1 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2 * bug fix: defining classes whose accessors are methods on existing
3 generic functions in other (locked) packages no longer signals
4 bogus package lock violations. (reported by François-René Rideau)
5 * bug fix: special variables as DEFMETHOD parameters no longer have
6 associated bogus type declarations. (reported by David Wragg and
8 * bug fix: read-write consistency on streams of element-type
9 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
10 * bug fix: redefiniton of the only method of a generic function with
11 no DEFGENERIC no longer emits a full WARNING. In addition,
12 redefinition of generic functions with no DEFGENERIC to an
13 incompatible lambda list now signals an error. (thanks to Zach
15 * bug fix: DEFGENERIC now works even when there's a function of the
16 same name in an enclosing lexical environment. (thanks to Zach
18 * fixed compiler failure, caused by instrumenting code during
19 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
20 * optimization: added loop analysis and improved register allocation
21 * fixed some bugs revealed by Paul Dietz' test suite:
22 ** POSITION on displaced vectors with non-zero displacement
23 returns the right answer.
24 ** (SIMPLE-STRING) is a valid type specifier for sequence
26 ** *PRINT-LEVEL* handling for slotless structures is pedantically
28 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
29 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
31 ** The justification version of the FORMAT ~< directive treats
32 non-zero minpad parameter correctly.
34 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
35 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
36 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
37 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
38 the supported interface.
39 * new feature: Single-stepping of code compiled with DEBUG 2 or
40 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
41 * new feature: saving cores with foreign code loaded is now
42 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
43 Timothy Moore's work for CMUCL)
44 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
45 parameters to * instead of NIL if no initform is supplied.
46 (reported by Johan Bockgaard)
47 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
48 correctly when FUNCALL appears as the car of the form. Note:
49 despite this FUNCALL forms are not currently subject to
50 compiler-macro expansion. (port of Raymond Toy's fix for the
51 same from CMUCL, reported by Johan Bockgaard)
52 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
53 (thanks for Teemu Kalvas)
54 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
55 case instead of terminating the iteration (thanks for Julian
56 Squires, Sean Champ and Raymond Toy)
57 * bug fix: incorrect expansion of defgeneric that caused a style
58 warning. (thanks for Zach Beane)
59 * bug fix: slot accessor effective method computation works properly
60 for classes with multiple non-standard applicable methods on
61 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
62 * on x86 compiler supports stack allocation of results of LIST and
63 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
64 CMUCL implementation by Gerd Moellmann)
65 * fixed some bugs revealed by Paul Dietz' test suite:
66 ** FORMAT strings with both the ~<~:;~> form of the justification
67 directive and pretty-printing directives cause an error.
69 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
70 * incompatible change: the internal functions
71 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
72 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
73 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
74 instead of the old functions.
75 * new feature: on platforms where "dladdr" is available foreign
76 function names now appear in backtraces. (based on Helmut Eller's
78 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
79 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
81 * documentation: ASDF manual is now distributed with SBCL.
82 * bug fix: SBCL can now load its contributed modules with REQUIRE
83 even if the system-provided entries have been removed from
84 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
86 * bug fix: backtraces involving undefined functions or assembly
87 routines are more informative. (thanks to Brian Downing)
88 * bug fix: mutually referent alien structures now work correctly.
89 (reported by Rick Taube)
90 * bug fix: structures defined by WITH-ALIEN can be referred to
91 within other definitions in the same WITH-ALIEN.
92 * bug fix: division operators (MOD, TRUNCATE and the like) with
93 constant zero divisors and integer dividends no longer generate
95 * bug fix: provide default methods for INPUT-STREAM-P and
96 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
97 * bug fix: improve the detection and resolution of MOP metacycles
98 (where effective method computation depends on the generic function
99 having its effective method computed). (reported by Bruno Haible)
100 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
101 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
102 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
103 * bug fix: pathnames with relative directory components can be
104 represented relative to default pathnames.
105 * optimization: in taking the GCD of bignums, reduce the two bignums
106 to approximately the same size (using Euclid's algorithm) before
107 applying the more sophisticated binary GCD. (thanks to Juho
109 * optimization: COUNT on bitvectors now operates word-at-a-time.
110 * optimization: ASH with a positive, but not necessarily constant,
111 (leftwards) shift, when in a modular context, is compiled to a
113 * fixed some bugs revealed by Paul Dietz' test suite:
114 ** FORMAT variable parameters ("~V<char>") are defaulted properly
115 if the corresponding argument is NIL.
116 ** FORMAT directives accepting a minpad parameter treat negative
118 ** literal commas as character parameters to FORMAT directives are
120 ** literal spaces directly after ~<Newline> directives within a
121 format-logical-block (~:< ~@:>) do not induce :FILL-style
122 conditional newlines.
123 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
124 as specified: it no longer includes conditional newlines).
125 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
127 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
128 :PER-LINE-PREFIX argument does not evaluate to a string.
130 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
131 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
132 the manual for details; for now, package locks can be disabled by
133 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
134 if no major problems are found then it is likely that they will be
135 compiled in unconditionally.
136 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
137 unsupported operators on all platforms. To load a shared library
138 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
139 object file, link it into a shared library outside of SBCL and
140 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
142 * bug workaround: ROOM T and the GC were not working together
143 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
144 reporting facility aren't preserved. That reporting has been
145 disabled, so now until and unless someone figures out how to make it
146 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
147 * minor incompatible change: as threatened around sbcl-0.8.0, the
148 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
149 The symbols are also exported from SB-PCL for backwards
150 compatibility, but more so than before SB-PCL should be treated as
151 an implementation-internal package.
152 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
154 * the SB-SPROF contrib now works on (most) non-x86 architectures.
155 It is known as of this release not to work on the Alpha, however.
156 * fixed bug #167: errors signalled due to illegal syntax in method
157 bodies are now more legible.
158 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
159 designators and can hence be used with TYPEP.
160 * fixed bug #333: CHECK-TYPE now ensures that the type error
161 signalled, if any, has the right object to be accessed by
162 TYPE-ERROR-DATUM. (reported by Tony Martinez)
163 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
164 5.1.2.3 for multiple-value place subforms. (reported by Kalle
166 * fixed bug #334: programmatic addition of slots using specialized
167 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
168 and :ALLOCATION :CLASS slots.
169 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
170 float exponents. (rereported by Peter Seibel)
171 * fixed a bug: #\Space (and other whitespace characters) are no
172 longer considered to be macro characters in standard syntax by
174 * fixed bug: initialization of condition class metaobjects no longer
175 causes an instance of the condition to be created. (reported by
177 * fixed bug: it is now possible to have slots such that
178 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
179 description is neither :INSTANCE nor :CLASS.
180 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
181 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
182 SLOT-VALUE-USING-CLASS), and no longer causes errors with
183 non-standard SLOT-DEFINITION-ALLOCATION values.
184 * fixed bugs: various race conditions handling exiting threads.
185 CL-PPCRE's thread torture test now passes.
186 * fixed arguably-a-bug: GC time proportional to number of threads in
187 system even when most of them are idle
188 * optimization: improved performance of BIT and SBIT on bit-vectors.
189 * .fasl file incompatibility: The fasl file version number has
190 been incremented because of changes associated with package locks.
192 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
193 * minor incompatible change: the system no longer provides
194 optimization hints (conditions which are TYPEP
195 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
196 mode; these hints are still emitted when higher SPEED optimization
198 * new contrib module: a sampling profiler (profiling by statistical
199 sampling, rather than by instrumenting functions) is available as
200 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
201 * the behaviour of the standard function ED is now customizeable by
202 third parties through a hook variable: see ED's documentation
203 string for information on the protocol.
204 * the compiler no longer emits efficiency notes for (FUNCALL X)
205 when the type of X is uncertain under default optimization
207 * fixed bug 276: mutating a binding of a specialized parameter to a
208 method to something that is not TYPEP the specializer is now
210 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
211 exhibit the required behaviour on the x86 platform. (thanks to
212 Peter van Eynde, Eric Marsden and Bruno Haible)
213 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
214 even for difficult arguments. (reported by Peter Graves)
215 * fixed bug 141a: the backquote printer now descends quoted
217 * fixed another bug in backquote printing: no more destructive
218 modification of the form's list structure. (reported by Brian
220 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
221 move between its address being taken and the call to
222 interrupt_thread, fixing a crashing race condition.
223 * the SB-POSIX contrib implementation has been adjusted so that it
224 no longer exhibits ridiculously poor performance when constructing
225 instances corresponding to C structs.
227 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
228 * minor incompatible change: the sb-grovel contrib now treats C
229 structures as alien (in the sense of SB-ALIEN) objects rather than
230 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
231 has implications for memory management of client code
232 (sb-grovel-returned objects must now be manually managed) and for
233 type safety (alien objects now have full types).
234 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
235 used to control emission of compiler diagnostics, rather than the
236 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
237 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
238 quality should be considered deprecated.
239 * install.sh now installs the user manual as well
240 * (not quite a new documentable feature, but worth considering in
241 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
242 beginnings of a semantically meaningful condition hierarchy is
243 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
245 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
246 designator as the defaults argument.
247 * fixed bug: Displaced arrays whose displaced-to array has become
248 too small now cause ARRAY-DIMENSION to signal an error, providing
249 for safer bounds-checking. (reported by Bruno Haible)
250 * fixed bug: DEFCLASS slot definitions with identical :READER and
251 :WRITER names now signal a reasonable error. (reported by Thomas
253 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
255 * fixed bug: Invalid slot specification errors now print correctly.
256 (thanks to Zach Beane)
257 * fixed bug 320: Shared to local slot value transfers in class
258 redefinitions now happen corrently from superclasses as
259 well. (reported by Bruno Haible)
260 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
262 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
263 improper lists correctly. (reported by Bruno Haible)
264 * fixed bug 313: source-transform for <fun-name> was erroneously
265 applied to a call of a value of a variable with name <fun-name>.
266 (reported by Antonio Menezes Leitao)
267 * fixed bug 307: The obsolete instance protocol ensures that
268 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
269 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
270 remove dead unknown-values globs from the middle of the stack.
271 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
272 easier to resume long computations after using *BREAK-ON-SIGNALS*
273 to diagnose and fix failures. (thanks to Nikodemus Siivola)
274 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
275 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
276 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
277 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
278 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
280 * fixed a (fairly theoretical) bug in string printing: if
281 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
282 does not have array-element-type equal to the most general string
284 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
285 function name. (thanks to Nikodemus Siivola)
286 * fixed bug: compile-time format string checker failed on
287 non-closed ~{. (reported by Thomas F Burdick)
288 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
289 don't warn on legal loop code involving more than one aggregate
290 boolean. (thanks to Nikodemus Siivola)
291 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
292 in keyword position to MAKE-PACKAGE was not being reported as
294 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
295 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
297 * fixed bugs 280 and 312: the checking for multiple definitions in a
298 file is less likely to become confused by uses of inline
300 * fixed bug: the #S reader macro performs the keyword coercion
301 specified for slot names. (reported by Kalle Niemitalo)
302 * fixed bug: lambda lists may contain symbols whose names start with
303 & but are not lambda-list-keywords; their occurrence triggers a
305 * fixed bug 321: define-method-combination argument lambda lists do
306 not cause invalid code to be generated when &optional or &aux
307 variables have default values. (reported by Bruno Haible)
308 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
309 have CLOS classes; this appears also to have fixed reported
310 instability in stack exhaustion detection.
311 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
312 MUFFLE-WARNING when no associated restart is present is now
314 * optimization: rearranged the expansion of various defining macros
315 so that each expands into only one top-level form in a
316 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
318 * optimization: used a previously little-used slot in symbols to
319 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
321 * fixed some bugs revealed by Paul Dietz' test suite:
322 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
324 ** Fixed an optimization bug related to inheritance of initforms
325 from local to shared slots.
326 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
327 ** CAST optimizer forgot to flush argument derived type.
328 ** print/read consistency on floats is now orders of magnitude
329 more likely. (thanks also to Bruno Haible for a similar report
331 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
332 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
333 for objects of type REAL. Make it so.
334 ** SXHASH is specified (infelicitously) to respect similarity,
335 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
336 it so. (thanks to Markus Ziegler)
337 ** on the Alpha, the compiler succeeds in compiling functions
338 returning a known number of arguments greater than 63.
339 ** fixed handling of invalid NIL arguments in keyword position
341 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
343 ** provided a readably-printable representation for RANDOM-STATE
345 ** ensured that pathnames, where they have a namestring, always
346 print using #P"..." syntax.
348 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
349 * Support for the forthcoming 2.0 version of the NetBSD kernel
350 running on x86 hardware has been added. (thanks to Perry
351 E. Metzger most immediately, and others for their past work)
352 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
353 current release version; SBCL's OpenBSD support had been broken
354 since about the time of OpenBSD's switch to ELF binary format.)
355 * [placeholder for DX summary]
356 ** user code with &REST lists declared dynamic-extent, under high
357 speed or space and low safety and debug optimization policy.
358 * The manual has been converted to Texinfo format and the debugger
359 chapter from the cmucl manual has been added.
360 * A facility has been added to extract documentation strings from
361 sbcl and store them as Texinfo-formatted snippets for inclusion in
362 the manual (via Texinfo's @include directive)
363 * bug fix: compiler emitted division in optimized DEREF. (thanks for
364 the test case to Dave Roberts)
365 * bug fix: multidimensional simple arrays loaded from FASLs had fill
366 pointers. (reported by Sean Ross)
367 * bug fix: PROFILE output is printed nicely even for large numerical
368 values. (thanks to Zach Beane)
369 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
370 greater than 32 handle EOF correctly.
