works on CONDITION objects.
2132 * The debugger now handles errors which arise when trying to print
2133 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2135 * The build system now uses an additional file, customize-target-features.lisp,
2136 to allow local modifications to the target *FEATURES* list. (The point of
2137 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2138 debugging features enabled, without having to worry about propagating it
2139 into everyone's system when I do a "cvs update".) When no
2140 customize-target-features.lisp file exists, the target *FEATURES* list
2141 should be constructed the same way as before.
2142 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2143 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2144 warn about behavior which is undefined under the ANSI spec. Note: This
2145 is specified by ANSI, but it's not very popular with programmers.
2146 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2147 macro in the SBCL sources for an example of a workaround which you
2148 might use to make portable ANSI-standard code which does what you want.
2149 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2150 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2151 when it's used as a non-toplevel form. (This is required in order
2152 to implement the DEFCONSTANT-EQX macro.)
2153 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2154 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2155 funny things, due to bugs in the implementation of EVAL-WHEN
2156 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2157 non-toplevel DEFCONSTANTs are uncommon.)
2158 * The core file version number and fasl file version number have been
2159 incremented, because the old noncompliant DEFCONSTANT behavior involved
2160 calling functions which no longer exist, and because I also took the
2161 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2162 * fixed bug 1 (error handling before read-eval-print loop starts), and
2163 redid debugger restarts and related debugger commands somewhat while
2165 ** The QUIT debugger command is gone, since it did something
2166 rather different than the SB-EXT:QUIT command, and since it never
2167 worked properly outside the main toplevel read/eval/print loop.
2168 Invoking the new TOPLEVEL restart provides the same functionality.
2169 ** The GO debugger command is also gone, since you can just invoke
2170 the CONTINUE restart directly instead.
2171 ** The TOP debugger command is also gone, since it's redundant with the
2172 FRAME 0 command, and since it interfered with abbreviations for the
2174 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2176 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2177 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2178 some porting work by Martin Atzmueller)
2179 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2180 with a boatload of patches which he ported to SBCL. Now that those
2182 ** The system tries to make sure that its low-priority messages
2183 are prefixed by semicolons, to help people who like to use
2184 syntax highlighting in their ILISP buffer. (This patch
2185 was originally due to Raymond Toy.)
2186 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2187 patches originally written by Raymond Toy.
2188 ** The compiler understands coercion between single-value and
2189 multiple-VALUES type expressions better, getting rid of some very
2190 weird behavior, thanks to patches originally by Robert MacLachlan
2191 and Douglas Crosher.
2192 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2193 lambda lists, thanks to a patch originally by Pierre Mai.
2194 ** The system no longer bogusly warns about "abbreviated type
2196 ** The compiler gets less confused by inlining and RETURN-FROM,
2197 thanks to some patches originally by Tim Moore.
2198 ** The system no longer hangs when dumping circular lists to fasl
2199 files, thanks to a patch originally from Douglas Crosher.
2200 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2201 undefined function error.
2202 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2203 in structure slot initforms) for now, documented workaround instead:-|
2204 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2205 * fixed bug 5: added stubs for various Gray stream functions called
2206 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2207 installed, at least appropriate type errors are generated
2208 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2209 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2210 consistently in DEFMETHOD forms.
2211 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2212 it doesn't seem to affect SBCL after all
2213 * The C runtime system now builds with better optimization and many
2214 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2216 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2218 * The system is now under CVS at SourceForge (instead of the
2219 CVS repository on my home machine).
2220 * The new signal handling code has been tweaked to treat register
2221 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2222 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2223 downstream, e.g. in debug-int.lisp, has implicit dependencies
2224 on the unsignedness of integer representation of machine words,
2225 and that caused the system to bomb out with infinite regress
2226 when trying to recover from type errors involving signed values,
2227 e.g. (BUTLAST '(1 2 3) -1).)
2228 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2229 (This was one of the bugs Peter Van Eynde reported back in July.)
2230 * The system now uses code inspired by Colin Walters' O(N)
2231 implementation of MAP (from the cmucl-imp@cons.org mailing
2232 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2233 inline the MAP operation, and there is more than one
2234 sequence argument to the MAP call (so that it can't just
2235 do ETYPECASE once and for all based on the type of the
2236 single sequence argument). (The old non-inline implementation
2237 of the general M-argument sequence-of-length-N case required
2238 O(M*N*N) time when any of the sequence arguments were LISTs.)
2239 * The QUIT :UNIX-CODE keyword argument has been renamed to
2240 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2241 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2243 * Raymond Wiker's patches to port dynamic loading from Linux to
2244 FreeBSD have been added.
2245 * The BUGS file is now more nearly up to date, thanks in large part
2246 to Martin Atzmueller's review of it.
