1 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2 * enhancement: saving cores with foreign code loaded is now
3 supported on x86/NetBSD and sparc/Linux in addition to the previously
5 * bug fix: on some platforms repeated installations caused multiple
6 copies of HTML documentation to be installed -- should not happen
7 any more. (reported by Stefan Scholl)
8 * bug fix: parsing self-recursive alien record types multiple times
9 no longer causes infinite recursion. (reported by Thomas F. Burdick,
10 original patch by Helmut Eller for CMUCL)
11 * bug fix: stack-exhaustion detection works now on NetBSD as well.
12 (thanks to Richard Kreuter)
13 * bug fix: defining classes whose accessors are methods on existing
14 generic functions in other (locked) packages no longer signals
15 bogus package lock violations. (reported by François-René Rideau)
16 * bug fix: special variables as DEFMETHOD parameters no longer have
17 associated bogus type declarations. (reported by David Wragg and
19 * bug fix: read-write consistency on streams of element-type
20 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
21 * bug fix: redefiniton of the only method of a generic function with
22 no DEFGENERIC no longer emits a full WARNING. In addition,
23 redefinition of generic functions with no DEFGENERIC to an
24 incompatible lambda list now signals an error. (thanks to Zach
26 * bug fix: DEFGENERIC now works even when there's a function of the
27 same name in an enclosing lexical environment. (thanks to Zach
29 * fixed compiler failure, caused by instrumenting code during
30 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
31 * optimization: added loop analysis and improved register allocation
32 * fixed some bugs revealed by Paul Dietz' test suite:
33 ** POSITION on displaced vectors with non-zero displacement
34 returns the right answer.
35 ** (SIMPLE-STRING) is a valid type specifier for sequence
37 ** *PRINT-LEVEL* handling for slotless structures is pedantically
39 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
40 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
42 ** The justification version of the FORMAT ~< directive treats
43 non-zero minpad parameter correctly.
45 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
46 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
47 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
48 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
49 the supported interface.
50 * new feature: Single-stepping of code compiled with DEBUG 2 or
51 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
52 * new feature: saving cores with foreign code loaded is now
53 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
54 Timothy Moore's work for CMUCL)
55 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
56 parameters to * instead of NIL if no initform is supplied.
57 (reported by Johan Bockgaard)
58 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
59 correctly when FUNCALL appears as the car of the form. Note:
60 despite this FUNCALL forms are not currently subject to
61 compiler-macro expansion. (port of Raymond Toy's fix for the
62 same from CMUCL, reported by Johan Bockgaard)
63 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
64 (thanks for Teemu Kalvas)
65 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
66 case instead of terminating the iteration (thanks for Julian
67 Squires, Sean Champ and Raymond Toy)
68 * bug fix: incorrect expansion of defgeneric that caused a style
69 warning. (thanks for Zach Beane)
70 * bug fix: slot accessor effective method computation works properly
71 for classes with multiple non-standard applicable methods on
72 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
73 * on x86 compiler supports stack allocation of results of LIST and
74 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
75 CMUCL implementation by Gerd Moellmann)
76 * fixed some bugs revealed by Paul Dietz' test suite:
77 ** FORMAT strings with both the ~<~:;~> form of the justification
78 directive and pretty-printing directives cause an error.
80 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
81 * incompatible change: the internal functions
82 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
83 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
84 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
85 instead of the old functions.
86 * new feature: on platforms where "dladdr" is available foreign
87 function names now appear in backtraces. (based on Helmut Eller's
89 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
90 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
92 * documentation: ASDF manual is now distributed with SBCL.
93 * bug fix: SBCL can now load its contributed modules with REQUIRE
94 even if the system-provided entries have been removed from
95 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
97 * bug fix: backtraces involving undefined functions or assembly
98 routines are more informative. (thanks to Brian Downing)
99 * bug fix: mutually referent alien structures now work correctly.
100 (reported by Rick Taube)
101 * bug fix: structures defined by WITH-ALIEN can be referred to
102 within other definitions in the same WITH-ALIEN.
103 * bug fix: division operators (MOD, TRUNCATE and the like) with
104 constant zero divisors and integer dividends no longer generate
106 * bug fix: provide default methods for INPUT-STREAM-P and
107 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
108 * bug fix: improve the detection and resolution of MOP metacycles
109 (where effective method computation depends on the generic function
110 having its effective method computed). (reported by Bruno Haible)
111 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
112 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
113 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
114 * bug fix: pathnames with relative directory components can be
115 represented relative to default pathnames.
116 * optimization: in taking the GCD of bignums, reduce the two bignums
117 to approximately the same size (using Euclid's algorithm) before
118 applying the more sophisticated binary GCD. (thanks to Juho
120 * optimization: COUNT on bitvectors now operates word-at-a-time.
121 * optimization: ASH with a positive, but not necessarily constant,
122 (leftwards) shift, when in a modular context, is compiled to a
124 * fixed some bugs revealed by Paul Dietz' test suite:
125 ** FORMAT variable parameters ("~V<char>") are defaulted properly
126 if the corresponding argument is NIL.
127 ** FORMAT directives accepting a minpad parameter treat negative
129 ** literal commas as character parameters to FORMAT directives are
131 ** literal spaces directly after ~<Newline> directives within a
132 format-logical-block (~:< ~@:>) do not induce :FILL-style
133 conditional newlines.
134 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
135 as specified: it no longer includes conditional newlines).
136 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
138 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
139 :PER-LINE-PREFIX argument does not evaluate to a string.
141 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
142 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
143 the manual for details; for now, package locks can be disabled by
144 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
145 if no major problems are found then it is likely that they will be
146 compiled in unconditionally.
147 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
148 unsupported operators on all platforms. To load a shared library
149 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
150 object file, link it into a shared library outside of SBCL and
151 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
153 * bug workaround: ROOM T and the GC were not working together
154 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
155 reporting facility aren't preserved. That reporting has been
156 disabled, so now until and unless someone figures out how to make it
157 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
158 * minor incompatible change: as threatened around sbcl-0.8.0, the
159 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
160 The symbols are also exported from SB-PCL for backwards
161 compatibility, but more so than before SB-PCL should be treated as
162 an implementation-internal package.
163 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
165 * the SB-SPROF contrib now works on (most) non-x86 architectures.
166 It is known as of this release not to work on the Alpha, however.
167 * fixed bug #167: errors signalled due to illegal syntax in method
168 bodies are now more legible.
169 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
170 designators and can hence be used with TYPEP.
171 * fixed bug #333: CHECK-TYPE now ensures that the type error
172 signalled, if any, has the right object to be accessed by
173 TYPE-ERROR-DATUM. (reported by Tony Martinez)
174 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
175 5.1.2.3 for multiple-value place subforms. (reported by Kalle
177 * fixed bug #334: programmatic addition of slots using specialized
178 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
179 and :ALLOCATION :CLASS slots.
180 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
181 float exponents. (rereported by Peter Seibel)
182 * fixed a bug: #\Space (and other whitespace characters) are no
183 longer considered to be macro characters in standard syntax by
185 * fixed bug: initialization of condition class metaobjects no longer
186 causes an instance of the condition to be created. (reported by
188 * fixed bug: it is now possible to have slots such that
189 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
190 description is neither :INSTANCE nor :CLASS.
191 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
192 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
193 SLOT-VALUE-USING-CLASS), and no longer causes errors with
194 non-standard SLOT-DEFINITION-ALLOCATION values.
195 * fixed bugs: various race conditions handling exiting threads.
196 CL-PPCRE's thread torture test now passes.
197 * fixed arguably-a-bug: GC time proportional to number of threads in
198 system even when most of them are idle
199 * optimization: improved performance of BIT and SBIT on bit-vectors.
200 * .fasl file incompatibility: The fasl file version number has
201 been incremented because of changes associated with package locks.
203 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
204 * minor incompatible change: the system no longer provides
205 optimization hints (conditions which are TYPEP
206 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
207 mode; these hints are still emitted when higher SPEED optimization
209 * new contrib module: a sampling profiler (profiling by statistical
210 sampling, rather than by instrumenting functions) is available as
211 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
212 * the behaviour of the standard function ED is now customizeable by
213 third parties through a hook variable: see ED's documentation
214 string for information on the protocol.
215 * the compiler no longer emits efficiency notes for (FUNCALL X)
216 when the type of X is uncertain under default optimization
218 * fixed bug 276: mutating a binding of a specialized parameter to a
219 method to something that is not TYPEP the specializer is now
221 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
222 exhibit the required behaviour on the x86 platform. (thanks to
223 Peter van Eynde, Eric Marsden and Bruno Haible)
224 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
225 even for difficult arguments. (reported by Peter Graves)
226 * fixed bug 141a: the backquote printer now descends quoted
228 * fixed another bug in backquote printing: no more destructive
229 modification of the form's list structure. (reported by Brian
231 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
232 move between its address being taken and the call to
233 interrupt_thread, fixing a crashing race condition.
234 * the SB-POSIX contrib implementation has been adjusted so that it
235 no longer exhibits ridiculously poor performance when constructing
236 instances corresponding to C structs.
238 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
239 * minor incompatible change: the sb-grovel contrib now treats C
240 structures as alien (in the sense of SB-ALIEN) objects rather than
241 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
242 has implications for memory management of client code
243 (sb-grovel-returned objects must now be manually managed) and for
244 type safety (alien objects now have full types).
245 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
246 used to control emission of compiler diagnostics, rather than the
247 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
248 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
249 quality should be considered deprecated.
250 * install.sh now installs the user manual as well
251 * (not quite a new documentable feature, but worth considering in
252 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
253 beginnings of a semantically meaningful condition hierarchy is
254 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
256 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
257 designator as the defaults argument.
258 * fixed bug: Displaced arrays whose displaced-to array has become
259 too small now cause ARRAY-DIMENSION to signal an error, providing
260 for safer bounds-checking. (reported by Bruno Haible)
261 * fixed bug: DEFCLASS slot definitions with identical :READER and
262 :WRITER names now signal a reasonable error. (reported by Thomas
264 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
266 * fixed bug: Invalid slot specification errors now print correctly.
267 (thanks to Zach Beane)
268 * fixed bug 320: Shared to local slot value transfers in class
269 redefinitions now happen corrently from superclasses as
270 well. (reported by Bruno Haible)
271 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
273 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
274 improper lists correctly. (reported by Bruno Haible)
275 * fixed bug 313: source-transform for <fun-name> was erroneously
276 applied to a call of a value of a variable with name <fun-name>.
277 (reported by Antonio Menezes Leitao)
278 * fixed bug 307: The obsolete instance protocol ensures that
279 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
280 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
281 remove dead unknown-values globs from the middle of the stack.
282 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
283 easier to resume long computations after using *BREAK-ON-SIGNALS*
284 to diagnose and fix failures. (thanks to Nikodemus Siivola)
285 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
286 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
287 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
288 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
289 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
291 * fixed a (fairly theoretical) bug in string printing: if
292 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
293 does not have array-element-type equal to the most general string
295 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
296 function name. (thanks to Nikodemus Siivola)
297 * fixed bug: compile-time format string checker failed on
298 non-closed ~{. (reported by Thomas F Burdick)
299 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
300 don't warn on legal loop code involving more than one aggregate
301 boolean. (thanks to Nikodemus Siivola)
302 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
303 in keyword position to MAKE-PACKAGE was not being reported as
305 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
306 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
308 * fixed bugs 280 and 312: the checking for multiple definitions in a
309 file is less likely to become confused by uses of inline
311 * fixed bug: the #S reader macro performs the keyword coercion
312 specified for slot names. (reported by Kalle Niemitalo)
313 * fixed bug: lambda lists may contain symbols whose names start with
314 & but are not lambda-list-keywords; their occurrence triggers a
316 * fixed bug 321: define-method-combination argument lambda lists do
317 not cause invalid code to be generated when &optional or &aux
318 variables have default values. (reported by Bruno Haible)
319 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
320 have CLOS classes; this appears also to have fixed reported
321 instability in stack exhaustion detection.
322 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
323 MUFFLE-WARNING when no associated restart is present is now
325 * optimization: rearranged the expansion of various defining macros
326 so that each expands into only one top-level form in a
327 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
329 * optimization: used a previously little-used slot in symbols to
330 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
332 * fixed some bugs revealed by Paul Dietz' test suite:
333 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
335 ** Fixed an optimization bug related to inheritance of initforms
336 from local to shared slots.
337 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
338 ** CAST optimizer forgot to flush argument derived type.
339 ** print/read consistency on floats is now orders of magnitude
340 more likely. (thanks also to Bruno Haible for a similar report
342 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
343 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
344 for objects of type REAL. Make it so.
345 ** SXHASH is specified (infelicitously) to respect similarity,
346 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
347 it so. (thanks to Markus Ziegler)
348 ** on the Alpha, the compiler succeeds in compiling functions
349 returning a known number of arguments greater than 63.
350 ** fixed handling of invalid NIL arguments in keyword position
352 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
354 ** provided a readably-printable representation for RANDOM-STATE
356 ** ensured that pathnames, where they have a namestring, always
357 print using #P"..." syntax.
359 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
360 * Support for the forthcoming 2.0 version of the NetBSD kernel
361 running on x86 hardware has been added. (thanks to Perry
362 E. Metzger most immediately, and others for their past work)
363 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
364 current release version; SBCL's OpenBSD support had been broken
365 since about the time of OpenBSD's switch to ELF binary format.)
366 * [placeholder for DX summary]
367 ** user code with &REST lists declared dynamic-extent, under high
368 speed or space and low safety and debug optimization policy.
369 * The manual has been converted to Texinfo format and the debugger
370 chapter from the cmucl manual has been added.