371 * bug fix: on X86 an immediate argument of the IMUL instruction is
372 correctly printed in disassembly. (thanks to Lutz Euler)
373 * bug fix: class slots in redefined classes preserve their old
374 values. (thanks to Bruno Haible and Nikodemus Siivola)
375 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
376 succeeds. (reported by Marco Baringer)
377 * fixed some bugs revealed by Paul Dietz' test suite:
378 ** READ-SEQUENCE now works on ECHO-STREAMs.
379 ** RATIONALIZE works more according to its specification. (thanks
382 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
383 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
384 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
385 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
386 should matter to you only if you rebind the printer control
387 variables and then find you want different bindings in the
388 debugger than in the ordinary execution of your program.)
389 * The runtime build system has been tweaked to support building
390 (on SPARC/SunOS) using a C compiler which invokes Sun's own
391 assembler and linker. (thanks to Nikodemus Siivola)
392 * Unbound, undefined, undeclared variables now trigger full
393 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
394 more often programmer error than deliberate exploitation of undefined
396 * optimization: the hash algorithm for strings has changed to one
397 that is less vulnerable to spurious collisions. (thanks to Juho
399 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
400 needless bounds checking. (thanks to Juho Snellman)
401 * optimization: implemented multiplication as a modular
402 (UNSIGNED-BYTE 32) operation on the PPC backend.
403 * fixed some bugs revealed by Paul Dietz' test suite:
404 ** ADJUST-ARRAY now copies the datum in a zero rank array if
406 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
407 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
408 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
410 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
412 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
414 ** ECHO-STREAMs no longer attempt to echo the end of file value to
415 their output stream on EOF from read.
416 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
417 have been read to end-of-file.
418 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
420 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
421 description of determination of which consecutive characters
423 ** Printing the "Space" character with escaping on now yields "#\\ ",
424 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
425 ** Reading floating-point numbers with *READ-BASE* set to a number
426 less than 10 works correctly.
427 ** Reading floating-point numbers with *READ-BASE* set to a number
428 more than 10 works correctly.
429 ** Printing with *PRINT-READABLY* targets the standard readtable, not
430 the readtable currently in effect.
432 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
433 * minor incompatible change: parsing of namestrings on a physical
434 (Unix) host has changed; numbers after the final #\. in a
435 namestring are no longer interpreted as a version field. This is
436 intented to be largely invisible to the user, except that the
437 meaning of the namestring "*.*.*" has changed: it now refers to a
438 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
439 should usually be replaced by
440 (make-pathname :name :wild :type :wild :version :wild)
441 with the added benefit that this is more likely to be portable.
442 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
443 option to OPEN now signals an error if the file being opened
444 exists; this may have an impact on existing code.
445 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
446 signals received, on the PowerPC platforms (both Linux and
447 Darwin). (thanks to Pierre Mai for pointing out the location of
449 * several fixes on OS X: The system now builds and runs cleanly on
450 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
451 (thanks to Brian Mastenbrook)
452 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
453 second-resolution: integer multiples of 1/3600 between -24 and 24.
454 (thanks to Vincent Arkesteijn)
455 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
456 type when called with 1 argument; PEEK-CHAR checked type of
457 PEEK-TYPE only after having read first character from a
458 stream. (reported by Peter Graves)
459 * bug fix: the garbage collector now has much better locality
460 behaviour, and in particular no longer treats all memory as being
461 exhausted when many small objects point to each other in a deeply
463 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
464 recognized as being TYPEP their class.
465 * bug fix: the PUSHNEW documentation string has been corrected.
466 (thanks to Vincent Arkesteijn)
467 * bug fix: defaulting of the value for the last of an atypically
468 large number of multiple values being bound was not being
469 performed correctly on the Alpha or PPC platforms
470 * optimization: implemented multiplication as a modular
471 (UNSIGNED-BYTE 32) operation on the x86 backend.
472 * optimization: SEARCH on simple-base-strings can now be open-coded.
473 (see also contrib/compiler-extras.lisp for inspiration for
474 teaching the compiler about the Boyer-Moore algorithm).
475 * value, returned by MAX (and MIN) called with several EQUALP, but
476 not EQL, arguments now does not depend on compiler settings.
477 * fixed some bugs revealed by Paul Dietz' test suite:
478 ** in stack analysis liveness information is propagated from
479 non-local entry points.
480 ** pathwise CAST removing failed when the CAST node did not start
482 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
483 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
485 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
486 host is already defined.
487 ** RENAME-FILE works on streams instead of signalling an internal
489 ** PEEK-CHAR uses the current readtable when determining whether
490 or not a character is whitespace.
491 ** MERGE-PATHNAMES handles the case when the pathname does not
492 specify a name while the default-pathname specifies a version.
493 ** Pathnames now stand a better chance of respecting print/read
495 ** Attempting to use standardized file system operators with a
496 pathname with invalid :DIRECTORY components signals a
498 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
500 ** DIRECTORY on logical pathnames is more correct.
501 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
502 signal a TYPE-ERROR if their argument is not a stream.
503 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
504 designator argument does not designate a stream.
505 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
506 examining the synonym.
507 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
509 ** OPEN and WITH-OPEN-STREAM allow opening streams with
510 element-type larger than ([UN]SIGNED-BYTE 32).
512 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
513 * When built with the :SB-FUTEX feature, threaded builds now take
514 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
515 for faster/more reliable mutex and condition variable support.
516 * Incompatible change (but one you probably shouldn't have been using
517 anyway): the interface and code for arbitrating between multiple
518 threads in the same user session has been redesigned.
519 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
520 error when called without an explicit environment argument.
521 (thanks to Nikodemus Siivola)
522 * bug fix: buffered :DIRECTION :IO streams are less likely to become
523 confused about their position. (thanks to Adam Warner and Gerd
525 * bug fix: Pretty printing backquoted forms with unquotations in the
526 argument list position of various code constructs such as LAMBDA
527 now works correctly. (reported by Paul Dietz)
528 * bug fix: Pretty printing unquotations no longer loses all
529 stream position information.
530 * optimization: performance of string output streams is now less
531 poor for multiple small sequence writes.
532 * optimization: performance of CSUBTYPEP in the presence of complex
533 expressions involving CONS and NOT many times has been improved.
534 (reported by Paul Dietz)
535 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
537 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
538 streams for the REPL, for the debugger, and so on. (thanks to
540 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
541 * fixed some bugs revealed by Paul Dietz' test suite:
542 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
543 is ignored if it is too small, rather than propagating through
544 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
545 ** extremely complex negations of CONS types were not being
546 sufficiently canonicalized, leading to inconsistencies in
548 ** VALUES tranformer lost derived type.
550 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
551 * fixed a bootstrapping bug: the build process no longer assumes
552 that the various BOOLE-related constants have the same value in
553 host and target lisps. (noted by Paul Dietz' test suite on an
554 SBCL binary built from CLISP)
555 * The system can now be dynamically linked on the MIPS platform,
556 which enables dynamic loading of foreign code from Lisp. (thanks
557 to Ralf Baechle for discussions on the MIPS ABI)
558 * The system now records debugging information for its own source
559 files in a filesystem-position-independent manner, relative to
560 the "SYS" logical host.
561 * fixed a compiler bug: MV-LET convertion did not check references
562 to the "max args" entry point. (reported by Brian Downing)
563 * tweaked disassembly notes to be less confident about proclaiming
564 some instruction as an LRA. (thanks to Brian Downing)
565 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
566 now each have their own history, command character, and other
567 characteristics. (thanks to David Lichteblau)
568 * fixed some bugs revealed by Paul Dietz' test suite:
569 ** compiler failure in compiling LOGAND expressions including a
571 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
572 shift greater than 32.
573 ** FLUSH-DEST did not mark blocks for type check regeneration.
574 ** HANDLER-CASE failed to accept declarations in handler clauses
575 in some circumstances.
577 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
578 * New code in contrib/sb-introspect (still probably not entirely
579 stable yet) provides some support for smart Lisp development
580 environments like SLIME.
581 * The conditions signalled for errors occurring when loading .fasl
582 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
583 in a way which should help ASDF recover gracefully.
584 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
585 stuff has been cleaned up. If you code contrib/ stuff, this might
586 affect you, and you can look at contrib/README, contrib/STANDARDS,
587 and/or the 0.8.4.27 diff to check.
588 * In full calls the compiler now does not generate checks for declared
589 argument types for all arguments.
590 * various threading fixes
591 ** and some experimental patches which didn't make it into
592 the main tree for this release, but which are shipped in
593 contrib/experimental-thread.patch as a possible fix for some
594 failures (deadlock, spinning...) in GC-intensive multithreaded
596 * fixed PPC build problem (source code incompatibility of different
597 library versions): added offsetof() hackery which attempts to divine
598 where glibc maintainers put uc_mcontext today
599 * fixed bug 282: compiler does not trust type assertions while passing
600 arguments to a full call.
601 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
602 &OPTIONAL VALUES type parameter.
603 * fix bug 214: algorithm for noting rejected templates is now more
604 similar to that of template seletion. (also reported by rydis on
606 * fixed bug 141b: printing backquoted information readably and prettily
607 inserts a space where necessary.
608 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
609 (CHARACTER 10) are now reported as errors, rather than propagated
610 as unknown types. (reported by piso on #lisp)
611 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
612 with values NIL and :ERROR. (thanks to Milan Zamazal)
613 * fixed bug 191c: CLOS now does proper keyword argument checking as
614 described in CLHS 7.6.5 and 7.6.5.1.
615 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
616 longer raise an error; further, using a list as a for-as-arithmetic
617 counter now raises a meaningful error.
618 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
619 now understood by sequence creation functions such as MAKE-SEQUENCE
621 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
622 a STRING-INPUT-STREAM.
623 * compiler enhancement: SIGNUM is now better able to derive the type
625 * type declarations inside WITH-SLOTS are checked. (reported by
627 * fixed some bugs revealed by Paul Dietz' test suite:
628 ** incorrect optimization of TRUNCATE for a positive first
629 argument and negative second.
630 ** compiler failure in let-convertion during flushing dead code.
631 ** compiler failure while deriving type of TRUNCATE on an
632 interval, containing 0.
633 ** ASH of a negative bignum by a negative bignum count now returns
635 ** intersection of CONS types now canonicalizes properly, fixing
636 inconsistencies in SUBTYPEP.
638 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
639 * incompatible change: The --disable-debugger command line
640 option now clobbers the debugger at a more fundamental
641 level, by redefining #'INVOKE-DEBUGGER instead of by
642 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
643 is specified by ANSI to ignore *DEBUGGER-HOOK* and
644 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
645 enter the debugger REPL and then suffer recursive errors
646 because *DEBUG-IO* is also messed up in --disable-debugger mode;
647 while under the new system, BREAK in --disable-debugger mode
648 terminates the system just as an unhandled error would.
649 * fixed compiler performance when processing loops with a step >1;
650 * bug fix: DOCUMENTATION now retrieves generic function
651 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
652 support has been systematized, and now supports the methods
653 specified by ANSI, along with a default method and a method for
654 slot documentation. (reported by Nathan Froyd)
655 * bug fix: effective methods associated with a generic function are
656 no longer cached over a change of that generic function's method
657 combination. (reported by Andreas Fuchs)
658 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
659 names a non-existent directory. (thanks to Andreas Fuchs)
660 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
661 the PPC platform, lead to stack corruption; this has been fixed.
662 (reported by Rainer Joswig)
663 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
664 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
665 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
666 arithmetic in general, is now much more reliable on the PPC
668 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
669 platform now returns the right answer.
670 * optimization: restored some effective method precomputation in
671 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
672 precomputation is now tunable.
673 * optimization: compiler-internal data structure use has been
674 reviewed, and changes have been made that should improve the
675 performance of the compiler by about 20%.
676 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
677 simple-base-strings and simple-bit-vectors is improved.
678 * optimization: the optimization of 32-bit logical and arithmetic
679 functions introduced in version 0.8.3 on the x86 has been
680 implemented on the mips, ppc and sparc platforms; an
681 implementation of the same facility, but for 64-bit arithmetic,
682 has been added for the alpha.
683 * microoptimization: the compiler is better able to make use of the
684 x86 LEA instruction for multiplication by constants.
685 * bug fix: in some situations compiler did not report usage of
686 generic arithmetic in (SPEED 3) policy.
687 * bug 145b fix: compiler used wrong type specifier while converting
688 MEMBER-types to numeric.
689 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
691 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
692 index argument now works properly on the Alpha platform.
693 * bug fix: floating point exception treatment on the Alpha platform
695 * bug fix: FILE-POSITION works much better on string input and
696 output streams. (thanks to Nikodemus Siivola)
697 * bug fix: many threading/garbage collection symptoms sorted.
698 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
699 might be pseudo-atomic.
700 * internal change: Stopping for GC is now done with signals not
701 ptrace. GC is now done in whichever thread wanted it, instead of
703 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
705 * bug fix: result form in DO is not contained in the implicit
707 * incompatible change: ICR structure is changed; the value part of
708 CONTINUATION is now called LVAR; corresponding functions are
709 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
710 * added type deriver for ISQRT (thanks to Robert E. Brown).
711 * bug fix: better support for loading from the command line when an
712 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
714 * fixed some bugs revealed by Paul Dietz' test suite:
715 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
716 ** ROUND and FROUND now give the right answer when given very
717 small float arguments.
718 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
720 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
721 ** (GCD 0 <negative-integer>) returned <negative-integer>.
722 ** LCM should return a non-negative integer.
