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: parsing self-recursive alien record types multiple times
6 no longer causes infinite recursion. (reported by Thomas F. Burdick,
7 original patch by Helmut Eller for CMUCL)
8 * bug fix: stack-exhaustion detection works now on NetBSD as well.
9 (thanks to Richard Kreuter)
10 * bug fix: defining classes whose accessors are methods on existing
11 generic functions in other (locked) packages no longer signals
12 bogus package lock violations. (reported by François-René Rideau)
13 * bug fix: special variables as DEFMETHOD parameters no longer have
14 associated bogus type declarations. (reported by David Wragg and
16 * bug fix: read-write consistency on streams of element-type
17 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
18 * bug fix: redefiniton of the only method of a generic function with
19 no DEFGENERIC no longer emits a full WARNING. In addition,
20 redefinition of generic functions with no DEFGENERIC to an
21 incompatible lambda list now signals an error. (thanks to Zach
23 * bug fix: DEFGENERIC now works even when there's a function of the
24 same name in an enclosing lexical environment. (thanks to Zach
26 * fixed compiler failure, caused by instrumenting code during
27 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
28 * optimization: added loop analysis and improved register allocation
29 * fixed some bugs revealed by Paul Dietz' test suite:
30 ** POSITION on displaced vectors with non-zero displacement
31 returns the right answer.
32 ** (SIMPLE-STRING) is a valid type specifier for sequence
34 ** *PRINT-LEVEL* handling for slotless structures is pedantically
36 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
37 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
39 ** The justification version of the FORMAT ~< directive treats
40 non-zero minpad parameter correctly.
42 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
43 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
44 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
45 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
46 the supported interface.
47 * new feature: Single-stepping of code compiled with DEBUG 2 or
48 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
49 * new feature: saving cores with foreign code loaded is now
50 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
51 Timothy Moore's work for CMUCL)
52 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
53 parameters to * instead of NIL if no initform is supplied.
54 (reported by Johan Bockgaard)
55 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
56 correctly when FUNCALL appears as the car of the form. Note:
57 despite this FUNCALL forms are not currently subject to
58 compiler-macro expansion. (port of Raymond Toy's fix for the
59 same from CMUCL, reported by Johan Bockgaard)
60 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
61 (thanks for Teemu Kalvas)
62 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
63 case instead of terminating the iteration (thanks for Julian
64 Squires, Sean Champ and Raymond Toy)
65 * bug fix: incorrect expansion of defgeneric that caused a style
66 warning. (thanks for Zach Beane)
67 * bug fix: slot accessor effective method computation works properly
68 for classes with multiple non-standard applicable methods on
69 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
70 * on x86 compiler supports stack allocation of results of LIST and
71 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
72 CMUCL implementation by Gerd Moellmann)
73 * fixed some bugs revealed by Paul Dietz' test suite:
74 ** FORMAT strings with both the ~<~:;~> form of the justification
75 directive and pretty-printing directives cause an error.
77 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
78 * incompatible change: the internal functions
79 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
80 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
81 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
82 instead of the old functions.
83 * new feature: on platforms where "dladdr" is available foreign
84 function names now appear in backtraces. (based on Helmut Eller's
86 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
87 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
89 * documentation: ASDF manual is now distributed with SBCL.
90 * bug fix: SBCL can now load its contributed modules with REQUIRE
91 even if the system-provided entries have been removed from
92 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
94 * bug fix: backtraces involving undefined functions or assembly
95 routines are more informative. (thanks to Brian Downing)
96 * bug fix: mutually referent alien structures now work correctly.
97 (reported by Rick Taube)
98 * bug fix: structures defined by WITH-ALIEN can be referred to
99 within other definitions in the same WITH-ALIEN.
100 * bug fix: division operators (MOD, TRUNCATE and the like) with
101 constant zero divisors and integer dividends no longer generate
103 * bug fix: provide default methods for INPUT-STREAM-P and
104 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
105 * bug fix: improve the detection and resolution of MOP metacycles
106 (where effective method computation depends on the generic function
107 having its effective method computed). (reported by Bruno Haible)
108 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
109 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
110 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
111 * bug fix: pathnames with relative directory components can be
112 represented relative to default pathnames.
113 * optimization: in taking the GCD of bignums, reduce the two bignums
114 to approximately the same size (using Euclid's algorithm) before
115 applying the more sophisticated binary GCD. (thanks to Juho
117 * optimization: COUNT on bitvectors now operates word-at-a-time.
118 * optimization: ASH with a positive, but not necessarily constant,
119 (leftwards) shift, when in a modular context, is compiled to a
121 * fixed some bugs revealed by Paul Dietz' test suite:
122 ** FORMAT variable parameters ("~V<char>") are defaulted properly
123 if the corresponding argument is NIL.
124 ** FORMAT directives accepting a minpad parameter treat negative
126 ** literal commas as character parameters to FORMAT directives are
128 ** literal spaces directly after ~<Newline> directives within a
129 format-logical-block (~:< ~@:>) do not induce :FILL-style
130 conditional newlines.
131 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
132 as specified: it no longer includes conditional newlines).
133 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
135 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
136 :PER-LINE-PREFIX argument does not evaluate to a string.
138 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
139 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
140 the manual for details; for now, package locks can be disabled by
141 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
142 if no major problems are found then it is likely that they will be
143 compiled in unconditionally.
144 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
145 unsupported operators on all platforms. To load a shared library
146 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
147 object file, link it into a shared library outside of SBCL and
148 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
150 * bug workaround: ROOM T and the GC were not working together
151 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
152 reporting facility aren't preserved. That reporting has been
153 disabled, so now until and unless someone figures out how to make it
154 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
155 * minor incompatible change: as threatened around sbcl-0.8.0, the
156 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
157 The symbols are also exported from SB-PCL for backwards
158 compatibility, but more so than before SB-PCL should be treated as
159 an implementation-internal package.
160 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
162 * the SB-SPROF contrib now works on (most) non-x86 architectures.
163 It is known as of this release not to work on the Alpha, however.
164 * fixed bug #167: errors signalled due to illegal syntax in method
165 bodies are now more legible.
166 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
167 designators and can hence be used with TYPEP.
168 * fixed bug #333: CHECK-TYPE now ensures that the type error
169 signalled, if any, has the right object to be accessed by
170 TYPE-ERROR-DATUM. (reported by Tony Martinez)
171 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
172 5.1.2.3 for multiple-value place subforms. (reported by Kalle
174 * fixed bug #334: programmatic addition of slots using specialized
175 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
176 and :ALLOCATION :CLASS slots.
177 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
178 float exponents. (rereported by Peter Seibel)
179 * fixed a bug: #\Space (and other whitespace characters) are no
180 longer considered to be macro characters in standard syntax by
182 * fixed bug: initialization of condition class metaobjects no longer
183 causes an instance of the condition to be created. (reported by
185 * fixed bug: it is now possible to have slots such that
186 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
187 description is neither :INSTANCE nor :CLASS.
188 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
189 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
190 SLOT-VALUE-USING-CLASS), and no longer causes errors with
191 non-standard SLOT-DEFINITION-ALLOCATION values.
192 * fixed bugs: various race conditions handling exiting threads.
193 CL-PPCRE's thread torture test now passes.
194 * fixed arguably-a-bug: GC time proportional to number of threads in
195 system even when most of them are idle
196 * optimization: improved performance of BIT and SBIT on bit-vectors.
197 * .fasl file incompatibility: The fasl file version number has
198 been incremented because of changes associated with package locks.
200 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
201 * minor incompatible change: the system no longer provides
202 optimization hints (conditions which are TYPEP
203 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
204 mode; these hints are still emitted when higher SPEED optimization
206 * new contrib module: a sampling profiler (profiling by statistical
207 sampling, rather than by instrumenting functions) is available as
208 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
209 * the behaviour of the standard function ED is now customizeable by
210 third parties through a hook variable: see ED's documentation
211 string for information on the protocol.
212 * the compiler no longer emits efficiency notes for (FUNCALL X)
213 when the type of X is uncertain under default optimization
215 * fixed bug 276: mutating a binding of a specialized parameter to a
216 method to something that is not TYPEP the specializer is now
218 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
219 exhibit the required behaviour on the x86 platform. (thanks to
220 Peter van Eynde, Eric Marsden and Bruno Haible)
221 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
222 even for difficult arguments. (reported by Peter Graves)
223 * fixed bug 141a: the backquote printer now descends quoted
225 * fixed another bug in backquote printing: no more destructive
226 modification of the form's list structure. (reported by Brian
228 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
229 move between its address being taken and the call to
230 interrupt_thread, fixing a crashing race condition.
231 * the SB-POSIX contrib implementation has been adjusted so that it
232 no longer exhibits ridiculously poor performance when constructing
233 instances corresponding to C structs.
235 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
236 * minor incompatible change: the sb-grovel contrib now treats C
237 structures as alien (in the sense of SB-ALIEN) objects rather than
238 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
239 has implications for memory management of client code
240 (sb-grovel-returned objects must now be manually managed) and for
241 type safety (alien objects now have full types).
242 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
243 used to control emission of compiler diagnostics, rather than the
244 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
245 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
246 quality should be considered deprecated.
247 * install.sh now installs the user manual as well
248 * (not quite a new documentable feature, but worth considering in
249 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
250 beginnings of a semantically meaningful condition hierarchy is
251 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
253 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
254 designator as the defaults argument.
255 * fixed bug: Displaced arrays whose displaced-to array has become
256 too small now cause ARRAY-DIMENSION to signal an error, providing
257 for safer bounds-checking. (reported by Bruno Haible)
258 * fixed bug: DEFCLASS slot definitions with identical :READER and
259 :WRITER names now signal a reasonable error. (reported by Thomas
261 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
263 * fixed bug: Invalid slot specification errors now print correctly.
264 (thanks to Zach Beane)
265 * fixed bug 320: Shared to local slot value transfers in class
266 redefinitions now happen corrently from superclasses as
267 well. (reported by Bruno Haible)
268 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
270 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
271 improper lists correctly. (reported by Bruno Haible)
272 * fixed bug 313: source-transform for <fun-name> was erroneously
273 applied to a call of a value of a variable with name <fun-name>.
274 (reported by Antonio Menezes Leitao)
275 * fixed bug 307: The obsolete instance protocol ensures that
276 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
277 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
278 remove dead unknown-values globs from the middle of the stack.
279 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
280 easier to resume long computations after using *BREAK-ON-SIGNALS*
281 to diagnose and fix failures. (thanks to Nikodemus Siivola)
282 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
283 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
284 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
285 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
286 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
288 * fixed a (fairly theoretical) bug in string printing: if
289 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
290 does not have array-element-type equal to the most general string
292 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
293 function name. (thanks to Nikodemus Siivola)
294 * fixed bug: compile-time format string checker failed on
295 non-closed ~{. (reported by Thomas F Burdick)
296 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
297 don't warn on legal loop code involving more than one aggregate
298 boolean. (thanks to Nikodemus Siivola)
299 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
300 in keyword position to MAKE-PACKAGE was not being reported as
302 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
303 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
305 * fixed bugs 280 and 312: the checking for multiple definitions in a
306 file is less likely to become confused by uses of inline
308 * fixed bug: the #S reader macro performs the keyword coercion
309 specified for slot names. (reported by Kalle Niemitalo)
310 * fixed bug: lambda lists may contain symbols whose names start with
311 & but are not lambda-list-keywords; their occurrence triggers a
313 * fixed bug 321: define-method-combination argument lambda lists do
314 not cause invalid code to be generated when &optional or &aux
315 variables have default values. (reported by Bruno Haible)
316 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
317 have CLOS classes; this appears also to have fixed reported
318 instability in stack exhaustion detection.
319 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
320 MUFFLE-WARNING when no associated restart is present is now
322 * optimization: rearranged the expansion of various defining macros
323 so that each expands into only one top-level form in a
324 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
326 * optimization: used a previously little-used slot in symbols to
327 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
329 * fixed some bugs revealed by Paul Dietz' test suite:
330 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
332 ** Fixed an optimization bug related to inheritance of initforms
333 from local to shared slots.
334 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
335 ** CAST optimizer forgot to flush argument derived type.
336 ** print/read consistency on floats is now orders of magnitude
337 more likely. (thanks also to Bruno Haible for a similar report
339 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
340 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
341 for objects of type REAL. Make it so.
342 ** SXHASH is specified (infelicitously) to respect similarity,
343 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
344 it so. (thanks to Markus Ziegler)
345 ** on the Alpha, the compiler succeeds in compiling functions
346 returning a known number of arguments greater than 63.
347 ** fixed handling of invalid NIL arguments in keyword position
349 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
351 ** provided a readably-printable representation for RANDOM-STATE
353 ** ensured that pathnames, where they have a namestring, always
354 print using #P"..." syntax.
356 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
357 * Support for the forthcoming 2.0 version of the NetBSD kernel
358 running on x86 hardware has been added. (thanks to Perry
359 E. Metzger most immediately, and others for their past work)
360 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
361 current release version; SBCL's OpenBSD support had been broken
362 since about the time of OpenBSD's switch to ELF binary format.)
363 * [placeholder for DX summary]
364 ** user code with &REST lists declared dynamic-extent, under high
365 speed or space and low safety and debug optimization policy.
366 * The manual has been converted to Texinfo format and the debugger
367 chapter from the cmucl manual has been added.
368 * A facility has been added to extract documentation strings from
369 sbcl and store them as Texinfo-formatted snippets for inclusion in
370 the manual (via Texinfo's @include directive)
371 * bug fix: compiler emitted division in optimized DEREF. (thanks for
372 the test case to Dave Roberts)
373 * bug fix: multidimensional simple arrays loaded from FASLs had fill
374 pointers. (reported by Sean Ross)
375 * bug fix: PROFILE output is printed nicely even for large numerical
376 values. (thanks to Zach Beane)
377 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
378 greater than 32 handle EOF correctly.