371 * A facility has been added to extract documentation strings from
372 sbcl and store them as Texinfo-formatted snippets for inclusion in
373 the manual (via Texinfo's @include directive)
374 * bug fix: compiler emitted division in optimized DEREF. (thanks for
375 the test case to Dave Roberts)
376 * bug fix: multidimensional simple arrays loaded from FASLs had fill
377 pointers. (reported by Sean Ross)
378 * bug fix: PROFILE output is printed nicely even for large numerical
379 values. (thanks to Zach Beane)
380 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
381 greater than 32 handle EOF correctly.
382 * bug fix: on X86 an immediate argument of the IMUL instruction is
383 correctly printed in disassembly. (thanks to Lutz Euler)
384 * bug fix: class slots in redefined classes preserve their old
385 values. (thanks to Bruno Haible and Nikodemus Siivola)
386 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
387 succeeds. (reported by Marco Baringer)
388 * fixed some bugs revealed by Paul Dietz' test suite:
389 ** READ-SEQUENCE now works on ECHO-STREAMs.
390 ** RATIONALIZE works more according to its specification. (thanks
393 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
394 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
395 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
396 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
397 should matter to you only if you rebind the printer control
398 variables and then find you want different bindings in the
399 debugger than in the ordinary execution of your program.)
400 * The runtime build system has been tweaked to support building
401 (on SPARC/SunOS) using a C compiler which invokes Sun's own
402 assembler and linker. (thanks to Nikodemus Siivola)
403 * Unbound, undefined, undeclared variables now trigger full
404 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
405 more often programmer error than deliberate exploitation of undefined
407 * optimization: the hash algorithm for strings has changed to one
408 that is less vulnerable to spurious collisions. (thanks to Juho
410 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
411 needless bounds checking. (thanks to Juho Snellman)
412 * optimization: implemented multiplication as a modular
413 (UNSIGNED-BYTE 32) operation on the PPC backend.
414 * fixed some bugs revealed by Paul Dietz' test suite:
415 ** ADJUST-ARRAY now copies the datum in a zero rank array if
417 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
418 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
419 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
421 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
423 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
425 ** ECHO-STREAMs no longer attempt to echo the end of file value to
426 their output stream on EOF from read.
427 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
428 have been read to end-of-file.
429 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
431 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
432 description of determination of which consecutive characters
434 ** Printing the "Space" character with escaping on now yields "#\\ ",
435 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
436 ** Reading floating-point numbers with *READ-BASE* set to a number
437 less than 10 works correctly.
438 ** Reading floating-point numbers with *READ-BASE* set to a number
439 more than 10 works correctly.
440 ** Printing with *PRINT-READABLY* targets the standard readtable, not
441 the readtable currently in effect.
443 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
444 * minor incompatible change: parsing of namestrings on a physical
445 (Unix) host has changed; numbers after the final #\. in a
446 namestring are no longer interpreted as a version field. This is
447 intented to be largely invisible to the user, except that the
448 meaning of the namestring "*.*.*" has changed: it now refers to a
449 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
450 should usually be replaced by
451 (make-pathname :name :wild :type :wild :version :wild)
452 with the added benefit that this is more likely to be portable.
453 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
454 option to OPEN now signals an error if the file being opened
455 exists; this may have an impact on existing code.
456 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
457 signals received, on the PowerPC platforms (both Linux and
458 Darwin). (thanks to Pierre Mai for pointing out the location of
460 * several fixes on OS X: The system now builds and runs cleanly on
461 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
462 (thanks to Brian Mastenbrook)
463 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
464 second-resolution: integer multiples of 1/3600 between -24 and 24.
465 (thanks to Vincent Arkesteijn)
466 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
467 type when called with 1 argument; PEEK-CHAR checked type of
468 PEEK-TYPE only after having read first character from a
469 stream. (reported by Peter Graves)
470 * bug fix: the garbage collector now has much better locality
471 behaviour, and in particular no longer treats all memory as being
472 exhausted when many small objects point to each other in a deeply
474 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
475 recognized as being TYPEP their class.
476 * bug fix: the PUSHNEW documentation string has been corrected.
477 (thanks to Vincent Arkesteijn)
478 * bug fix: defaulting of the value for the last of an atypically
479 large number of multiple values being bound was not being
480 performed correctly on the Alpha or PPC platforms
481 * optimization: implemented multiplication as a modular
482 (UNSIGNED-BYTE 32) operation on the x86 backend.
483 * optimization: SEARCH on simple-base-strings can now be open-coded.
484 (see also contrib/compiler-extras.lisp for inspiration for
485 teaching the compiler about the Boyer-Moore algorithm).
486 * value, returned by MAX (and MIN) called with several EQUALP, but
487 not EQL, arguments now does not depend on compiler settings.
488 * fixed some bugs revealed by Paul Dietz' test suite:
489 ** in stack analysis liveness information is propagated from
490 non-local entry points.
491 ** pathwise CAST removing failed when the CAST node did not start
493 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
494 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
496 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
497 host is already defined.
498 ** RENAME-FILE works on streams instead of signalling an internal
500 ** PEEK-CHAR uses the current readtable when determining whether
501 or not a character is whitespace.
502 ** MERGE-PATHNAMES handles the case when the pathname does not
503 specify a name while the default-pathname specifies a version.
504 ** Pathnames now stand a better chance of respecting print/read
506 ** Attempting to use standardized file system operators with a
507 pathname with invalid :DIRECTORY components signals a
509 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
511 ** DIRECTORY on logical pathnames is more correct.
512 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
513 signal a TYPE-ERROR if their argument is not a stream.
514 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
515 designator argument does not designate a stream.
516 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
517 examining the synonym.
518 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
520 ** OPEN and WITH-OPEN-STREAM allow opening streams with
521 element-type larger than ([UN]SIGNED-BYTE 32).
523 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
524 * When built with the :SB-FUTEX feature, threaded builds now take
525 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
526 for faster/more reliable mutex and condition variable support.
527 * Incompatible change (but one you probably shouldn't have been using
528 anyway): the interface and code for arbitrating between multiple
529 threads in the same user session has been redesigned.
530 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
531 error when called without an explicit environment argument.
532 (thanks to Nikodemus Siivola)
533 * bug fix: buffered :DIRECTION :IO streams are less likely to become
534 confused about their position. (thanks to Adam Warner and Gerd
536 * bug fix: Pretty printing backquoted forms with unquotations in the
537 argument list position of various code constructs such as LAMBDA
538 now works correctly. (reported by Paul Dietz)
539 * bug fix: Pretty printing unquotations no longer loses all
540 stream position information.
541 * optimization: performance of string output streams is now less
542 poor for multiple small sequence writes.
543 * optimization: performance of CSUBTYPEP in the presence of complex
544 expressions involving CONS and NOT many times has been improved.
545 (reported by Paul Dietz)
546 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
548 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
549 streams for the REPL, for the debugger, and so on. (thanks to
551 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
552 * fixed some bugs revealed by Paul Dietz' test suite:
553 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
554 is ignored if it is too small, rather than propagating through
555 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
556 ** extremely complex negations of CONS types were not being
557 sufficiently canonicalized, leading to inconsistencies in
559 ** VALUES tranformer lost derived type.
561 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
562 * fixed a bootstrapping bug: the build process no longer assumes
563 that the various BOOLE-related constants have the same value in
564 host and target lisps. (noted by Paul Dietz' test suite on an
565 SBCL binary built from CLISP)
566 * The system can now be dynamically linked on the MIPS platform,
567 which enables dynamic loading of foreign code from Lisp. (thanks
568 to Ralf Baechle for discussions on the MIPS ABI)
569 * The system now records debugging information for its own source
570 files in a filesystem-position-independent manner, relative to
571 the "SYS" logical host.
572 * fixed a compiler bug: MV-LET convertion did not check references
573 to the "max args" entry point. (reported by Brian Downing)
574 * tweaked disassembly notes to be less confident about proclaiming
575 some instruction as an LRA. (thanks to Brian Downing)
576 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
577 now each have their own history, command character, and other
578 characteristics. (thanks to David Lichteblau)
579 * fixed some bugs revealed by Paul Dietz' test suite:
580 ** compiler failure in compiling LOGAND expressions including a
582 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
583 shift greater than 32.
584 ** FLUSH-DEST did not mark blocks for type check regeneration.
585 ** HANDLER-CASE failed to accept declarations in handler clauses
586 in some circumstances.
588 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
589 * New code in contrib/sb-introspect (still probably not entirely
590 stable yet) provides some support for smart Lisp development
591 environments like SLIME.
592 * The conditions signalled for errors occurring when loading .fasl
593 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
594 in a way which should help ASDF recover gracefully.
595 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
596 stuff has been cleaned up. If you code contrib/ stuff, this might
597 affect you, and you can look at contrib/README, contrib/STANDARDS,
598 and/or the 0.8.4.27 diff to check.
599 * In full calls the compiler now does not generate checks for declared
600 argument types for all arguments.
601 * various threading fixes
602 ** and some experimental patches which didn't make it into
603 the main tree for this release, but which are shipped in
604 contrib/experimental-thread.patch as a possible fix for some
605 failures (deadlock, spinning...) in GC-intensive multithreaded
607 * fixed PPC build problem (source code incompatibility of different
608 library versions): added offsetof() hackery which attempts to divine
609 where glibc maintainers put uc_mcontext today
610 * fixed bug 282: compiler does not trust type assertions while passing
611 arguments to a full call.
612 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
613 &OPTIONAL VALUES type parameter.
614 * fix bug 214: algorithm for noting rejected templates is now more
615 similar to that of template seletion. (also reported by rydis on
617 * fixed bug 141b: printing backquoted information readably and prettily
618 inserts a space where necessary.
619 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
620 (CHARACTER 10) are now reported as errors, rather than propagated
621 as unknown types. (reported by piso on #lisp)
622 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
623 with values NIL and :ERROR. (thanks to Milan Zamazal)
624 * fixed bug 191c: CLOS now does proper keyword argument checking as
625 described in CLHS 7.6.5 and 7.6.5.1.
626 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
627 longer raise an error; further, using a list as a for-as-arithmetic
628 counter now raises a meaningful error.
629 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
630 now understood by sequence creation functions such as MAKE-SEQUENCE
632 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
633 a STRING-INPUT-STREAM.
634 * compiler enhancement: SIGNUM is now better able to derive the type
636 * type declarations inside WITH-SLOTS are checked. (reported by
638 * fixed some bugs revealed by Paul Dietz' test suite:
639 ** incorrect optimization of TRUNCATE for a positive first
640 argument and negative second.
641 ** compiler failure in let-convertion during flushing dead code.
642 ** compiler failure while deriving type of TRUNCATE on an
643 interval, containing 0.
644 ** ASH of a negative bignum by a negative bignum count now returns
646 ** intersection of CONS types now canonicalizes properly, fixing
647 inconsistencies in SUBTYPEP.
649 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
650 * incompatible change: The --disable-debugger command line
651 option now clobbers the debugger at a more fundamental
652 level, by redefining #'INVOKE-DEBUGGER instead of by
653 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
654 is specified by ANSI to ignore *DEBUGGER-HOOK* and
655 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
656 enter the debugger REPL and then suffer recursive errors
657 because *DEBUG-IO* is also messed up in --disable-debugger mode;
658 while under the new system, BREAK in --disable-debugger mode
659 terminates the system just as an unhandled error would.
660 * fixed compiler performance when processing loops with a step >1;
661 * bug fix: DOCUMENTATION now retrieves generic function
662 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
663 support has been systematized, and now supports the methods
664 specified by ANSI, along with a default method and a method for
665 slot documentation. (reported by Nathan Froyd)
666 * bug fix: effective methods associated with a generic function are
667 no longer cached over a change of that generic function's method
668 combination. (reported by Andreas Fuchs)
669 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
670 names a non-existent directory. (thanks to Andreas Fuchs)
671 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
672 the PPC platform, lead to stack corruption; this has been fixed.
673 (reported by Rainer Joswig)
674 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
675 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
676 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
677 arithmetic in general, is now much more reliable on the PPC
679 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
680 platform now returns the right answer.
681 * optimization: restored some effective method precomputation in
682 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
683 precomputation is now tunable.
684 * optimization: compiler-internal data structure use has been
685 reviewed, and changes have been made that should improve the
686 performance of the compiler by about 20%.
687 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
688 simple-base-strings and simple-bit-vectors is improved.
689 * optimization: the optimization of 32-bit logical and arithmetic
690 functions introduced in version 0.8.3 on the x86 has been
691 implemented on the mips, ppc and sparc platforms; an
692 implementation of the same facility, but for 64-bit arithmetic,
693 has been added for the alpha.
694 * microoptimization: the compiler is better able to make use of the
695 x86 LEA instruction for multiplication by constants.
696 * bug fix: in some situations compiler did not report usage of
697 generic arithmetic in (SPEED 3) policy.
698 * bug 145b fix: compiler used wrong type specifier while converting
699 MEMBER-types to numeric.
700 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
702 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
703 index argument now works properly on the Alpha platform.
704 * bug fix: floating point exception treatment on the Alpha platform
706 * bug fix: FILE-POSITION works much better on string input and
707 output streams. (thanks to Nikodemus Siivola)
708 * bug fix: many threading/garbage collection symptoms sorted.
709 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
710 might be pseudo-atomic.
711 * internal change: Stopping for GC is now done with signals not
712 ptrace. GC is now done in whichever thread wanted it, instead of
714 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
716 * bug fix: result form in DO is not contained in the implicit
718 * incompatible change: ICR structure is changed; the value part of
719 CONTINUATION is now called LVAR; corresponding functions are
720 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
721 * added type deriver for ISQRT (thanks to Robert E. Brown).