723 ** PARSE-INTEGER returned the index of a terminator instead of the
724 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
725 ** PARSE-INTEGER returned an incorrect index being applied to a
727 ** LCM with two arguments of 0 returns 0 rather than signalling
729 ** unsigned addition of a 32-bit constant with the high bit set no
730 longer causes an internal compiler error on the x86.
731 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
732 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
733 with negative last argument.
734 ** byte specifiers with zero size and position no longer cause
735 an error during type derivation.
736 ** bignum multiplication on the Alpha platform now returns the
738 * porting: The system now builds on SuSE AMD64, although it still
739 generates a 32-bit binary.
740 * .fasl file incompatibility: The fasl file version number has
741 been incremented (because of the changes to internal compiler
742 data structures referred to above).
744 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
745 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
746 more accurately, on the Darwin kernel running on PowerPC hardware.
747 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
748 * Compiler code deletion notes now signal a condition of type
749 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
750 an associated MUFFLE-WARNING restart.
751 * The compiler now performs limited argument count validation of
752 constant format strings in FORMAT, and where appropriate in ERROR,
753 CERROR and WARN. (thanks to Gerd Moellmann)
754 * New ASDF-INSTALL contrib can be used for automatic download and
755 installation of third-party Lisp code from CCLAN or other sites
757 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
758 which forces another thread to execute a function supplied by the
760 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
761 now accept and act upon their :ELEMENT-TYPE keyword argument.
762 (reported by Martin Atzmueller, Edi Weitz)
763 * bug fix: FILE-POSITION now accepts position designators up to
764 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
765 is the greater. (thanks to Patrik Nordebo)
766 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
767 Kalle Olavi Niemitalo)
768 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
769 printable. (reported by Eric Marsden)
770 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
771 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
772 * x86 bug fix in control stack exhaustion checking: now shows backtrace
773 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
774 (thanks to Stig Sandoe)
775 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
776 UNREAD-CHAR now correctly includes the unread character in the
777 target sequence. (thanks to Gerd Moellmann)
778 * bug fix in threaded builds: the system can now be suspended and
779 resumed by shell job control with minimal disruption.
780 * bug fixes in times and timezones >2038AD
781 * better handling of "where is GNU make?" problem in build scripts
782 (thanks to Nikodemus Siivola)
783 * new optimization: inside a named function any reference to a
784 function with the same name is considered to be a self-reference;
785 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
786 optimization quality.
787 * new optimization on x86: logical functions and + now have
788 optimized (UNSIGNED-BYTE 32) versions, which are automatically
789 used when the result is truncated to 32 bits.
790 * VALUES declaration is partially enabled.
791 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
792 * fixed some bugs revealed by Paul Dietz' test suite:
793 ** The system now obeys the constraint imposed by
794 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
795 types form a lattice under type intersection.
796 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
797 ** ASSOC now ignores NIL elements in an alist.
798 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
799 and (1+ MOST-POSITIVE-FIXNUM) answers.
800 ** The addition of a method with invalid qualifiers to a generic
801 function does not cause an error to be signalled immediately;
802 a warning is signalled, and the error is generated only on
803 calling the generic function.
804 * changed .fasl file version number, in part to add type codes for
805 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
806 obscure ANSI requirements
808 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
809 * fixed bug 148: failure to inline-expand a local function left
810 garbage, confusing the compiler.
811 * fixed bugs 3cd: structure slot readers perform type check if the
812 slot can have an invalid value (i.e. it is either not initialized
813 or can be written with a less specific slot writer).
814 * bug fix: the compiler now traps array references to elements off
815 the end of an array; previously, the bounds checking in some
816 circumstances could go off-by-one.
817 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
819 * type declarations for array element types now obey the description
820 on the CLHS page "Declaration TYPE", as per discussions on
821 sbcl-help around 2003-05-08. This means that a declaration
822 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
823 declaration, all references to BAR will be asserted or assumed
824 (with THE, so dependent on compiler policy) to involve objects of
825 type FOO. Note that no such declaration is implied in
826 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
827 * declared types of functions from the "Conditions"
828 chapter. (reported by Paul Dietz)
829 * bug fix: CERROR accepts a function as its first argument.
830 * bug fix: NTH an NTHCDR accept a bignum as index
831 arguments. (reported by Adam Warner)
832 * optimization: character compare routines now optimize comparing
833 against a constant character. (reported by Gilbert Baumann)
834 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
835 argument now works properly on the MIPS platform.
836 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
837 * repeated evaluation of the same DEFSTRUCT, a slot of which is
838 declared to have a functional type, does not cause an error
840 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
841 many :DEFAULT-INITARGS. (reported by Istvan Marko)
842 * fixed bug: if last continuation of a deleted block has a
843 destination, this destination should be deleted too. (reported by
845 * fixed a bug in the bootstrap process: the host compiler's values
846 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
847 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
848 test case from Patrik Nordebo)
849 * improved the ability of the disassembler on the PPC platform to
850 provide helpful disassembly notes.
851 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
852 the class in more cases than previously.
853 * bug fix: FILE-POSITION now understands :START and :END for
854 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
855 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
856 8) data. (thanks to David Lichteblau)
857 * bug fix: it is possible to add a method to a generic function
859 * bug fix: reader failed to signal END-OF-FILE inside an
860 object representation. (reported by Nikodemus Siivola)
861 * fixed some bugs revealed by Paul Dietz' test suite:
862 ** LAST and [N]BUTLAST should accept a bignum.
863 ** condition slot accessors are methods.
864 ** (VECTOR NIL) is a subtype of STRING.
866 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
867 * minor incompatible change: some nonsensical specialized lambda
868 lists (used in DEFMETHOD) which were previously ignored now signal
870 * minor incompatible change: the system is now aware of the types of
871 variables in the COMMON-LISP package, and will signal errors for
872 most violations of these type constraints (where previously they
873 were silently accepted).
874 * minor incompatible change: COMPILE-FILE now uses the freedom
875 afforded (ANSI 3.2.2.3) to use derived function types for
876 functions defined in the same file. This also permits the system
877 to warn on static type mismatches and function
878 redefinition. (Currently it does not work with high DEBUG level.)
879 * minor incompatible change: VALUES declaration is disabled.
880 * When issuing notes, the compiler now signals a condition of type
881 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
882 restart for use in user handlers. It is expected that the
883 COMPILER-NOTE condition will eventually become a condition
884 supertype to a hierarchy of note types, which will then be
885 handleable in a similar fashion. However, at the moment, no such
886 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
887 but it's an implementation detail, not a classification for the
889 * Changes in type checking closed the following bugs:
890 ** type checking of unused values (192b, 194d, 203);
891 ** template selection based on unsafe type assertions (192c, 236);
892 ** type checking in branches (194bc).
893 * A short form of VALUES type specifier has ANSI meaning (it has
894 increased the number of situations when SBCL cannot perform type
896 * fixed bug in DEFSTRUCT: once again, naming structure slots with
897 keywords or constants is permissible.
898 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
899 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
900 classes. (thanks to Antonio Martinez)
901 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
902 outputs FRESH-LINE or TERPRI, and no longer converts its stream
903 argument to a pretty-print stream. Instead, it leaves any such
904 operations to DESCRIBE-OBJECT methods.
905 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
907 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
908 lambda lists. (thanks to David Lichteblau)
909 * bug fix: Defining a generic function with a :METHOD-CLASS being a
910 subclass of STANDARD-METHOD no longer causes stack exhaustion.
911 (thanks to Gerd Moellmann)
912 * fixed bug 246: increased compilation speed of long
913 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
915 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
917 * DEFCONSTANT now throws a condition of type
918 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
919 constant to a non-EQL value; CONTINUE and ABORT restarts
920 respectively change and preserve the value.
921 * fixed bug 63: The code walker, part of the implementation of CLOS,
922 is now better at handling symbol macros.
923 * bug fix: There is no longer an internal implementation type named
924 CL:LENGTH. (reported by Raymond Toy)
925 * bug fix: In macro-like defining macros/special operators the
926 implicit block does not enclose the lambda list.
927 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
928 enclosed in parentheses) are not suitable as type specifiers, and
929 their use properly signals an error now.
930 * bug fix: An argument count mismatch for a type specifier in code
931 being compiled no longer causes an unhandled error at compile
932 time, but signals a compile-time warning.
933 * fixed simple vector readable printing
934 * bug fix: DESCRIBE takes more care over whether the class
935 precedence list slot of a class is bound before accessing it.
936 (reported by Markus Krummenacker)
937 * bug fix: FORMATTER can successfully compile pretty-printer format
938 strings which use variants of the ~* directive inside.
939 * bug fix: SEARCH now applies its TEST predicate to the elements of
940 the arguments in the correct order. (thanks to Wolfhard Buss)
941 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
942 in type inference. (thanks to Robert E. Brown)
943 * bug fix: Reading in symbols with an explicit package name of ""
944 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
945 * fixed some bugs revealed by Paul Dietz' test suite:
946 ** NIL is now allowed as a structure slot name.
947 ** Arbitrary numbers, not just REALs, are allowed in certain
948 circumstances in LOOP for-as-arithmetic clauses.
949 ** Multiple class redefinitions before slot access no longer
951 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
952 association between the name and a class.
953 ** Generic functions with non-standard method-combination and over
954 five methods all of which return constants no longer return NIL
955 after the first few invocations. (thanks to Gerd Moellmann)
956 ** CALL-NEXT-METHOD with no arguments now passes the original
957 values of the arguments, even in the presence of assignment.
958 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
960 ** DISASSEMBLE works with closures and funcallable instances.
961 ** ADD-METHOD now returns the generic function, not the new method.
962 ** FIND-METHOD signals an error if the lengths of the specializers
963 is incompatible with the generic function, even if the ERRORP
965 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
966 which its argument is a member.
967 ** DEFCLASS only redefines the class named by its class-name
968 argument if that name is the proper name of the class;
969 otherwise, it creates a new class.
970 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
971 of the UNBOUND-SLOT condition to the name of the slot.
972 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
973 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
974 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
975 SLOT-MAKUNBOUND in the specified fashion.
977 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
978 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
979 cross-compilation host. As a consequence, we can now bootstrap our
980 way up to SBCL starting with a bare gcc toolchain and human-readable
981 source code (first the source to CLISP, then the source to SBCL).
982 * A contributed module containing a partial implementation of the
983 simple-streams interface has been included. (thanks to Rudi
985 * A contributed module implementing the RFC1321 Message Digest
986 Algorithm, known as MD5, has been included.
987 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
988 no longer has any effect, as the code controlled by this feature
989 has been deleted. (As far as we know, no-one has ever built using
990 this feature, and its semantics were confused in any case).
991 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
992 work on conditions (as required by the ANSI specification),
993 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
994 expected behaviour on conditions. Users should note, however,
995 that such behaviour is not required by the ANSI specification,
996 and so use of this behaviour may render their code unportable.
997 * fixed some bugs revealed by Paul Dietz' test suite:
998 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1000 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1001 ** SXHASH on condition objects no longer returns NIL.
1002 ** :ALLOCATION :CLASS slots are better treated; their values are
1003 updated on class redefinition, and initforms inherited from
1004 superclasses are applied.
1005 ** REMOVE-METHOD returns its generic function argument even when
1006 no method was removed.
1007 ** SHARED-INITIALIZE now initializes the values of the requested
1008 slots, including those with :ALLOCATION :CLASS.
1009 ** ALLOCATE-INSTANCE now works on structure classes defined via
1010 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1012 ** SLOT-EXISTS-P now works on conditions, as well as structures
1014 ** MAKE-LOAD-FORM now has the required methods on
1015 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1016 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1017 keyword, and now implements the SLOT-NAMES argument.
1018 ** methods with &OPTIONAL arguments no longer allow too many
1019 arguments to be passed in the call without error.
1020 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1021 option is consistent with the required arguments of the generic
1022 function lambda list.
1023 * bug fix: REQUIRE accepts a string designator. (Thanks to
1025 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1026 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1027 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1028 * bug fix: The compiler now checks for duplicated variables in macro
1030 * bug fix: SETQ on globals returns the correct value.
1031 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1032 not a COMPILER-ERROR (followed by some other strange error on
1033 choosing the CONTINUE restart).
1034 * bug fix: make.sh and friends are now more consistent in the way that
1035 they look for GNU "make".
1037 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1038 * experimental native threads support (on x86 Linux >=2.4 only).
1039 This is not compiled in by default: you need to add :SB-THREAD to
1040 the target features. See the "Beyond ANSI" chapter of the manual
1042 * fix for longstanding nonANSIism: The old distinction between
1043 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1044 The return value from CL:FIND-CLASS is now a CLOS class, and
1045 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1046 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1047 * An interface to the MetaObject Protocol, as described in Kiczales,
1048 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1049 MIT Press, 1991, is available from the SB-MOP package.
1050 * incompatible change: the SB-PCL package should now be considered
1051 a private implementation detail, and no longer a semi-private MOP
1053 * minor incompatible change: due to rearrangement for threads, the
1054 control stack and binding stack are now allocated at arbitrary
1055 addresses instead of being hardcoded per-port. Users affected by
1056 this probably have to be doing advanced things with shared
1057 libraries, and will know who they are.
1058 * minor incompatible change: Previously, all --eval forms used were
1059 processed with READ before any of them were processed with EVAL.
1060 Now each --eval form is processed with both READ and EVAL before
1061 the next --eval form is processed. (Thus package operations like
1062 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1063 work as the user might reasonably expect.)
1064 * minor incompatible change: *STANDARD-INPUT* is now only an
1065 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1067 * minor incompatible change: Y-OR-N-P is now character-oriented,
1068 not line oriented. Also, YES-OR-NO-P now works without errors.