379 * bug fix: on X86 an immediate argument of the IMUL instruction is
380 correctly printed in disassembly. (thanks to Lutz Euler)
381 * bug fix: class slots in redefined classes preserve their old
382 values. (thanks to Bruno Haible and Nikodemus Siivola)
383 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
384 succeeds. (reported by Marco Baringer)
385 * fixed some bugs revealed by Paul Dietz' test suite:
386 ** READ-SEQUENCE now works on ECHO-STREAMs.
387 ** RATIONALIZE works more according to its specification. (thanks
390 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
391 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
392 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
393 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
394 should matter to you only if you rebind the printer control
395 variables and then find you want different bindings in the
396 debugger than in the ordinary execution of your program.)
397 * The runtime build system has been tweaked to support building
398 (on SPARC/SunOS) using a C compiler which invokes Sun's own
399 assembler and linker. (thanks to Nikodemus Siivola)
400 * Unbound, undefined, undeclared variables now trigger full
401 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
402 more often programmer error than deliberate exploitation of undefined
404 * optimization: the hash algorithm for strings has changed to one
405 that is less vulnerable to spurious collisions. (thanks to Juho
407 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
408 needless bounds checking. (thanks to Juho Snellman)
409 * optimization: implemented multiplication as a modular
410 (UNSIGNED-BYTE 32) operation on the PPC backend.
411 * fixed some bugs revealed by Paul Dietz' test suite:
412 ** ADJUST-ARRAY now copies the datum in a zero rank array if
414 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
415 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
416 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
418 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
420 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
422 ** ECHO-STREAMs no longer attempt to echo the end of file value to
423 their output stream on EOF from read.
424 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
425 have been read to end-of-file.
426 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
428 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
429 description of determination of which consecutive characters
431 ** Printing the "Space" character with escaping on now yields "#\\ ",
432 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
433 ** Reading floating-point numbers with *READ-BASE* set to a number
434 less than 10 works correctly.
435 ** Reading floating-point numbers with *READ-BASE* set to a number
436 more than 10 works correctly.
437 ** Printing with *PRINT-READABLY* targets the standard readtable, not
438 the readtable currently in effect.
440 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
441 * minor incompatible change: parsing of namestrings on a physical
442 (Unix) host has changed; numbers after the final #\. in a
443 namestring are no longer interpreted as a version field. This is
444 intented to be largely invisible to the user, except that the
445 meaning of the namestring "*.*.*" has changed: it now refers to a
446 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
447 should usually be replaced by
448 (make-pathname :name :wild :type :wild :version :wild)
449 with the added benefit that this is more likely to be portable.
450 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
451 option to OPEN now signals an error if the file being opened
452 exists; this may have an impact on existing code.
453 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
454 signals received, on the PowerPC platforms (both Linux and
455 Darwin). (thanks to Pierre Mai for pointing out the location of
457 * several fixes on OS X: The system now builds and runs cleanly on
458 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
459 (thanks to Brian Mastenbrook)
460 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
461 second-resolution: integer multiples of 1/3600 between -24 and 24.
462 (thanks to Vincent Arkesteijn)
463 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
464 type when called with 1 argument; PEEK-CHAR checked type of
465 PEEK-TYPE only after having read first character from a
466 stream. (reported by Peter Graves)
467 * bug fix: the garbage collector now has much better locality
468 behaviour, and in particular no longer treats all memory as being
469 exhausted when many small objects point to each other in a deeply
471 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
472 recognized as being TYPEP their class.
473 * bug fix: the PUSHNEW documentation string has been corrected.
474 (thanks to Vincent Arkesteijn)
475 * bug fix: defaulting of the value for the last of an atypically
476 large number of multiple values being bound was not being
477 performed correctly on the Alpha or PPC platforms
478 * optimization: implemented multiplication as a modular
479 (UNSIGNED-BYTE 32) operation on the x86 backend.
480 * optimization: SEARCH on simple-base-strings can now be open-coded.
481 (see also contrib/compiler-extras.lisp for inspiration for
482 teaching the compiler about the Boyer-Moore algorithm).
483 * value, returned by MAX (and MIN) called with several EQUALP, but
484 not EQL, arguments now does not depend on compiler settings.
485 * fixed some bugs revealed by Paul Dietz' test suite:
486 ** in stack analysis liveness information is propagated from
487 non-local entry points.
488 ** pathwise CAST removing failed when the CAST node did not start
490 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
491 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
493 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
494 host is already defined.
495 ** RENAME-FILE works on streams instead of signalling an internal
497 ** PEEK-CHAR uses the current readtable when determining whether
498 or not a character is whitespace.
499 ** MERGE-PATHNAMES handles the case when the pathname does not
500 specify a name while the default-pathname specifies a version.
501 ** Pathnames now stand a better chance of respecting print/read
503 ** Attempting to use standardized file system operators with a
504 pathname with invalid :DIRECTORY components signals a
506 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
508 ** DIRECTORY on logical pathnames is more correct.
509 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
510 signal a TYPE-ERROR if their argument is not a stream.
511 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
512 designator argument does not designate a stream.
513 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
514 examining the synonym.
515 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
517 ** OPEN and WITH-OPEN-STREAM allow opening streams with
518 element-type larger than ([UN]SIGNED-BYTE 32).
520 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
521 * When built with the :SB-FUTEX feature, threaded builds now take
522 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
523 for faster/more reliable mutex and condition variable support.
524 * Incompatible change (but one you probably shouldn't have been using
525 anyway): the interface and code for arbitrating between multiple
526 threads in the same user session has been redesigned.
527 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
528 error when called without an explicit environment argument.
529 (thanks to Nikodemus Siivola)
530 * bug fix: buffered :DIRECTION :IO streams are less likely to become
531 confused about their position. (thanks to Adam Warner and Gerd
533 * bug fix: Pretty printing backquoted forms with unquotations in the
534 argument list position of various code constructs such as LAMBDA
535 now works correctly. (reported by Paul Dietz)
536 * bug fix: Pretty printing unquotations no longer loses all
537 stream position information.
538 * optimization: performance of string output streams is now less
539 poor for multiple small sequence writes.
540 * optimization: performance of CSUBTYPEP in the presence of complex
541 expressions involving CONS and NOT many times has been improved.
542 (reported by Paul Dietz)
543 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
545 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
546 streams for the REPL, for the debugger, and so on. (thanks to
548 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
549 * fixed some bugs revealed by Paul Dietz' test suite:
550 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
551 is ignored if it is too small, rather than propagating through
552 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
553 ** extremely complex negations of CONS types were not being
554 sufficiently canonicalized, leading to inconsistencies in
556 ** VALUES tranformer lost derived type.
558 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
559 * fixed a bootstrapping bug: the build process no longer assumes
560 that the various BOOLE-related constants have the same value in
561 host and target lisps. (noted by Paul Dietz' test suite on an
562 SBCL binary built from CLISP)
563 * The system can now be dynamically linked on the MIPS platform,
564 which enables dynamic loading of foreign code from Lisp. (thanks
565 to Ralf Baechle for discussions on the MIPS ABI)
566 * The system now records debugging information for its own source
567 files in a filesystem-position-independent manner, relative to
568 the "SYS" logical host.
569 * fixed a compiler bug: MV-LET convertion did not check references
570 to the "max args" entry point. (reported by Brian Downing)
571 * tweaked disassembly notes to be less confident about proclaiming
572 some instruction as an LRA. (thanks to Brian Downing)
573 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
574 now each have their own history, command character, and other
575 characteristics. (thanks to David Lichteblau)
576 * fixed some bugs revealed by Paul Dietz' test suite:
577 ** compiler failure in compiling LOGAND expressions including a
579 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
580 shift greater than 32.
581 ** FLUSH-DEST did not mark blocks for type check regeneration.
582 ** HANDLER-CASE failed to accept declarations in handler clauses
583 in some circumstances.
585 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
586 * New code in contrib/sb-introspect (still probably not entirely
587 stable yet) provides some support for smart Lisp development
588 environments like SLIME.
589 * The conditions signalled for errors occurring when loading .fasl
590 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
591 in a way which should help ASDF recover gracefully.
592 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
593 stuff has been cleaned up. If you code contrib/ stuff, this might
594 affect you, and you can look at contrib/README, contrib/STANDARDS,
595 and/or the 0.8.4.27 diff to check.
596 * In full calls the compiler now does not generate checks for declared
597 argument types for all arguments.
598 * various threading fixes
599 ** and some experimental patches which didn't make it into
600 the main tree for this release, but which are shipped in
601 contrib/experimental-thread.patch as a possible fix for some
602 failures (deadlock, spinning...) in GC-intensive multithreaded
604 * fixed PPC build problem (source code incompatibility of different
605 library versions): added offsetof() hackery which attempts to divine
606 where glibc maintainers put uc_mcontext today
607 * fixed bug 282: compiler does not trust type assertions while passing
608 arguments to a full call.
609 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
610 &OPTIONAL VALUES type parameter.
611 * fix bug 214: algorithm for noting rejected templates is now more
612 similar to that of template seletion. (also reported by rydis on
614 * fixed bug 141b: printing backquoted information readably and prettily
615 inserts a space where necessary.
616 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
617 (CHARACTER 10) are now reported as errors, rather than propagated
618 as unknown types. (reported by piso on #lisp)
619 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
620 with values NIL and :ERROR. (thanks to Milan Zamazal)
621 * fixed bug 191c: CLOS now does proper keyword argument checking as
622 described in CLHS 7.6.5 and 7.6.5.1.
623 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
624 longer raise an error; further, using a list as a for-as-arithmetic
625 counter now raises a meaningful error.
626 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
627 now understood by sequence creation functions such as MAKE-SEQUENCE
629 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
630 a STRING-INPUT-STREAM.
631 * compiler enhancement: SIGNUM is now better able to derive the type
633 * type declarations inside WITH-SLOTS are checked. (reported by
635 * fixed some bugs revealed by Paul Dietz' test suite:
636 ** incorrect optimization of TRUNCATE for a positive first
637 argument and negative second.
638 ** compiler failure in let-convertion during flushing dead code.
639 ** compiler failure while deriving type of TRUNCATE on an
640 interval, containing 0.
641 ** ASH of a negative bignum by a negative bignum count now returns
643 ** intersection of CONS types now canonicalizes properly, fixing
644 inconsistencies in SUBTYPEP.
646 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
647 * incompatible change: The --disable-debugger command line
648 option now clobbers the debugger at a more fundamental
649 level, by redefining #'INVOKE-DEBUGGER instead of by
650 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
651 is specified by ANSI to ignore *DEBUGGER-HOOK* and
652 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
653 enter the debugger REPL and then suffer recursive errors
654 because *DEBUG-IO* is also messed up in --disable-debugger mode;
655 while under the new system, BREAK in --disable-debugger mode
656 terminates the system just as an unhandled error would.
657 * fixed compiler performance when processing loops with a step >1;
658 * bug fix: DOCUMENTATION now retrieves generic function
659 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
660 support has been systematized, and now supports the methods
661 specified by ANSI, along with a default method and a method for
662 slot documentation. (reported by Nathan Froyd)
663 * bug fix: effective methods associated with a generic function are
664 no longer cached over a change of that generic function's method
665 combination. (reported by Andreas Fuchs)
666 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
667 names a non-existent directory. (thanks to Andreas Fuchs)
668 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
669 the PPC platform, lead to stack corruption; this has been fixed.
670 (reported by Rainer Joswig)
671 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
672 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
673 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
674 arithmetic in general, is now much more reliable on the PPC
676 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
677 platform now returns the right answer.
678 * optimization: restored some effective method precomputation in
679 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
680 precomputation is now tunable.
681 * optimization: compiler-internal data structure use has been
682 reviewed, and changes have been made that should improve the
683 performance of the compiler by about 20%.
684 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
685 simple-base-strings and simple-bit-vectors is improved.
686 * optimization: the optimization of 32-bit logical and arithmetic
687 functions introduced in version 0.8.3 on the x86 has been
688 implemented on the mips, ppc and sparc platforms; an
689 implementation of the same facility, but for 64-bit arithmetic,
690 has been added for the alpha.
691 * microoptimization: the compiler is better able to make use of the
692 x86 LEA instruction for multiplication by constants.
693 * bug fix: in some situations compiler did not report usage of
694 generic arithmetic in (SPEED 3) policy.
695 * bug 145b fix: compiler used wrong type specifier while converting
696 MEMBER-types to numeric.
697 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
699 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
700 index argument now works properly on the Alpha platform.
701 * bug fix: floating point exception treatment on the Alpha platform
703 * bug fix: FILE-POSITION works much better on string input and
704 output streams. (thanks to Nikodemus Siivola)
705 * bug fix: many threading/garbage collection symptoms sorted.
706 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
707 might be pseudo-atomic.
708 * internal change: Stopping for GC is now done with signals not
709 ptrace. GC is now done in whichever thread wanted it, instead of
711 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
713 * bug fix: result form in DO is not contained in the implicit
715 * incompatible change: ICR structure is changed; the value part of
716 CONTINUATION is now called LVAR; corresponding functions are
717 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
718 * added type deriver for ISQRT (thanks to Robert E. Brown).
719 * bug fix: better support for loading from the command line when an
720 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
722 * fixed some bugs revealed by Paul Dietz' test suite:
723 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
724 ** ROUND and FROUND now give the right answer when given very
725 small float arguments.