722 * bug fix: better support for loading from the command line when an
723 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
725 * fixed some bugs revealed by Paul Dietz' test suite:
726 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
727 ** ROUND and FROUND now give the right answer when given very
728 small float arguments.
729 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
731 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
732 ** (GCD 0 <negative-integer>) returned <negative-integer>.
733 ** LCM should return a non-negative integer.
734 ** PARSE-INTEGER returned the index of a terminator instead of the
735 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
736 ** PARSE-INTEGER returned an incorrect index being applied to a
738 ** LCM with two arguments of 0 returns 0 rather than signalling
740 ** unsigned addition of a 32-bit constant with the high bit set no
741 longer causes an internal compiler error on the x86.
742 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
743 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
744 with negative last argument.
745 ** byte specifiers with zero size and position no longer cause
746 an error during type derivation.
747 ** bignum multiplication on the Alpha platform now returns the
749 * porting: The system now builds on SuSE AMD64, although it still
750 generates a 32-bit binary.
751 * .fasl file incompatibility: The fasl file version number has
752 been incremented (because of the changes to internal compiler
753 data structures referred to above).
755 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
756 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
757 more accurately, on the Darwin kernel running on PowerPC hardware.
758 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
759 * Compiler code deletion notes now signal a condition of type
760 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
761 an associated MUFFLE-WARNING restart.
762 * The compiler now performs limited argument count validation of
763 constant format strings in FORMAT, and where appropriate in ERROR,
764 CERROR and WARN. (thanks to Gerd Moellmann)
765 * New ASDF-INSTALL contrib can be used for automatic download and
766 installation of third-party Lisp code from CCLAN or other sites
768 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
769 which forces another thread to execute a function supplied by the
771 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
772 now accept and act upon their :ELEMENT-TYPE keyword argument.
773 (reported by Martin Atzmueller, Edi Weitz)
774 * bug fix: FILE-POSITION now accepts position designators up to
775 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
776 is the greater. (thanks to Patrik Nordebo)
777 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
778 Kalle Olavi Niemitalo)
779 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
780 printable. (reported by Eric Marsden)
781 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
782 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
783 * x86 bug fix in control stack exhaustion checking: now shows backtrace
784 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
785 (thanks to Stig Sandoe)
786 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
787 UNREAD-CHAR now correctly includes the unread character in the
788 target sequence. (thanks to Gerd Moellmann)
789 * bug fix in threaded builds: the system can now be suspended and
790 resumed by shell job control with minimal disruption.
791 * bug fixes in times and timezones >2038AD
792 * better handling of "where is GNU make?" problem in build scripts
793 (thanks to Nikodemus Siivola)
794 * new optimization: inside a named function any reference to a
795 function with the same name is considered to be a self-reference;
796 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
797 optimization quality.
798 * new optimization on x86: logical functions and + now have
799 optimized (UNSIGNED-BYTE 32) versions, which are automatically
800 used when the result is truncated to 32 bits.
801 * VALUES declaration is partially enabled.
802 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
803 * fixed some bugs revealed by Paul Dietz' test suite:
804 ** The system now obeys the constraint imposed by
805 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
806 types form a lattice under type intersection.
807 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
808 ** ASSOC now ignores NIL elements in an alist.
809 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
810 and (1+ MOST-POSITIVE-FIXNUM) answers.
811 ** The addition of a method with invalid qualifiers to a generic
812 function does not cause an error to be signalled immediately;
813 a warning is signalled, and the error is generated only on
814 calling the generic function.
815 * changed .fasl file version number, in part to add type codes for
816 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
817 obscure ANSI requirements
819 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
820 * fixed bug 148: failure to inline-expand a local function left
821 garbage, confusing the compiler.
822 * fixed bugs 3cd: structure slot readers perform type check if the
823 slot can have an invalid value (i.e. it is either not initialized
824 or can be written with a less specific slot writer).
825 * bug fix: the compiler now traps array references to elements off
826 the end of an array; previously, the bounds checking in some
827 circumstances could go off-by-one.
828 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
830 * type declarations for array element types now obey the description
831 on the CLHS page "Declaration TYPE", as per discussions on
832 sbcl-help around 2003-05-08. This means that a declaration
833 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
834 declaration, all references to BAR will be asserted or assumed
835 (with THE, so dependent on compiler policy) to involve objects of
836 type FOO. Note that no such declaration is implied in
837 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
838 * declared types of functions from the "Conditions"
839 chapter. (reported by Paul Dietz)
840 * bug fix: CERROR accepts a function as its first argument.
841 * bug fix: NTH an NTHCDR accept a bignum as index
842 arguments. (reported by Adam Warner)
843 * optimization: character compare routines now optimize comparing
844 against a constant character. (reported by Gilbert Baumann)
845 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
846 argument now works properly on the MIPS platform.
847 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
848 * repeated evaluation of the same DEFSTRUCT, a slot of which is
849 declared to have a functional type, does not cause an error
851 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
852 many :DEFAULT-INITARGS. (reported by Istvan Marko)
853 * fixed bug: if last continuation of a deleted block has a
854 destination, this destination should be deleted too. (reported by
856 * fixed a bug in the bootstrap process: the host compiler's values
857 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
858 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
859 test case from Patrik Nordebo)
860 * improved the ability of the disassembler on the PPC platform to
861 provide helpful disassembly notes.
862 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
863 the class in more cases than previously.
864 * bug fix: FILE-POSITION now understands :START and :END for
865 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
866 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
867 8) data. (thanks to David Lichteblau)
868 * bug fix: it is possible to add a method to a generic function
870 * bug fix: reader failed to signal END-OF-FILE inside an
871 object representation. (reported by Nikodemus Siivola)
872 * fixed some bugs revealed by Paul Dietz' test suite:
873 ** LAST and [N]BUTLAST should accept a bignum.
874 ** condition slot accessors are methods.
875 ** (VECTOR NIL) is a subtype of STRING.
877 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
878 * minor incompatible change: some nonsensical specialized lambda
879 lists (used in DEFMETHOD) which were previously ignored now signal
881 * minor incompatible change: the system is now aware of the types of
882 variables in the COMMON-LISP package, and will signal errors for
883 most violations of these type constraints (where previously they
884 were silently accepted).
885 * minor incompatible change: COMPILE-FILE now uses the freedom
886 afforded (ANSI 3.2.2.3) to use derived function types for
887 functions defined in the same file. This also permits the system
888 to warn on static type mismatches and function
889 redefinition. (Currently it does not work with high DEBUG level.)
890 * minor incompatible change: VALUES declaration is disabled.
891 * When issuing notes, the compiler now signals a condition of type
892 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
893 restart for use in user handlers. It is expected that the
894 COMPILER-NOTE condition will eventually become a condition
895 supertype to a hierarchy of note types, which will then be
896 handleable in a similar fashion. However, at the moment, no such
897 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
898 but it's an implementation detail, not a classification for the
900 * Changes in type checking closed the following bugs:
901 ** type checking of unused values (192b, 194d, 203);
902 ** template selection based on unsafe type assertions (192c, 236);
903 ** type checking in branches (194bc).
904 * A short form of VALUES type specifier has ANSI meaning (it has
905 increased the number of situations when SBCL cannot perform type
907 * fixed bug in DEFSTRUCT: once again, naming structure slots with
908 keywords or constants is permissible.
909 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
910 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
911 classes. (thanks to Antonio Martinez)
912 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
913 outputs FRESH-LINE or TERPRI, and no longer converts its stream
914 argument to a pretty-print stream. Instead, it leaves any such
915 operations to DESCRIBE-OBJECT methods.
916 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
918 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
919 lambda lists. (thanks to David Lichteblau)
920 * bug fix: Defining a generic function with a :METHOD-CLASS being a
921 subclass of STANDARD-METHOD no longer causes stack exhaustion.
922 (thanks to Gerd Moellmann)
923 * fixed bug 246: increased compilation speed of long
924 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
926 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
928 * DEFCONSTANT now throws a condition of type
929 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
930 constant to a non-EQL value; CONTINUE and ABORT restarts
931 respectively change and preserve the value.
932 * fixed bug 63: The code walker, part of the implementation of CLOS,
933 is now better at handling symbol macros.
934 * bug fix: There is no longer an internal implementation type named
935 CL:LENGTH. (reported by Raymond Toy)
936 * bug fix: In macro-like defining macros/special operators the
937 implicit block does not enclose the lambda list.
938 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
939 enclosed in parentheses) are not suitable as type specifiers, and
940 their use properly signals an error now.
941 * bug fix: An argument count mismatch for a type specifier in code
942 being compiled no longer causes an unhandled error at compile
943 time, but signals a compile-time warning.
944 * fixed simple vector readable printing
945 * bug fix: DESCRIBE takes more care over whether the class
946 precedence list slot of a class is bound before accessing it.
947 (reported by Markus Krummenacker)
948 * bug fix: FORMATTER can successfully compile pretty-printer format
949 strings which use variants of the ~* directive inside.
950 * bug fix: SEARCH now applies its TEST predicate to the elements of
951 the arguments in the correct order. (thanks to Wolfhard Buss)
952 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
953 in type inference. (thanks to Robert E. Brown)
954 * bug fix: Reading in symbols with an explicit package name of ""
955 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
956 * fixed some bugs revealed by Paul Dietz' test suite:
957 ** NIL is now allowed as a structure slot name.
958 ** Arbitrary numbers, not just REALs, are allowed in certain
959 circumstances in LOOP for-as-arithmetic clauses.
960 ** Multiple class redefinitions before slot access no longer
962 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
963 association between the name and a class.
964 ** Generic functions with non-standard method-combination and over
965 five methods all of which return constants no longer return NIL
966 after the first few invocations. (thanks to Gerd Moellmann)
967 ** CALL-NEXT-METHOD with no arguments now passes the original
968 values of the arguments, even in the presence of assignment.
969 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
971 ** DISASSEMBLE works with closures and funcallable instances.
972 ** ADD-METHOD now returns the generic function, not the new method.
973 ** FIND-METHOD signals an error if the lengths of the specializers
974 is incompatible with the generic function, even if the ERRORP
976 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
977 which its argument is a member.
978 ** DEFCLASS only redefines the class named by its class-name
979 argument if that name is the proper name of the class;
980 otherwise, it creates a new class.
981 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
982 of the UNBOUND-SLOT condition to the name of the slot.
983 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
984 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
985 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
986 SLOT-MAKUNBOUND in the specified fashion.
988 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
989 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
990 cross-compilation host. As a consequence, we can now bootstrap our
991 way up to SBCL starting with a bare gcc toolchain and human-readable
992 source code (first the source to CLISP, then the source to SBCL).
993 * A contributed module containing a partial implementation of the
994 simple-streams interface has been included. (thanks to Rudi
996 * A contributed module implementing the RFC1321 Message Digest
997 Algorithm, known as MD5, has been included.
998 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
999 no longer has any effect, as the code controlled by this feature
1000 has been deleted. (As far as we know, no-one has ever built using
1001 this feature, and its semantics were confused in any case).
1002 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1003 work on conditions (as required by the ANSI specification),
1004 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1005 expected behaviour on conditions. Users should note, however,
1006 that such behaviour is not required by the ANSI specification,
1007 and so use of this behaviour may render their code unportable.
1008 * fixed some bugs revealed by Paul Dietz' test suite:
1009 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1011 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1012 ** SXHASH on condition objects no longer returns NIL.
1013 ** :ALLOCATION :CLASS slots are better treated; their values are
1014 updated on class redefinition, and initforms inherited from
1015 superclasses are applied.
1016 ** REMOVE-METHOD returns its generic function argument even when
1017 no method was removed.
1018 ** SHARED-INITIALIZE now initializes the values of the requested
1019 slots, including those with :ALLOCATION :CLASS.
1020 ** ALLOCATE-INSTANCE now works on structure classes defined via
1021 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1023 ** SLOT-EXISTS-P now works on conditions, as well as structures
1025 ** MAKE-LOAD-FORM now has the required methods on
1026 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1027 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1028 keyword, and now implements the SLOT-NAMES argument.
1029 ** methods with &OPTIONAL arguments no longer allow too many
1030 arguments to be passed in the call without error.
1031 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1032 option is consistent with the required arguments of the generic
1033 function lambda list.
1034 * bug fix: REQUIRE accepts a string designator. (Thanks to
1036 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1037 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1038 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1039 * bug fix: The compiler now checks for duplicated variables in macro
1041 * bug fix: SETQ on globals returns the correct value.
1042 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1043 not a COMPILER-ERROR (followed by some other strange error on
1044 choosing the CONTINUE restart).
1045 * bug fix: make.sh and friends are now more consistent in the way that
1046 they look for GNU "make".
1048 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1049 * experimental native threads support (on x86 Linux >=2.4 only).
1050 This is not compiled in by default: you need to add :SB-THREAD to
1051 the target features. See the "Beyond ANSI" chapter of the manual
1053 * fix for longstanding nonANSIism: The old distinction between
1054 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1055 The return value from CL:FIND-CLASS is now a CLOS class, and
1056 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1057 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1058 * An interface to the MetaObject Protocol, as described in Kiczales,
1059 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1060 MIT Press, 1991, is available from the SB-MOP package.
1061 * incompatible change: the SB-PCL package should now be considered
1062 a private implementation detail, and no longer a semi-private MOP
1064 * minor incompatible change: due to rearrangement for threads, the
1065 control stack and binding stack are now allocated at arbitrary
1066 addresses instead of being hardcoded per-port. Users affected by
1067 this probably have to be doing advanced things with shared
1068 libraries, and will know who they are.
1069 * minor incompatible change: Previously, all --eval forms used were
1070 processed with READ before any of them were processed with EVAL.