1069 (thanks to Antonio Martinez)
1070 * sb-aclrepl module improvements: an integrated inspector, added
1071 repl features, and a bug fix to :trace command.
1072 * Known functions, which cannot be open coded by the backend, are
1073 considered to be able to check types of their arguments. (fixing
1074 a bug report by Nathan J. Froyd)
1075 * fixed a bug in computing method discriminating functions: It is
1076 now possible to define methods specialized on classes which have
1077 forward-referenced superclasses. (thanks to Gerd Moellmann)
1078 * fixed evaluation order in optional entries (reported by Gilbert
1080 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1081 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1082 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1083 required-by-AMOP NAME argument, as well as CLASS and
1084 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1085 * fixed bug 20: DEFMETHOD can define methods using names that are
1086 not the proper names of classes to designate class specializers.
1087 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1088 Unix file descriptors, instead of blowing up. (thanks to Antonio
1090 * Garbage collection refactoring: user-visible change is that a
1091 call to the GC function during WITHOUT-GCING will not do garbage
1092 collection until the end of the WITHOUT-GCING. If you were doing
1093 this you were probably losing anyway.
1094 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1095 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1096 TYPEP the latter but not the former.
1097 * The compiler issues a full WARNING for calls to undefined functions
1098 with names from the CL package.
1099 * MAP-INTO for a vector destination is open coded. (reported by
1100 Brian Downing on c.l.l)
1101 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1102 documentation string.
1103 * fixed some bugs revealed by Paul Dietz' test suite:
1104 ** COPY-ALIST now signals an error if its argument is a dotted
1106 ** Condition slots are now accessed more correctly in the presence
1107 of multiple initargs for a given slot.
1108 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1109 exclude restarts of the same name associated with a different
1111 ** DEFCLASS of forward-referenced classes with another
1112 forward-referenced class in the superclasses list no longer
1114 ** Condition slots are now initialized once each, not multiple
1115 times. (thanks to Gerd Moellmann)
1116 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1118 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1119 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1120 arguments contain duplicated elements.
1121 ** RESTART-CASE understands local macros.
1122 ** RESTART-CASE associates exactly its own restarts with a condition.
1123 ** ENDP in safe mode checks its argument to be of type LIST.
1124 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1125 in question is unbound.
1126 ** Optimized MAKE-INSTANCE functions no longer cause internal
1127 assertion failures in the presence of duplicate initargs.
1128 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1129 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1130 propagated even in the ctor optimized implementation of
1132 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1134 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1135 * a better implementation of SXHASH on (simple) bit vectors,
1136 measured both in execution speed and in distribution of results
1137 over the positive fixnums, has been installed. Likewise, a better
1138 implementation of EQUAL for simple bit vectors is now available.
1139 * fixed CEILING optimization for a divisor of form 2^k.
1140 * fixed bug 240 (emitting extra style warnings "using the lexical
1141 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1142 by Antonio Martinez)
1143 * fixed SXHASH, giving different results for NIL depending on type
1144 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1145 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1146 variable. (found by Rolf Wester)
1147 * SBCL does not ignore type declarations for special
1148 variables. (reported by rif on c.l.l 2003-03-05)
1149 * some bug fixes in contrib/sb-aclrepl/
1150 * fixed some bugs revealed by Paul Dietz' test suite:
1151 ** a bug in the CONS type specifier, whereby the CAR and CDR
1152 types got intertwined, has been fixed;
1153 ** the type system is now able to reason about the interaction
1154 between INTEGER and RATIO types more completely;
1155 ** APPEND, [N]REVERSE and NRECONC check that those their
1156 arguments, which must be proper lists, are really so;
1157 ** An array specialized to be unable to hold elements has been
1158 implemented, as required -- yes, really -- by ANSI;
1159 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1160 on malformed property lists;
1162 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1163 * incompatible packaging change: in line with Unix convention,
1164 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1165 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1166 * REQUIRE and PROVIDE are now optionally capable of doing something
1167 useful. See the documentation string for REQUIRE.
1168 * infrastructure for a managed SBCL contrib system: contributed
1169 modules in this release include:
1170 ** the ASDF system definition facility;
1171 ** an interface to the BSD Sockets API;
1172 ** an ACL-like convenience interface to the repl;
1173 (thanks to Kevin Rosenberg)
1174 ** an implementation of ROTATE-BYTE, with efficient implementation
1176 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1177 gives a closer approximation to the right answer than previously.
1178 (thanks to Raymond Toy)
1179 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1180 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1181 optional environment arguments, as required by ANSI.
1182 * fixed bugs in other functions taking environment objects, allowing
1183 calls with an explicit NIL environment argument to be compiled
1185 * fixed bug 228: primary return values from
1186 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1187 COMPILE or FUNCTION.
1188 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1189 :TYPE will no longer signal errors on innocuous objects.
1190 * fixed bug 231b: SETQ is better at respecting type declarations in
1191 the lexical environment.
1192 * fixed a bug in DEFCLASS: classes named by symbols with no or
1193 unprintable packages can now be defined.
1194 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1195 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1196 * fixed compiler failure related to checking types of functions.
1197 (reported by Robert E. Brown)
1198 * the compiler is now much more consistent in its error-checking
1199 treatment of bounding index arguments to sequence functions: in
1200 (SAFETY 3) code, errors will be signalled in almost all cases if
1201 invalid sequence bounding indices are passed to functions defined
1202 by ANSI to operate on sequences.
1203 * fixed a bug in the build procedure: documentation of SBCL-specific
1204 packages is now preserved and available in the final Lisp image.
1205 * lifted FDEFINITION lookup out of loops in the implementation of
1206 many list operations. (thanks to Robert E. Brown)
1207 * fixed a bug in the reader: the #n# reader macro now works for
1208 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1209 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1210 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1211 E. Brown and Tony Martinez respectively)
1212 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1213 SLOT-BOUNDP: in particular, it is now able to optimize them much
1214 better, and is now not vulnerable to having packages renamed.
1215 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1216 always called when a slot is not present in an instance. (thanks
1218 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1219 longer signal an error on unreading a character following EOF on
1220 the previous constituent stream. (thanks to Tony Martinez)
1221 * fixed some bugs revealed by Paul Dietz' test suite:
1222 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1223 not just nonnegative fixnums;
1224 ** the logical bit-array operators such as BIT-AND now accept an
1225 explicit NIL for their "opt-arg" argument (to indicate a
1226 freshly-consed result bit-array);
1227 ** ELT now signals an error on an invalid sequence index in safe
1229 ** the type system is now cleverer about negations of numeric
1230 types, and consequently understands the BIGNUM and RATIO types
1232 ** the type system is now cleverer about the interaction between
1233 INTEGER and RATIO types: while bugs still remain, many more
1234 cases are accurately computed;
1235 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1236 if it is in the last clause;
1237 ** CONSTANTLY now correctly returns a side-effect-free function in
1239 ** DECLARE is no longer treated as a special-operator; in
1240 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1241 * incremented fasl file version number due to the change in the
1242 DEFSTRUCT-SLOT-DESCRIPTION structure.
1244 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1245 * minor incompatible change: code processed by the "interpreter" or
1246 EVAL now has a compilation optimization policy of (DEBUG 2)
1247 (changed from (DEBUG 1)) to improve debuggability of interactive
1248 development, and to allow the use of the debug RETURN command in
1250 * an experimental implementation of the RETURN command for the
1251 debugger has been included. (thanks to Frederik Kuivinen)
1252 * fixed bug 62: constraints were not propagated into a loop.
1253 * fixed bug in embedded calls of SORT (reported and investigated by
1255 * fixed some bugs revealed by Paul Dietz' test suite:
1256 ** printing and reading of arrays with some dimensions having
1257 length 0 (thanks to Gerd Moellmann);
1258 ** BOA constructor with &AUX argument without a default value does
1259 not cause a type error;
1260 ** CONSTANTP now returns true for all self-evaluating objects.
1262 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1263 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1264 accessors that are related by inheritance, as specified in the
1265 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1266 Valtteri Vuorikoski)
1267 * The compiler is now able to inline functions that were defined in
1268 a complex lexical environment (e.g. inside a MACROLET).
1269 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1271 * Support for the upcoming FreeBSD-5.0 release has been included.
1272 (thanks to Dag-Erling Smorgrav)
1273 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1274 effect when it is not in a toplevel context.
1275 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1276 stand a better chance of being correct. (thanks to Gerd
1278 * fixed bug in COERCE, which now signals an error on coercing a
1279 rational to a bounded real type which excludes the expected
1281 * The compiler is now able to derive types more accurately from the
1282 COERCE and COMPILE functions.
1283 * fixed bug 223: functional binding is considered to be constant
1284 only for symbols in the CL package.
1285 * fixed bug 231: SETQ did not check the type of a variable being set
1286 (reported by Robert E. Brown)
1287 * A new optimization for MAKE-INSTANCE has been included, fixing
1288 various bugs (including relating to :ALLOCATION :CLASS slots and
1289 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1291 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1292 ** As required by ANSI, LOOP now disallows anonymous collection
1293 clauses such as COLLECT I in conjunction with aggregate boolean
1294 clauses such as THEREIS (= I 1);
1295 ** LOOP now signals an error when any variable is reused in the
1296 same loop (including the potentially useful construct analogous
1297 to WITH A = 1 WITH A = (1+ A);
1298 ** IT is only a special loop symbol within the first clause of a
1299 conditional loop clause;
1300 ** LOOP with a typed iteration variable over a hashtable now
1301 signals a type error iff it should.
1302 * fixed some other bugs revealed by Paul Dietz' test suite:
1303 ** FILE-STREAM now names the class previously known as FD-STREAM;
1304 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1305 argument) no longer signals an error;
1306 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1307 of the slot symbol, rather than using the current package
1308 ((:CONC-NAME "") continues to intern the slot's name in the
1310 * incremented fasl file version number, because of the incompatible
1311 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1312 of the new implementation of DEFINE-COMPILER-MACRO.
1314 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1315 * Support for building SBCL for MIPS platforms running in
1316 little-endian mode has now been checked in, and basic
1317 functionality on said platforms verified.
1318 * minor incompatible change: PCL now records the pathname of a file
1319 in which methods and the like are defined, rather than its
1321 * minor incompatible change: TRUENAME now considers the truename of
1322 a file naming a directory to be the pathname with :DIRECTORY
1323 component indicating that directory.
1324 * minor incompatible change: a NAMED clause in the extended form of
1325 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1326 reason for the previous behaviour is unclear.
1327 * more systematization and improvement of CLOS and MOP conformance
1328 in PCL (thanks to Gerd Moellman and Pierre Mai):
1329 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1331 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1332 primary methods with no specializers;
1333 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1335 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1336 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1337 FORWARD-REFERENCED-CLASSes; error reporting on
1338 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1340 ** SXHASH on CLOS instances now uses a slot internal to the
1341 instance to return different numbers on distinct instances,
1342 while preserving the same return value through invocations of
1344 ** DEFMETHOD signals errors when methods with longer incongruent
1345 lambda lists are added to generic functions;
1346 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1347 CLASS, as specified in AMOP;
1348 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1349 on the order returned by the primary method for classes of
1350 class STANDARD-CLASS;
1351 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1352 * fixed some bugs shown by Paul Dietz' test suite:
1353 ** DOLIST puts its body in TAGBODY;
1354 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1356 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1357 value producing form;
1358 ** if more variables are given to PROGV than values, extra
1359 variables are bound and made to have no value;
1360 ** NSUBSTITUTE on list arguments gets the right answer with
1362 ** ELT signals an error of type TYPE-ERROR when the index argument
1363 is not a valid sequence index;
1364 ** LOOP signals (at macroexpansion time) an error of type
1365 PROGRAM-ERROR when duplicate variable names are found;
1366 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1367 ** FUNCALL of special-operators now cause an error of type
1369 ** PSETQ now works as required in the presence of side-effecting
1370 symbol-macro places;
1371 ** NCONC accepts any object as its last argument;
1372 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1374 ** loop-for-as-package does not require a package to be explicitely
1376 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1378 * fixed bug 166: compiler preserves "there is a way to go"
1379 invariant when deleting code.
1380 * fixed bug 172: macro lambda lists with required arguments after
1381 &REST arguments now cause an error to be signalled. (thanks to
1383 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1384 bug. (thanks to Matthew Danish)
1385 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1387 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1388 arguments are lexically rebound. (thanks to Gerd Moellmann and
1390 * fixed bug 194: error messages are now more informative when there
1391 is no primary method applicable in a call to a generic
1392 function. (thanks to Gerd Moellmann)
1393 * fixed bug in command line argument checking (thanks to Julian
1395 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1396 (thanks to Matthew Danish)
1397 * incremented fasl file version number, because of the
1398 SXHASH-related changes in the layout of CLOS data structures
1400 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1401 * minor incompatible change: The runtime (the Unix executable named
1402 "sbcl") is now much pickier about the .core files it will load.
1403 Essentially it now requires .core files to descend from the same
1404 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1405 as the runtime does. (The intent is to prevent the crashes which
1406 can occur, and which can even be reported as mysterious failures,
1407 when people patch the sources or change the build parameters
1408 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1409 sbcl and .core files.)
1410 * fixed bug: VALUES-LIST is no longer optimized away.
1411 * fixed bug 142: The FFI conversion of C string values to Lisp
1412 string values no longer conses excessively. (thanks to Nathan
1413 Froyd porting Raymond Toy's fix to CMU CL)
1414 * began to systematize and improve MOP conformance in PCL (thanks to
1415 Nathan Froyd, Gerd Moellman and Pierre Mai):
1416 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1418 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1419 ** FINALIZE-INHERITANCE is now called on class finalization;
1420 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1421 argument precedence order.