2247 * The debugger now flushes standard output streams before it begins
2248 its output ("debugger invoked" and so forth).
2249 * The core version number and fasl file version number have both
2250 been incremented, because of incompatible changes in the layout
2252 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2253 used to not be called for a saved Lisp image.)
2254 * Martin Atzmueller's version of a patch to fix a compiler crash,
2255 as posted on sbcl-devel 13 September 2000, has been installed.
2256 * Instead of installing Martin Atzmueller's patch for the
2257 compiler transform for SUBSEQ, I deleted the compiler transform,
2258 and transforms for some similar consing operations.
2259 * A bug in signal handling which kept TRACE from working on OpenBSD
2261 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2262 away by constant folding
2263 * The system now defines its address space constants in one place
2264 (in the Lisp sources), and propagates them automatically elsewhere
2265 (through GENESIS and the sbcl.h file). Therefore, patching the
2266 address map is less unnecessarily tedious and error-prone. The
2267 Lisp names of address space constants have also been systematized.
2268 * CVS tags like dollar-Header-dollar have been removed from
2269 the sources, because they have never saved me trouble and
2270 they've been source of trouble working with patches and other
2271 diff-related operations.
2272 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2273 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2275 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2277 * The system has been ported to OpenBSD.
2278 * The system now compiles with a simple "sh make.sh" on the systems
2279 that it's supported on. I.e., now you no longer need to tweak
2280 text in the target-features.lisp-expr and symlinks in src/runtime/
2281 by hand, the make.sh takes care of it for you.
2282 * The system is no longer so grossly inefficient when compiling code
2283 involving vectors implemented as general (not simple) vectors (VECTOR T),
2284 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2285 which dares to use the various sequence functions on non-simple
2286 vectors, takes less of a performance hit.
2287 * There is now a primitive type predicate VECTOR-T-P
2288 to test for the (VECTOR T) type, so that e.g.
2289 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2290 can now be compiled with some semblance of efficiency. (The old code
2291 turned the type declaration into a full call to %TYPEP at runtime!)
2292 * AREF on (VECTOR T) is still not fast, since it's still compiled
2293 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2294 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2295 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2296 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2297 without making full calls to SUBTYPEP at runtime.
2298 (Some analogous efficiency issues for non-simple vectors specialized to
2299 element types other than T, or for non-simple multidimensional arrays,
2300 have not been addressed. They could almost certainly be handled the
2301 same way if anyone is motivated to do so.)
2302 * The changes in array handling break binary compatibility, so
2303 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2304 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2305 instead of (VALUES T T).
2306 * By following the instructions that Dan Barlow posted to sbcl-devel
2307 on 2 July 2000, I was able to enable primitive dynamic object
2308 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2309 functionality is not implemented (since it calls ld to resolve
2310 library references automatically, requiring RUN-PROGRAM for its
2311 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2312 try to resolve library references) is now supported.
2313 * The system now flushes the standard output streams when it terminates,
2314 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2315 them at several other probably-convenient times, e.g. in each pass of
2316 the toplevel read-eval-print loop, and after evaluating a form given
2317 as an "--eval" command-line option. (These changes were motivated by a
2318 discussion of stream flushing issues on cmucl-imp in August 2000.)
2319 * The source transform for TYPEP of array types no longer assumes
2320 that an array whose element type is a not-yet-defined type
2321 is implemented as an array of T, but instead punts, so that the
2322 type will be interpreted at runtime.
2323 * There is now some support for cross-compiling in make.sh: each of
2324 the phases of make.sh has its own script. (This should be transparent
2325 to people doing ordinary, non-cross-compile builds.)
2326 * Since my laptop doesn't have hundreds of megabytes of memory like
2327 my desktop machine, I became more motivated to do some items on
2328 my to-do list in order to reduce the size of the system a little:
2329 ** Arrange for various needed-only-at-cold-init things to be
2330 uninterned after cold init. To support this, those things have
2331 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2332 symbols with such names are now uninterned after cold init).
2333 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2334 fasl files for cold load.
2335 ** Remove the old compiler structure pooling code (which used to
2336 be conditional on the target feature :SB-ALLOC) completely.