1071 Now each --eval form is processed with both READ and EVAL before
1072 the next --eval form is processed. (Thus package operations like
1073 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1074 work as the user might reasonably expect.)
1075 * minor incompatible change: *STANDARD-INPUT* is now only an
1076 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1078 * minor incompatible change: Y-OR-N-P is now character-oriented,
1079 not line oriented. Also, YES-OR-NO-P now works without errors.
1080 (thanks to Antonio Martinez)
1081 * sb-aclrepl module improvements: an integrated inspector, added
1082 repl features, and a bug fix to :trace command.
1083 * Known functions, which cannot be open coded by the backend, are
1084 considered to be able to check types of their arguments. (fixing
1085 a bug report by Nathan J. Froyd)
1086 * fixed a bug in computing method discriminating functions: It is
1087 now possible to define methods specialized on classes which have
1088 forward-referenced superclasses. (thanks to Gerd Moellmann)
1089 * fixed evaluation order in optional entries (reported by Gilbert
1091 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1092 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1093 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1094 required-by-AMOP NAME argument, as well as CLASS and
1095 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1096 * fixed bug 20: DEFMETHOD can define methods using names that are
1097 not the proper names of classes to designate class specializers.
1098 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1099 Unix file descriptors, instead of blowing up. (thanks to Antonio
1101 * Garbage collection refactoring: user-visible change is that a
1102 call to the GC function during WITHOUT-GCING will not do garbage
1103 collection until the end of the WITHOUT-GCING. If you were doing
1104 this you were probably losing anyway.
1105 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1106 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1107 TYPEP the latter but not the former.
1108 * The compiler issues a full WARNING for calls to undefined functions
1109 with names from the CL package.
1110 * MAP-INTO for a vector destination is open coded. (reported by
1111 Brian Downing on c.l.l)
1112 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1113 documentation string.
1114 * fixed some bugs revealed by Paul Dietz' test suite:
1115 ** COPY-ALIST now signals an error if its argument is a dotted
1117 ** Condition slots are now accessed more correctly in the presence
1118 of multiple initargs for a given slot.
1119 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1120 exclude restarts of the same name associated with a different
1122 ** DEFCLASS of forward-referenced classes with another
1123 forward-referenced class in the superclasses list no longer
1125 ** Condition slots are now initialized once each, not multiple
1126 times. (thanks to Gerd Moellmann)
1127 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1129 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1130 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1131 arguments contain duplicated elements.
1132 ** RESTART-CASE understands local macros.
1133 ** RESTART-CASE associates exactly its own restarts with a condition.
1134 ** ENDP in safe mode checks its argument to be of type LIST.
1135 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1136 in question is unbound.
1137 ** Optimized MAKE-INSTANCE functions no longer cause internal
1138 assertion failures in the presence of duplicate initargs.
1139 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1140 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1141 propagated even in the ctor optimized implementation of
1143 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1145 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1146 * a better implementation of SXHASH on (simple) bit vectors,
1147 measured both in execution speed and in distribution of results
1148 over the positive fixnums, has been installed. Likewise, a better
1149 implementation of EQUAL for simple bit vectors is now available.
1150 * fixed CEILING optimization for a divisor of form 2^k.
1151 * fixed bug 240 (emitting extra style warnings "using the lexical
1152 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1153 by Antonio Martinez)
1154 * fixed SXHASH, giving different results for NIL depending on type
1155 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1156 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1157 variable. (found by Rolf Wester)
1158 * SBCL does not ignore type declarations for special
1159 variables. (reported by rif on c.l.l 2003-03-05)
1160 * some bug fixes in contrib/sb-aclrepl/
1161 * fixed some bugs revealed by Paul Dietz' test suite:
1162 ** a bug in the CONS type specifier, whereby the CAR and CDR
1163 types got intertwined, has been fixed;
1164 ** the type system is now able to reason about the interaction
1165 between INTEGER and RATIO types more completely;
1166 ** APPEND, [N]REVERSE and NRECONC check that those their
1167 arguments, which must be proper lists, are really so;
1168 ** An array specialized to be unable to hold elements has been
1169 implemented, as required -- yes, really -- by ANSI;
1170 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1171 on malformed property lists;
1173 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1174 * incompatible packaging change: in line with Unix convention,
1175 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1176 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1177 * REQUIRE and PROVIDE are now optionally capable of doing something
1178 useful. See the documentation string for REQUIRE.
1179 * infrastructure for a managed SBCL contrib system: contributed
1180 modules in this release include:
1181 ** the ASDF system definition facility;
1182 ** an interface to the BSD Sockets API;
1183 ** an ACL-like convenience interface to the repl;
1184 (thanks to Kevin Rosenberg)
1185 ** an implementation of ROTATE-BYTE, with efficient implementation
1187 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1188 gives a closer approximation to the right answer than previously.
1189 (thanks to Raymond Toy)
1190 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1191 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1192 optional environment arguments, as required by ANSI.
1193 * fixed bugs in other functions taking environment objects, allowing
1194 calls with an explicit NIL environment argument to be compiled
1196 * fixed bug 228: primary return values from
1197 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1198 COMPILE or FUNCTION.
1199 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1200 :TYPE will no longer signal errors on innocuous objects.
1201 * fixed bug 231b: SETQ is better at respecting type declarations in
1202 the lexical environment.
1203 * fixed a bug in DEFCLASS: classes named by symbols with no or
1204 unprintable packages can now be defined.
1205 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1206 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1207 * fixed compiler failure related to checking types of functions.
1208 (reported by Robert E. Brown)
1209 * the compiler is now much more consistent in its error-checking
1210 treatment of bounding index arguments to sequence functions: in
1211 (SAFETY 3) code, errors will be signalled in almost all cases if
1212 invalid sequence bounding indices are passed to functions defined
1213 by ANSI to operate on sequences.
1214 * fixed a bug in the build procedure: documentation of SBCL-specific
1215 packages is now preserved and available in the final Lisp image.
1216 * lifted FDEFINITION lookup out of loops in the implementation of
1217 many list operations. (thanks to Robert E. Brown)
1218 * fixed a bug in the reader: the #n# reader macro now works for
1219 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1220 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1221 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1222 E. Brown and Tony Martinez respectively)
1223 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1224 SLOT-BOUNDP: in particular, it is now able to optimize them much
1225 better, and is now not vulnerable to having packages renamed.
1226 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1227 always called when a slot is not present in an instance. (thanks
1229 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1230 longer signal an error on unreading a character following EOF on
1231 the previous constituent stream. (thanks to Tony Martinez)
1232 * fixed some bugs revealed by Paul Dietz' test suite:
1233 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1234 not just nonnegative fixnums;
1235 ** the logical bit-array operators such as BIT-AND now accept an
1236 explicit NIL for their "opt-arg" argument (to indicate a
1237 freshly-consed result bit-array);
1238 ** ELT now signals an error on an invalid sequence index in safe
1240 ** the type system is now cleverer about negations of numeric
1241 types, and consequently understands the BIGNUM and RATIO types
1243 ** the type system is now cleverer about the interaction between
1244 INTEGER and RATIO types: while bugs still remain, many more
1245 cases are accurately computed;
1246 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1247 if it is in the last clause;
1248 ** CONSTANTLY now correctly returns a side-effect-free function in
1250 ** DECLARE is no longer treated as a special-operator; in
1251 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1252 * incremented fasl file version number due to the change in the
1253 DEFSTRUCT-SLOT-DESCRIPTION structure.
1255 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1256 * minor incompatible change: code processed by the "interpreter" or
1257 EVAL now has a compilation optimization policy of (DEBUG 2)
1258 (changed from (DEBUG 1)) to improve debuggability of interactive
1259 development, and to allow the use of the debug RETURN command in
1261 * an experimental implementation of the RETURN command for the
1262 debugger has been included. (thanks to Frederik Kuivinen)
1263 * fixed bug 62: constraints were not propagated into a loop.
1264 * fixed bug in embedded calls of SORT (reported and investigated by
1266 * fixed some bugs revealed by Paul Dietz' test suite:
1267 ** printing and reading of arrays with some dimensions having
1268 length 0 (thanks to Gerd Moellmann);
1269 ** BOA constructor with &AUX argument without a default value does
1270 not cause a type error;
1271 ** CONSTANTP now returns true for all self-evaluating objects.
1273 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1274 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1275 accessors that are related by inheritance, as specified in the
1276 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1277 Valtteri Vuorikoski)
1278 * The compiler is now able to inline functions that were defined in
1279 a complex lexical environment (e.g. inside a MACROLET).
1280 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1282 * Support for the upcoming FreeBSD-5.0 release has been included.
1283 (thanks to Dag-Erling Smorgrav)
1284 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1285 effect when it is not in a toplevel context.
1286 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1287 stand a better chance of being correct. (thanks to Gerd
1289 * fixed bug in COERCE, which now signals an error on coercing a
1290 rational to a bounded real type which excludes the expected
1292 * The compiler is now able to derive types more accurately from the
1293 COERCE and COMPILE functions.
1294 * fixed bug 223: functional binding is considered to be constant
1295 only for symbols in the CL package.
1296 * fixed bug 231: SETQ did not check the type of a variable being set
1297 (reported by Robert E. Brown)
1298 * A new optimization for MAKE-INSTANCE has been included, fixing
1299 various bugs (including relating to :ALLOCATION :CLASS slots and
1300 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1302 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1303 ** As required by ANSI, LOOP now disallows anonymous collection
1304 clauses such as COLLECT I in conjunction with aggregate boolean
1305 clauses such as THEREIS (= I 1);
1306 ** LOOP now signals an error when any variable is reused in the
1307 same loop (including the potentially useful construct analogous
1308 to WITH A = 1 WITH A = (1+ A);
1309 ** IT is only a special loop symbol within the first clause of a
1310 conditional loop clause;
1311 ** LOOP with a typed iteration variable over a hashtable now
1312 signals a type error iff it should.
1313 * fixed some other bugs revealed by Paul Dietz' test suite:
1314 ** FILE-STREAM now names the class previously known as FD-STREAM;
1315 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1316 argument) no longer signals an error;
1317 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1318 of the slot symbol, rather than using the current package
1319 ((:CONC-NAME "") continues to intern the slot's name in the
1321 * incremented fasl file version number, because of the incompatible
1322 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1323 of the new implementation of DEFINE-COMPILER-MACRO.
1325 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1326 * Support for building SBCL for MIPS platforms running in
1327 little-endian mode has now been checked in, and basic
1328 functionality on said platforms verified.
1329 * minor incompatible change: PCL now records the pathname of a file
1330 in which methods and the like are defined, rather than its
1332 * minor incompatible change: TRUENAME now considers the truename of
1333 a file naming a directory to be the pathname with :DIRECTORY
1334 component indicating that directory.
1335 * minor incompatible change: a NAMED clause in the extended form of
1336 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1337 reason for the previous behaviour is unclear.
1338 * more systematization and improvement of CLOS and MOP conformance
1339 in PCL (thanks to Gerd Moellman and Pierre Mai):
1340 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1342 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1343 primary methods with no specializers;
1344 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1346 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1347 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1348 FORWARD-REFERENCED-CLASSes; error reporting on
1349 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1351 ** SXHASH on CLOS instances now uses a slot internal to the
1352 instance to return different numbers on distinct instances,
1353 while preserving the same return value through invocations of
1355 ** DEFMETHOD signals errors when methods with longer incongruent
1356 lambda lists are added to generic functions;
1357 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1358 CLASS, as specified in AMOP;
1359 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1360 on the order returned by the primary method for classes of
1361 class STANDARD-CLASS;
1362 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1363 * fixed some bugs shown by Paul Dietz' test suite:
1364 ** DOLIST puts its body in TAGBODY;
1365 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1367 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1368 value producing form;
1369 ** if more variables are given to PROGV than values, extra
1370 variables are bound and made to have no value;
1371 ** NSUBSTITUTE on list arguments gets the right answer with
1373 ** ELT signals an error of type TYPE-ERROR when the index argument
1374 is not a valid sequence index;
1375 ** LOOP signals (at macroexpansion time) an error of type
1376 PROGRAM-ERROR when duplicate variable names are found;
1377 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1378 ** FUNCALL of special-operators now cause an error of type
1380 ** PSETQ now works as required in the presence of side-effecting
1381 symbol-macro places;
1382 ** NCONC accepts any object as its last argument;
1383 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1385 ** loop-for-as-package does not require a package to be explicitely
1387 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1389 * fixed bug 166: compiler preserves "there is a way to go"
1390 invariant when deleting code.
1391 * fixed bug 172: macro lambda lists with required arguments after
1392 &REST arguments now cause an error to be signalled. (thanks to
1394 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1395 bug. (thanks to Matthew Danish)
1396 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1398 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1399 arguments are lexically rebound. (thanks to Gerd Moellmann and
1401 * fixed bug 194: error messages are now more informative when there
1402 is no primary method applicable in a call to a generic
1403 function. (thanks to Gerd Moellmann)
1404 * fixed bug in command line argument checking (thanks to Julian
1406 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1407 (thanks to Matthew Danish)
1408 * incremented fasl file version number, because of the
1409 SXHASH-related changes in the layout of CLOS data structures
1411 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1412 * minor incompatible change: The runtime (the Unix executable named
1413 "sbcl") is now much pickier about the .core files it will load.
1414 Essentially it now requires .core files to descend from the same
1415 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1416 as the runtime does. (The intent is to prevent the crashes which
1417 can occur, and which can even be reported as mysterious failures,
1418 when people patch the sources or change the build parameters
1419 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1420 sbcl and .core files.)
1421 * fixed bug: VALUES-LIST is no longer optimized away.