1422 * fixed bug 202: The compiler no longer fails on functions whose
1423 derived types contradict their declared type.
1424 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1425 so it can be non-toplevel.
1426 * The fasl file version number has changed (because of the new
1427 implementation of DEFMACRO).
1428 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1429 safe code, that any length requirement by their type-specifier
1430 argument is valid. The exceptions to this are described in bug
1432 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1433 functions now check the types of their inputs as required by ANSI.
1434 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1435 introduced symbol is DECLAREd to be SPECIAL.
1436 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1437 * fixed bug: :COUNT argument to sequence functions may be negative
1438 * fixed bug: body of DO-SYMBOLS may contain declarations
1439 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1440 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1441 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1442 symbol macro only once
1443 * fixed printing of call frame when argument list is unavailable
1444 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1445 * compiler no longer signals WARNING on unknown keyword
1448 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1449 * A beta-quality port to the mips architecture running Linux,
1450 based on the old CMUCL backend, has been made. It has been tested
1451 on a big-endian kernel, and works sufficiently well to be able to
1452 rebuild itself; it has not been tested in little-endian mode.
1453 * fixed an inconsistency between gencgc.c and purify.c which made
1454 dumping/loading .core files unreliable
1455 * fixed bug 120a: The compiler now deals correctly with IFs where
1456 the consequent is the same as the alternative, instead of
1457 misderiving the return type. (thanks to Alexey Dejneka)
1458 * fixed bug 113: Logical pathnames are now dumpable (the logical
1459 host is resolved at load-time, throwing an error if it is not
1461 * fixed bug 174: FORMAT's error message is slightly clearer when a
1462 non-printing character is used in a format directive.
1463 * fixed several bugs in compiler checking of type declarations, i.e.
1464 violations of the Python "declarations are assertions" principle
1465 (thanks to Alexey Dejneka)
1466 * fixed several bugs in PCL's error checking (thanks to Gerd
1468 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1470 * fixed bug in compilation of functions as first class values
1471 (thanks to Antonio Martinez-Shotton)
1472 * The compiler's handling TYPE-ERRORs which it can prove will
1473 inevitably happen at runtime has been cleaned up and corrected
1474 in several ways. (thanks to Alexey Dejneka)
1475 * improved argument type checking for various basic arithmetic
1476 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1477 lavished on them in the past that they can be compiled in many
1478 ways in different special cases
1479 * fixed bug 181: compiler checks validity of user supplied type
1481 * cleaned up code flushing in optimization: Function calls which
1482 should signal errors for safety purposes (e.g. which ANSI says
1483 should signal errors when their arguments are of incorrect type)
1484 are no longer optimized away.
1485 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1486 * incremented fasl file version number, because changes in the
1487 implementation of sequence functions like COERCE caused
1488 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1489 in old inline expansions) to become undefined. (Actually these
1490 changes were later undone, so we might very well be binary
1491 compatible with 0.7.7 after all, but leaving the version number
1492 incremented seemed like the simplest and most conservative
1495 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1496 * An alpha-quality port to the parisc architecture running Linux,
1497 based on the old CMUCL backend, has been made. This, even more so
1498 than the other backends, should be considered still a work in
1499 progress; known problems include that the Linux kernel in 64-bit
1500 mode does not propagate the correct sigcontext structure to
1501 userspace, and consequently SBCL on a parisc64 kernel will not
1503 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1504 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1505 suppressed.) Also now that the compiler is looking at declarations
1506 in the environment, it checks optimization declarations as well,
1507 and suppresses inlining when (> DEBUG SPEED).
1508 * More fixes have been made to treatment of floating point exception
1509 treatment and other Unix signals. In particular, floating point
1510 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1511 * The detection and handling of control stack exhaustion (infinite
1512 or very deeply nested recursion) has changed. Stack exhaustion
1513 detection is now done by write-protecting pages at the OS level
1514 and applies at all optimization settings; when found, a
1515 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1516 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1517 be caught using IGNORE-ERRORS.
1518 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1519 correctly when transforming recursive LABELS functions to LETs.
1520 (thanks to Alexey Dejneka)
1521 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1522 that are names of constants or global variables.
1523 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1524 alien routines with docstrings.
1525 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1526 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1528 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1529 (thanks to Eric Marsden)
1530 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1531 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1532 * Bug fix: The SPARC backend can now compile functions involving
1533 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1534 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1535 to a functions expecting a PATHNAME-DESIGNATOR.
1536 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1537 lambda lists. (thanks to Alexey Dejneka)
1538 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1539 functions so that EXPORTing the name of the class doesn't cause
1540 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1541 to Antonio Martinez for reporting this)
1542 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1543 constant-folded without failing an assertion. (thanks to Einar
1544 Floystad Dorum for reporting this)
1545 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1546 types are now dealt with more correctly by the compiler.
1547 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1548 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1549 behaviour analogous to RENAME-FILE. This puts its behaviour more
1550 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1552 * The fasl file version number has changed again. (because of the
1553 bug fix involving the names of PCL MAKE-INSTANCE functions)
1555 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1556 * bug fix: Floating point exceptions are treated much more
1557 consistently on the x86/Linux and PPC/Linux platforms.
1558 * Array initialization with :INITIAL-ELEMENT is now much faster for
1559 cases when the compiler cannot open code the array creation, but
1560 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1561 array accesses have also seen a speed increase.
1562 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1563 specified. (This was at the root of some bad interactions between
1564 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1566 * bug fix: Internal error arguments for undefined functions are now
1567 computed correctly on the PPC/Linux platform.
1568 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1569 Raymond Toy's patch for CMU CL)
1570 * Support for the Solaris 9 operating environment has been included
1571 (thanks to Daniel Merritt)
1572 * A very ugly but hopefully complete draft of the missing FFI chapter
1573 of the manual has been created by reformatting the corresponding
1574 CMU CL manual chapter into (currently very ugly and incoherent)
1575 DocBook and bringing it up to date for SBCL behavior. Thus, the
1576 manual is now essentially complete, at least by my extreme
1577 once-and-only-once standards, whereby it's acceptable to refer to
1578 the doc strings of SB-EXT functions as the primary documentation.
1579 * The fasl file version number has changed again, due to cleanup of
1580 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1581 is no longer a static symbol.)
1583 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1584 * SBCL now builds with OpenMCL (version 0.12) as the
1585 cross-compilation host; also, more progress has been made toward
1586 bootstrapping under CLISP.
1587 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1589 * bug 158 fixed: The compiler can now deal with integer loop
1590 increments different from 1; fixing this turned out also to fix
1592 * bug 169 fixed: no more bogus warnings about using lexical bindings
1593 despite the presence of perfectly good SPECIAL declarations (thanks
1594 to David Lichteblau)
1595 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1596 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1597 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1599 * bug 180 fixed: Method combination specifications no longer ignore
1600 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1601 * bug fix: Structure type predicate functions now check their argument
1602 count as they should.
1603 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1604 correctly. (thanks to Pierre Mai)
1605 * minor incompatible change: The --noprogrammer option is deprecated
1606 in favor of the new --disable-debugger option, which is very similar.
1607 (The major difference is that it takes effect at a slightly different
1608 time at startup, causing handling of errors in --sysinit and
1609 --userinit files will be affected differently.) The
1610 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1611 been added to allow this functionality to be controlled from ordinary
1612 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1613 maintainers, who might want to run non-interactive scripts to
1614 build SBCL cores which will later be used interactively.)
1615 * minor incompatible change: The LOAD function no longer, when given
1616 a wild pathname to load, loads all files matching that pathname.
1617 Instead, an error of type FILE-ERROR is signalled.
1619 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1620 * bug 147 fixed: The compiler preserves its block link/count
1621 invariants more correctly now so that it doesn't crash. (thanks
1623 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1625 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1626 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1627 no-value-for-this-character as NIL (as specified by ANSI).
1628 * HOST-NAMESTRING on physical pathnames now returns a string that is
1629 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1630 (thanks to Christophe Rhodes)
1631 * The Alpha port handles icache flushing more correctly. (thanks to
1633 * More progress has been made toward bootstrapping under CLISP. (thanks
1634 to Christophe Rhodes)
1635 * The fasl file format has changed again, because dynamic loading
1636 on OpenBSD (which has non-ELF object files) motivated some cleanups
1637 in the way that foreign symbols are transformed and passed around.
1638 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1639 is no longer treated as whitespace by the reader, but instead as
1640 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1641 |A
\7fB|, instead of A as it used to.
1643 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1644 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1645 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1646 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1647 this as of 0.7.1.45, but was left out of the previous news section
1648 (thanks to Dan Barlow)
1649 * SBCL now runs on the Solaris operating system on SPARC architectures
1650 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1651 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1652 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1653 * SPARC backend cleanups, allowing builds of cores optimized for V8
1654 and V9 SPARCS, and also emission of code targeted to a particular
1655 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1657 * SBCL is closer to bootstrapping under CLISP, thanks to various
1658 fixes by Christophe Rhodes.
1659 * The fasl file format has changed again, to allow the compiler's
1660 INFO database to support symbol macros.
1661 * The user manual (in doc/) is formatted into HTML more nicely.
1662 (thanks to coreythomas)
1663 * The system is smarter about SUBTYPEP relationships, especially
1664 those involving NOT types (including types such as ATOM which are
1665 represented internally using NOT types). Thus SUBTYPEP is less
1666 likely to return (VALUES NIL NIL) in general, and in particular
1667 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1669 * The fasl file format has changed again, because the internal
1670 representation of types now includes a new slot to support the new
1671 SUBTYPEP-of-NOT-types logic.
1672 * (not a change in the main branch of SBCL, but a related prototype
1673 which can hopefully be merged into the main branch of SBCL in the
1674 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1675 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1676 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1677 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1678 dispatch macro character combinations. (thanks to Alexey Dejneka)
1679 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1680 that superficially look like logical namestrings correctly.
1681 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1683 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1684 default supplied-p) for &optional and &key arguments. (thanks to
1687 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1688 * incompatible change: The compiler is now less aggressive about
1689 tail call optimization, doing it only when (> SPACE DEBUG) or
1690 (> SPEED DEBUG). (This is an incompatible change because there are
1691 programs which relied on the old CMU-CL-style behavior to optimize
1692 away their unbounded recursion which will now die of stack overflow.)
1693 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1694 for non-GENCGC systems has been increased to 20M (since that
1695 seems much closer to the likely performance optimum for modern
1696 systems than the old 4M value was)
1697 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1698 and *DEBUG-PRINT-LEVEL*
1699 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1700 of CMU CL's support for SPARC, and various endianness and other
1701 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1702 * new syntactic sugar for the Unix command line: --load foo.bar is now
1703 an alternate notation for --eval '(load "foo.bar")'.
1705 ** The system now detects stack overflow and handles it gracefully,
1706 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1707 optimization settings. (This is a good thing in general, and
1708 its introduction in this version should be particularly timely
1709 for anyone whose code fails because of suppression of tail
1711 ** The system now hunts for the C variable "environ" in a more
1712 devious way, to avoid segfaults when the C library version differs
1713 between compile time and run time. (thanks to Christophe Rhodes)
1714 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1715 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1716 ** The compiler no longer issues bogus style warnings for undefined
1717 classes in the same source file as the DEFCLASSes which defined
1718 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1720 ** fixes in CONDITION class precedence list for undefined function
1721 errors (thanks to Alexei Dejneka)
1722 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1723 correctly. (thanks to Dan Barlow)
1724 ** portability fixes aiming at bootstrapping under CLISP (thanks
1725 to Dave McDonald and Christophe Rhodes)
1726 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1727 ** fixes in type translation and and type inference (thanks to
1729 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1730 ** various fixes in the new ports (thanks to Dan Barlow)
1731 * several changes related to debugging:
1732 ** suppression of tail recursion, as noted above
1733 ** stack overflow detection, as noted above
1734 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1735 is now the default. (For some time encapsulation has been more
1736 reliable than the breakpoint-based :ENCAPSULATE NIL
1737 implementation, at least on X86 systems; and I just noticed that
1738 encapsulation also seems closer to the spirit of the ANSI
1741 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1743 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1744 up properly again. (There was a packaging bug in 0.7.0 which
1745 left their definitions in SB-SYS::LOAD-FOREIGN and
1746 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1747 vital for most things which interface to C-level interfaces,
1748 like extensions working with sockets or databases or
1749 Perl-compatible regexes or whatever, and the need to fix
1750 this bug is the main reason that 0.7.1 was released so
1752 ** DEFGENERIC is now choosier about the methods it redefines, so that
1753 reLOADing a previously-LOADed file containing DEFGENERICs does
1754 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1755 cycle now works as it should. (thanks to Alexey Dejneka)
1756 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1757 fixed by Christophe Rhodes. (He actually submitted this patch
1758 months ago, and I delayed until after 0.7.0.)
1759 ** Bug 111 (internal compiler confusion about runtime checks on
1760 FUNCTION types) was fixed by Alexey Dejneka.
1761 * Some internal cleanups (getting rid of variables which aren't
1762 needed now that the byte interpreter is gone) caused the fasl
1763 file format number to change again.
1765 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1766 * major incompatible change: The default fasl file extension, i.e. the
1767 default extension for files produced by COMPILE-FILE, has changed
1768 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1770 ** There are many changes in the implementation of the compiler.