2337 ** Redo the representation of some data in cold init to be more compact.
2338 (I also looked into supporting byte compiled code at bootstrap time,
2339 which would probably reduce the size of the system a lot, but that
2340 looked too complicated, so I punted for now.)
2341 * The maximum signal nesting depth in the src/runtime/ support code has
2342 been reduced from 4096 to 256. (I don't know any reason for the very
2343 large old value. If the new smaller value turns out to break something,
2344 I'll probably just bump it back up.)
2345 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2347 * Many, many bugs reported by Peter Van Eynde have been added to
2348 the BUGS list; some have even been fixed.
2349 * While enabling dynamic object file loading, I tried to make the
2350 code easier to understand, renaming various functions and variables
2351 with less ambiguous names, and changing some function calling
2352 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2353 * While trying to figure out how to do the OpenBSD port, I tried to
2354 clean up some of the code in src/runtime/. In particular, I dropped
2355 support for non-POSIX signal handling, added various comments,
2356 tweaked the code to reduce the number of compilation warnings, and
2357 renamed some files to increase consistency.
2358 * To support the new automatic configuration functionality in make.sh,
2359 the source file target-features.lisp-expr has been replaced with the
2360 source file base-target-features.lisp-expr and the machine-generated
2361 file local-target-features.lisp-expr.
2362 * fixed a stupid quoting error in make.sh so that using CMU CL
2363 "lisp -batch" as cross-compilation host works again
2365 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2367 * DESCRIBE no longer tries to call itself recursively to describe
2368 bound/fbound values, so that it no longer fails on symbols which are
2369 bound to themselves (like keywords, T, and NIL).
2370 * DESCRIBE now works on generic functions.
2371 * The printer now prints less-screwed-up representations of closures
2372 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2373 * A private symbol is used instead of the :EMPTY keyword previously
2374 used to mark empty slots in hash tables. Thus
2375 (DEFVAR *HT* (MAKE-HASH-TABLE))
2376 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2377 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2378 now does what ANSI says that it should. (You can still get
2379 similar noncompliant behavior if bang on the hash table
2380 implementation with all the symbols you get back from
2381 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2382 This breaks binary compatibility, since tests for equality to
2383 :EMPTY are wired into things like the macroexpansion of
2384 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2386 * There's now a minimal placeholder implementation for CL:STEP,
2387 as required by ANSI.
2388 * An obscure bug in the interaction of the normal compiler, the byte
2389 compiler, inlining, and structure predicates has been patched
2390 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2391 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2393 * Missing ordinary arguments in a macro call are now detected even
2394 when the macro lambda list contains &KEY or &REST.
2395 * The debugger no longer complains about encountering the top of the
2396 stack when you type "FRAME 0" to explicitly instruct it to go to
2397 the top of the stack. And it now prints the frame you request even
2398 if it's the current frame (instead of saying "You are here.").
2399 * As specified by ANSI, the system now always prints keywords
2400 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2402 * The default initial SIZE of HASH-TABLEs is now smaller.
2403 * Type information from CLOS class dispatch is now propagated
2404 into DEFMETHOD bodies, so that e.g.
2405 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2407 is now basically equivalent to
2408 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2409 (DECLARE (TYPE SINGLE-FLOAT X))
2411 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2412 operation, without having to do run-time type dispatch.
2413 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2414 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2415 * Since I don't seem to be making big file reorganizations very often
2416 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2417 is overflowing my ability to conveniently back them up), I've finally
2418 checked the system into CVS. (The CVS repository is on my home system,
2419 not at SourceForge -- putting it on SourceForge might come later.)
2420 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2421 high-level GC-NOTIFY-FOO functions send their output. (There's
2422 still very little control of where low-level verbose GC functions
2423 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2424 less than it used to -- the GC-NOTIFY-FOO functions are now under
2425 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2426 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2427 in only one place in the source code, and propagates it automatically
2428 everywhere that it's needed. Thus e.g. when I bump the version from
2429 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2430 * The C source files now include boilerplate legalese and documentation
2431 at the head of each file (just as the Lisp source files already did).
2432 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2433 to his page will be replaced with a link to his new CLiki service.
2435 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2437 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2438 * The build process now looks for GNU make under the default name "gmake",
2439 instead of "make" as it used to. If GNU make is not available as "gmake"
2440 on your system, you can change this default behavior by setting the
2441 GNUMAKE environment variable.
2442 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2443 can build without error under CMU CL.