1422 * fixed bug 142: The FFI conversion of C string values to Lisp
1423 string values no longer conses excessively. (thanks to Nathan
1424 Froyd porting Raymond Toy's fix to CMU CL)
1425 * began to systematize and improve MOP conformance in PCL (thanks to
1426 Nathan Froyd, Gerd Moellman and Pierre Mai):
1427 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1429 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1430 ** FINALIZE-INHERITANCE is now called on class finalization;
1431 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1432 argument precedence order.
1433 * fixed bug 202: The compiler no longer fails on functions whose
1434 derived types contradict their declared type.
1435 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1436 so it can be non-toplevel.
1437 * The fasl file version number has changed (because of the new
1438 implementation of DEFMACRO).
1439 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1440 safe code, that any length requirement by their type-specifier
1441 argument is valid. The exceptions to this are described in bug
1443 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1444 functions now check the types of their inputs as required by ANSI.
1445 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1446 introduced symbol is DECLAREd to be SPECIAL.
1447 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1448 * fixed bug: :COUNT argument to sequence functions may be negative
1449 * fixed bug: body of DO-SYMBOLS may contain declarations
1450 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1451 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1452 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1453 symbol macro only once
1454 * fixed printing of call frame when argument list is unavailable
1455 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1456 * compiler no longer signals WARNING on unknown keyword
1459 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1460 * A beta-quality port to the mips architecture running Linux,
1461 based on the old CMUCL backend, has been made. It has been tested
1462 on a big-endian kernel, and works sufficiently well to be able to
1463 rebuild itself; it has not been tested in little-endian mode.
1464 * fixed an inconsistency between gencgc.c and purify.c which made
1465 dumping/loading .core files unreliable
1466 * fixed bug 120a: The compiler now deals correctly with IFs where
1467 the consequent is the same as the alternative, instead of
1468 misderiving the return type. (thanks to Alexey Dejneka)
1469 * fixed bug 113: Logical pathnames are now dumpable (the logical
1470 host is resolved at load-time, throwing an error if it is not
1472 * fixed bug 174: FORMAT's error message is slightly clearer when a
1473 non-printing character is used in a format directive.
1474 * fixed several bugs in compiler checking of type declarations, i.e.
1475 violations of the Python "declarations are assertions" principle
1476 (thanks to Alexey Dejneka)
1477 * fixed several bugs in PCL's error checking (thanks to Gerd
1479 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1481 * fixed bug in compilation of functions as first class values
1482 (thanks to Antonio Martinez-Shotton)
1483 * The compiler's handling TYPE-ERRORs which it can prove will
1484 inevitably happen at runtime has been cleaned up and corrected
1485 in several ways. (thanks to Alexey Dejneka)
1486 * improved argument type checking for various basic arithmetic
1487 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1488 lavished on them in the past that they can be compiled in many
1489 ways in different special cases
1490 * fixed bug 181: compiler checks validity of user supplied type
1492 * cleaned up code flushing in optimization: Function calls which
1493 should signal errors for safety purposes (e.g. which ANSI says
1494 should signal errors when their arguments are of incorrect type)
1495 are no longer optimized away.
1496 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1497 * incremented fasl file version number, because changes in the
1498 implementation of sequence functions like COERCE caused
1499 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1500 in old inline expansions) to become undefined. (Actually these
1501 changes were later undone, so we might very well be binary
1502 compatible with 0.7.7 after all, but leaving the version number
1503 incremented seemed like the simplest and most conservative
1506 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1507 * An alpha-quality port to the parisc architecture running Linux,
1508 based on the old CMUCL backend, has been made. This, even more so
1509 than the other backends, should be considered still a work in
1510 progress; known problems include that the Linux kernel in 64-bit
1511 mode does not propagate the correct sigcontext structure to
1512 userspace, and consequently SBCL on a parisc64 kernel will not
1514 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1515 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1516 suppressed.) Also now that the compiler is looking at declarations
1517 in the environment, it checks optimization declarations as well,
1518 and suppresses inlining when (> DEBUG SPEED).
1519 * More fixes have been made to treatment of floating point exception
1520 treatment and other Unix signals. In particular, floating point
1521 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1522 * The detection and handling of control stack exhaustion (infinite
1523 or very deeply nested recursion) has changed. Stack exhaustion
1524 detection is now done by write-protecting pages at the OS level
1525 and applies at all optimization settings; when found, a
1526 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1527 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1528 be caught using IGNORE-ERRORS.
1529 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1530 correctly when transforming recursive LABELS functions to LETs.
1531 (thanks to Alexey Dejneka)
1532 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1533 that are names of constants or global variables.
1534 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1535 alien routines with docstrings.
1536 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1537 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1539 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1540 (thanks to Eric Marsden)
1541 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1542 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1543 * Bug fix: The SPARC backend can now compile functions involving
1544 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1545 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1546 to a functions expecting a PATHNAME-DESIGNATOR.
1547 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1548 lambda lists. (thanks to Alexey Dejneka)
1549 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1550 functions so that EXPORTing the name of the class doesn't cause
1551 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1552 to Antonio Martinez for reporting this)
1553 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1554 constant-folded without failing an assertion. (thanks to Einar
1555 Floystad Dorum for reporting this)
1556 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1557 types are now dealt with more correctly by the compiler.
1558 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1559 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1560 behaviour analogous to RENAME-FILE. This puts its behaviour more
1561 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1563 * The fasl file version number has changed again. (because of the
1564 bug fix involving the names of PCL MAKE-INSTANCE functions)
1566 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1567 * bug fix: Floating point exceptions are treated much more
1568 consistently on the x86/Linux and PPC/Linux platforms.
1569 * Array initialization with :INITIAL-ELEMENT is now much faster for
1570 cases when the compiler cannot open code the array creation, but
1571 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1572 array accesses have also seen a speed increase.
1573 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1574 specified. (This was at the root of some bad interactions between
1575 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1577 * bug fix: Internal error arguments for undefined functions are now
1578 computed correctly on the PPC/Linux platform.
1579 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1580 Raymond Toy's patch for CMU CL)
1581 * Support for the Solaris 9 operating environment has been included
1582 (thanks to Daniel Merritt)
1583 * A very ugly but hopefully complete draft of the missing FFI chapter
1584 of the manual has been created by reformatting the corresponding
1585 CMU CL manual chapter into (currently very ugly and incoherent)
1586 DocBook and bringing it up to date for SBCL behavior. Thus, the
1587 manual is now essentially complete, at least by my extreme
1588 once-and-only-once standards, whereby it's acceptable to refer to
1589 the doc strings of SB-EXT functions as the primary documentation.
1590 * The fasl file version number has changed again, due to cleanup of
1591 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1592 is no longer a static symbol.)
1594 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1595 * SBCL now builds with OpenMCL (version 0.12) as the
1596 cross-compilation host; also, more progress has been made toward
1597 bootstrapping under CLISP.
1598 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1600 * bug 158 fixed: The compiler can now deal with integer loop
1601 increments different from 1; fixing this turned out also to fix
1603 * bug 169 fixed: no more bogus warnings about using lexical bindings
1604 despite the presence of perfectly good SPECIAL declarations (thanks
1605 to David Lichteblau)
1606 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1607 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1608 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1610 * bug 180 fixed: Method combination specifications no longer ignore
1611 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1612 * bug fix: Structure type predicate functions now check their argument
1613 count as they should.
1614 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1615 correctly. (thanks to Pierre Mai)
1616 * minor incompatible change: The --noprogrammer option is deprecated
1617 in favor of the new --disable-debugger option, which is very similar.
1618 (The major difference is that it takes effect at a slightly different
1619 time at startup, causing handling of errors in --sysinit and
1620 --userinit files will be affected differently.) The
1621 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1622 been added to allow this functionality to be controlled from ordinary
1623 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1624 maintainers, who might want to run non-interactive scripts to
1625 build SBCL cores which will later be used interactively.)
1626 * minor incompatible change: The LOAD function no longer, when given
1627 a wild pathname to load, loads all files matching that pathname.
1628 Instead, an error of type FILE-ERROR is signalled.
1630 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1631 * bug 147 fixed: The compiler preserves its block link/count
1632 invariants more correctly now so that it doesn't crash. (thanks
1634 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1636 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1637 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1638 no-value-for-this-character as NIL (as specified by ANSI).
1639 * HOST-NAMESTRING on physical pathnames now returns a string that is
1640 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1641 (thanks to Christophe Rhodes)
1642 * The Alpha port handles icache flushing more correctly. (thanks to
1644 * More progress has been made toward bootstrapping under CLISP. (thanks
1645 to Christophe Rhodes)
1646 * The fasl file format has changed again, because dynamic loading
1647 on OpenBSD (which has non-ELF object files) motivated some cleanups
1648 in the way that foreign symbols are transformed and passed around.
1649 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1650 is no longer treated as whitespace by the reader, but instead as
1651 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1652 |A
\7fB|, instead of A as it used to.
1654 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1655 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1656 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1657 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1658 this as of 0.7.1.45, but was left out of the previous news section
1659 (thanks to Dan Barlow)
1660 * SBCL now runs on the Solaris operating system on SPARC architectures
1661 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1662 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1663 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1664 * SPARC backend cleanups, allowing builds of cores optimized for V8
1665 and V9 SPARCS, and also emission of code targeted to a particular
1666 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1668 * SBCL is closer to bootstrapping under CLISP, thanks to various
1669 fixes by Christophe Rhodes.
1670 * The fasl file format has changed again, to allow the compiler's
1671 INFO database to support symbol macros.
1672 * The user manual (in doc/) is formatted into HTML more nicely.
1673 (thanks to coreythomas)
1674 * The system is smarter about SUBTYPEP relationships, especially
1675 those involving NOT types (including types such as ATOM which are
1676 represented internally using NOT types). Thus SUBTYPEP is less
1677 likely to return (VALUES NIL NIL) in general, and in particular
1678 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1680 * The fasl file format has changed again, because the internal
1681 representation of types now includes a new slot to support the new
1682 SUBTYPEP-of-NOT-types logic.
1683 * (not a change in the main branch of SBCL, but a related prototype
1684 which can hopefully be merged into the main branch of SBCL in the
1685 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1686 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1687 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1688 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1689 dispatch macro character combinations. (thanks to Alexey Dejneka)
1690 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1691 that superficially look like logical namestrings correctly.
1692 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1694 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1695 default supplied-p) for &optional and &key arguments. (thanks to
1698 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1699 * incompatible change: The compiler is now less aggressive about
1700 tail call optimization, doing it only when (> SPACE DEBUG) or
1701 (> SPEED DEBUG). (This is an incompatible change because there are
1702 programs which relied on the old CMU-CL-style behavior to optimize
1703 away their unbounded recursion which will now die of stack overflow.)
1704 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1705 for non-GENCGC systems has been increased to 20M (since that
1706 seems much closer to the likely performance optimum for modern
1707 systems than the old 4M value was)
1708 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1709 and *DEBUG-PRINT-LEVEL*
1710 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1711 of CMU CL's support for SPARC, and various endianness and other
1712 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1713 * new syntactic sugar for the Unix command line: --load foo.bar is now
1714 an alternate notation for --eval '(load "foo.bar")'.
1716 ** The system now detects stack overflow and handles it gracefully,
1717 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1718 optimization settings. (This is a good thing in general, and
1719 its introduction in this version should be particularly timely
1720 for anyone whose code fails because of suppression of tail
1722 ** The system now hunts for the C variable "environ" in a more
1723 devious way, to avoid segfaults when the C library version differs
1724 between compile time and run time. (thanks to Christophe Rhodes)
1725 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1726 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1727 ** The compiler no longer issues bogus style warnings for undefined
1728 classes in the same source file as the DEFCLASSes which defined
1729 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1731 ** fixes in CONDITION class precedence list for undefined function
1732 errors (thanks to Alexei Dejneka)
1733 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1734 correctly. (thanks to Dan Barlow)
1735 ** portability fixes aiming at bootstrapping under CLISP (thanks
1736 to Dave McDonald and Christophe Rhodes)
1737 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1738 ** fixes in type translation and and type inference (thanks to
1740 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1741 ** various fixes in the new ports (thanks to Dan Barlow)
1742 * several changes related to debugging:
1743 ** suppression of tail recursion, as noted above
1744 ** stack overflow detection, as noted above
1745 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1746 is now the default. (For some time encapsulation has been more
1747 reliable than the breakpoint-based :ENCAPSULATE NIL
1748 implementation, at least on X86 systems; and I just noticed that
1749 encapsulation also seems closer to the spirit of the ANSI
1752 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1754 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1755 up properly again. (There was a packaging bug in 0.7.0 which
1756 left their definitions in SB-SYS::LOAD-FOREIGN and
1757 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1758 vital for most things which interface to C-level interfaces,
1759 like extensions working with sockets or databases or
1760 Perl-compatible regexes or whatever, and the need to fix
1761 this bug is the main reason that 0.7.1 was released so
1763 ** DEFGENERIC is now choosier about the methods it redefines, so that
1764 reLOADing a previously-LOADed file containing DEFGENERICs does
1765 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1766 cycle now works as it should. (thanks to Alexey Dejneka)
1767 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1768 fixed by Christophe Rhodes. (He actually submitted this patch
1769 months ago, and I delayed until after 0.7.0.)
1770 ** Bug 111 (internal compiler confusion about runtime checks on
1771 FUNCTION types) was fixed by Alexey Dejneka.
1772 * Some internal cleanups (getting rid of variables which aren't
1773 needed now that the byte interpreter is gone) caused the fasl
1774 file format number to change again.
1776 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1777 * major incompatible change: The default fasl file extension, i.e. the
1778 default extension for files produced by COMPILE-FILE, has changed
1779 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1781 ** There are many changes in the implementation of the compiler.