1771 SBCL is now essentially a compiler-only implementation of ANSI
1772 Common Lisp. EVAL still "interprets" a few special cases, but
1773 almost all the interesting cases are handled by creating
1774 a LAMBDA expression, calling COMPILE on it, then calling
1775 FUNCALL on the result.
1776 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1777 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1778 Since the code is newer, there might still be some new bugs
1779 (though not as many as before Martin Atzmueller's fixes:-). But
1780 the new code is substantially simpler and clearer, and hopefully
1781 any remaining bugs will be simpler, less fundamental, and more
1782 fixable then the bugs in the old code.
1783 ** The revised compiler is still a little unsteady on its feet.
1785 *** The debugging information it produces (particularly the names
1786 of FUNCTION objects) is sometimes much less useful than what
1787 the old compiler produced.
1788 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1789 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1790 within a MACROLET) has been temporarily weakened.
1791 ** There are new compiler optimizations for various functions:
1792 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1793 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1794 *** the math functions TRUNCATE, FLOOR, and CEILING
1795 *** the function-of-all-trades COERCE
1796 Mostly these should be transparent, but there's one
1797 potentially-annoying problem (bug 117): when the compiler
1798 inline-expands a function and does type analysis on the result,
1799 it can create control paths which have type mismatches, and
1800 when it can't prove that those control paths aren't taken,
1801 it will issue WARNINGs about the type mismatches. This is
1802 a particular problem in practice for the new sequence functions.
1803 It's not clear how this should be fixed, and for now, a
1804 workaround is given in the entry for 117 in the BUGS file.
1805 ** (Because of the interaction between the two previous items --
1806 occasional inlining problems and new inline expansions -- some
1807 of the new sequence function optimizations won't really kick in
1808 completely until debugging information, and then inlining, are
1809 straightened out in some future version.)
1810 * minor incompatible changes:
1811 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1812 DIRECTORY no longer implicitly promotes NIL slots of its
1813 pathname argument to :WILD. In particular, when you ask for the
1814 contents of a directory (which you used to be able to do without
1815 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1816 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1817 ** changes in behavior that ANSI explicitly defines to be
1818 implementation dependent:
1819 *** The new compiler-only implementation still conforms with ANSI,
1820 but acts a little different than before. Besides the obvious
1821 changes in performance tradeoffs (that the cost per form passed
1822 to EVAL has gone up, and the cost per form executed by EVAL
1823 has gone down), the behavior of the system changes a little
1824 because there are no longer any interpreted function objects.
1825 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1826 e.g. doing COMPILE on the output of interactive DEFUN is
1828 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1830 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1831 has changed from (:CL) to NIL.
1832 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1833 e.g. #\Newline and #\Tab, don't have names specified in the
1834 ANSI Common Lisp standard, is now based on their ASCII symbolic
1835 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1836 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1837 are no longer used for output.
1838 ** changes in internal implementation constants:
1839 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1840 4 million. (If your application spends a lot of time GCing and
1841 you have a lot of RAM, you might want to experiment with
1842 increasing it even more.)
1843 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1844 However, almost all old code should still continue to work without
1845 immediate update, as SB-C-CALL is now a (deprecated) nickname
1847 ** Old operator names in the style DEF-FOO are now deprecated in
1848 favor of new corresponding names DEFINE-FOO, for consistency with
1849 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1850 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1851 internal symbols, but a few supported extensions like
1852 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1853 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1854 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1855 etc. as you get deeper into recursive calls to the debugger
1856 command loop, instead of the old "5]", "5]]", "5]]]"
1857 sequence. (I was motivated to do this when squabbles between
1858 ILISP and SBCL left me very deeply nested in the debugger. In the
1859 short term, this change will probably provoke more ILISP/SBCL
1860 squabbles, but hopefully it will be an improvement in the long run.)
1861 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1862 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1863 function, since declaiming return types involving aliens is
1864 (1) annoyingly messy to do by hand and (2) vital to efficient
1865 compilation of code which calls such functions.
1866 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1867 longer reexported by the SB-EXT package. They're solely useful
1868 for alien code, so it seems more logical that you should get
1869 them from the SB-ALIEN package, not in SB-EXT.
1870 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1871 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1872 features. Instead, the code that they used to control is always
1873 built into the system.
1874 * many other bug fixes
1875 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1876 advantage of the new EVAL-WHEN stuff and to clean them up in
1877 general, and they are now more ANSI-compliant in a number of
1878 ways. Martin Atzmueller is responsible for a lot of this.
1879 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1881 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1882 *** correct ERROR type for various file operations
1883 *** some fixes for Lisp streams
1884 *** DEFMETHOD syntax checking
1885 *** changing old weird representation of debug information as
1886 strings (which, among their other deficiencies, don't transform
1887 correctly when you rename packages, and don't change their
1888 print representation when you change things like *PACKAGE*
1889 and *PRINT-LENGTH*) to symbols and lists of symbols
1890 He also made several improvements and fixed several bugs in DESCRIBE.
1891 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1893 *** misbehavior of WRITE-STRING/WRITE-LINE
1894 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1895 and several other LOOP problems as well
1896 *** DIRECTORY when similar filenames are present
1897 *** DEFGENERIC with :METHOD options
1898 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1899 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1900 *** argument ordering in FIND with :TEST option
1901 *** mishandled package designator argument in APROPOS-LIST
1902 *** various problems in the backquote readmacro
1903 *** a bug in APROPOS
1904 *** probably some others that I'm not describing very well here,
1905 since the CVS log documents them by reference to sbcl-devel
1906 messages, and the SourceForge archives aren't working well.:-(
1907 ** Dan Barlow improved the Alpha port (and is making progress on the
1908 PPC port, for those of you who think different).
1909 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
1910 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
1911 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
1912 continues to work on the SPARC port (for those of you in a position
1913 to look down upon our little PC-compatible boxes from a great height).
1914 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
1915 and uses that copy, rather than the current dynamic value, when
1916 it's trying to decide whether to truncate output. Thus e.g.
1917 (let ((*print-lines* 50))
1918 (pprint-logical-block (stream nil)
1920 (let ((*print-lines* 8))
1921 (print (aref possiblybigthings i) stream)))))
1922 should now truncate the logical block only at 50 lines, instead of
1923 often truncating it at 8 lines, as it did before.
1924 * The doc/cmucl/ directory, containing old CMU CL documentation
1925 from the time of the fork, is no longer part of the base system.
1926 SourceForge has shut down its anonymous FTP service, and with it
1927 my original plan for distributing the old CMU CL documentation
1928 there. For now, if you need these files you can download an old
1929 SBCL source release and extract them from it.
1930 * The fasl file version number changed again, for dozens of reasons,
1931 some of which are apparent above.
1933 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
1934 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
1935 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
1936 patch, so that DISASSEMBLE output is much nicer.
1937 * The code in the SB-PROFILE package now seems reasonably stable.
1938 I still haven't decided what the final interface should look like
1939 (I'd like PROFILE to interact cleanly with TRACE, since both
1940 facilities use function encapsulation) but if you have a need
1941 for profiling now, you can probably use it successfully with
1942 the current CMU-CL-style interface.
1943 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
1944 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
1945 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
1946 link now returns the dangling link itself, and for similar
1947 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
1948 link itself. (In these cases the old code signalled an error and
1949 looped endlessly, respectively.) Thus, DIRECTORY now works even
1950 in the presence of dangling and cyclic symbolic links.
1951 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
1952 is now a supported extension again, since the consensus on
1953 sbcl-devel was that it can be useful for ordinary development
1954 work, not just for debugging SBCL itself.
1955 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
1956 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
1957 that currently-defined functions might be redefined later with
1958 different return types.
1959 * Hash tables can be printed readably, as inspired by CMU CL code
1960 of Eric Marsden and SBCL code of Martin Atzmueller.
1961 * better error handling in CLOS method combination, thanks to
1962 Martin Atzmueller porting Pierre Mai's CMU CL patches
1963 * more overflow fixes for >16Mbyte I/O buffers
1964 * A bug in READ has been fixed, so that now a single Ctrl-D
1965 character suffices to cause end-of-file on character streams.
1966 In particular, now you only need one Ctrl-D at the command
1967 line (not two) to exit SBCL.
1968 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
1970 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
1971 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
1972 does the right thing.
1973 * The implementation of some type tests, especially for CONDITION
1974 types, is now tidier and maybe faster, due to CMU CL code
1975 originally by Douglas Crosher, ported by Martin Atzmueller.
1976 * Some math functions have been fixed, and there are new
1977 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
1978 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
1979 * (There are also some new optimizers in contrib/*-extras.lisp. Those
1980 aren't built into sbcl-0.6.13, but are a sneak preview of what's
1981 likely to be built into sbcl-0.7.0.)
1982 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
1983 CL, ported to SBCL by Martin Atzmueller)
1984 * DESCRIBE now gives more information in some cases. (Pierre Mai's
1985 patch to CMU CL, ported to SBCL by Martin Atzmueller)
1986 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
1987 * There's a new slam.sh hack to shorten the edit/compile/debug
1988 cycle for low-level changes to SBCL itself, and a new
1989 :SB-AFTER-XC-CORE target feature to control the generation of
1990 the after-xc.core file needed by slam.sh.
1991 * minor incompatible change: The ENTRY-POINTS &KEY argument to
1992 COMPILE-FILE is no longer supported, so that now every function
1993 gets an entry point, so that block compilation looks a little
1994 more like the plain vanilla ANSI section 3.2.2.3 scheme.
1995 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
1996 returns the number of bytes consed since the system started,
1997 rather than the number consed since the first time the function
1998 was called. (The new definition parallels ANSI functions like
1999 CL:GET-INTERNAL-RUN-TIME.)
2000 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2001 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2002 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2003 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2004 * Fasl file version numbers are now independent of the target CPU,
2005 since historically most system changes which required version
2006 number changes have affected all CPUs equally. Similarly,
2007 the byte fasl file version is now equal to the ordinary
2010 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2011 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2012 is no longer recognized. I apologize for this, because it was
2013 listed in SB-EXT as a supported extension, but I found that
2014 its existing behavior was poorly specified, as well as incorrectly
2015 specified, and it looked like too much of a mess to straighten it
2016 out. I have enough on my hands trying to get ANSI stuff to work..
2017 * many patches ported from CMU CL by Martin Atzmueller, with
2018 half a dozen bug fixes in pretty-printing and the debugger, and
2019 half a dozen others elsewhere
2020 * fixed bug 13: Floating point infinities are now supported again.
2021 They might still be a little bit flaky, but thanks to bug reports
2022 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2023 as flaky as they were.
2024 * The --noprogrammer command line option is now supported. (Its
2025 behavior is slightly different in detail from what the old man
2026 page claimed it would do, but it's still appropriate under the
2027 same circumstances that the man page talks about.)
2028 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2029 are now supported, and enabled by default. Thus, the compiler can
2030 handle many floating point and complex operations much less
2031 inefficiently. (Thus e.g. you can implement a complex FFT
2033 * The compiler now detects type mismatches between DECLAIM FTYPE
2034 and DEFUN better, and implements CHECK-TYPE more correctly, and
2035 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2036 thanks to patches from Martin Atzmueller.
2037 * various fixes to make the cross-compiler more portable to
2038 ANSI-conforming-but-different cross-compilation hosts (notably
2039 Lispworks for Windows, following bug reports from Arthur Lemmens)
2040 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2041 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2042 fixed thanks to Pierre Mai's CMU CL patches.
2043 * fixes to keep the system from overflowing internal counters when
2044 it tries to use i/o buffers larger than 16M bytes
2045 * fixed bug 45a: Various internal functions required to support
2046 complex special functions have been merged from CMU CL sources.
2047 (When I was first setting up SBCL, I misunderstood a compile-time
2048 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2049 * improved support for type intersection and union, fixing bug 12
2050 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2051 more obscure bugs as well
2052 * some steps toward byte-compiling non-performance-critical
2053 parts of the system, courtesy of patches from Martin Atzmueller
2054 * Christophe Rhodes has made some debian packages of sbcl at
2055 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2056 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2057 stable, but are nonetheless usable. When he's ready, I'd be happy
2058 to add them to the SourceForge "File Releases" section. (And if
2059 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2060 * new fasl file format version number (because of changes in
2061 internal representation of (OR ..) types to accommodate the new
2062 support for (AND ..) types, among other things)
2064 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2065 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2067 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2069 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2070 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2071 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2072 * DESCRIBE now works on structure objects again.
2073 * Most function call argument type mismatches are now handled as
2074 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2075 know whether the function will be redefined before the call is
2076 executed. (The compiler could flag local calls with full WARNINGs,
2077 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2078 it doesn't keep track of enough information to know whether calls
2079 are local in this sense.)
2080 * Compiler output is now more verbose, with messages truncated
2081 later than before. (There should be some supported way for users
2082 to override the default verbosity, but I haven't decided how to
2083 provide it yet, so this behavior is still controlled by the internal
2084 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2085 src/compiler/ir1util.lisp.)
2086 * Fasl file format version numbers have increased again, because
2087 support for the Gray streams extension changes the layout of the
2088 system's STREAM objects.
2089 * The Gray subclassable streams extension now works, thanks to a
2090 patch from Martin Atzmueller.
2091 * The full LOAD-FOREIGN extension (not just the primitive
2092 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2093 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2094 but like most other programs, it defaults to copying the Unix
2095 environment from the original process instead of starting the
2096 new process in an empty environment.
2097 * Extensions which manipulate the Unix environment now support
2098 an :ENVIRONMENT keyword option which doesn't smash case or
2099 do other bad things. The CMU-CL-style :ENV option is retained
2100 for porting convenience.
2101 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2102 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2104 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2106 * A patch from Martin Atzmueller seems to have solved the SIGINT
2107 problem, and as far as we know, signal-handling now works cleanly.
2108 (If you find any new bugs, please report them!)