726 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
728 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
729 ** (GCD 0 <negative-integer>) returned <negative-integer>.
730 ** LCM should return a non-negative integer.
731 ** PARSE-INTEGER returned the index of a terminator instead of the
732 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
733 ** PARSE-INTEGER returned an incorrect index being applied to a
735 ** LCM with two arguments of 0 returns 0 rather than signalling
737 ** unsigned addition of a 32-bit constant with the high bit set no
738 longer causes an internal compiler error on the x86.
739 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
740 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
741 with negative last argument.
742 ** byte specifiers with zero size and position no longer cause
743 an error during type derivation.
744 ** bignum multiplication on the Alpha platform now returns the
746 * porting: The system now builds on SuSE AMD64, although it still
747 generates a 32-bit binary.
748 * .fasl file incompatibility: The fasl file version number has
749 been incremented (because of the changes to internal compiler
750 data structures referred to above).
752 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
753 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
754 more accurately, on the Darwin kernel running on PowerPC hardware.
755 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
756 * Compiler code deletion notes now signal a condition of type
757 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
758 an associated MUFFLE-WARNING restart.
759 * The compiler now performs limited argument count validation of
760 constant format strings in FORMAT, and where appropriate in ERROR,
761 CERROR and WARN. (thanks to Gerd Moellmann)
762 * New ASDF-INSTALL contrib can be used for automatic download and
763 installation of third-party Lisp code from CCLAN or other sites
765 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
766 which forces another thread to execute a function supplied by the
768 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
769 now accept and act upon their :ELEMENT-TYPE keyword argument.
770 (reported by Martin Atzmueller, Edi Weitz)
771 * bug fix: FILE-POSITION now accepts position designators up to
772 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
773 is the greater. (thanks to Patrik Nordebo)
774 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
775 Kalle Olavi Niemitalo)
776 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
777 printable. (reported by Eric Marsden)
778 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
779 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
780 * x86 bug fix in control stack exhaustion checking: now shows backtrace
781 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
782 (thanks to Stig Sandoe)
783 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
784 UNREAD-CHAR now correctly includes the unread character in the
785 target sequence. (thanks to Gerd Moellmann)
786 * bug fix in threaded builds: the system can now be suspended and
787 resumed by shell job control with minimal disruption.
788 * bug fixes in times and timezones >2038AD
789 * better handling of "where is GNU make?" problem in build scripts
790 (thanks to Nikodemus Siivola)
791 * new optimization: inside a named function any reference to a
792 function with the same name is considered to be a self-reference;
793 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
794 optimization quality.
795 * new optimization on x86: logical functions and + now have
796 optimized (UNSIGNED-BYTE 32) versions, which are automatically
797 used when the result is truncated to 32 bits.
798 * VALUES declaration is partially enabled.
799 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
800 * fixed some bugs revealed by Paul Dietz' test suite:
801 ** The system now obeys the constraint imposed by
802 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
803 types form a lattice under type intersection.
804 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
805 ** ASSOC now ignores NIL elements in an alist.
806 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
807 and (1+ MOST-POSITIVE-FIXNUM) answers.
808 ** The addition of a method with invalid qualifiers to a generic
809 function does not cause an error to be signalled immediately;
810 a warning is signalled, and the error is generated only on
811 calling the generic function.
812 * changed .fasl file version number, in part to add type codes for
813 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
814 obscure ANSI requirements
816 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
817 * fixed bug 148: failure to inline-expand a local function left
818 garbage, confusing the compiler.
819 * fixed bugs 3cd: structure slot readers perform type check if the
820 slot can have an invalid value (i.e. it is either not initialized
821 or can be written with a less specific slot writer).
822 * bug fix: the compiler now traps array references to elements off
823 the end of an array; previously, the bounds checking in some
824 circumstances could go off-by-one.
825 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
827 * type declarations for array element types now obey the description
828 on the CLHS page "Declaration TYPE", as per discussions on
829 sbcl-help around 2003-05-08. This means that a declaration
830 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
831 declaration, all references to BAR will be asserted or assumed
832 (with THE, so dependent on compiler policy) to involve objects of
833 type FOO. Note that no such declaration is implied in
834 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
835 * declared types of functions from the "Conditions"
836 chapter. (reported by Paul Dietz)
837 * bug fix: CERROR accepts a function as its first argument.
838 * bug fix: NTH an NTHCDR accept a bignum as index
839 arguments. (reported by Adam Warner)
840 * optimization: character compare routines now optimize comparing
841 against a constant character. (reported by Gilbert Baumann)
842 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
843 argument now works properly on the MIPS platform.
844 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
845 * repeated evaluation of the same DEFSTRUCT, a slot of which is
846 declared to have a functional type, does not cause an error
848 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
849 many :DEFAULT-INITARGS. (reported by Istvan Marko)
850 * fixed bug: if last continuation of a deleted block has a
851 destination, this destination should be deleted too. (reported by
853 * fixed a bug in the bootstrap process: the host compiler's values
854 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
855 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
856 test case from Patrik Nordebo)
857 * improved the ability of the disassembler on the PPC platform to
858 provide helpful disassembly notes.
859 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
860 the class in more cases than previously.
861 * bug fix: FILE-POSITION now understands :START and :END for
862 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
863 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
864 8) data. (thanks to David Lichteblau)
865 * bug fix: it is possible to add a method to a generic function
867 * bug fix: reader failed to signal END-OF-FILE inside an
868 object representation. (reported by Nikodemus Siivola)
869 * fixed some bugs revealed by Paul Dietz' test suite:
870 ** LAST and [N]BUTLAST should accept a bignum.
871 ** condition slot accessors are methods.
872 ** (VECTOR NIL) is a subtype of STRING.
874 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
875 * minor incompatible change: some nonsensical specialized lambda
876 lists (used in DEFMETHOD) which were previously ignored now signal
878 * minor incompatible change: the system is now aware of the types of
879 variables in the COMMON-LISP package, and will signal errors for
880 most violations of these type constraints (where previously they
881 were silently accepted).
882 * minor incompatible change: COMPILE-FILE now uses the freedom
883 afforded (ANSI 3.2.2.3) to use derived function types for
884 functions defined in the same file. This also permits the system
885 to warn on static type mismatches and function
886 redefinition. (Currently it does not work with high DEBUG level.)
887 * minor incompatible change: VALUES declaration is disabled.
888 * When issuing notes, the compiler now signals a condition of type
889 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
890 restart for use in user handlers. It is expected that the
891 COMPILER-NOTE condition will eventually become a condition
892 supertype to a hierarchy of note types, which will then be
893 handleable in a similar fashion. However, at the moment, no such
894 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
895 but it's an implementation detail, not a classification for the
897 * Changes in type checking closed the following bugs:
898 ** type checking of unused values (192b, 194d, 203);
899 ** template selection based on unsafe type assertions (192c, 236);
900 ** type checking in branches (194bc).
901 * A short form of VALUES type specifier has ANSI meaning (it has
902 increased the number of situations when SBCL cannot perform type
904 * fixed bug in DEFSTRUCT: once again, naming structure slots with
905 keywords or constants is permissible.
906 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
907 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
908 classes. (thanks to Antonio Martinez)
909 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
910 outputs FRESH-LINE or TERPRI, and no longer converts its stream
911 argument to a pretty-print stream. Instead, it leaves any such
912 operations to DESCRIBE-OBJECT methods.
913 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
915 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
916 lambda lists. (thanks to David Lichteblau)
917 * bug fix: Defining a generic function with a :METHOD-CLASS being a
918 subclass of STANDARD-METHOD no longer causes stack exhaustion.
919 (thanks to Gerd Moellmann)
920 * fixed bug 246: increased compilation speed of long
921 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
923 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
925 * DEFCONSTANT now throws a condition of type
926 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
927 constant to a non-EQL value; CONTINUE and ABORT restarts
928 respectively change and preserve the value.
929 * fixed bug 63: The code walker, part of the implementation of CLOS,
930 is now better at handling symbol macros.
931 * bug fix: There is no longer an internal implementation type named
932 CL:LENGTH. (reported by Raymond Toy)
933 * bug fix: In macro-like defining macros/special operators the
934 implicit block does not enclose the lambda list.
935 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
936 enclosed in parentheses) are not suitable as type specifiers, and
937 their use properly signals an error now.
938 * bug fix: An argument count mismatch for a type specifier in code
939 being compiled no longer causes an unhandled error at compile
940 time, but signals a compile-time warning.
941 * fixed simple vector readable printing
942 * bug fix: DESCRIBE takes more care over whether the class
943 precedence list slot of a class is bound before accessing it.
944 (reported by Markus Krummenacker)
945 * bug fix: FORMATTER can successfully compile pretty-printer format
946 strings which use variants of the ~* directive inside.
947 * bug fix: SEARCH now applies its TEST predicate to the elements of
948 the arguments in the correct order. (thanks to Wolfhard Buss)
949 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
950 in type inference. (thanks to Robert E. Brown)
951 * bug fix: Reading in symbols with an explicit package name of ""
952 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
953 * fixed some bugs revealed by Paul Dietz' test suite:
954 ** NIL is now allowed as a structure slot name.
955 ** Arbitrary numbers, not just REALs, are allowed in certain
956 circumstances in LOOP for-as-arithmetic clauses.
957 ** Multiple class redefinitions before slot access no longer
959 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
960 association between the name and a class.
961 ** Generic functions with non-standard method-combination and over
962 five methods all of which return constants no longer return NIL
963 after the first few invocations. (thanks to Gerd Moellmann)
964 ** CALL-NEXT-METHOD with no arguments now passes the original
965 values of the arguments, even in the presence of assignment.
966 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
968 ** DISASSEMBLE works with closures and funcallable instances.
969 ** ADD-METHOD now returns the generic function, not the new method.
970 ** FIND-METHOD signals an error if the lengths of the specializers
971 is incompatible with the generic function, even if the ERRORP
973 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
974 which its argument is a member.
975 ** DEFCLASS only redefines the class named by its class-name
976 argument if that name is the proper name of the class;
977 otherwise, it creates a new class.
978 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
979 of the UNBOUND-SLOT condition to the name of the slot.
980 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
981 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
982 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
983 SLOT-MAKUNBOUND in the specified fashion.
985 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
986 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
987 cross-compilation host. As a consequence, we can now bootstrap our
988 way up to SBCL starting with a bare gcc toolchain and human-readable
989 source code (first the source to CLISP, then the source to SBCL).
990 * A contributed module containing a partial implementation of the
991 simple-streams interface has been included. (thanks to Rudi
993 * A contributed module implementing the RFC1321 Message Digest
994 Algorithm, known as MD5, has been included.
995 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
996 no longer has any effect, as the code controlled by this feature
997 has been deleted. (As far as we know, no-one has ever built using
998 this feature, and its semantics were confused in any case).
999 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1000 work on conditions (as required by the ANSI specification),
1001 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1002 expected behaviour on conditions. Users should note, however,
1003 that such behaviour is not required by the ANSI specification,
1004 and so use of this behaviour may render their code unportable.
1005 * fixed some bugs revealed by Paul Dietz' test suite:
1006 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1008 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1009 ** SXHASH on condition objects no longer returns NIL.
1010 ** :ALLOCATION :CLASS slots are better treated; their values are
1011 updated on class redefinition, and initforms inherited from
1012 superclasses are applied.
1013 ** REMOVE-METHOD returns its generic function argument even when
1014 no method was removed.
1015 ** SHARED-INITIALIZE now initializes the values of the requested
1016 slots, including those with :ALLOCATION :CLASS.
1017 ** ALLOCATE-INSTANCE now works on structure classes defined via
1018 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1020 ** SLOT-EXISTS-P now works on conditions, as well as structures
1022 ** MAKE-LOAD-FORM now has the required methods on
1023 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1024 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1025 keyword, and now implements the SLOT-NAMES argument.
1026 ** methods with &OPTIONAL arguments no longer allow too many
1027 arguments to be passed in the call without error.
1028 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1029 option is consistent with the required arguments of the generic
1030 function lambda list.
1031 * bug fix: REQUIRE accepts a string designator. (Thanks to
1033 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1034 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1035 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1036 * bug fix: The compiler now checks for duplicated variables in macro
1038 * bug fix: SETQ on globals returns the correct value.
1039 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1040 not a COMPILER-ERROR (followed by some other strange error on
1041 choosing the CONTINUE restart).
1042 * bug fix: make.sh and friends are now more consistent in the way that
1043 they look for GNU "make".
1045 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1046 * experimental native threads support (on x86 Linux >=2.4 only).
1047 This is not compiled in by default: you need to add :SB-THREAD to
1048 the target features. See the "Beyond ANSI" chapter of the manual
1050 * fix for longstanding nonANSIism: The old distinction between
1051 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1052 The return value from CL:FIND-CLASS is now a CLOS class, and
1053 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1054 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1055 * An interface to the MetaObject Protocol, as described in Kiczales,
1056 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1057 MIT Press, 1991, is available from the SB-MOP package.
1058 * incompatible change: the SB-PCL package should now be considered
1059 a private implementation detail, and no longer a semi-private MOP
1061 * minor incompatible change: due to rearrangement for threads, the
1062 control stack and binding stack are now allocated at arbitrary
1063 addresses instead of being hardcoded per-port. Users affected by
1064 this probably have to be doing advanced things with shared
1065 libraries, and will know who they are.
1066 * minor incompatible change: Previously, all --eval forms used were
1067 processed with READ before any of them were processed with EVAL.
1068 Now each --eval form is processed with both READ and EVAL before
1069 the next --eval form is processed. (Thus package operations like
1070 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1071 work as the user might reasonably expect.)