2445 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2447 * There is now a partial SBCL user manual (with some new text and some
2448 text cribbed from the CMU CL manual).
2449 * The beginnings of a profiler have been added (starting with the
2450 CMU CL profiler and simplifying and cleaning up). Eventually the
2451 main interface should be through the TRACE macro, but for now,
2452 it's still accessed through vaguely CMU-CL-style functions and macros
2453 exported from the package SB-PROFILE.
2454 * Some problems left over from porting CMU CL to the new
2455 cross-compilation bootstrap process have been cleaned up:
2456 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2457 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2458 part when getting rid of the compiler *BACKEND* stuff.)
2459 ** The value of *NULL-TYPE* was screwed up, because it was
2460 being initialized before the type system knew the final
2461 definition of the 'NULL type. This screwed up several key
2462 optimizations in the compiler, causing inefficiency in all sorts
2463 of places. (I found it because I wanted to understand why
2464 GET-INTERNAL-RUN-TIME was consing.)
2465 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2466 PROCLAIM FTYPE information. Unfortunately this broke binary
2467 compatibility again, since now the forms output by DEFGENERIC
2468 to refer to functions which didn't exist in 0.6.3.
2469 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2470 can use the new (as of 0.6.3) transform for SOME into MAP into
2472 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2473 values to (INTEGER 0 1000000), so that the time code will no longer
2474 occasionally get blown up by Linux returning 1000000 microseconds
2475 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2476 its output conform to the ANSI spec. (Alas, this makes its output
2477 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2478 * A full call to MAP NIL with a single sequence argument no longer conses.
2479 * fixes to problems pointed out by Martin Atzmueller:
2480 * The manual page no longer talks about multiprocessing as though
2481 it were currently supported.
2482 * The ILISP support patches have been removed from the distribution,
2483 because as of version 5.10.1, ILISP now supports SBCL without us
2484 having to maintain patches.
2485 * added a modified version of Raymond Toy's recent CMU CL patch for
2486 EQUALP comparison of HASH-TABLE
2488 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2490 * The system still can't cross-compile itself with
2491 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2492 entails), but at least it can compile more of itself that way
2493 than it used to be able to, and various buglets which were uncovered
2494 by trying to cross-compile itself that way have now been fixed.
2495 * This release breaks binary compatibility again. This time
2496 at least I've incremented the FASL file format version to 2, so that the
2497 problem can be detected reliably instead of just causing weird errors.
2498 * various new style warnings:
2499 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2500 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2501 ** using the lexical binding of a variable named in the *FOO* style
2502 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2503 as specified by ANSI.
2504 * *RANDOM-STATE* is no longer automatically initialized from
2505 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2506 default behavior of the system is to repeat its behavior every time
2507 it's run. If you'd like to change this behavior, you can always
2508 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2509 old convention there was no comparably easy way to get the system to
2510 repeat its behavior every time it was run.
2511 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2512 FTYPE declarations has been removed, in favor of their ANSI
2513 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2514 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2515 the types of their sequence arguments can be determined at compile time.
2516 This is done through a new open code expansion for MAP which eliminates
2517 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2518 argument types can be determined at compile time.
2519 * The optimizer now transforms COERCE into an identity operation when it
2520 can prove that the coerced object is already of the correct type. (This
2521 can be a win for machine generated code, including the output of other
2522 optimization transforms, such as the MAP transform above.)
2523 * Credit information has been moved from source file headers into CREDITS.
2524 * Source file headers have been made more standard.
2525 * The CASE macro now compiles without complaining even when it has
2528 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2530 * (Note that the way that the PCL macroexpansions were rewritten
2531 to accommodate the change in DEFGENERIC below breaks binary
2532 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2533 not run under sbcl-0.6.2. Once we get out of alpha releases,
2534 i.e. hit release 1.0.0, we'll probably try to maintain binary
2535 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2536 and sbcl-1.4.4. Until then, however, it might be fairly common
2537 for maintenance releases to break binary compatibility.)
2538 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2540 * The reporting of unhandled signals has been changed to print some
2541 explanatory text as well as the report form. (Previously only
2542 the report form was printed.)
2543 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2544 it defines, so that the compiler no longer issues undefined function
2545 warnings for compiled-but-not-yet-loaded generic functions.
2546 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2547 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2548 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2549 section of the ANSI spec are supported.
2550 * The "" nickname for the "KEYWORD" package has been removed.