2109 * The system no longer defaults Lisp source file names to types
2110 ".l", ".cl", or ".lsp", but only to ".lisp".
2111 * The compiler no longer uses special default file extensions for
2112 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2113 seems to expect a single default extension for all compiled code,
2114 and there's no compelling reason to try to stretch the standard
2115 to allow two different extensions.) Instead, byte-compiled files
2116 default to the same extension as native-compiled files.
2117 * Fasl file format version numbers have increased again, because
2118 a rearrangement of internal implementation packages made some
2119 dumped symbols in old fasl files unreadable in new cores.
2120 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2121 many fewer weird special cases.
2122 * Bug #17 (differing COMPILE-FILE behavior between logical and
2123 physical pathnames) has been fixed, and some related misbehavior too,
2124 thanks to a patch from Martin Atzmueller.
2125 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2126 by Tim Moore, ported to SBCL by Martin Atzmueller.
2127 * Martin Atzmueller fixed several filesystem-related problems,
2128 including bug #36, in part by porting CMU CL patches, which were
2129 written in part by Paul Werkowski.
2130 * More compiler warnings in src/runtime/ are gone, thanks to
2131 more patches from Martin Atzmueller.
2132 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2135 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2137 * DESCRIBE now works on CONDITION objects.
2138 * The debugger now handles errors which arise when trying to print
2139 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2141 * The build system now uses an additional file, customize-target-features.lisp,
2142 to allow local modifications to the target *FEATURES* list. (The point of
2143 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2144 debugging features enabled, without having to worry about propagating it
2145 into everyone's system when I do a "cvs update".) When no
2146 customize-target-features.lisp file exists, the target *FEATURES* list
2147 should be constructed the same way as before.
2148 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2149 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2150 warn about behavior which is undefined under the ANSI spec. Note: This
2151 is specified by ANSI, but it's not very popular with programmers.
2152 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2153 macro in the SBCL sources for an example of a workaround which you
2154 might use to make portable ANSI-standard code which does what you want.
2155 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2156 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2157 when it's used as a non-toplevel form. (This is required in order
2158 to implement the DEFCONSTANT-EQX macro.)
2159 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2160 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2161 funny things, due to bugs in the implementation of EVAL-WHEN
2162 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2163 non-toplevel DEFCONSTANTs are uncommon.)
2164 * The core file version number and fasl file version number have been
2165 incremented, because the old noncompliant DEFCONSTANT behavior involved
2166 calling functions which no longer exist, and because I also took the
2167 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2168 * fixed bug 1 (error handling before read-eval-print loop starts), and
2169 redid debugger restarts and related debugger commands somewhat while
2171 ** The QUIT debugger command is gone, since it did something
2172 rather different than the SB-EXT:QUIT command, and since it never
2173 worked properly outside the main toplevel read/eval/print loop.
2174 Invoking the new TOPLEVEL restart provides the same functionality.
2175 ** The GO debugger command is also gone, since you can just invoke
2176 the CONTINUE restart directly instead.
2177 ** The TOP debugger command is also gone, since it's redundant with the
2178 FRAME 0 command, and since it interfered with abbreviations for the
2180 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2182 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2183 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2184 some porting work by Martin Atzmueller)
2185 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2186 with a boatload of patches which he ported to SBCL. Now that those
2188 ** The system tries to make sure that its low-priority messages
2189 are prefixed by semicolons, to help people who like to use
2190 syntax highlighting in their ILISP buffer. (This patch
2191 was originally due to Raymond Toy.)
2192 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2193 patches originally written by Raymond Toy.
2194 ** The compiler understands coercion between single-value and
2195 multiple-VALUES type expressions better, getting rid of some very
2196 weird behavior, thanks to patches originally by Robert MacLachlan
2197 and Douglas Crosher.
2198 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2199 lambda lists, thanks to a patch originally by Pierre Mai.
2200 ** The system no longer bogusly warns about "abbreviated type
2202 ** The compiler gets less confused by inlining and RETURN-FROM,
2203 thanks to some patches originally by Tim Moore.
2204 ** The system no longer hangs when dumping circular lists to fasl
2205 files, thanks to a patch originally from Douglas Crosher.
2206 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2207 undefined function error.
2208 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2209 in structure slot initforms) for now, documented workaround instead:-|
2210 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2211 * fixed bug 5: added stubs for various Gray stream functions called
2212 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2213 installed, at least appropriate type errors are generated
2214 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2215 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2216 consistently in DEFMETHOD forms.
2217 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2218 it doesn't seem to affect SBCL after all
2219 * The C runtime system now builds with better optimization and many
2220 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2222 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2224 * The system is now under CVS at SourceForge (instead of the
2225 CVS repository on my home machine).
2226 * The new signal handling code has been tweaked to treat register
2227 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2228 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2229 downstream, e.g. in debug-int.lisp, has implicit dependencies
2230 on the unsignedness of integer representation of machine words,
2231 and that caused the system to bomb out with infinite regress
2232 when trying to recover from type errors involving signed values,
2233 e.g. (BUTLAST '(1 2 3) -1).)
2234 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2235 (This was one of the bugs Peter Van Eynde reported back in July.)
2236 * The system now uses code inspired by Colin Walters' O(N)
2237 implementation of MAP (from the cmucl-imp@cons.org mailing
2238 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2239 inline the MAP operation, and there is more than one
2240 sequence argument to the MAP call (so that it can't just
2241 do ETYPECASE once and for all based on the type of the
2242 single sequence argument). (The old non-inline implementation
2243 of the general M-argument sequence-of-length-N case required
2244 O(M*N*N) time when any of the sequence arguments were LISTs.)
2245 * The QUIT :UNIX-CODE keyword argument has been renamed to
2246 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2247 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2249 * Raymond Wiker's patches to port dynamic loading from Linux to
2250 FreeBSD have been added.
2251 * The BUGS file is now more nearly up to date, thanks in large part
2252 to Martin Atzmueller's review of it.
2253 * The debugger now flushes standard output streams before it begins
2254 its output ("debugger invoked" and so forth).
2255 * The core version number and fasl file version number have both
2256 been incremented, because of incompatible changes in the layout
2258 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2259 used to not be called for a saved Lisp image.)
2260 * Martin Atzmueller's version of a patch to fix a compiler crash,
2261 as posted on sbcl-devel 13 September 2000, has been installed.
2262 * Instead of installing Martin Atzmueller's patch for the
2263 compiler transform for SUBSEQ, I deleted the compiler transform,
2264 and transforms for some similar consing operations.
2265 * A bug in signal handling which kept TRACE from working on OpenBSD
2267 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2268 away by constant folding
2269 * The system now defines its address space constants in one place
2270 (in the Lisp sources), and propagates them automatically elsewhere
2271 (through GENESIS and the sbcl.h file). Therefore, patching the
2272 address map is less unnecessarily tedious and error-prone. The
2273 Lisp names of address space constants have also been systematized.
2274 * CVS tags like dollar-Header-dollar have been removed from
2275 the sources, because they have never saved me trouble and
2276 they've been source of trouble working with patches and other
2277 diff-related operations.
2278 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2279 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2281 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2283 * The system has been ported to OpenBSD.
2284 * The system now compiles with a simple "sh make.sh" on the systems
2285 that it's supported on. I.e., now you no longer need to tweak
2286 text in the target-features.lisp-expr and symlinks in src/runtime/
2287 by hand, the make.sh takes care of it for you.
2288 * The system is no longer so grossly inefficient when compiling code
2289 involving vectors implemented as general (not simple) vectors (VECTOR T),
2290 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2291 which dares to use the various sequence functions on non-simple
2292 vectors, takes less of a performance hit.
2293 * There is now a primitive type predicate VECTOR-T-P
2294 to test for the (VECTOR T) type, so that e.g.
2295 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2296 can now be compiled with some semblance of efficiency. (The old code
2297 turned the type declaration into a full call to %TYPEP at runtime!)
2298 * AREF on (VECTOR T) is still not fast, since it's still compiled
2299 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2300 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2301 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2302 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2303 without making full calls to SUBTYPEP at runtime.
2304 (Some analogous efficiency issues for non-simple vectors specialized to
2305 element types other than T, or for non-simple multidimensional arrays,
2306 have not been addressed. They could almost certainly be handled the
2307 same way if anyone is motivated to do so.)
2308 * The changes in array handling break binary compatibility, so
2309 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2310 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2311 instead of (VALUES T T).
2312 * By following the instructions that Dan Barlow posted to sbcl-devel
2313 on 2 July 2000, I was able to enable primitive dynamic object
2314 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2315 functionality is not implemented (since it calls ld to resolve
2316 library references automatically, requiring RUN-PROGRAM for its
2317 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2318 try to resolve library references) is now supported.
2319 * The system now flushes the standard output streams when it terminates,
2320 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2321 them at several other probably-convenient times, e.g. in each pass of
2322 the toplevel read-eval-print loop, and after evaluating a form given
2323 as an "--eval" command-line option. (These changes were motivated by a
2324 discussion of stream flushing issues on cmucl-imp in August 2000.)
2325 * The source transform for TYPEP of array types no longer assumes
2326 that an array whose element type is a not-yet-defined type
2327 is implemented as an array of T, but instead punts, so that the
2328 type will be interpreted at runtime.
2329 * There is now some support for cross-compiling in make.sh: each of
2330 the phases of make.sh has its own script. (This should be transparent
2331 to people doing ordinary, non-cross-compile builds.)
2332 * Since my laptop doesn't have hundreds of megabytes of memory like
2333 my desktop machine, I became more motivated to do some items on
2334 my to-do list in order to reduce the size of the system a little:
2335 ** Arrange for various needed-only-at-cold-init things to be
2336 uninterned after cold init. To support this, those things have
2337 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2338 symbols with such names are now uninterned after cold init).
2339 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2340 fasl files for cold load.
2341 ** Remove the old compiler structure pooling code (which used to
2342 be conditional on the target feature :SB-ALLOC) completely.
2343 ** Redo the representation of some data in cold init to be more compact.
2344 (I also looked into supporting byte compiled code at bootstrap time,
2345 which would probably reduce the size of the system a lot, but that
2346 looked too complicated, so I punted for now.)
2347 * The maximum signal nesting depth in the src/runtime/ support code has
2348 been reduced from 4096 to 256. (I don't know any reason for the very
2349 large old value. If the new smaller value turns out to break something,
2350 I'll probably just bump it back up.)
2351 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2353 * Many, many bugs reported by Peter Van Eynde have been added to
2354 the BUGS list; some have even been fixed.
2355 * While enabling dynamic object file loading, I tried to make the
2356 code easier to understand, renaming various functions and variables
2357 with less ambiguous names, and changing some function calling
2358 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2359 * While trying to figure out how to do the OpenBSD port, I tried to
2360 clean up some of the code in src/runtime/. In particular, I dropped
2361 support for non-POSIX signal handling, added various comments,
2362 tweaked the code to reduce the number of compilation warnings, and
2363 renamed some files to increase consistency.
2364 * To support the new automatic configuration functionality in make.sh,
2365 the source file target-features.lisp-expr has been replaced with the
2366 source file base-target-features.lisp-expr and the machine-generated
2367 file local-target-features.lisp-expr.
2368 * fixed a stupid quoting error in make.sh so that using CMU CL
2369 "lisp -batch" as cross-compilation host works again
2371 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2373 * DESCRIBE no longer tries to call itself recursively to describe
2374 bound/fbound values, so that it no longer fails on symbols which are
2375 bound to themselves (like keywords, T, and NIL).
2376 * DESCRIBE now works on generic functions.
2377 * The printer now prints less-screwed-up representations of closures
2378 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2379 * A private symbol is used instead of the :EMPTY keyword previously
2380 used to mark empty slots in hash tables. Thus
2381 (DEFVAR *HT* (MAKE-HASH-TABLE))
2382 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2383 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2384 now does what ANSI says that it should. (You can still get
2385 similar noncompliant behavior if bang on the hash table
2386 implementation with all the symbols you get back from
2387 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2388 This breaks binary compatibility, since tests for equality to
2389 :EMPTY are wired into things like the macroexpansion of
2390 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2392 * There's now a minimal placeholder implementation for CL:STEP,
2393 as required by ANSI.
2394 * An obscure bug in the interaction of the normal compiler, the byte
2395 compiler, inlining, and structure predicates has been patched
2396 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2397 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2399 * Missing ordinary arguments in a macro call are now detected even
2400 when the macro lambda list contains &KEY or &REST.
2401 * The debugger no longer complains about encountering the top of the
2402 stack when you type "FRAME 0" to explicitly instruct it to go to
2403 the top of the stack. And it now prints the frame you request even
2404 if it's the current frame (instead of saying "You are here.").
2405 * As specified by ANSI, the system now always prints keywords
2406 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2408 * The default initial SIZE of HASH-TABLEs is now smaller.
2409 * Type information from CLOS class dispatch is now propagated
2410 into DEFMETHOD bodies, so that e.g.
2411 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2413 is now basically equivalent to
2414 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2415 (DECLARE (TYPE SINGLE-FLOAT X))
2417 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2418 operation, without having to do run-time type dispatch.
2419 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2420 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2421 * Since I don't seem to be making big file reorganizations very often
2422 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2423 is overflowing my ability to conveniently back them up), I've finally
2424 checked the system into CVS. (The CVS repository is on my home system,
2425 not at SourceForge -- putting it on SourceForge might come later.)
2426 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2427 high-level GC-NOTIFY-FOO functions send their output. (There's
2428 still very little control of where low-level verbose GC functions
2429 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2430 less than it used to -- the GC-NOTIFY-FOO functions are now under
2431 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2432 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2433 in only one place in the source code, and propagates it automatically
2434 everywhere that it's needed. Thus e.g. when I bump the version from
2435 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2436 * The C source files now include boilerplate legalese and documentation
2437 at the head of each file (just as the Lisp source files already did).