1072 * minor incompatible change: *STANDARD-INPUT* is now only an
1073 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1075 * minor incompatible change: Y-OR-N-P is now character-oriented,
1076 not line oriented. Also, YES-OR-NO-P now works without errors.
1077 (thanks to Antonio Martinez)
1078 * sb-aclrepl module improvements: an integrated inspector, added
1079 repl features, and a bug fix to :trace command.
1080 * Known functions, which cannot be open coded by the backend, are
1081 considered to be able to check types of their arguments. (fixing
1082 a bug report by Nathan J. Froyd)
1083 * fixed a bug in computing method discriminating functions: It is
1084 now possible to define methods specialized on classes which have
1085 forward-referenced superclasses. (thanks to Gerd Moellmann)
1086 * fixed evaluation order in optional entries (reported by Gilbert
1088 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1089 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1090 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1091 required-by-AMOP NAME argument, as well as CLASS and
1092 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1093 * fixed bug 20: DEFMETHOD can define methods using names that are
1094 not the proper names of classes to designate class specializers.
1095 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1096 Unix file descriptors, instead of blowing up. (thanks to Antonio
1098 * Garbage collection refactoring: user-visible change is that a
1099 call to the GC function during WITHOUT-GCING will not do garbage
1100 collection until the end of the WITHOUT-GCING. If you were doing
1101 this you were probably losing anyway.
1102 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1103 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1104 TYPEP the latter but not the former.
1105 * The compiler issues a full WARNING for calls to undefined functions
1106 with names from the CL package.
1107 * MAP-INTO for a vector destination is open coded. (reported by
1108 Brian Downing on c.l.l)
1109 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1110 documentation string.
1111 * fixed some bugs revealed by Paul Dietz' test suite:
1112 ** COPY-ALIST now signals an error if its argument is a dotted
1114 ** Condition slots are now accessed more correctly in the presence
1115 of multiple initargs for a given slot.
1116 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1117 exclude restarts of the same name associated with a different
1119 ** DEFCLASS of forward-referenced classes with another
1120 forward-referenced class in the superclasses list no longer
1122 ** Condition slots are now initialized once each, not multiple
1123 times. (thanks to Gerd Moellmann)
1124 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1126 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1127 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1128 arguments contain duplicated elements.
1129 ** RESTART-CASE understands local macros.
1130 ** RESTART-CASE associates exactly its own restarts with a condition.
1131 ** ENDP in safe mode checks its argument to be of type LIST.
1132 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1133 in question is unbound.
1134 ** Optimized MAKE-INSTANCE functions no longer cause internal
1135 assertion failures in the presence of duplicate initargs.
1136 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1137 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1138 propagated even in the ctor optimized implementation of
1140 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1142 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1143 * a better implementation of SXHASH on (simple) bit vectors,
1144 measured both in execution speed and in distribution of results
1145 over the positive fixnums, has been installed. Likewise, a better
1146 implementation of EQUAL for simple bit vectors is now available.
1147 * fixed CEILING optimization for a divisor of form 2^k.
1148 * fixed bug 240 (emitting extra style warnings "using the lexical
1149 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1150 by Antonio Martinez)
1151 * fixed SXHASH, giving different results for NIL depending on type
1152 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1153 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1154 variable. (found by Rolf Wester)
1155 * SBCL does not ignore type declarations for special
1156 variables. (reported by rif on c.l.l 2003-03-05)
1157 * some bug fixes in contrib/sb-aclrepl/
1158 * fixed some bugs revealed by Paul Dietz' test suite:
1159 ** a bug in the CONS type specifier, whereby the CAR and CDR
1160 types got intertwined, has been fixed;
1161 ** the type system is now able to reason about the interaction
1162 between INTEGER and RATIO types more completely;
1163 ** APPEND, [N]REVERSE and NRECONC check that those their
1164 arguments, which must be proper lists, are really so;
1165 ** An array specialized to be unable to hold elements has been
1166 implemented, as required -- yes, really -- by ANSI;
1167 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1168 on malformed property lists;
1170 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1171 * incompatible packaging change: in line with Unix convention,
1172 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1173 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1174 * REQUIRE and PROVIDE are now optionally capable of doing something
1175 useful. See the documentation string for REQUIRE.
1176 * infrastructure for a managed SBCL contrib system: contributed
1177 modules in this release include:
1178 ** the ASDF system definition facility;
1179 ** an interface to the BSD Sockets API;
1180 ** an ACL-like convenience interface to the repl;
1181 (thanks to Kevin Rosenberg)
1182 ** an implementation of ROTATE-BYTE, with efficient implementation
1184 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1185 gives a closer approximation to the right answer than previously.
1186 (thanks to Raymond Toy)
1187 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1188 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1189 optional environment arguments, as required by ANSI.
1190 * fixed bugs in other functions taking environment objects, allowing
1191 calls with an explicit NIL environment argument to be compiled
1193 * fixed bug 228: primary return values from
1194 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1195 COMPILE or FUNCTION.
1196 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1197 :TYPE will no longer signal errors on innocuous objects.
1198 * fixed bug 231b: SETQ is better at respecting type declarations in
1199 the lexical environment.
1200 * fixed a bug in DEFCLASS: classes named by symbols with no or
1201 unprintable packages can now be defined.
1202 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1203 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1204 * fixed compiler failure related to checking types of functions.
1205 (reported by Robert E. Brown)
1206 * the compiler is now much more consistent in its error-checking
1207 treatment of bounding index arguments to sequence functions: in
1208 (SAFETY 3) code, errors will be signalled in almost all cases if
1209 invalid sequence bounding indices are passed to functions defined
1210 by ANSI to operate on sequences.
1211 * fixed a bug in the build procedure: documentation of SBCL-specific
1212 packages is now preserved and available in the final Lisp image.
1213 * lifted FDEFINITION lookup out of loops in the implementation of
1214 many list operations. (thanks to Robert E. Brown)
1215 * fixed a bug in the reader: the #n# reader macro now works for
1216 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1217 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1218 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1219 E. Brown and Tony Martinez respectively)
1220 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1221 SLOT-BOUNDP: in particular, it is now able to optimize them much
1222 better, and is now not vulnerable to having packages renamed.
1223 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1224 always called when a slot is not present in an instance. (thanks
1226 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1227 longer signal an error on unreading a character following EOF on
1228 the previous constituent stream. (thanks to Tony Martinez)
1229 * fixed some bugs revealed by Paul Dietz' test suite:
1230 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1231 not just nonnegative fixnums;
1232 ** the logical bit-array operators such as BIT-AND now accept an
1233 explicit NIL for their "opt-arg" argument (to indicate a
1234 freshly-consed result bit-array);
1235 ** ELT now signals an error on an invalid sequence index in safe
1237 ** the type system is now cleverer about negations of numeric
1238 types, and consequently understands the BIGNUM and RATIO types
1240 ** the type system is now cleverer about the interaction between
1241 INTEGER and RATIO types: while bugs still remain, many more
1242 cases are accurately computed;
1243 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1244 if it is in the last clause;
1245 ** CONSTANTLY now correctly returns a side-effect-free function in
1247 ** DECLARE is no longer treated as a special-operator; in
1248 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1249 * incremented fasl file version number due to the change in the
1250 DEFSTRUCT-SLOT-DESCRIPTION structure.
1252 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1253 * minor incompatible change: code processed by the "interpreter" or
1254 EVAL now has a compilation optimization policy of (DEBUG 2)
1255 (changed from (DEBUG 1)) to improve debuggability of interactive
1256 development, and to allow the use of the debug RETURN command in
1258 * an experimental implementation of the RETURN command for the
1259 debugger has been included. (thanks to Frederik Kuivinen)
1260 * fixed bug 62: constraints were not propagated into a loop.
1261 * fixed bug in embedded calls of SORT (reported and investigated by
1263 * fixed some bugs revealed by Paul Dietz' test suite:
1264 ** printing and reading of arrays with some dimensions having
1265 length 0 (thanks to Gerd Moellmann);
1266 ** BOA constructor with &AUX argument without a default value does
1267 not cause a type error;
1268 ** CONSTANTP now returns true for all self-evaluating objects.
1270 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1271 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1272 accessors that are related by inheritance, as specified in the
1273 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1274 Valtteri Vuorikoski)
1275 * The compiler is now able to inline functions that were defined in
1276 a complex lexical environment (e.g. inside a MACROLET).
1277 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1279 * Support for the upcoming FreeBSD-5.0 release has been included.
1280 (thanks to Dag-Erling Smorgrav)
1281 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1282 effect when it is not in a toplevel context.
1283 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1284 stand a better chance of being correct. (thanks to Gerd
1286 * fixed bug in COERCE, which now signals an error on coercing a
1287 rational to a bounded real type which excludes the expected
1289 * The compiler is now able to derive types more accurately from the
1290 COERCE and COMPILE functions.
1291 * fixed bug 223: functional binding is considered to be constant
1292 only for symbols in the CL package.
1293 * fixed bug 231: SETQ did not check the type of a variable being set
1294 (reported by Robert E. Brown)
1295 * A new optimization for MAKE-INSTANCE has been included, fixing
1296 various bugs (including relating to :ALLOCATION :CLASS slots and
1297 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1299 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1300 ** As required by ANSI, LOOP now disallows anonymous collection
1301 clauses such as COLLECT I in conjunction with aggregate boolean
1302 clauses such as THEREIS (= I 1);
1303 ** LOOP now signals an error when any variable is reused in the
1304 same loop (including the potentially useful construct analogous
1305 to WITH A = 1 WITH A = (1+ A);
1306 ** IT is only a special loop symbol within the first clause of a
1307 conditional loop clause;
1308 ** LOOP with a typed iteration variable over a hashtable now
1309 signals a type error iff it should.
1310 * fixed some other bugs revealed by Paul Dietz' test suite:
1311 ** FILE-STREAM now names the class previously known as FD-STREAM;
1312 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1313 argument) no longer signals an error;
1314 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1315 of the slot symbol, rather than using the current package
1316 ((:CONC-NAME "") continues to intern the slot's name in the
1318 * incremented fasl file version number, because of the incompatible
1319 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1320 of the new implementation of DEFINE-COMPILER-MACRO.
1322 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1323 * Support for building SBCL for MIPS platforms running in
1324 little-endian mode has now been checked in, and basic
1325 functionality on said platforms verified.
1326 * minor incompatible change: PCL now records the pathname of a file
1327 in which methods and the like are defined, rather than its
1329 * minor incompatible change: TRUENAME now considers the truename of
1330 a file naming a directory to be the pathname with :DIRECTORY
1331 component indicating that directory.
1332 * minor incompatible change: a NAMED clause in the extended form of
1333 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1334 reason for the previous behaviour is unclear.
1335 * more systematization and improvement of CLOS and MOP conformance
1336 in PCL (thanks to Gerd Moellman and Pierre Mai):
1337 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1339 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1340 primary methods with no specializers;
1341 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1343 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1344 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1345 FORWARD-REFERENCED-CLASSes; error reporting on
1346 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1348 ** SXHASH on CLOS instances now uses a slot internal to the
1349 instance to return different numbers on distinct instances,
1350 while preserving the same return value through invocations of
1352 ** DEFMETHOD signals errors when methods with longer incongruent
1353 lambda lists are added to generic functions;
1354 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1355 CLASS, as specified in AMOP;
1356 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1357 on the order returned by the primary method for classes of
1358 class STANDARD-CLASS;
1359 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1360 * fixed some bugs shown by Paul Dietz' test suite:
1361 ** DOLIST puts its body in TAGBODY;
1362 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1364 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1365 value producing form;
1366 ** if more variables are given to PROGV than values, extra
1367 variables are bound and made to have no value;
1368 ** NSUBSTITUTE on list arguments gets the right answer with
1370 ** ELT signals an error of type TYPE-ERROR when the index argument
1371 is not a valid sequence index;
1372 ** LOOP signals (at macroexpansion time) an error of type
1373 PROGRAM-ERROR when duplicate variable names are found;
1374 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1375 ** FUNCALL of special-operators now cause an error of type
1377 ** PSETQ now works as required in the presence of side-effecting
1378 symbol-macro places;
1379 ** NCONC accepts any object as its last argument;
1380 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1382 ** loop-for-as-package does not require a package to be explicitely
1384 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1386 * fixed bug 166: compiler preserves "there is a way to go"
1387 invariant when deleting code.
1388 * fixed bug 172: macro lambda lists with required arguments after
1389 &REST arguments now cause an error to be signalled. (thanks to
1391 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1392 bug. (thanks to Matthew Danish)
1393 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1395 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1396 arguments are lexically rebound. (thanks to Gerd Moellmann and
1398 * fixed bug 194: error messages are now more informative when there
1399 is no primary method applicable in a call to a generic
1400 function. (thanks to Gerd Moellmann)
1401 * fixed bug in command line argument checking (thanks to Julian
1403 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1404 (thanks to Matthew Danish)
1405 * incremented fasl file version number, because of the
1406 SXHASH-related changes in the layout of CLOS data structures
1408 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1409 * minor incompatible change: The runtime (the Unix executable named
1410 "sbcl") is now much pickier about the .core files it will load.
1411 Essentially it now requires .core files to descend from the same
1412 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1413 as the runtime does. (The intent is to prevent the crashes which
1414 can occur, and which can even be reported as mysterious failures,
1415 when people patch the sources or change the build parameters
1416 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1417 sbcl and .core files.)
1418 * fixed bug: VALUES-LIST is no longer optimized away.