1782 SBCL is now essentially a compiler-only implementation of ANSI
1783 Common Lisp. EVAL still "interprets" a few special cases, but
1784 almost all the interesting cases are handled by creating
1785 a LAMBDA expression, calling COMPILE on it, then calling
1786 FUNCALL on the result.
1787 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1788 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1789 Since the code is newer, there might still be some new bugs
1790 (though not as many as before Martin Atzmueller's fixes:-). But
1791 the new code is substantially simpler and clearer, and hopefully
1792 any remaining bugs will be simpler, less fundamental, and more
1793 fixable then the bugs in the old code.
1794 ** The revised compiler is still a little unsteady on its feet.
1796 *** The debugging information it produces (particularly the names
1797 of FUNCTION objects) is sometimes much less useful than what
1798 the old compiler produced.
1799 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1800 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1801 within a MACROLET) has been temporarily weakened.
1802 ** There are new compiler optimizations for various functions:
1803 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1804 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1805 *** the math functions TRUNCATE, FLOOR, and CEILING
1806 *** the function-of-all-trades COERCE
1807 Mostly these should be transparent, but there's one
1808 potentially-annoying problem (bug 117): when the compiler
1809 inline-expands a function and does type analysis on the result,
1810 it can create control paths which have type mismatches, and
1811 when it can't prove that those control paths aren't taken,
1812 it will issue WARNINGs about the type mismatches. This is
1813 a particular problem in practice for the new sequence functions.
1814 It's not clear how this should be fixed, and for now, a
1815 workaround is given in the entry for 117 in the BUGS file.
1816 ** (Because of the interaction between the two previous items --
1817 occasional inlining problems and new inline expansions -- some
1818 of the new sequence function optimizations won't really kick in
1819 completely until debugging information, and then inlining, are
1820 straightened out in some future version.)
1821 * minor incompatible changes:
1822 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1823 DIRECTORY no longer implicitly promotes NIL slots of its
1824 pathname argument to :WILD. In particular, when you ask for the
1825 contents of a directory (which you used to be able to do without
1826 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1827 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1828 ** changes in behavior that ANSI explicitly defines to be
1829 implementation dependent:
1830 *** The new compiler-only implementation still conforms with ANSI,
1831 but acts a little different than before. Besides the obvious
1832 changes in performance tradeoffs (that the cost per form passed
1833 to EVAL has gone up, and the cost per form executed by EVAL
1834 has gone down), the behavior of the system changes a little
1835 because there are no longer any interpreted function objects.
1836 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1837 e.g. doing COMPILE on the output of interactive DEFUN is
1839 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1841 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1842 has changed from (:CL) to NIL.
1843 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1844 e.g. #\Newline and #\Tab, don't have names specified in the
1845 ANSI Common Lisp standard, is now based on their ASCII symbolic
1846 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1847 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1848 are no longer used for output.
1849 ** changes in internal implementation constants:
1850 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1851 4 million. (If your application spends a lot of time GCing and
1852 you have a lot of RAM, you might want to experiment with
1853 increasing it even more.)
1854 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1855 However, almost all old code should still continue to work without
1856 immediate update, as SB-C-CALL is now a (deprecated) nickname
1858 ** Old operator names in the style DEF-FOO are now deprecated in
1859 favor of new corresponding names DEFINE-FOO, for consistency with
1860 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1861 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1862 internal symbols, but a few supported extensions like
1863 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1864 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1865 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1866 etc. as you get deeper into recursive calls to the debugger
1867 command loop, instead of the old "5]", "5]]", "5]]]"
1868 sequence. (I was motivated to do this when squabbles between
1869 ILISP and SBCL left me very deeply nested in the debugger. In the
1870 short term, this change will probably provoke more ILISP/SBCL
1871 squabbles, but hopefully it will be an improvement in the long run.)
1872 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1873 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1874 function, since declaiming return types involving aliens is
1875 (1) annoyingly messy to do by hand and (2) vital to efficient
1876 compilation of code which calls such functions.
1877 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1878 longer reexported by the SB-EXT package. They're solely useful
1879 for alien code, so it seems more logical that you should get
1880 them from the SB-ALIEN package, not in SB-EXT.
1881 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1882 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1883 features. Instead, the code that they used to control is always
1884 built into the system.
1885 * many other bug fixes
1886 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1887 advantage of the new EVAL-WHEN stuff and to clean them up in
1888 general, and they are now more ANSI-compliant in a number of
1889 ways. Martin Atzmueller is responsible for a lot of this.
1890 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1892 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1893 *** correct ERROR type for various file operations
1894 *** some fixes for Lisp streams
1895 *** DEFMETHOD syntax checking
1896 *** changing old weird representation of debug information as
1897 strings (which, among their other deficiencies, don't transform
1898 correctly when you rename packages, and don't change their
1899 print representation when you change things like *PACKAGE*
1900 and *PRINT-LENGTH*) to symbols and lists of symbols
1901 He also made several improvements and fixed several bugs in DESCRIBE.
1902 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1904 *** misbehavior of WRITE-STRING/WRITE-LINE
1905 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1906 and several other LOOP problems as well
1907 *** DIRECTORY when similar filenames are present
1908 *** DEFGENERIC with :METHOD options
1909 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1910 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1911 *** argument ordering in FIND with :TEST option
1912 *** mishandled package designator argument in APROPOS-LIST
1913 *** various problems in the backquote readmacro
1914 *** a bug in APROPOS
1915 *** probably some others that I'm not describing very well here,
1916 since the CVS log documents them by reference to sbcl-devel
1917 messages, and the SourceForge archives aren't working well.:-(
1918 ** Dan Barlow improved the Alpha port (and is making progress on the
1919 PPC port, for those of you who think different).
1920 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
1921 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
1922 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
1923 continues to work on the SPARC port (for those of you in a position
1924 to look down upon our little PC-compatible boxes from a great height).
1925 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
1926 and uses that copy, rather than the current dynamic value, when
1927 it's trying to decide whether to truncate output. Thus e.g.
1928 (let ((*print-lines* 50))
1929 (pprint-logical-block (stream nil)
1931 (let ((*print-lines* 8))
1932 (print (aref possiblybigthings i) stream)))))
1933 should now truncate the logical block only at 50 lines, instead of
1934 often truncating it at 8 lines, as it did before.
1935 * The doc/cmucl/ directory, containing old CMU CL documentation
1936 from the time of the fork, is no longer part of the base system.
1937 SourceForge has shut down its anonymous FTP service, and with it
1938 my original plan for distributing the old CMU CL documentation
1939 there. For now, if you need these files you can download an old
1940 SBCL source release and extract them from it.
1941 * The fasl file version number changed again, for dozens of reasons,
1942 some of which are apparent above.
1944 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
1945 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
1946 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
1947 patch, so that DISASSEMBLE output is much nicer.
1948 * The code in the SB-PROFILE package now seems reasonably stable.
1949 I still haven't decided what the final interface should look like
1950 (I'd like PROFILE to interact cleanly with TRACE, since both
1951 facilities use function encapsulation) but if you have a need
1952 for profiling now, you can probably use it successfully with
1953 the current CMU-CL-style interface.
1954 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
1955 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
1956 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
1957 link now returns the dangling link itself, and for similar
1958 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
1959 link itself. (In these cases the old code signalled an error and
1960 looped endlessly, respectively.) Thus, DIRECTORY now works even
1961 in the presence of dangling and cyclic symbolic links.
1962 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
1963 is now a supported extension again, since the consensus on
1964 sbcl-devel was that it can be useful for ordinary development
1965 work, not just for debugging SBCL itself.
1966 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
1967 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
1968 that currently-defined functions might be redefined later with
1969 different return types.
1970 * Hash tables can be printed readably, as inspired by CMU CL code
1971 of Eric Marsden and SBCL code of Martin Atzmueller.
1972 * better error handling in CLOS method combination, thanks to
1973 Martin Atzmueller porting Pierre Mai's CMU CL patches
1974 * more overflow fixes for >16Mbyte I/O buffers
1975 * A bug in READ has been fixed, so that now a single Ctrl-D
1976 character suffices to cause end-of-file on character streams.
1977 In particular, now you only need one Ctrl-D at the command
1978 line (not two) to exit SBCL.
1979 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
1981 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
1982 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
1983 does the right thing.
1984 * The implementation of some type tests, especially for CONDITION
1985 types, is now tidier and maybe faster, due to CMU CL code
1986 originally by Douglas Crosher, ported by Martin Atzmueller.
1987 * Some math functions have been fixed, and there are new
1988 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
1989 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
1990 * (There are also some new optimizers in contrib/*-extras.lisp. Those
1991 aren't built into sbcl-0.6.13, but are a sneak preview of what's
1992 likely to be built into sbcl-0.7.0.)
1993 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
1994 CL, ported to SBCL by Martin Atzmueller)
1995 * DESCRIBE now gives more information in some cases. (Pierre Mai's
1996 patch to CMU CL, ported to SBCL by Martin Atzmueller)
1997 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
1998 * There's a new slam.sh hack to shorten the edit/compile/debug
1999 cycle for low-level changes to SBCL itself, and a new
2000 :SB-AFTER-XC-CORE target feature to control the generation of
2001 the after-xc.core file needed by slam.sh.
2002 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2003 COMPILE-FILE is no longer supported, so that now every function
2004 gets an entry point, so that block compilation looks a little
2005 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2006 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2007 returns the number of bytes consed since the system started,
2008 rather than the number consed since the first time the function
2009 was called. (The new definition parallels ANSI functions like
2010 CL:GET-INTERNAL-RUN-TIME.)
2011 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2012 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2013 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2014 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2015 * Fasl file version numbers are now independent of the target CPU,
2016 since historically most system changes which required version
2017 number changes have affected all CPUs equally. Similarly,
2018 the byte fasl file version is now equal to the ordinary
2021 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2022 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2023 is no longer recognized. I apologize for this, because it was
2024 listed in SB-EXT as a supported extension, but I found that
2025 its existing behavior was poorly specified, as well as incorrectly
2026 specified, and it looked like too much of a mess to straighten it
2027 out. I have enough on my hands trying to get ANSI stuff to work..
2028 * many patches ported from CMU CL by Martin Atzmueller, with
2029 half a dozen bug fixes in pretty-printing and the debugger, and
2030 half a dozen others elsewhere
2031 * fixed bug 13: Floating point infinities are now supported again.
2032 They might still be a little bit flaky, but thanks to bug reports
2033 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2034 as flaky as they were.
2035 * The --noprogrammer command line option is now supported. (Its
2036 behavior is slightly different in detail from what the old man
2037 page claimed it would do, but it's still appropriate under the
2038 same circumstances that the man page talks about.)
2039 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2040 are now supported, and enabled by default. Thus, the compiler can
2041 handle many floating point and complex operations much less
2042 inefficiently. (Thus e.g. you can implement a complex FFT
2044 * The compiler now detects type mismatches between DECLAIM FTYPE
2045 and DEFUN better, and implements CHECK-TYPE more correctly, and
2046 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2047 thanks to patches from Martin Atzmueller.
2048 * various fixes to make the cross-compiler more portable to
2049 ANSI-conforming-but-different cross-compilation hosts (notably
2050 Lispworks for Windows, following bug reports from Arthur Lemmens)
2051 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2052 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2053 fixed thanks to Pierre Mai's CMU CL patches.
2054 * fixes to keep the system from overflowing internal counters when
2055 it tries to use i/o buffers larger than 16M bytes
2056 * fixed bug 45a: Various internal functions required to support
2057 complex special functions have been merged from CMU CL sources.
2058 (When I was first setting up SBCL, I misunderstood a compile-time
2059 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2060 * improved support for type intersection and union, fixing bug 12
2061 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2062 more obscure bugs as well
2063 * some steps toward byte-compiling non-performance-critical
2064 parts of the system, courtesy of patches from Martin Atzmueller
2065 * Christophe Rhodes has made some debian packages of sbcl at
2066 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2067 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2068 stable, but are nonetheless usable. When he's ready, I'd be happy
2069 to add them to the SourceForge "File Releases" section. (And if
2070 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2071 * new fasl file format version number (because of changes in
2072 internal representation of (OR ..) types to accommodate the new
2073 support for (AND ..) types, among other things)
2075 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2076 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2078 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2080 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2081 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2082 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2083 * DESCRIBE now works on structure objects again.
2084 * Most function call argument type mismatches are now handled as
2085 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2086 know whether the function will be redefined before the call is
2087 executed. (The compiler could flag local calls with full WARNINGs,
2088 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2089 it doesn't keep track of enough information to know whether calls
2090 are local in this sense.)
2091 * Compiler output is now more verbose, with messages truncated
2092 later than before. (There should be some supported way for users
2093 to override the default verbosity, but I haven't decided how to
2094 provide it yet, so this behavior is still controlled by the internal
2095 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2096 src/compiler/ir1util.lisp.)
2097 * Fasl file format version numbers have increased again, because
2098 support for the Gray streams extension changes the layout of the
2099 system's STREAM objects.
2100 * The Gray subclassable streams extension now works, thanks to a
2101 patch from Martin Atzmueller.
2102 * The full LOAD-FOREIGN extension (not just the primitive
2103 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2104 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2105 but like most other programs, it defaults to copying the Unix
2106 environment from the original process instead of starting the
2107 new process in an empty environment.
2108 * Extensions which manipulate the Unix environment now support
2109 an :ENVIRONMENT keyword option which doesn't smash case or
2110 do other bad things. The CMU-CL-style :ENV option is retained
2111 for porting convenience.
2112 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2113 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2115 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2117 * A patch from Martin Atzmueller seems to have solved the SIGINT
2118 problem, and as far as we know, signal-handling now works cleanly.