2438 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2439 to his page will be replaced with a link to his new CLiki service.
2441 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2443 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2444 * The build process now looks for GNU make under the default name "gmake",
2445 instead of "make" as it used to. If GNU make is not available as "gmake"
2446 on your system, you can change this default behavior by setting the
2447 GNUMAKE environment variable.
2448 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2449 can build without error under CMU CL.
2451 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2453 * There is now a partial SBCL user manual (with some new text and some
2454 text cribbed from the CMU CL manual).
2455 * The beginnings of a profiler have been added (starting with the
2456 CMU CL profiler and simplifying and cleaning up). Eventually the
2457 main interface should be through the TRACE macro, but for now,
2458 it's still accessed through vaguely CMU-CL-style functions and macros
2459 exported from the package SB-PROFILE.
2460 * Some problems left over from porting CMU CL to the new
2461 cross-compilation bootstrap process have been cleaned up:
2462 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2463 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2464 part when getting rid of the compiler *BACKEND* stuff.)
2465 ** The value of *NULL-TYPE* was screwed up, because it was
2466 being initialized before the type system knew the final
2467 definition of the 'NULL type. This screwed up several key
2468 optimizations in the compiler, causing inefficiency in all sorts
2469 of places. (I found it because I wanted to understand why
2470 GET-INTERNAL-RUN-TIME was consing.)
2471 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2472 PROCLAIM FTYPE information. Unfortunately this broke binary
2473 compatibility again, since now the forms output by DEFGENERIC
2474 to refer to functions which didn't exist in 0.6.3.
2475 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2476 can use the new (as of 0.6.3) transform for SOME into MAP into
2478 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2479 values to (INTEGER 0 1000000), so that the time code will no longer
2480 occasionally get blown up by Linux returning 1000000 microseconds
2481 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2482 its output conform to the ANSI spec. (Alas, this makes its output
2483 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2484 * A full call to MAP NIL with a single sequence argument no longer conses.
2485 * fixes to problems pointed out by Martin Atzmueller:
2486 * The manual page no longer talks about multiprocessing as though
2487 it were currently supported.
2488 * The ILISP support patches have been removed from the distribution,
2489 because as of version 5.10.1, ILISP now supports SBCL without us
2490 having to maintain patches.
2491 * added a modified version of Raymond Toy's recent CMU CL patch for
2492 EQUALP comparison of HASH-TABLE
2494 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2496 * The system still can't cross-compile itself with
2497 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2498 entails), but at least it can compile more of itself that way
2499 than it used to be able to, and various buglets which were uncovered
2500 by trying to cross-compile itself that way have now been fixed.
2501 * This release breaks binary compatibility again. This time
2502 at least I've incremented the FASL file format version to 2, so that the
2503 problem can be detected reliably instead of just causing weird errors.
2504 * various new style warnings:
2505 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2506 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2507 ** using the lexical binding of a variable named in the *FOO* style
2508 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2509 as specified by ANSI.
2510 * *RANDOM-STATE* is no longer automatically initialized from
2511 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2512 default behavior of the system is to repeat its behavior every time
2513 it's run. If you'd like to change this behavior, you can always
2514 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2515 old convention there was no comparably easy way to get the system to
2516 repeat its behavior every time it was run.
2517 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2518 FTYPE declarations has been removed, in favor of their ANSI
2519 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2520 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2521 the types of their sequence arguments can be determined at compile time.
2522 This is done through a new open code expansion for MAP which eliminates
2523 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2524 argument types can be determined at compile time.
2525 * The optimizer now transforms COERCE into an identity operation when it
2526 can prove that the coerced object is already of the correct type. (This
2527 can be a win for machine generated code, including the output of other
2528 optimization transforms, such as the MAP transform above.)
2529 * Credit information has been moved from source file headers into CREDITS.
2530 * Source file headers have been made more standard.
2531 * The CASE macro now compiles without complaining even when it has
2534 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2536 * (Note that the way that the PCL macroexpansions were rewritten
2537 to accommodate the change in DEFGENERIC below breaks binary
2538 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2539 not run under sbcl-0.6.2. Once we get out of alpha releases,
2540 i.e. hit release 1.0.0, we'll probably try to maintain binary
2541 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2542 and sbcl-1.4.4. Until then, however, it might be fairly common
2543 for maintenance releases to break binary compatibility.)
2544 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2546 * The reporting of unhandled signals has been changed to print some
2547 explanatory text as well as the report form. (Previously only
2548 the report form was printed.)
2549 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2550 it defines, so that the compiler no longer issues undefined function
2551 warnings for compiled-but-not-yet-loaded generic functions.
2552 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2553 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2554 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2555 section of the ANSI spec are supported.
2556 * The "" nickname for the "KEYWORD" package has been removed.
2557 The reader still handles symbol tokens which begin with a package marker
2558 as keywords, but it doesn't expose its mechanism for doing so in the
2559 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2560 * The system now issues STYLE-WARNINGs for contradictory TYPE
2561 proclamations. (Warnings for contradictory FTYPE proclamations would
2562 be nice too, but those can't be done usefully unless the type system
2563 is made smarter about FUNCTION types.)
2564 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2565 systematized, so that "*target-*.lisp is supposed to exist only on the
2566 target and imply that there's a related file which exists on the
2567 host, and *host-*.lisp is supposed to exist only on the host and imply
2568 that there's a related file which exists on the target. This involves a
2569 lot of renaming. Hopefully the acute confusion caused by the renaming
2570 will be justified by the reduction in chronic confusion..
2571 ** runtime-type.lisp -> early-target-type.lisp
2572 ** target-type.lisp -> late-target-type.lisp
2573 ** early-host-format.lisp -> early-format.lisp
2574 ** late-host-format.lisp -> late-format.lisp
2575 ** host-error.lisp -> misc-error.lisp
2576 ** early-error.lisp -> early-target-error.lisp
2577 ** late-error.lisp -> late-target-error.lisp
2578 ** host-defboot.lisp -> early-defboot.lisp
2579 ** code/misc.lisp -> code/target-misc.lisp
2580 ** code/host-misc.lisp -> code/misc.lisp
2581 ** code/numbers.lisp -> code/target-numbers.lisp
2582 ** code/early-numbers.lisp -> numbers.lisp
2583 ** early-host-type.lisp -> early-type.lisp
2584 ** late-host-type.lisp -> late-type.lisp
2585 ** host-typep.lisp -> typep.lisp
2586 ** load.lisp -> target-load.lisp
2587 ** host-load.lisp -> load.lisp
2588 ** host-disassem.lisp -> disassem.lisp
2589 ** host-insts.lisp -> insts.lisp
2590 ** byte-comp.lisp -> target-byte-comp.lisp
2591 ** host-byte-comp.lisp -> byte-comp.lisp
2592 ** host-signal.lisp -> signal.lisp
2593 ** host-defstruct.lisp -> defstruct.lisp
2594 ** late-target-type.lisp -> deftypes-for-target.lisp
2595 Furthermore, several other previously target-only files foo.lisp (e.g.
2596 hash-table.lisp and random.lisp) have been split into a target-and-host
2597 foo.lisp file and a target-only target-foo.lisp file, with their key type
2598 definitions in the target-and-host part, so that the cross-compiler will
2599 know more about target types.
2600 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2601 gone away. In their place are various *BACKEND-FOO* variables
2602 corresponding to the slots of the old structure.
2603 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2604 into the target SBCL has been fixed.
2605 * The chill.lisp system for loading cold code into a running SBCL
2607 * Support for the CMU CL "scavenger hook" extension has been removed.
2608 (It was undocumented and unused in the CMU CL sources that SBCL was
2609 derived from, and stale in sbcl-0.6.1.)
2610 * Various errors in the cross-compiler type system were detected
2611 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2612 (enabling various consistency checks). Many of them were fixed,
2613 but some hard problems remain, so the compiler is back to
2614 running without *TYPE-SYSTEM-INITIALIZED* for now.
2615 * As part of the cross-compiler type system cleanup, I implemented
2616 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2617 * I have started adding UNCROSS calls throughout the type system
2618 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2619 on cross-compiler input files will be able to go away, and various
2621 * CONSTANTP now returns true for quoted forms (as explicitly required
2624 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2626 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2627 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2628 (MAKE-LIST -1). In the longer term, it ought to become true
2629 that declarations are assertions even at SAFETY 1. For now, it's not
2630 quite true even at SAFETY 3, but it's at least more nearly true..
2631 (Note that this change seems to increases the size of the system by
2632 O(5%) and to decrease the speed of the compiler by 20% or more.)
2633 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2634 for the problem of printing dozens of lines of distracting information
2635 about low-level system machinery as part of the top stack frame
2636 on entry to the debugger when an undefined function was called.
2637 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2638 is not reset to COMMON-LISP-USER.
2639 * Compilation of stuff related to dyncount.lisp has been made conditional
2640 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2641 smaller. The various dyncount-related symbols have been moved into
2642 a new "SB-DYNCOUNT" package.
2643 * tty-inspect.lisp has been renamed to inspect.lisp.
2644 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2645 feature has gone away. (When we eventually port to other flavors of
2646 libc and/or Unix, we'll try to make the differences between flavors
2647 invisible at the user level.)
2648 * Various other *FEATURES* tags, and/or their associated conditionals,
2649 have been removed if obsolescent, or given better documentation, or
2650 sometimes given more-mnemonic names.
2652 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2654 * tidied up "make.sh" script
2655 * tidied up system directory structure
2656 * better "clean.sh" behavior
2657 * added doc/FOR-CMUCL-DEVELOPERS
2658 * many many small tweaks to output format, e.g. removing possibly-confusing
2659 trailing #\. character in DESCRIBE-INSTANCE
2660 * (EQUALP #\A 'A) no longer signals an error.
2661 * new hashing code, including EQUALP hashing
2662 * tidied up Lisp initialization and toplevel
2663 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2664 * command line argument processing
2665 * added POSIX-GETENV function to deal with Unix-ish environment variables
2666 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2667 terminating SBCL on EOF
2668 * non-verbose GC by default
2669 * There is no more "sbcl" shell script; the sbcl file is now the C
2670 runtime executable (just like CMU CL).
2671 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2673 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2674 symbol and package information as Lisp native symbol and package objects
2675 instead of strings naming symbols and strings naming packages. This way,
2676 whenever packages are renamed (as in warm init), debug information is
2677 transformed along with everything else.
2678 * tweaked the optimization policy declarations which control the building
2679 of SBCL itself. Now, among other things, the system no longer saves
2680 source location debugging information. (This helps two problems at once
2681 by reducing SBCL size and by keeping SBCL from trying to look for its
2682 sources -- which may not exist -- when reporting errors.)
2683 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2684 debugging and testing purposes
2685 * cleaned up printing, making the printer call PRINT-OBJECT for
2686 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2687 methods, giving nearly-ANSI behavior
2688 * converted almost all special variables to use *FOO* naming convention
2689 * deleted PARSE-TIME functionality, since it can be done portably
2690 * moved some files out of cold init into warm init
2691 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2693 * regularized formatting of source files
2694 * added an install.sh script
2695 * fixed ridiculous memory usage of cross-compiler by making
2696 compiler/alloc.lisp not try to do pooling unless it can hook
2697 itself into the GC of the cross-compilation host. Now the system
2698 builds nicely on my old laptop.
2699 * added :SB-ALLOC in target-features.lisp-expr
2700 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2701 was not implemented)
2702 * re-did condition handling and note reporting in the compiler. Notes
2703 are no longer handled by signalling conditions. Style warnings
2704 and warnings are handled more correctly and reported in such a way
2705 that it's easy to find one or the other in your output (so that you
2706 can e.g. figure out which of many problems caused COMPILE-FILE to
2708 * changed the severity of several compiler warnings from full WARNING
2709 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2710 compiler note reporting so that it doesn't use the condition system
2711 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2712 COMPILE-FILE command)
2713 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2714 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2715 no longer supported, since their implementation was deeply intertwingled
2716 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2717 * removed bogus "support" for compiler macros named (SETF FOO), and
2718 removed the compiler macro for SETF INFO (but only after making a fool
2719 of myself on the cmucl-imp mailing list by posting a bogus patch for
2720 DEFINE-COMPILER-MACRO..)
2721 * Compiled files containing forms which have side effects on the Lisp
2722 reader (such as DEFPACKAGE forms) are now handled more correctly.
2723 (Compiler queuing of top level lambdas has been suppressed by setting
2724 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2725 * deleted various currently-unused source files, e.g. gengc.lisp. They
2726 may be added back at some point e.g. when porting to other architectures,
2727 but until they are it's distracting to distribute them and to try to
2729 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2730 there were so many of them they're just distractions, and UNCROSS is
2731 known to be able to handle the current sources
2732 * moved PROFILE functionality into TRACE, so that it will be clear
2733 how the wrapping and unwrapping of functions when you profile them
2734 interacts with the wrapping and unwrapping of functions when you
2735 trace them. (Actually, the functionality isn't there yet, but at least
2736 the interface specification is there. Hopefully, the functionality will
2737 arrive with some future maintenance release.)
2738 * removed host-oops.lisp
2739 * changed signature of QUIT function to allow UNIX-CODE argument
2740 * fixed READ-SEQUENCE bug
2741 * tweaked verbose GC output so that it looks more like the progress
2742 output that ANSI specifies for functions like LOAD
2743 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2744 * added <http://sbcl.sourceforge.com> to the banner information printed by