1419 * fixed bug 142: The FFI conversion of C string values to Lisp
1420 string values no longer conses excessively. (thanks to Nathan
1421 Froyd porting Raymond Toy's fix to CMU CL)
1422 * began to systematize and improve MOP conformance in PCL (thanks to
1423 Nathan Froyd, Gerd Moellman and Pierre Mai):
1424 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1426 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1427 ** FINALIZE-INHERITANCE is now called on class finalization;
1428 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1429 argument precedence order.
1430 * fixed bug 202: The compiler no longer fails on functions whose
1431 derived types contradict their declared type.
1432 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1433 so it can be non-toplevel.
1434 * The fasl file version number has changed (because of the new
1435 implementation of DEFMACRO).
1436 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1437 safe code, that any length requirement by their type-specifier
1438 argument is valid. The exceptions to this are described in bug
1440 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1441 functions now check the types of their inputs as required by ANSI.
1442 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1443 introduced symbol is DECLAREd to be SPECIAL.
1444 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1445 * fixed bug: :COUNT argument to sequence functions may be negative
1446 * fixed bug: body of DO-SYMBOLS may contain declarations
1447 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1448 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1449 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1450 symbol macro only once
1451 * fixed printing of call frame when argument list is unavailable
1452 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1453 * compiler no longer signals WARNING on unknown keyword
1456 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1457 * A beta-quality port to the mips architecture running Linux,
1458 based on the old CMUCL backend, has been made. It has been tested
1459 on a big-endian kernel, and works sufficiently well to be able to
1460 rebuild itself; it has not been tested in little-endian mode.
1461 * fixed an inconsistency between gencgc.c and purify.c which made
1462 dumping/loading .core files unreliable
1463 * fixed bug 120a: The compiler now deals correctly with IFs where
1464 the consequent is the same as the alternative, instead of
1465 misderiving the return type. (thanks to Alexey Dejneka)
1466 * fixed bug 113: Logical pathnames are now dumpable (the logical
1467 host is resolved at load-time, throwing an error if it is not
1469 * fixed bug 174: FORMAT's error message is slightly clearer when a
1470 non-printing character is used in a format directive.
1471 * fixed several bugs in compiler checking of type declarations, i.e.
1472 violations of the Python "declarations are assertions" principle
1473 (thanks to Alexey Dejneka)
1474 * fixed several bugs in PCL's error checking (thanks to Gerd
1476 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1478 * fixed bug in compilation of functions as first class values
1479 (thanks to Antonio Martinez-Shotton)
1480 * The compiler's handling TYPE-ERRORs which it can prove will
1481 inevitably happen at runtime has been cleaned up and corrected
1482 in several ways. (thanks to Alexey Dejneka)
1483 * improved argument type checking for various basic arithmetic
1484 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1485 lavished on them in the past that they can be compiled in many
1486 ways in different special cases
1487 * fixed bug 181: compiler checks validity of user supplied type
1489 * cleaned up code flushing in optimization: Function calls which
1490 should signal errors for safety purposes (e.g. which ANSI says
1491 should signal errors when their arguments are of incorrect type)
1492 are no longer optimized away.
1493 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1494 * incremented fasl file version number, because changes in the
1495 implementation of sequence functions like COERCE caused
1496 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1497 in old inline expansions) to become undefined. (Actually these
1498 changes were later undone, so we might very well be binary
1499 compatible with 0.7.7 after all, but leaving the version number
1500 incremented seemed like the simplest and most conservative
1503 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1504 * An alpha-quality port to the parisc architecture running Linux,
1505 based on the old CMUCL backend, has been made. This, even more so
1506 than the other backends, should be considered still a work in
1507 progress; known problems include that the Linux kernel in 64-bit
1508 mode does not propagate the correct sigcontext structure to
1509 userspace, and consequently SBCL on a parisc64 kernel will not
1511 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1512 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1513 suppressed.) Also now that the compiler is looking at declarations
1514 in the environment, it checks optimization declarations as well,
1515 and suppresses inlining when (> DEBUG SPEED).
1516 * More fixes have been made to treatment of floating point exception
1517 treatment and other Unix signals. In particular, floating point
1518 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1519 * The detection and handling of control stack exhaustion (infinite
1520 or very deeply nested recursion) has changed. Stack exhaustion
1521 detection is now done by write-protecting pages at the OS level
1522 and applies at all optimization settings; when found, a
1523 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1524 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1525 be caught using IGNORE-ERRORS.
1526 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1527 correctly when transforming recursive LABELS functions to LETs.
1528 (thanks to Alexey Dejneka)
1529 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1530 that are names of constants or global variables.
1531 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1532 alien routines with docstrings.
1533 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1534 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1536 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1537 (thanks to Eric Marsden)
1538 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1539 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1540 * Bug fix: The SPARC backend can now compile functions involving
1541 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1542 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1543 to a functions expecting a PATHNAME-DESIGNATOR.
1544 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1545 lambda lists. (thanks to Alexey Dejneka)
1546 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1547 functions so that EXPORTing the name of the class doesn't cause
1548 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1549 to Antonio Martinez for reporting this)
1550 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1551 constant-folded without failing an assertion. (thanks to Einar
1552 Floystad Dorum for reporting this)
1553 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1554 types are now dealt with more correctly by the compiler.
1555 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1556 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1557 behaviour analogous to RENAME-FILE. This puts its behaviour more
1558 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1560 * The fasl file version number has changed again. (because of the
1561 bug fix involving the names of PCL MAKE-INSTANCE functions)
1563 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1564 * bug fix: Floating point exceptions are treated much more
1565 consistently on the x86/Linux and PPC/Linux platforms.
1566 * Array initialization with :INITIAL-ELEMENT is now much faster for
1567 cases when the compiler cannot open code the array creation, but
1568 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1569 array accesses have also seen a speed increase.
1570 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1571 specified. (This was at the root of some bad interactions between
1572 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1574 * bug fix: Internal error arguments for undefined functions are now
1575 computed correctly on the PPC/Linux platform.
1576 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1577 Raymond Toy's patch for CMU CL)
1578 * Support for the Solaris 9 operating environment has been included
1579 (thanks to Daniel Merritt)
1580 * A very ugly but hopefully complete draft of the missing FFI chapter
1581 of the manual has been created by reformatting the corresponding
1582 CMU CL manual chapter into (currently very ugly and incoherent)
1583 DocBook and bringing it up to date for SBCL behavior. Thus, the
1584 manual is now essentially complete, at least by my extreme
1585 once-and-only-once standards, whereby it's acceptable to refer to
1586 the doc strings of SB-EXT functions as the primary documentation.
1587 * The fasl file version number has changed again, due to cleanup of
1588 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1589 is no longer a static symbol.)
1591 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1592 * SBCL now builds with OpenMCL (version 0.12) as the
1593 cross-compilation host; also, more progress has been made toward
1594 bootstrapping under CLISP.
1595 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1597 * bug 158 fixed: The compiler can now deal with integer loop
1598 increments different from 1; fixing this turned out also to fix
1600 * bug 169 fixed: no more bogus warnings about using lexical bindings
1601 despite the presence of perfectly good SPECIAL declarations (thanks
1602 to David Lichteblau)
1603 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1604 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1605 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1607 * bug 180 fixed: Method combination specifications no longer ignore
1608 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1609 * bug fix: Structure type predicate functions now check their argument
1610 count as they should.
1611 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1612 correctly. (thanks to Pierre Mai)
1613 * minor incompatible change: The --noprogrammer option is deprecated
1614 in favor of the new --disable-debugger option, which is very similar.
1615 (The major difference is that it takes effect at a slightly different
1616 time at startup, causing handling of errors in --sysinit and
1617 --userinit files will be affected differently.) The
1618 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1619 been added to allow this functionality to be controlled from ordinary
1620 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1621 maintainers, who might want to run non-interactive scripts to
1622 build SBCL cores which will later be used interactively.)
1623 * minor incompatible change: The LOAD function no longer, when given
1624 a wild pathname to load, loads all files matching that pathname.
1625 Instead, an error of type FILE-ERROR is signalled.
1627 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1628 * bug 147 fixed: The compiler preserves its block link/count
1629 invariants more correctly now so that it doesn't crash. (thanks
1631 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1633 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1634 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1635 no-value-for-this-character as NIL (as specified by ANSI).
1636 * HOST-NAMESTRING on physical pathnames now returns a string that is
1637 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1638 (thanks to Christophe Rhodes)
1639 * The Alpha port handles icache flushing more correctly. (thanks to
1641 * More progress has been made toward bootstrapping under CLISP. (thanks
1642 to Christophe Rhodes)
1643 * The fasl file format has changed again, because dynamic loading
1644 on OpenBSD (which has non-ELF object files) motivated some cleanups
1645 in the way that foreign symbols are transformed and passed around.
1646 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1647 is no longer treated as whitespace by the reader, but instead as
1648 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1649 |A
\7fB|, instead of A as it used to.
1651 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1652 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1653 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1654 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1655 this as of 0.7.1.45, but was left out of the previous news section
1656 (thanks to Dan Barlow)
1657 * SBCL now runs on the Solaris operating system on SPARC architectures
1658 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1659 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1660 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1661 * SPARC backend cleanups, allowing builds of cores optimized for V8
1662 and V9 SPARCS, and also emission of code targeted to a particular
1663 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1665 * SBCL is closer to bootstrapping under CLISP, thanks to various
1666 fixes by Christophe Rhodes.
1667 * The fasl file format has changed again, to allow the compiler's
1668 INFO database to support symbol macros.
1669 * The user manual (in doc/) is formatted into HTML more nicely.
1670 (thanks to coreythomas)
1671 * The system is smarter about SUBTYPEP relationships, especially
1672 those involving NOT types (including types such as ATOM which are
1673 represented internally using NOT types). Thus SUBTYPEP is less
1674 likely to return (VALUES NIL NIL) in general, and in particular
1675 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1677 * The fasl file format has changed again, because the internal
1678 representation of types now includes a new slot to support the new
1679 SUBTYPEP-of-NOT-types logic.
1680 * (not a change in the main branch of SBCL, but a related prototype
1681 which can hopefully be merged into the main branch of SBCL in the
1682 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1683 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1684 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1685 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1686 dispatch macro character combinations. (thanks to Alexey Dejneka)
1687 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1688 that superficially look like logical namestrings correctly.
1689 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1691 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1692 default supplied-p) for &optional and &key arguments. (thanks to
1695 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1696 * incompatible change: The compiler is now less aggressive about
1697 tail call optimization, doing it only when (> SPACE DEBUG) or
1698 (> SPEED DEBUG). (This is an incompatible change because there are
1699 programs which relied on the old CMU-CL-style behavior to optimize
1700 away their unbounded recursion which will now die of stack overflow.)
1701 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1702 for non-GENCGC systems has been increased to 20M (since that
1703 seems much closer to the likely performance optimum for modern
1704 systems than the old 4M value was)
1705 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1706 and *DEBUG-PRINT-LEVEL*
1707 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1708 of CMU CL's support for SPARC, and various endianness and other
1709 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1710 * new syntactic sugar for the Unix command line: --load foo.bar is now
1711 an alternate notation for --eval '(load "foo.bar")'.
1713 ** The system now detects stack overflow and handles it gracefully,
1714 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1715 optimization settings. (This is a good thing in general, and
1716 its introduction in this version should be particularly timely
1717 for anyone whose code fails because of suppression of tail
1719 ** The system now hunts for the C variable "environ" in a more
1720 devious way, to avoid segfaults when the C library version differs
1721 between compile time and run time. (thanks to Christophe Rhodes)
1722 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1723 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1724 ** The compiler no longer issues bogus style warnings for undefined
1725 classes in the same source file as the DEFCLASSes which defined
1726 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1728 ** fixes in CONDITION class precedence list for undefined function
1729 errors (thanks to Alexei Dejneka)
1730 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1731 correctly. (thanks to Dan Barlow)
1732 ** portability fixes aiming at bootstrapping under CLISP (thanks
1733 to Dave McDonald and Christophe Rhodes)
1734 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1735 ** fixes in type translation and and type inference (thanks to
1737 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1738 ** various fixes in the new ports (thanks to Dan Barlow)
1739 * several changes related to debugging:
1740 ** suppression of tail recursion, as noted above
1741 ** stack overflow detection, as noted above
1742 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1743 is now the default. (For some time encapsulation has been more
1744 reliable than the breakpoint-based :ENCAPSULATE NIL
1745 implementation, at least on X86 systems; and I just noticed that
1746 encapsulation also seems closer to the spirit of the ANSI
1749 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1751 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1752 up properly again. (There was a packaging bug in 0.7.0 which
1753 left their definitions in SB-SYS::LOAD-FOREIGN and
1754 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1755 vital for most things which interface to C-level interfaces,
1756 like extensions working with sockets or databases or
1757 Perl-compatible regexes or whatever, and the need to fix
1758 this bug is the main reason that 0.7.1 was released so
1760 ** DEFGENERIC is now choosier about the methods it redefines, so that
1761 reLOADing a previously-LOADed file containing DEFGENERICs does
1762 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1763 cycle now works as it should. (thanks to Alexey Dejneka)
1764 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1765 fixed by Christophe Rhodes. (He actually submitted this patch
1766 months ago, and I delayed until after 0.7.0.)
1767 ** Bug 111 (internal compiler confusion about runtime checks on
1768 FUNCTION types) was fixed by Alexey Dejneka.
1769 * Some internal cleanups (getting rid of variables which aren't
1770 needed now that the byte interpreter is gone) caused the fasl
1771 file format number to change again.
1773 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1774 * major incompatible change: The default fasl file extension, i.e. the
1775 default extension for files produced by COMPILE-FILE, has changed
1776 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1778 ** There are many changes in the implementation of the compiler.