2119 (If you find any new bugs, please report them!)
2120 * The system no longer defaults Lisp source file names to types
2121 ".l", ".cl", or ".lsp", but only to ".lisp".
2122 * The compiler no longer uses special default file extensions for
2123 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2124 seems to expect a single default extension for all compiled code,
2125 and there's no compelling reason to try to stretch the standard
2126 to allow two different extensions.) Instead, byte-compiled files
2127 default to the same extension as native-compiled files.
2128 * Fasl file format version numbers have increased again, because
2129 a rearrangement of internal implementation packages made some
2130 dumped symbols in old fasl files unreadable in new cores.
2131 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2132 many fewer weird special cases.
2133 * Bug #17 (differing COMPILE-FILE behavior between logical and
2134 physical pathnames) has been fixed, and some related misbehavior too,
2135 thanks to a patch from Martin Atzmueller.
2136 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2137 by Tim Moore, ported to SBCL by Martin Atzmueller.
2138 * Martin Atzmueller fixed several filesystem-related problems,
2139 including bug #36, in part by porting CMU CL patches, which were
2140 written in part by Paul Werkowski.
2141 * More compiler warnings in src/runtime/ are gone, thanks to
2142 more patches from Martin Atzmueller.
2143 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2146 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2148 * DESCRIBE now works on CONDITION objects.
2149 * The debugger now handles errors which arise when trying to print
2150 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2152 * The build system now uses an additional file, customize-target-features.lisp,
2153 to allow local modifications to the target *FEATURES* list. (The point of
2154 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2155 debugging features enabled, without having to worry about propagating it
2156 into everyone's system when I do a "cvs update".) When no
2157 customize-target-features.lisp file exists, the target *FEATURES* list
2158 should be constructed the same way as before.
2159 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2160 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2161 warn about behavior which is undefined under the ANSI spec. Note: This
2162 is specified by ANSI, but it's not very popular with programmers.
2163 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2164 macro in the SBCL sources for an example of a workaround which you
2165 might use to make portable ANSI-standard code which does what you want.
2166 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2167 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2168 when it's used as a non-toplevel form. (This is required in order
2169 to implement the DEFCONSTANT-EQX macro.)
2170 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2171 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2172 funny things, due to bugs in the implementation of EVAL-WHEN
2173 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2174 non-toplevel DEFCONSTANTs are uncommon.)
2175 * The core file version number and fasl file version number have been
2176 incremented, because the old noncompliant DEFCONSTANT behavior involved
2177 calling functions which no longer exist, and because I also took the
2178 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2179 * fixed bug 1 (error handling before read-eval-print loop starts), and
2180 redid debugger restarts and related debugger commands somewhat while
2182 ** The QUIT debugger command is gone, since it did something
2183 rather different than the SB-EXT:QUIT command, and since it never
2184 worked properly outside the main toplevel read/eval/print loop.
2185 Invoking the new TOPLEVEL restart provides the same functionality.
2186 ** The GO debugger command is also gone, since you can just invoke
2187 the CONTINUE restart directly instead.
2188 ** The TOP debugger command is also gone, since it's redundant with the
2189 FRAME 0 command, and since it interfered with abbreviations for the
2191 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2193 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2194 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2195 some porting work by Martin Atzmueller)
2196 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2197 with a boatload of patches which he ported to SBCL. Now that those
2199 ** The system tries to make sure that its low-priority messages
2200 are prefixed by semicolons, to help people who like to use
2201 syntax highlighting in their ILISP buffer. (This patch
2202 was originally due to Raymond Toy.)
2203 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2204 patches originally written by Raymond Toy.
2205 ** The compiler understands coercion between single-value and
2206 multiple-VALUES type expressions better, getting rid of some very
2207 weird behavior, thanks to patches originally by Robert MacLachlan
2208 and Douglas Crosher.
2209 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2210 lambda lists, thanks to a patch originally by Pierre Mai.
2211 ** The system no longer bogusly warns about "abbreviated type
2213 ** The compiler gets less confused by inlining and RETURN-FROM,
2214 thanks to some patches originally by Tim Moore.
2215 ** The system no longer hangs when dumping circular lists to fasl
2216 files, thanks to a patch originally from Douglas Crosher.
2217 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2218 undefined function error.
2219 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2220 in structure slot initforms) for now, documented workaround instead:-|
2221 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2222 * fixed bug 5: added stubs for various Gray stream functions called
2223 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2224 installed, at least appropriate type errors are generated
2225 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2226 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2227 consistently in DEFMETHOD forms.
2228 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2229 it doesn't seem to affect SBCL after all
2230 * The C runtime system now builds with better optimization and many
2231 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2233 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2235 * The system is now under CVS at SourceForge (instead of the
2236 CVS repository on my home machine).
2237 * The new signal handling code has been tweaked to treat register
2238 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2239 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2240 downstream, e.g. in debug-int.lisp, has implicit dependencies
2241 on the unsignedness of integer representation of machine words,
2242 and that caused the system to bomb out with infinite regress
2243 when trying to recover from type errors involving signed values,
2244 e.g. (BUTLAST '(1 2 3) -1).)
2245 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2246 (This was one of the bugs Peter Van Eynde reported back in July.)
2247 * The system now uses code inspired by Colin Walters' O(N)
2248 implementation of MAP (from the cmucl-imp@cons.org mailing
2249 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2250 inline the MAP operation, and there is more than one
2251 sequence argument to the MAP call (so that it can't just
2252 do ETYPECASE once and for all based on the type of the
2253 single sequence argument). (The old non-inline implementation
2254 of the general M-argument sequence-of-length-N case required
2255 O(M*N*N) time when any of the sequence arguments were LISTs.)
2256 * The QUIT :UNIX-CODE keyword argument has been renamed to
2257 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2258 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2260 * Raymond Wiker's patches to port dynamic loading from Linux to
2261 FreeBSD have been added.
2262 * The BUGS file is now more nearly up to date, thanks in large part
2263 to Martin Atzmueller's review of it.
2264 * The debugger now flushes standard output streams before it begins
2265 its output ("debugger invoked" and so forth).
2266 * The core version number and fasl file version number have both
2267 been incremented, because of incompatible changes in the layout
2269 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2270 used to not be called for a saved Lisp image.)
2271 * Martin Atzmueller's version of a patch to fix a compiler crash,
2272 as posted on sbcl-devel 13 September 2000, has been installed.
2273 * Instead of installing Martin Atzmueller's patch for the
2274 compiler transform for SUBSEQ, I deleted the compiler transform,
2275 and transforms for some similar consing operations.
2276 * A bug in signal handling which kept TRACE from working on OpenBSD
2278 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2279 away by constant folding
2280 * The system now defines its address space constants in one place
2281 (in the Lisp sources), and propagates them automatically elsewhere
2282 (through GENESIS and the sbcl.h file). Therefore, patching the
2283 address map is less unnecessarily tedious and error-prone. The
2284 Lisp names of address space constants have also been systematized.
2285 * CVS tags like dollar-Header-dollar have been removed from
2286 the sources, because they have never saved me trouble and
2287 they've been source of trouble working with patches and other
2288 diff-related operations.
2289 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2290 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2292 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2294 * The system has been ported to OpenBSD.
2295 * The system now compiles with a simple "sh make.sh" on the systems
2296 that it's supported on. I.e., now you no longer need to tweak
2297 text in the target-features.lisp-expr and symlinks in src/runtime/
2298 by hand, the make.sh takes care of it for you.
2299 * The system is no longer so grossly inefficient when compiling code
2300 involving vectors implemented as general (not simple) vectors (VECTOR T),
2301 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2302 which dares to use the various sequence functions on non-simple
2303 vectors, takes less of a performance hit.
2304 * There is now a primitive type predicate VECTOR-T-P
2305 to test for the (VECTOR T) type, so that e.g.
2306 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2307 can now be compiled with some semblance of efficiency. (The old code
2308 turned the type declaration into a full call to %TYPEP at runtime!)
2309 * AREF on (VECTOR T) is still not fast, since it's still compiled
2310 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2311 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2312 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2313 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2314 without making full calls to SUBTYPEP at runtime.
2315 (Some analogous efficiency issues for non-simple vectors specialized to
2316 element types other than T, or for non-simple multidimensional arrays,
2317 have not been addressed. They could almost certainly be handled the
2318 same way if anyone is motivated to do so.)
2319 * The changes in array handling break binary compatibility, so
2320 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2321 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2322 instead of (VALUES T T).
2323 * By following the instructions that Dan Barlow posted to sbcl-devel
2324 on 2 July 2000, I was able to enable primitive dynamic object
2325 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2326 functionality is not implemented (since it calls ld to resolve
2327 library references automatically, requiring RUN-PROGRAM for its
2328 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2329 try to resolve library references) is now supported.
2330 * The system now flushes the standard output streams when it terminates,
2331 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2332 them at several other probably-convenient times, e.g. in each pass of
2333 the toplevel read-eval-print loop, and after evaluating a form given
2334 as an "--eval" command-line option. (These changes were motivated by a
2335 discussion of stream flushing issues on cmucl-imp in August 2000.)
2336 * The source transform for TYPEP of array types no longer assumes
2337 that an array whose element type is a not-yet-defined type
2338 is implemented as an array of T, but instead punts, so that the
2339 type will be interpreted at runtime.
2340 * There is now some support for cross-compiling in make.sh: each of
2341 the phases of make.sh has its own script. (This should be transparent
2342 to people doing ordinary, non-cross-compile builds.)
2343 * Since my laptop doesn't have hundreds of megabytes of memory like
2344 my desktop machine, I became more motivated to do some items on
2345 my to-do list in order to reduce the size of the system a little:
2346 ** Arrange for various needed-only-at-cold-init things to be
2347 uninterned after cold init. To support this, those things have
2348 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2349 symbols with such names are now uninterned after cold init).
2350 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2351 fasl files for cold load.
2352 ** Remove the old compiler structure pooling code (which used to
2353 be conditional on the target feature :SB-ALLOC) completely.
2354 ** Redo the representation of some data in cold init to be more compact.
2355 (I also looked into supporting byte compiled code at bootstrap time,
2356 which would probably reduce the size of the system a lot, but that
2357 looked too complicated, so I punted for now.)
2358 * The maximum signal nesting depth in the src/runtime/ support code has
2359 been reduced from 4096 to 256. (I don't know any reason for the very
2360 large old value. If the new smaller value turns out to break something,
2361 I'll probably just bump it back up.)
2362 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2364 * Many, many bugs reported by Peter Van Eynde have been added to
2365 the BUGS list; some have even been fixed.
2366 * While enabling dynamic object file loading, I tried to make the
2367 code easier to understand, renaming various functions and variables
2368 with less ambiguous names, and changing some function calling
2369 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2370 * While trying to figure out how to do the OpenBSD port, I tried to
2371 clean up some of the code in src/runtime/. In particular, I dropped
2372 support for non-POSIX signal handling, added various comments,
2373 tweaked the code to reduce the number of compilation warnings, and
2374 renamed some files to increase consistency.
2375 * To support the new automatic configuration functionality in make.sh,
2376 the source file target-features.lisp-expr has been replaced with the
2377 source file base-target-features.lisp-expr and the machine-generated
2378 file local-target-features.lisp-expr.
2379 * fixed a stupid quoting error in make.sh so that using CMU CL
2380 "lisp -batch" as cross-compilation host works again
2382 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2384 * DESCRIBE no longer tries to call itself recursively to describe
2385 bound/fbound values, so that it no longer fails on symbols which are
2386 bound to themselves (like keywords, T, and NIL).
2387 * DESCRIBE now works on generic functions.
2388 * The printer now prints less-screwed-up representations of closures
2389 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2390 * A private symbol is used instead of the :EMPTY keyword previously
2391 used to mark empty slots in hash tables. Thus
2392 (DEFVAR *HT* (MAKE-HASH-TABLE))
2393 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2394 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2395 now does what ANSI says that it should. (You can still get
2396 similar noncompliant behavior if bang on the hash table
2397 implementation with all the symbols you get back from
2398 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2399 This breaks binary compatibility, since tests for equality to
2400 :EMPTY are wired into things like the macroexpansion of
2401 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2403 * There's now a minimal placeholder implementation for CL:STEP,
2404 as required by ANSI.
2405 * An obscure bug in the interaction of the normal compiler, the byte
2406 compiler, inlining, and structure predicates has been patched
2407 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2408 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2410 * Missing ordinary arguments in a macro call are now detected even
2411 when the macro lambda list contains &KEY or &REST.
2412 * The debugger no longer complains about encountering the top of the
2413 stack when you type "FRAME 0" to explicitly instruct it to go to
2414 the top of the stack. And it now prints the frame you request even
2415 if it's the current frame (instead of saying "You are here.").
2416 * As specified by ANSI, the system now always prints keywords
2417 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2419 * The default initial SIZE of HASH-TABLEs is now smaller.
2420 * Type information from CLOS class dispatch is now propagated
2421 into DEFMETHOD bodies, so that e.g.
2422 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2424 is now basically equivalent to
2425 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2426 (DECLARE (TYPE SINGLE-FLOAT X))
2428 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2429 operation, without having to do run-time type dispatch.
2430 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2431 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2432 * Since I don't seem to be making big file reorganizations very often
2433 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2434 is overflowing my ability to conveniently back them up), I've finally
2435 checked the system into CVS. (The CVS repository is on my home system,
2436 not at SourceForge -- putting it on SourceForge might come later.)