2551 The reader still handles symbol tokens which begin with a package marker
2552 as keywords, but it doesn't expose its mechanism for doing so in the
2553 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2554 * The system now issues STYLE-WARNINGs for contradictory TYPE
2555 proclamations. (Warnings for contradictory FTYPE proclamations would
2556 be nice too, but those can't be done usefully unless the type system
2557 is made smarter about FUNCTION types.)
2558 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2559 systematized, so that "*target-*.lisp is supposed to exist only on the
2560 target and imply that there's a related file which exists on the
2561 host, and *host-*.lisp is supposed to exist only on the host and imply
2562 that there's a related file which exists on the target. This involves a
2563 lot of renaming. Hopefully the acute confusion caused by the renaming
2564 will be justified by the reduction in chronic confusion..
2565 ** runtime-type.lisp -> early-target-type.lisp
2566 ** target-type.lisp -> late-target-type.lisp
2567 ** early-host-format.lisp -> early-format.lisp
2568 ** late-host-format.lisp -> late-format.lisp
2569 ** host-error.lisp -> misc-error.lisp
2570 ** early-error.lisp -> early-target-error.lisp
2571 ** late-error.lisp -> late-target-error.lisp
2572 ** host-defboot.lisp -> early-defboot.lisp
2573 ** code/misc.lisp -> code/target-misc.lisp
2574 ** code/host-misc.lisp -> code/misc.lisp
2575 ** code/numbers.lisp -> code/target-numbers.lisp
2576 ** code/early-numbers.lisp -> numbers.lisp
2577 ** early-host-type.lisp -> early-type.lisp
2578 ** late-host-type.lisp -> late-type.lisp
2579 ** host-typep.lisp -> typep.lisp
2580 ** load.lisp -> target-load.lisp
2581 ** host-load.lisp -> load.lisp
2582 ** host-disassem.lisp -> disassem.lisp
2583 ** host-insts.lisp -> insts.lisp
2584 ** byte-comp.lisp -> target-byte-comp.lisp
2585 ** host-byte-comp.lisp -> byte-comp.lisp
2586 ** host-signal.lisp -> signal.lisp
2587 ** host-defstruct.lisp -> defstruct.lisp
2588 ** late-target-type.lisp -> deftypes-for-target.lisp
2589 Furthermore, several other previously target-only files foo.lisp (e.g.
2590 hash-table.lisp and random.lisp) have been split into a target-and-host
2591 foo.lisp file and a target-only target-foo.lisp file, with their key type
2592 definitions in the target-and-host part, so that the cross-compiler will
2593 know more about target types.
2594 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2595 gone away. In their place are various *BACKEND-FOO* variables
2596 corresponding to the slots of the old structure.
2597 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2598 into the target SBCL has been fixed.
2599 * The chill.lisp system for loading cold code into a running SBCL
2601 * Support for the CMU CL "scavenger hook" extension has been removed.
2602 (It was undocumented and unused in the CMU CL sources that SBCL was
2603 derived from, and stale in sbcl-0.6.1.)
2604 * Various errors in the cross-compiler type system were detected
2605 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2606 (enabling various consistency checks). Many of them were fixed,
2607 but some hard problems remain, so the compiler is back to
2608 running without *TYPE-SYSTEM-INITIALIZED* for now.
2609 * As part of the cross-compiler type system cleanup, I implemented
2610 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2611 * I have started adding UNCROSS calls throughout the type system
2612 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2613 on cross-compiler input files will be able to go away, and various
2615 * CONSTANTP now returns true for quoted forms (as explicitly required
2618 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2620 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2621 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2622 (MAKE-LIST -1). In the longer term, it ought to become true
2623 that declarations are assertions even at SAFETY 1. For now, it's not
2624 quite true even at SAFETY 3, but it's at least more nearly true..
2625 (Note that this change seems to increases the size of the system by
2626 O(5%) and to decrease the speed of the compiler by 20% or more.)
2627 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2628 for the problem of printing dozens of lines of distracting information
2629 about low-level system machinery as part of the top stack frame
2630 on entry to the debugger when an undefined function was called.
2631 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2632 is not reset to COMMON-LISP-USER.
2633 * Compilation of stuff related to dyncount.lisp has been made conditional
2634 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2635 smaller. The various dyncount-related symbols have been moved into
2636 a new "SB-DYNCOUNT" package.
2637 * tty-inspect.lisp has been renamed to inspect.lisp.
2638 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2639 feature has gone away. (When we eventually port to other flavors of
2640 libc and/or Unix, we'll try to make the differences between flavors
2641 invisible at the user level.)