1779 SBCL is now essentially a compiler-only implementation of ANSI
1780 Common Lisp. EVAL still "interprets" a few special cases, but
1781 almost all the interesting cases are handled by creating
1782 a LAMBDA expression, calling COMPILE on it, then calling
1783 FUNCALL on the result.
1784 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1785 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1786 Since the code is newer, there might still be some new bugs
1787 (though not as many as before Martin Atzmueller's fixes:-). But
1788 the new code is substantially simpler and clearer, and hopefully
1789 any remaining bugs will be simpler, less fundamental, and more
1790 fixable then the bugs in the old code.
1791 ** The revised compiler is still a little unsteady on its feet.
1793 *** The debugging information it produces (particularly the names
1794 of FUNCTION objects) is sometimes much less useful than what
1795 the old compiler produced.
1796 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1797 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1798 within a MACROLET) has been temporarily weakened.
1799 ** There are new compiler optimizations for various functions:
1800 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1801 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1802 *** the math functions TRUNCATE, FLOOR, and CEILING
1803 *** the function-of-all-trades COERCE
1804 Mostly these should be transparent, but there's one
1805 potentially-annoying problem (bug 117): when the compiler
1806 inline-expands a function and does type analysis on the result,
1807 it can create control paths which have type mismatches, and
1808 when it can't prove that those control paths aren't taken,
1809 it will issue WARNINGs about the type mismatches. This is
1810 a particular problem in practice for the new sequence functions.
1811 It's not clear how this should be fixed, and for now, a
1812 workaround is given in the entry for 117 in the BUGS file.
1813 ** (Because of the interaction between the two previous items --
1814 occasional inlining problems and new inline expansions -- some
1815 of the new sequence function optimizations won't really kick in
1816 completely until debugging information, and then inlining, are
1817 straightened out in some future version.)
1818 * minor incompatible changes:
1819 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1820 DIRECTORY no longer implicitly promotes NIL slots of its
1821 pathname argument to :WILD. In particular, when you ask for the
1822 contents of a directory (which you used to be able to do without
1823 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1824 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1825 ** changes in behavior that ANSI explicitly defines to be
1826 implementation dependent:
1827 *** The new compiler-only implementation still conforms with ANSI,
1828 but acts a little different than before. Besides the obvious
1829 changes in performance tradeoffs (that the cost per form passed
1830 to EVAL has gone up, and the cost per form executed by EVAL
1831 has gone down), the behavior of the system changes a little
1832 because there are no longer any interpreted function objects.
1833 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1834 e.g. doing COMPILE on the output of interactive DEFUN is
1836 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1838 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1839 has changed from (:CL) to NIL.
1840 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1841 e.g. #\Newline and #\Tab, don't have names specified in the
1842 ANSI Common Lisp standard, is now based on their ASCII symbolic
1843 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1844 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1845 are no longer used for output.
1846 ** changes in internal implementation constants:
1847 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1848 4 million. (If your application spends a lot of time GCing and
1849 you have a lot of RAM, you might want to experiment with
1850 increasing it even more.)
1851 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1852 However, almost all old code should still continue to work without
1853 immediate update, as SB-C-CALL is now a (deprecated) nickname
1855 ** Old operator names in the style DEF-FOO are now deprecated in
1856 favor of new corresponding names DEFINE-FOO, for consistency with
1857 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1858 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1859 internal symbols, but a few supported extensions like
1860 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1861 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1862 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1863 etc. as you get deeper into recursive calls to the debugger
1864 command loop, instead of the old "5]", "5]]", "5]]]"
1865 sequence. (I was motivated to do this when squabbles between
1866 ILISP and SBCL left me very deeply nested in the debugger. In the
1867 short term, this change will probably provoke more ILISP/SBCL
1868 squabbles, but hopefully it will be an improvement in the long run.)
1869 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1870 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1871 function, since declaiming return types involving aliens is
1872 (1) annoyingly messy to do by hand and (2) vital to efficient
1873 compilation of code which calls such functions.
1874 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1875 longer reexported by the SB-EXT package. They're solely useful
1876 for alien code, so it seems more logical that you should get
1877 them from the SB-ALIEN package, not in SB-EXT.
1878 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1879 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1880 features. Instead, the code that they used to control is always
1881 built into the system.
1882 * many other bug fixes
1883 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1884 advantage of the new EVAL-WHEN stuff and to clean them up in
1885 general, and they are now more ANSI-compliant in a number of
1886 ways. Martin Atzmueller is responsible for a lot of this.
1887 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1889 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1890 *** correct ERROR type for various file operations
1891 *** some fixes for Lisp streams
1892 *** DEFMETHOD syntax checking
1893 *** changing old weird representation of debug information as
1894 strings (which, among their other deficiencies, don't transform
1895 correctly when you rename packages, and don't change their
1896 print representation when you change things like *PACKAGE*
1897 and *PRINT-LENGTH*) to symbols and lists of symbols
1898 He also made several improvements and fixed several bugs in DESCRIBE.
1899 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1901 *** misbehavior of WRITE-STRING/WRITE-LINE
1902 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1903 and several other LOOP problems as well
1904 *** DIRECTORY when similar filenames are present
1905 *** DEFGENERIC with :METHOD options
1906 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1907 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1908 *** argument ordering in FIND with :TEST option
1909 *** mishandled package designator argument in APROPOS-LIST
1910 *** various problems in the backquote readmacro
1911 *** a bug in APROPOS
1912 *** probably some others that I'm not describing very well here,
1913 since the CVS log documents them by reference to sbcl-devel
1914 messages, and the SourceForge archives aren't working well.:-(
1915 ** Dan Barlow improved the Alpha port (and is making progress on the
1916 PPC port, for those of you who think different).
1917 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
1918 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
1919 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
1920 continues to work on the SPARC port (for those of you in a position
1921 to look down upon our little PC-compatible boxes from a great height).
1922 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
1923 and uses that copy, rather than the current dynamic value, when
1924 it's trying to decide whether to truncate output. Thus e.g.
1925 (let ((*print-lines* 50))
1926 (pprint-logical-block (stream nil)
1928 (let ((*print-lines* 8))
1929 (print (aref possiblybigthings i) stream)))))
1930 should now truncate the logical block only at 50 lines, instead of
1931 often truncating it at 8 lines, as it did before.
1932 * The doc/cmucl/ directory, containing old CMU CL documentation
1933 from the time of the fork, is no longer part of the base system.
1934 SourceForge has shut down its anonymous FTP service, and with it
1935 my original plan for distributing the old CMU CL documentation
1936 there. For now, if you need these files you can download an old
1937 SBCL source release and extract them from it.
1938 * The fasl file version number changed again, for dozens of reasons,
1939 some of which are apparent above.
1941 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
1942 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
1943 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
1944 patch, so that DISASSEMBLE output is much nicer.
1945 * The code in the SB-PROFILE package now seems reasonably stable.
1946 I still haven't decided what the final interface should look like
1947 (I'd like PROFILE to interact cleanly with TRACE, since both
1948 facilities use function encapsulation) but if you have a need
1949 for profiling now, you can probably use it successfully with
1950 the current CMU-CL-style interface.
1951 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
1952 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
1953 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
1954 link now returns the dangling link itself, and for similar
1955 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
1956 link itself. (In these cases the old code signalled an error and
1957 looped endlessly, respectively.) Thus, DIRECTORY now works even
1958 in the presence of dangling and cyclic symbolic links.
1959 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
1960 is now a supported extension again, since the consensus on
1961 sbcl-devel was that it can be useful for ordinary development
1962 work, not just for debugging SBCL itself.
1963 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
1964 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
1965 that currently-defined functions might be redefined later with
1966 different return types.
1967 * Hash tables can be printed readably, as inspired by CMU CL code
1968 of Eric Marsden and SBCL code of Martin Atzmueller.
1969 * better error handling in CLOS method combination, thanks to
1970 Martin Atzmueller porting Pierre Mai's CMU CL patches
1971 * more overflow fixes for >16Mbyte I/O buffers
1972 * A bug in READ has been fixed, so that now a single Ctrl-D
1973 character suffices to cause end-of-file on character streams.
1974 In particular, now you only need one Ctrl-D at the command
1975 line (not two) to exit SBCL.
1976 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
1978 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
1979 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
1980 does the right thing.
1981 * The implementation of some type tests, especially for CONDITION
1982 types, is now tidier and maybe faster, due to CMU CL code
1983 originally by Douglas Crosher, ported by Martin Atzmueller.
1984 * Some math functions have been fixed, and there are new
1985 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
1986 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
1987 * (There are also some new optimizers in contrib/*-extras.lisp. Those
1988 aren't built into sbcl-0.6.13, but are a sneak preview of what's
1989 likely to be built into sbcl-0.7.0.)
1990 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
1991 CL, ported to SBCL by Martin Atzmueller)
1992 * DESCRIBE now gives more information in some cases. (Pierre Mai's
1993 patch to CMU CL, ported to SBCL by Martin Atzmueller)
1994 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
1995 * There's a new slam.sh hack to shorten the edit/compile/debug
1996 cycle for low-level changes to SBCL itself, and a new
1997 :SB-AFTER-XC-CORE target feature to control the generation of
1998 the after-xc.core file needed by slam.sh.
1999 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2000 COMPILE-FILE is no longer supported, so that now every function
2001 gets an entry point, so that block compilation looks a little
2002 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2003 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2004 returns the number of bytes consed since the system started,
2005 rather than the number consed since the first time the function
2006 was called. (The new definition parallels ANSI functions like
2007 CL:GET-INTERNAL-RUN-TIME.)
2008 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2009 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2010 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2011 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2012 * Fasl file version numbers are now independent of the target CPU,
2013 since historically most system changes which required version
2014 number changes have affected all CPUs equally. Similarly,
2015 the byte fasl file version is now equal to the ordinary
2018 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2019 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2020 is no longer recognized. I apologize for this, because it was
2021 listed in SB-EXT as a supported extension, but I found that
2022 its existing behavior was poorly specified, as well as incorrectly
2023 specified, and it looked like too much of a mess to straighten it
2024 out. I have enough on my hands trying to get ANSI stuff to work..
2025 * many patches ported from CMU CL by Martin Atzmueller, with
2026 half a dozen bug fixes in pretty-printing and the debugger, and
2027 half a dozen others elsewhere
2028 * fixed bug 13: Floating point infinities are now supported again.
2029 They might still be a little bit flaky, but thanks to bug reports
2030 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2031 as flaky as they were.
2032 * The --noprogrammer command line option is now supported. (Its
2033 behavior is slightly different in detail from what the old man
2034 page claimed it would do, but it's still appropriate under the
2035 same circumstances that the man page talks about.)
2036 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2037 are now supported, and enabled by default. Thus, the compiler can
2038 handle many floating point and complex operations much less
2039 inefficiently. (Thus e.g. you can implement a complex FFT
2041 * The compiler now detects type mismatches between DECLAIM FTYPE
2042 and DEFUN better, and implements CHECK-TYPE more correctly, and
2043 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2044 thanks to patches from Martin Atzmueller.
2045 * various fixes to make the cross-compiler more portable to
2046 ANSI-conforming-but-different cross-compilation hosts (notably
2047 Lispworks for Windows, following bug reports from Arthur Lemmens)
2048 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2049 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2050 fixed thanks to Pierre Mai's CMU CL patches.
2051 * fixes to keep the system from overflowing internal counters when
2052 it tries to use i/o buffers larger than 16M bytes
2053 * fixed bug 45a: Various internal functions required to support
2054 complex special functions have been merged from CMU CL sources.
2055 (When I was first setting up SBCL, I misunderstood a compile-time
2056 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2057 * improved support for type intersection and union, fixing bug 12
2058 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2059 more obscure bugs as well
2060 * some steps toward byte-compiling non-performance-critical
2061 parts of the system, courtesy of patches from Martin Atzmueller
2062 * Christophe Rhodes has made some debian packages of sbcl at
2063 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2064 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2065 stable, but are nonetheless usable. When he's ready, I'd be happy
2066 to add them to the SourceForge "File Releases" section. (And if
2067 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2068 * new fasl file format version number (because of changes in
2069 internal representation of (OR ..) types to accommodate the new
2070 support for (AND ..) types, among other things)
2072 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2073 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2075 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2077 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2078 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2079 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2080 * DESCRIBE now works on structure objects again.
2081 * Most function call argument type mismatches are now handled as
2082 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2083 know whether the function will be redefined before the call is
2084 executed. (The compiler could flag local calls with full WARNINGs,
2085 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2086 it doesn't keep track of enough information to know whether calls
2087 are local in this sense.)
2088 * Compiler output is now more verbose, with messages truncated
2089 later than before. (There should be some supported way for users
2090 to override the default verbosity, but I haven't decided how to
2091 provide it yet, so this behavior is still controlled by the internal
2092 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2093 src/compiler/ir1util.lisp.)
2094 * Fasl file format version numbers have increased again, because
2095 support for the Gray streams extension changes the layout of the
2096 system's STREAM objects.
2097 * The Gray subclassable streams extension now works, thanks to a
2098 patch from Martin Atzmueller.
2099 * The full LOAD-FOREIGN extension (not just the primitive
2100 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2101 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2102 but like most other programs, it defaults to copying the Unix
2103 environment from the original process instead of starting the
2104 new process in an empty environment.
2105 * Extensions which manipulate the Unix environment now support
2106 an :ENVIRONMENT keyword option which doesn't smash case or
2107 do other bad things. The CMU-CL-style :ENV option is retained
2108 for porting convenience.
2109 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2110 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2112 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2114 * A patch from Martin Atzmueller seems to have solved the SIGINT
2115 problem, and as far as we know, signal-handling now works cleanly.
2116 (If you find any new bugs, please report them!)