2437 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2438 high-level GC-NOTIFY-FOO functions send their output. (There's
2439 still very little control of where low-level verbose GC functions
2440 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2441 less than it used to -- the GC-NOTIFY-FOO functions are now under
2442 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2443 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2444 in only one place in the source code, and propagates it automatically
2445 everywhere that it's needed. Thus e.g. when I bump the version from
2446 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2447 * The C source files now include boilerplate legalese and documentation
2448 at the head of each file (just as the Lisp source files already did).
2449 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2450 to his page will be replaced with a link to his new CLiki service.
2452 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2454 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2455 * The build process now looks for GNU make under the default name "gmake",
2456 instead of "make" as it used to. If GNU make is not available as "gmake"
2457 on your system, you can change this default behavior by setting the
2458 GNUMAKE environment variable.
2459 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2460 can build without error under CMU CL.
2462 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2464 * There is now a partial SBCL user manual (with some new text and some
2465 text cribbed from the CMU CL manual).
2466 * The beginnings of a profiler have been added (starting with the
2467 CMU CL profiler and simplifying and cleaning up). Eventually the
2468 main interface should be through the TRACE macro, but for now,
2469 it's still accessed through vaguely CMU-CL-style functions and macros
2470 exported from the package SB-PROFILE.
2471 * Some problems left over from porting CMU CL to the new
2472 cross-compilation bootstrap process have been cleaned up:
2473 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2474 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2475 part when getting rid of the compiler *BACKEND* stuff.)
2476 ** The value of *NULL-TYPE* was screwed up, because it was
2477 being initialized before the type system knew the final
2478 definition of the 'NULL type. This screwed up several key
2479 optimizations in the compiler, causing inefficiency in all sorts
2480 of places. (I found it because I wanted to understand why
2481 GET-INTERNAL-RUN-TIME was consing.)
2482 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2483 PROCLAIM FTYPE information. Unfortunately this broke binary
2484 compatibility again, since now the forms output by DEFGENERIC
2485 to refer to functions which didn't exist in 0.6.3.
2486 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2487 can use the new (as of 0.6.3) transform for SOME into MAP into
2489 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2490 values to (INTEGER 0 1000000), so that the time code will no longer
2491 occasionally get blown up by Linux returning 1000000 microseconds
2492 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2493 its output conform to the ANSI spec. (Alas, this makes its output
2494 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2495 * A full call to MAP NIL with a single sequence argument no longer conses.
2496 * fixes to problems pointed out by Martin Atzmueller:
2497 * The manual page no longer talks about multiprocessing as though
2498 it were currently supported.
2499 * The ILISP support patches have been removed from the distribution,
2500 because as of version 5.10.1, ILISP now supports SBCL without us
2501 having to maintain patches.
2502 * added a modified version of Raymond Toy's recent CMU CL patch for
2503 EQUALP comparison of HASH-TABLE
2505 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2507 * The system still can't cross-compile itself with
2508 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2509 entails), but at least it can compile more of itself that way
2510 than it used to be able to, and various buglets which were uncovered
2511 by trying to cross-compile itself that way have now been fixed.
2512 * This release breaks binary compatibility again. This time
2513 at least I've incremented the FASL file format version to 2, so that the
2514 problem can be detected reliably instead of just causing weird errors.
2515 * various new style warnings:
2516 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2517 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2518 ** using the lexical binding of a variable named in the *FOO* style
2519 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2520 as specified by ANSI.
2521 * *RANDOM-STATE* is no longer automatically initialized from
2522 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2523 default behavior of the system is to repeat its behavior every time
2524 it's run. If you'd like to change this behavior, you can always
2525 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2526 old convention there was no comparably easy way to get the system to
2527 repeat its behavior every time it was run.
2528 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2529 FTYPE declarations has been removed, in favor of their ANSI
2530 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2531 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2532 the types of their sequence arguments can be determined at compile time.
2533 This is done through a new open code expansion for MAP which eliminates
2534 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2535 argument types can be determined at compile time.
2536 * The optimizer now transforms COERCE into an identity operation when it
2537 can prove that the coerced object is already of the correct type. (This
2538 can be a win for machine generated code, including the output of other
2539 optimization transforms, such as the MAP transform above.)
2540 * Credit information has been moved from source file headers into CREDITS.
2541 * Source file headers have been made more standard.
2542 * The CASE macro now compiles without complaining even when it has
2545 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2547 * (Note that the way that the PCL macroexpansions were rewritten
2548 to accommodate the change in DEFGENERIC below breaks binary
2549 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2550 not run under sbcl-0.6.2. Once we get out of alpha releases,
2551 i.e. hit release 1.0.0, we'll probably try to maintain binary
2552 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2553 and sbcl-1.4.4. Until then, however, it might be fairly common
2554 for maintenance releases to break binary compatibility.)
2555 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2557 * The reporting of unhandled signals has been changed to print some
2558 explanatory text as well as the report form. (Previously only
2559 the report form was printed.)
2560 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2561 it defines, so that the compiler no longer issues undefined function
2562 warnings for compiled-but-not-yet-loaded generic functions.
2563 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2564 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2565 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2566 section of the ANSI spec are supported.
2567 * The "" nickname for the "KEYWORD" package has been removed.
2568 The reader still handles symbol tokens which begin with a package marker
2569 as keywords, but it doesn't expose its mechanism for doing so in the
2570 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2571 * The system now issues STYLE-WARNINGs for contradictory TYPE
2572 proclamations. (Warnings for contradictory FTYPE proclamations would
2573 be nice too, but those can't be done usefully unless the type system
2574 is made smarter about FUNCTION types.)
2575 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2576 systematized, so that "*target-*.lisp is supposed to exist only on the
2577 target and imply that there's a related file which exists on the
2578 host, and *host-*.lisp is supposed to exist only on the host and imply
2579 that there's a related file which exists on the target. This involves a
2580 lot of renaming. Hopefully the acute confusion caused by the renaming
2581 will be justified by the reduction in chronic confusion..
2582 ** runtime-type.lisp -> early-target-type.lisp
2583 ** target-type.lisp -> late-target-type.lisp
2584 ** early-host-format.lisp -> early-format.lisp
2585 ** late-host-format.lisp -> late-format.lisp
2586 ** host-error.lisp -> misc-error.lisp
2587 ** early-error.lisp -> early-target-error.lisp
2588 ** late-error.lisp -> late-target-error.lisp
2589 ** host-defboot.lisp -> early-defboot.lisp
2590 ** code/misc.lisp -> code/target-misc.lisp
2591 ** code/host-misc.lisp -> code/misc.lisp
2592 ** code/numbers.lisp -> code/target-numbers.lisp
2593 ** code/early-numbers.lisp -> numbers.lisp
2594 ** early-host-type.lisp -> early-type.lisp
2595 ** late-host-type.lisp -> late-type.lisp
2596 ** host-typep.lisp -> typep.lisp
2597 ** load.lisp -> target-load.lisp
2598 ** host-load.lisp -> load.lisp
2599 ** host-disassem.lisp -> disassem.lisp
2600 ** host-insts.lisp -> insts.lisp
2601 ** byte-comp.lisp -> target-byte-comp.lisp
2602 ** host-byte-comp.lisp -> byte-comp.lisp
2603 ** host-signal.lisp -> signal.lisp
2604 ** host-defstruct.lisp -> defstruct.lisp
2605 ** late-target-type.lisp -> deftypes-for-target.lisp
2606 Furthermore, several other previously target-only files foo.lisp (e.g.
2607 hash-table.lisp and random.lisp) have been split into a target-and-host
2608 foo.lisp file and a target-only target-foo.lisp file, with their key type
2609 definitions in the target-and-host part, so that the cross-compiler will
2610 know more about target types.
2611 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2612 gone away. In their place are various *BACKEND-FOO* variables
2613 corresponding to the slots of the old structure.
2614 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2615 into the target SBCL has been fixed.
2616 * The chill.lisp system for loading cold code into a running SBCL
2618 * Support for the CMU CL "scavenger hook" extension has been removed.
2619 (It was undocumented and unused in the CMU CL sources that SBCL was
2620 derived from, and stale in sbcl-0.6.1.)
2621 * Various errors in the cross-compiler type system were detected
2622 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2623 (enabling various consistency checks). Many of them were fixed,
2624 but some hard problems remain, so the compiler is back to
2625 running without *TYPE-SYSTEM-INITIALIZED* for now.
2626 * As part of the cross-compiler type system cleanup, I implemented
2627 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2628 * I have started adding UNCROSS calls throughout the type system
2629 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2630 on cross-compiler input files will be able to go away, and various
2632 * CONSTANTP now returns true for quoted forms (as explicitly required
2635 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2637 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2638 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2639 (MAKE-LIST -1). In the longer term, it ought to become true
2640 that declarations are assertions even at SAFETY 1. For now, it's not
2641 quite true even at SAFETY 3, but it's at least more nearly true..
2642 (Note that this change seems to increases the size of the system by
2643 O(5%) and to decrease the speed of the compiler by 20% or more.)
2644 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2645 for the problem of printing dozens of lines of distracting information
2646 about low-level system machinery as part of the top stack frame
2647 on entry to the debugger when an undefined function was called.
2648 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2649 is not reset to COMMON-LISP-USER.
2650 * Compilation of stuff related to dyncount.lisp has been made conditional
2651 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2652 smaller. The various dyncount-related symbols have been moved into
2653 a new "SB-DYNCOUNT" package.
2654 * tty-inspect.lisp has been renamed to inspect.lisp.
2655 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2656 feature has gone away. (When we eventually port to other flavors of
2657 libc and/or Unix, we'll try to make the differences between flavors
2658 invisible at the user level.)
2659 * Various other *FEATURES* tags, and/or their associated conditionals,
2660 have been removed if obsolescent, or given better documentation, or
2661 sometimes given more-mnemonic names.
2663 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2665 * tidied up "make.sh" script
2666 * tidied up system directory structure
2667 * better "clean.sh" behavior
2668 * added doc/FOR-CMUCL-DEVELOPERS
2669 * many many small tweaks to output format, e.g. removing possibly-confusing
2670 trailing #\. character in DESCRIBE-INSTANCE
2671 * (EQUALP #\A 'A) no longer signals an error.
2672 * new hashing code, including EQUALP hashing
2673 * tidied up Lisp initialization and toplevel
2674 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2675 * command line argument processing
2676 * added POSIX-GETENV function to deal with Unix-ish environment variables
2677 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2678 terminating SBCL on EOF
2679 * non-verbose GC by default
2680 * There is no more "sbcl" shell script; the sbcl file is now the C
2681 runtime executable (just like CMU CL).
2682 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2684 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2685 symbol and package information as Lisp native symbol and package objects
2686 instead of strings naming symbols and strings naming packages. This way,
2687 whenever packages are renamed (as in warm init), debug information is
2688 transformed along with everything else.
2689 * tweaked the optimization policy declarations which control the building
2690 of SBCL itself. Now, among other things, the system no longer saves
2691 source location debugging information. (This helps two problems at once
2692 by reducing SBCL size and by keeping SBCL from trying to look for its
2693 sources -- which may not exist -- when reporting errors.)
2694 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2695 debugging and testing purposes
2696 * cleaned up printing, making the printer call PRINT-OBJECT for
2697 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2698 methods, giving nearly-ANSI behavior
2699 * converted almost all special variables to use *FOO* naming convention
2700 * deleted PARSE-TIME functionality, since it can be done portably
2701 * moved some files out of cold init into warm init
2702 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2704 * regularized formatting of source files
2705 * added an install.sh script
2706 * fixed ridiculous memory usage of cross-compiler by making
2707 compiler/alloc.lisp not try to do pooling unless it can hook
2708 itself into the GC of the cross-compilation host. Now the system
2709 builds nicely on my old laptop.
2710 * added :SB-ALLOC in target-features.lisp-expr
2711 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2712 was not implemented)
2713 * re-did condition handling and note reporting in the compiler. Notes
2714 are no longer handled by signalling conditions. Style warnings
2715 and warnings are handled more correctly and reported in such a way
2716 that it's easy to find one or the other in your output (so that you
2717 can e.g. figure out which of many problems caused COMPILE-FILE to
2719 * changed the severity of several compiler warnings from full WARNING
2720 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2721 compiler note reporting so that it doesn't use the condition system
2722 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2723 COMPILE-FILE command)
2724 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2725 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2726 no longer supported, since their implementation was deeply intertwingled
2727 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2728 * removed bogus "support" for compiler macros named (SETF FOO), and
2729 removed the compiler macro for SETF INFO (but only after making a fool
2730 of myself on the cmucl-imp mailing list by posting a bogus patch for
2731 DEFINE-COMPILER-MACRO..)
2732 * Compiled files containing forms which have side effects on the Lisp
2733 reader (such as DEFPACKAGE forms) are now handled more correctly.
2734 (Compiler queuing of top level lambdas has been suppressed by setting
2735 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2736 * deleted various currently-unused source files, e.g. gengc.lisp. They
2737 may be added back at some point e.g. when porting to other architectures,
2738 but until they are it's distracting to distribute them and to try to
2740 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2741 there were so many of them they're just distractions, and UNCROSS is
2742 known to be able to handle the current sources
2743 * moved PROFILE functionality into TRACE, so that it will be clear
2744 how the wrapping and unwrapping of functions when you profile them
2745 interacts with the wrapping and unwrapping of functions when you
2746 trace them. (Actually, the functionality isn't there yet, but at least
2747 the interface specification is there. Hopefully, the functionality will
2748 arrive with some future maintenance release.)
2749 * removed host-oops.lisp
2750 * changed signature of QUIT function to allow UNIX-CODE argument
2751 * fixed READ-SEQUENCE bug
2752 * tweaked verbose GC output so that it looks more like the progress
2753 output that ANSI specifies for functions like LOAD
2754 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2755 * added <http://sbcl.sourceforge.com> to the banner information printed by