2642 * Various other *FEATURES* tags, and/or their associated conditionals,
2643 have been removed if obsolescent, or given better documentation, or
2644 sometimes given more-mnemonic names.
2646 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2648 * tidied up "make.sh" script
2649 * tidied up system directory structure
2650 * better "clean.sh" behavior
2651 * added doc/FOR-CMUCL-DEVELOPERS
2652 * many many small tweaks to output format, e.g. removing possibly-confusing
2653 trailing #\. character in DESCRIBE-INSTANCE
2654 * (EQUALP #\A 'A) no longer signals an error.
2655 * new hashing code, including EQUALP hashing
2656 * tidied up Lisp initialization and toplevel
2657 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2658 * command line argument processing
2659 * added POSIX-GETENV function to deal with Unix-ish environment variables
2660 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2661 terminating SBCL on EOF
2662 * non-verbose GC by default
2663 * There is no more "sbcl" shell script; the sbcl file is now the C
2664 runtime executable (just like CMU CL).
2665 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2667 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2668 symbol and package information as Lisp native symbol and package objects
2669 instead of strings naming symbols and strings naming packages. This way,
2670 whenever packages are renamed (as in warm init), debug information is
2671 transformed along with everything else.
2672 * tweaked the optimization policy declarations which control the building
2673 of SBCL itself. Now, among other things, the system no longer saves
2674 source location debugging information. (This helps two problems at once
2675 by reducing SBCL size and by keeping SBCL from trying to look for its
2676 sources -- which may not exist -- when reporting errors.)
2677 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2678 debugging and testing purposes
2679 * cleaned up printing, making the printer call PRINT-OBJECT for
2680 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2681 methods, giving nearly-ANSI behavior
2682 * converted almost all special variables to use *FOO* naming convention
2683 * deleted PARSE-TIME functionality, since it can be done portably
2684 * moved some files out of cold init into warm init
2685 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2687 * regularized formatting of source files
2688 * added an install.sh script
2689 * fixed ridiculous memory usage of cross-compiler by making
2690 compiler/alloc.lisp not try to do pooling unless it can hook
2691 itself into the GC of the cross-compilation host. Now the system
2692 builds nicely on my old laptop.
2693 * added :SB-ALLOC in target-features.lisp-expr
2694 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2695 was not implemented)
2696 * re-did condition handling and note reporting in the compiler. Notes
2697 are no longer handled by signalling conditions. Style warnings
2698 and warnings are handled more correctly and reported in such a way
2699 that it's easy to find one or the other in your output (so that you
2700 can e.g. figure out which of many problems caused COMPILE-FILE to
2702 * changed the severity of several compiler warnings from full WARNING
2703 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2704 compiler note reporting so that it doesn't use the condition system
2705 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2706 COMPILE-FILE command)
2707 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2708 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2709 no longer supported, since their implementation was deeply intertwingled
2710 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2711 * removed bogus "support" for compiler macros named (SETF FOO), and
2712 removed the compiler macro for SETF INFO (but only after making a fool
2713 of myself on the cmucl-imp mailing list by posting a bogus patch for
2714 DEFINE-COMPILER-MACRO..)
2715 * Compiled files containing forms which have side effects on the Lisp
2716 reader (such as DEFPACKAGE forms) are now handled more correctly.
2717 (Compiler queuing of top level lambdas has been suppressed by setting
2718 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2719 * deleted various currently-unused source files, e.g. gengc.lisp. They
2720 may be added back at some point e.g. when porting to other architectures,
2721 but until they are it's distracting to distribute them and to try to
2723 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2724 there were so many of them they're just distractions, and UNCROSS is
2725 known to be able to handle the current sources
2726 * moved PROFILE functionality into TRACE, so that it will be clear
2727 how the wrapping and unwrapping of functions when you profile them
2728 interacts with the wrapping and unwrapping of functions when you
2729 trace them. (Actually, the functionality isn't there yet, but at least
2730 the interface specification is there. Hopefully, the functionality will
2731 arrive with some future maintenance release.)
2732 * removed host-oops.lisp
2733 * changed signature of QUIT function to allow UNIX-CODE argument
2734 * fixed READ-SEQUENCE bug
2735 * tweaked verbose GC output so that it looks more like the progress
2736 output that ANSI specifies for functions like LOAD
2737 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2738 * added <http://sbcl.sourceforge.com> to the banner information printed by