2117 * The system no longer defaults Lisp source file names to types
2118 ".l", ".cl", or ".lsp", but only to ".lisp".
2119 * The compiler no longer uses special default file extensions for
2120 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2121 seems to expect a single default extension for all compiled code,
2122 and there's no compelling reason to try to stretch the standard
2123 to allow two different extensions.) Instead, byte-compiled files
2124 default to the same extension as native-compiled files.
2125 * Fasl file format version numbers have increased again, because
2126 a rearrangement of internal implementation packages made some
2127 dumped symbols in old fasl files unreadable in new cores.
2128 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2129 many fewer weird special cases.
2130 * Bug #17 (differing COMPILE-FILE behavior between logical and
2131 physical pathnames) has been fixed, and some related misbehavior too,
2132 thanks to a patch from Martin Atzmueller.
2133 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2134 by Tim Moore, ported to SBCL by Martin Atzmueller.
2135 * Martin Atzmueller fixed several filesystem-related problems,
2136 including bug #36, in part by porting CMU CL patches, which were
2137 written in part by Paul Werkowski.
2138 * More compiler warnings in src/runtime/ are gone, thanks to
2139 more patches from Martin Atzmueller.
2140 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2143 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2145 * DESCRIBE now works on CONDITION objects.
2146 * The debugger now handles errors which arise when trying to print
2147 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2149 * The build system now uses an additional file, customize-target-features.lisp,
2150 to allow local modifications to the target *FEATURES* list. (The point of
2151 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2152 debugging features enabled, without having to worry about propagating it
2153 into everyone's system when I do a "cvs update".) When no
2154 customize-target-features.lisp file exists, the target *FEATURES* list
2155 should be constructed the same way as before.
2156 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2157 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2158 warn about behavior which is undefined under the ANSI spec. Note: This
2159 is specified by ANSI, but it's not very popular with programmers.
2160 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2161 macro in the SBCL sources for an example of a workaround which you
2162 might use to make portable ANSI-standard code which does what you want.
2163 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2164 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2165 when it's used as a non-toplevel form. (This is required in order
2166 to implement the DEFCONSTANT-EQX macro.)
2167 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2168 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2169 funny things, due to bugs in the implementation of EVAL-WHEN
2170 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2171 non-toplevel DEFCONSTANTs are uncommon.)
2172 * The core file version number and fasl file version number have been
2173 incremented, because the old noncompliant DEFCONSTANT behavior involved
2174 calling functions which no longer exist, and because I also took the
2175 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2176 * fixed bug 1 (error handling before read-eval-print loop starts), and
2177 redid debugger restarts and related debugger commands somewhat while
2179 ** The QUIT debugger command is gone, since it did something
2180 rather different than the SB-EXT:QUIT command, and since it never
2181 worked properly outside the main toplevel read/eval/print loop.
2182 Invoking the new TOPLEVEL restart provides the same functionality.
2183 ** The GO debugger command is also gone, since you can just invoke
2184 the CONTINUE restart directly instead.
2185 ** The TOP debugger command is also gone, since it's redundant with the
2186 FRAME 0 command, and since it interfered with abbreviations for the
2188 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2190 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2191 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2192 some porting work by Martin Atzmueller)
2193 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2194 with a boatload of patches which he ported to SBCL. Now that those
2196 ** The system tries to make sure that its low-priority messages
2197 are prefixed by semicolons, to help people who like to use
2198 syntax highlighting in their ILISP buffer. (This patch
2199 was originally due to Raymond Toy.)
2200 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2201 patches originally written by Raymond Toy.
2202 ** The compiler understands coercion between single-value and
2203 multiple-VALUES type expressions better, getting rid of some very
2204 weird behavior, thanks to patches originally by Robert MacLachlan
2205 and Douglas Crosher.
2206 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2207 lambda lists, thanks to a patch originally by Pierre Mai.
2208 ** The system no longer bogusly warns about "abbreviated type
2210 ** The compiler gets less confused by inlining and RETURN-FROM,
2211 thanks to some patches originally by Tim Moore.
2212 ** The system no longer hangs when dumping circular lists to fasl
2213 files, thanks to a patch originally from Douglas Crosher.
2214 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2215 undefined function error.
2216 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2217 in structure slot initforms) for now, documented workaround instead:-|
2218 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2219 * fixed bug 5: added stubs for various Gray stream functions called
2220 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2221 installed, at least appropriate type errors are generated
2222 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2223 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2224 consistently in DEFMETHOD forms.
2225 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2226 it doesn't seem to affect SBCL after all
2227 * The C runtime system now builds with better optimization and many
2228 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2230 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2232 * The system is now under CVS at SourceForge (instead of the
2233 CVS repository on my home machine).
2234 * The new signal handling code has been tweaked to treat register
2235 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2236 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2237 downstream, e.g. in debug-int.lisp, has implicit dependencies
2238 on the unsignedness of integer representation of machine words,
2239 and that caused the system to bomb out with infinite regress
2240 when trying to recover from type errors involving signed values,
2241 e.g. (BUTLAST '(1 2 3) -1).)
2242 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2243 (This was one of the bugs Peter Van Eynde reported back in July.)
2244 * The system now uses code inspired by Colin Walters' O(N)
2245 implementation of MAP (from the cmucl-imp@cons.org mailing
2246 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2247 inline the MAP operation, and there is more than one
2248 sequence argument to the MAP call (so that it can't just
2249 do ETYPECASE once and for all based on the type of the
2250 single sequence argument). (The old non-inline implementation
2251 of the general M-argument sequence-of-length-N case required
2252 O(M*N*N) time when any of the sequence arguments were LISTs.)
2253 * The QUIT :UNIX-CODE keyword argument has been renamed to
2254 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2255 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2257 * Raymond Wiker's patches to port dynamic loading from Linux to
2258 FreeBSD have been added.
2259 * The BUGS file is now more nearly up to date, thanks in large part
2260 to Martin Atzmueller's review of it.
2261 * The debugger now flushes standard output streams before it begins
2262 its output ("debugger invoked" and so forth).
2263 * The core version number and fasl file version number have both
2264 been incremented, because of incompatible changes in the layout
2266 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2267 used to not be called for a saved Lisp image.)
2268 * Martin Atzmueller's version of a patch to fix a compiler crash,
2269 as posted on sbcl-devel 13 September 2000, has been installed.
2270 * Instead of installing Martin Atzmueller's patch for the
2271 compiler transform for SUBSEQ, I deleted the compiler transform,
2272 and transforms for some similar consing operations.
2273 * A bug in signal handling which kept TRACE from working on OpenBSD
2275 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2276 away by constant folding
2277 * The system now defines its address space constants in one place
2278 (in the Lisp sources), and propagates them automatically elsewhere
2279 (through GENESIS and the sbcl.h file). Therefore, patching the
2280 address map is less unnecessarily tedious and error-prone. The
2281 Lisp names of address space constants have also been systematized.
2282 * CVS tags like dollar-Header-dollar have been removed from
2283 the sources, because they have never saved me trouble and
2284 they've been source of trouble working with patches and other
2285 diff-related operations.
2286 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2287 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2289 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2291 * The system has been ported to OpenBSD.
2292 * The system now compiles with a simple "sh make.sh" on the systems
2293 that it's supported on. I.e., now you no longer need to tweak
2294 text in the target-features.lisp-expr and symlinks in src/runtime/
2295 by hand, the make.sh takes care of it for you.
2296 * The system is no longer so grossly inefficient when compiling code
2297 involving vectors implemented as general (not simple) vectors (VECTOR T),
2298 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2299 which dares to use the various sequence functions on non-simple
2300 vectors, takes less of a performance hit.
2301 * There is now a primitive type predicate VECTOR-T-P
2302 to test for the (VECTOR T) type, so that e.g.
2303 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2304 can now be compiled with some semblance of efficiency. (The old code
2305 turned the type declaration into a full call to %TYPEP at runtime!)
2306 * AREF on (VECTOR T) is still not fast, since it's still compiled
2307 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2308 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2309 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2310 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2311 without making full calls to SUBTYPEP at runtime.
2312 (Some analogous efficiency issues for non-simple vectors specialized to
2313 element types other than T, or for non-simple multidimensional arrays,
2314 have not been addressed. They could almost certainly be handled the
2315 same way if anyone is motivated to do so.)
2316 * The changes in array handling break binary compatibility, so
2317 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2318 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2319 instead of (VALUES T T).
2320 * By following the instructions that Dan Barlow posted to sbcl-devel
2321 on 2 July 2000, I was able to enable primitive dynamic object
2322 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2323 functionality is not implemented (since it calls ld to resolve
2324 library references automatically, requiring RUN-PROGRAM for its
2325 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2326 try to resolve library references) is now supported.
2327 * The system now flushes the standard output streams when it terminates,
2328 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2329 them at several other probably-convenient times, e.g. in each pass of
2330 the toplevel read-eval-print loop, and after evaluating a form given
2331 as an "--eval" command-line option. (These changes were motivated by a
2332 discussion of stream flushing issues on cmucl-imp in August 2000.)
2333 * The source transform for TYPEP of array types no longer assumes
2334 that an array whose element type is a not-yet-defined type
2335 is implemented as an array of T, but instead punts, so that the
2336 type will be interpreted at runtime.
2337 * There is now some support for cross-compiling in make.sh: each of
2338 the phases of make.sh has its own script. (This should be transparent
2339 to people doing ordinary, non-cross-compile builds.)
2340 * Since my laptop doesn't have hundreds of megabytes of memory like
2341 my desktop machine, I became more motivated to do some items on
2342 my to-do list in order to reduce the size of the system a little:
2343 ** Arrange for various needed-only-at-cold-init things to be
2344 uninterned after cold init. To support this, those things have
2345 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2346 symbols with such names are now uninterned after cold init).
2347 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2348 fasl files for cold load.
2349 ** Remove the old compiler structure pooling code (which used to
2350 be conditional on the target feature :SB-ALLOC) completely.
2351 ** Redo the representation of some data in cold init to be more compact.
2352 (I also looked into supporting byte compiled code at bootstrap time,
2353 which would probably reduce the size of the system a lot, but that
2354 looked too complicated, so I punted for now.)
2355 * The maximum signal nesting depth in the src/runtime/ support code has
2356 been reduced from 4096 to 256. (I don't know any reason for the very
2357 large old value. If the new smaller value turns out to break something,
2358 I'll probably just bump it back up.)
2359 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2361 * Many, many bugs reported by Peter Van Eynde have been added to
2362 the BUGS list; some have even been fixed.
2363 * While enabling dynamic object file loading, I tried to make the
2364 code easier to understand, renaming various functions and variables
2365 with less ambiguous names, and changing some function calling
2366 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2367 * While trying to figure out how to do the OpenBSD port, I tried to
2368 clean up some of the code in src/runtime/. In particular, I dropped
2369 support for non-POSIX signal handling, added various comments,
2370 tweaked the code to reduce the number of compilation warnings, and
2371 renamed some files to increase consistency.
2372 * To support the new automatic configuration functionality in make.sh,
2373 the source file target-features.lisp-expr has been replaced with the
2374 source file base-target-features.lisp-expr and the machine-generated
2375 file local-target-features.lisp-expr.
2376 * fixed a stupid quoting error in make.sh so that using CMU CL
2377 "lisp -batch" as cross-compilation host works again
2379 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2381 * DESCRIBE no longer tries to call itself recursively to describe
2382 bound/fbound values, so that it no longer fails on symbols which are
2383 bound to themselves (like keywords, T, and NIL).
2384 * DESCRIBE now works on generic functions.
2385 * The printer now prints less-screwed-up representations of closures
2386 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2387 * A private symbol is used instead of the :EMPTY keyword previously
2388 used to mark empty slots in hash tables. Thus
2389 (DEFVAR *HT* (MAKE-HASH-TABLE))
2390 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2391 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2392 now does what ANSI says that it should. (You can still get
2393 similar noncompliant behavior if bang on the hash table
2394 implementation with all the symbols you get back from
2395 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2396 This breaks binary compatibility, since tests for equality to
2397 :EMPTY are wired into things like the macroexpansion of
2398 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2400 * There's now a minimal placeholder implementation for CL:STEP,
2401 as required by ANSI.
2402 * An obscure bug in the interaction of the normal compiler, the byte
2403 compiler, inlining, and structure predicates has been patched
2404 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2405 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2407 * Missing ordinary arguments in a macro call are now detected even
2408 when the macro lambda list contains &KEY or &REST.
2409 * The debugger no longer complains about encountering the top of the
2410 stack when you type "FRAME 0" to explicitly instruct it to go to
2411 the top of the stack. And it now prints the frame you request even
2412 if it's the current frame (instead of saying "You are here.").
2413 * As specified by ANSI, the system now always prints keywords
2414 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2416 * The default initial SIZE of HASH-TABLEs is now smaller.
2417 * Type information from CLOS class dispatch is now propagated
2418 into DEFMETHOD bodies, so that e.g.
2419 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2421 is now basically equivalent to
2422 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2423 (DECLARE (TYPE SINGLE-FLOAT X))
2425 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2426 operation, without having to do run-time type dispatch.
2427 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2428 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2429 * Since I don't seem to be making big file reorganizations very often
2430 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2431 is overflowing my ability to conveniently back them up), I've finally
2432 checked the system into CVS. (The CVS repository is on my home system,
2433 not at SourceForge -- putting it on SourceForge might come later.)
2434 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2435 high-level GC-NOTIFY-FOO functions send their output. (There's
2436 still very little control of where low-level verbose GC functions
2437 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2438 less than it used to -- the GC-NOTIFY-FOO functions are now under
2439 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2440 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2441 in only one place in the source code, and propagates it automatically
2442 everywhere that it's needed. Thus e.g. when I bump the version from
2443 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2444 * The C source files now include boilerplate legalese and documentation
2445 at the head of each file (just as the Lisp source files already did).
2446 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2447 to his page will be replaced with a link to his new CLiki service.
2449 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2451 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2452 * The build process now looks for GNU make under the default name "gmake",
2453 instead of "make" as it used to. If GNU make is not available as "gmake"
2454 on your system, you can change this default behavior by setting the
2455 GNUMAKE environment variable.
2456 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2457 can build without error under CMU CL.
2459 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2461 * There is now a partial SBCL user manual (with some new text and some
2462 text cribbed from the CMU CL manual).
2463 * The beginnings of a profiler have been added (starting with the
2464 CMU CL profiler and simplifying and cleaning up). Eventually the
2465 main interface should be through the TRACE macro, but for now,
2466 it's still accessed through vaguely CMU-CL-style functions and macros
2467 exported from the package SB-PROFILE.
2468 * Some problems left over from porting CMU CL to the new
2469 cross-compilation bootstrap process have been cleaned up:
2470 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2471 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2472 part when getting rid of the compiler *BACKEND* stuff.)
2473 ** The value of *NULL-TYPE* was screwed up, because it was
2474 being initialized before the type system knew the final
2475 definition of the 'NULL type. This screwed up several key
2476 optimizations in the compiler, causing inefficiency in all sorts
2477 of places. (I found it because I wanted to understand why
2478 GET-INTERNAL-RUN-TIME was consing.)
2479 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2480 PROCLAIM FTYPE information. Unfortunately this broke binary
2481 compatibility again, since now the forms output by DEFGENERIC
2482 to refer to functions which didn't exist in 0.6.3.
2483 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2484 can use the new (as of 0.6.3) transform for SOME into MAP into
2486 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2487 values to (INTEGER 0 1000000), so that the time code will no longer
2488 occasionally get blown up by Linux returning 1000000 microseconds
2489 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2490 its output conform to the ANSI spec. (Alas, this makes its output
2491 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2492 * A full call to MAP NIL with a single sequence argument no longer conses.
2493 * fixes to problems pointed out by Martin Atzmueller:
2494 * The manual page no longer talks about multiprocessing as though
2495 it were currently supported.
2496 * The ILISP support patches have been removed from the distribution,
2497 because as of version 5.10.1, ILISP now supports SBCL without us
2498 having to maintain patches.
2499 * added a modified version of Raymond Toy's recent CMU CL patch for
2500 EQUALP comparison of HASH-TABLE
2502 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2504 * The system still can't cross-compile itself with
2505 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2506 entails), but at least it can compile more of itself that way
2507 than it used to be able to, and various buglets which were uncovered
2508 by trying to cross-compile itself that way have now been fixed.
2509 * This release breaks binary compatibility again. This time
2510 at least I've incremented the FASL file format version to 2, so that the
2511 problem can be detected reliably instead of just causing weird errors.
2512 * various new style warnings:
2513 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2514 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2515 ** using the lexical binding of a variable named in the *FOO* style
2516 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2517 as specified by ANSI.
2518 * *RANDOM-STATE* is no longer automatically initialized from
2519 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2520 default behavior of the system is to repeat its behavior every time
2521 it's run. If you'd like to change this behavior, you can always
2522 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2523 old convention there was no comparably easy way to get the system to
2524 repeat its behavior every time it was run.
2525 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2526 FTYPE declarations has been removed, in favor of their ANSI
2527 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2528 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2529 the types of their sequence arguments can be determined at compile time.
2530 This is done through a new open code expansion for MAP which eliminates
2531 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2532 argument types can be determined at compile time.
2533 * The optimizer now transforms COERCE into an identity operation when it
2534 can prove that the coerced object is already of the correct type. (This
2535 can be a win for machine generated code, including the output of other
2536 optimization transforms, such as the MAP transform above.)
2537 * Credit information has been moved from source file headers into CREDITS.
2538 * Source file headers have been made more standard.
2539 * The CASE macro now compiles without complaining even when it has
2542 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2544 * (Note that the way that the PCL macroexpansions were rewritten
2545 to accommodate the change in DEFGENERIC below breaks binary
2546 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2547 not run under sbcl-0.6.2. Once we get out of alpha releases,
2548 i.e. hit release 1.0.0, we'll probably try to maintain binary
2549 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2550 and sbcl-1.4.4. Until then, however, it might be fairly common
2551 for maintenance releases to break binary compatibility.)
2552 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2554 * The reporting of unhandled signals has been changed to print some
2555 explanatory text as well as the report form. (Previously only
2556 the report form was printed.)
2557 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2558 it defines, so that the compiler no longer issues undefined function
2559 warnings for compiled-but-not-yet-loaded generic functions.
2560 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2561 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2562 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2563 section of the ANSI spec are supported.
2564 * The "" nickname for the "KEYWORD" package has been removed.
2565 The reader still handles symbol tokens which begin with a package marker
2566 as keywords, but it doesn't expose its mechanism for doing so in the
2567 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2568 * The system now issues STYLE-WARNINGs for contradictory TYPE
2569 proclamations. (Warnings for contradictory FTYPE proclamations would
2570 be nice too, but those can't be done usefully unless the type system
2571 is made smarter about FUNCTION types.)
2572 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2573 systematized, so that "*target-*.lisp is supposed to exist only on the
2574 target and imply that there's a related file which exists on the
2575 host, and *host-*.lisp is supposed to exist only on the host and imply
2576 that there's a related file which exists on the target. This involves a
2577 lot of renaming. Hopefully the acute confusion caused by the renaming
2578 will be justified by the reduction in chronic confusion..
2579 ** runtime-type.lisp -> early-target-type.lisp
2580 ** target-type.lisp -> late-target-type.lisp
2581 ** early-host-format.lisp -> early-format.lisp
2582 ** late-host-format.lisp -> late-format.lisp
2583 ** host-error.lisp -> misc-error.lisp
2584 ** early-error.lisp -> early-target-error.lisp
2585 ** late-error.lisp -> late-target-error.lisp
2586 ** host-defboot.lisp -> early-defboot.lisp
2587 ** code/misc.lisp -> code/target-misc.lisp
2588 ** code/host-misc.lisp -> code/misc.lisp
2589 ** code/numbers.lisp -> code/target-numbers.lisp
2590 ** code/early-numbers.lisp -> numbers.lisp
2591 ** early-host-type.lisp -> early-type.lisp
2592 ** late-host-type.lisp -> late-type.lisp
2593 ** host-typep.lisp -> typep.lisp
2594 ** load.lisp -> target-load.lisp
2595 ** host-load.lisp -> load.lisp
2596 ** host-disassem.lisp -> disassem.lisp
2597 ** host-insts.lisp -> insts.lisp
2598 ** byte-comp.lisp -> target-byte-comp.lisp
2599 ** host-byte-comp.lisp -> byte-comp.lisp
2600 ** host-signal.lisp -> signal.lisp
2601 ** host-defstruct.lisp -> defstruct.lisp
2602 ** late-target-type.lisp -> deftypes-for-target.lisp
2603 Furthermore, several other previously target-only files foo.lisp (e.g.
2604 hash-table.lisp and random.lisp) have been split into a target-and-host
2605 foo.lisp file and a target-only target-foo.lisp file, with their key type
2606 definitions in the target-and-host part, so that the cross-compiler will
2607 know more about target types.
2608 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2609 gone away. In their place are various *BACKEND-FOO* variables
2610 corresponding to the slots of the old structure.
2611 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2612 into the target SBCL has been fixed.
2613 * The chill.lisp system for loading cold code into a running SBCL
2615 * Support for the CMU CL "scavenger hook" extension has been removed.
2616 (It was undocumented and unused in the CMU CL sources that SBCL was
2617 derived from, and stale in sbcl-0.6.1.)
2618 * Various errors in the cross-compiler type system were detected
2619 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2620 (enabling various consistency checks). Many of them were fixed,
2621 but some hard problems remain, so the compiler is back to
2622 running without *TYPE-SYSTEM-INITIALIZED* for now.
2623 * As part of the cross-compiler type system cleanup, I implemented
2624 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2625 * I have started adding UNCROSS calls throughout the type system
2626 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2627 on cross-compiler input files will be able to go away, and various
2629 * CONSTANTP now returns true for quoted forms (as explicitly required
2632 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2634 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2635 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2636 (MAKE-LIST -1). In the longer term, it ought to become true
2637 that declarations are assertions even at SAFETY 1. For now, it's not
2638 quite true even at SAFETY 3, but it's at least more nearly true..
2639 (Note that this change seems to increases the size of the system by
2640 O(5%) and to decrease the speed of the compiler by 20% or more.)
2641 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2642 for the problem of printing dozens of lines of distracting information
2643 about low-level system machinery as part of the top stack frame
2644 on entry to the debugger when an undefined function was called.
2645 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2646 is not reset to COMMON-LISP-USER.
2647 * Compilation of stuff related to dyncount.lisp has been made conditional
2648 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2649 smaller. The various dyncount-related symbols have been moved into
2650 a new "SB-DYNCOUNT" package.
2651 * tty-inspect.lisp has been renamed to inspect.lisp.
2652 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2653 feature has gone away. (When we eventually port to other flavors of
2654 libc and/or Unix, we'll try to make the differences between flavors
2655 invisible at the user level.)
2656 * Various other *FEATURES* tags, and/or their associated conditionals,
2657 have been removed if obsolescent, or given better documentation, or
2658 sometimes given more-mnemonic names.
2660 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2662 * tidied up "make.sh" script
2663 * tidied up system directory structure
2664 * better "clean.sh" behavior
2665 * added doc/FOR-CMUCL-DEVELOPERS
2666 * many many small tweaks to output format, e.g. removing possibly-confusing
2667 trailing #\. character in DESCRIBE-INSTANCE
2668 * (EQUALP #\A 'A) no longer signals an error.
2669 * new hashing code, including EQUALP hashing
2670 * tidied up Lisp initialization and toplevel
2671 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2672 * command line argument processing
2673 * added POSIX-GETENV function to deal with Unix-ish environment variables
2674 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2675 terminating SBCL on EOF
2676 * non-verbose GC by default
2677 * There is no more "sbcl" shell script; the sbcl file is now the C
2678 runtime executable (just like CMU CL).
2679 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2681 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2682 symbol and package information as Lisp native symbol and package objects
2683 instead of strings naming symbols and strings naming packages. This way,
2684 whenever packages are renamed (as in warm init), debug information is
2685 transformed along with everything else.
2686 * tweaked the optimization policy declarations which control the building
2687 of SBCL itself. Now, among other things, the system no longer saves
2688 source location debugging information. (This helps two problems at once
2689 by reducing SBCL size and by keeping SBCL from trying to look for its
2690 sources -- which may not exist -- when reporting errors.)
2691 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2692 debugging and testing purposes
2693 * cleaned up printing, making the printer call PRINT-OBJECT for
2694 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2695 methods, giving nearly-ANSI behavior
2696 * converted almost all special variables to use *FOO* naming convention
2697 * deleted PARSE-TIME functionality, since it can be done portably
2698 * moved some files out of cold init into warm init
2699 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2701 * regularized formatting of source files
2702 * added an install.sh script
2703 * fixed ridiculous memory usage of cross-compiler by making
2704 compiler/alloc.lisp not try to do pooling unless it can hook
2705 itself into the GC of the cross-compilation host. Now the system
2706 builds nicely on my old laptop.
2707 * added :SB-ALLOC in target-features.lisp-expr
2708 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2709 was not implemented)
2710 * re-did condition handling and note reporting in the compiler. Notes
2711 are no longer handled by signalling conditions. Style warnings
2712 and warnings are handled more correctly and reported in such a way
2713 that it's easy to find one or the other in your output (so that you
2714 can e.g. figure out which of many problems caused COMPILE-FILE to
2716 * changed the severity of several compiler warnings from full WARNING
2717 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2718 compiler note reporting so that it doesn't use the condition system
2719 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2720 COMPILE-FILE command)
2721 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2722 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2723 no longer supported, since their implementation was deeply intertwingled
2724 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2725 * removed bogus "support" for compiler macros named (SETF FOO), and
2726 removed the compiler macro for SETF INFO (but only after making a fool
2727 of myself on the cmucl-imp mailing list by posting a bogus patch for
2728 DEFINE-COMPILER-MACRO..)
2729 * Compiled files containing forms which have side effects on the Lisp
2730 reader (such as DEFPACKAGE forms) are now handled more correctly.
2731 (Compiler queuing of top level lambdas has been suppressed by setting
2732 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2733 * deleted various currently-unused source files, e.g. gengc.lisp. They
2734 may be added back at some point e.g. when porting to other architectures,
2735 but until they are it's distracting to distribute them and to try to
2737 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2738 there were so many of them they're just distractions, and UNCROSS is
2739 known to be able to handle the current sources
2740 * moved PROFILE functionality into TRACE, so that it will be clear
2741 how the wrapping and unwrapping of functions when you profile them
2742 interacts with the wrapping and unwrapping of functions when you
2743 trace them. (Actually, the functionality isn't there yet, but at least
2744 the interface specification is there. Hopefully, the functionality will
2745 arrive with some future maintenance release.)
2746 * removed host-oops.lisp
2747 * changed signature of QUIT function to allow UNIX-CODE argument
2748 * fixed READ-SEQUENCE bug
2749 * tweaked verbose GC output so that it looks more like the progress
2750 output that ANSI specifies for functions like LOAD
2751 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2752 * added <http://sbcl.sourceforge.com> to the banner information printed by