1 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2 * fixed some bugs revealed by Paul Dietz' test suite:
3 ** POSITION on displaced vectors with non-zero displacement
4 returns the right answer.
6 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
7 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
8 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
9 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
10 the supported interface.
11 * new feature: Single-stepping of code compiled with DEBUG 2 or
12 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
13 * new feature: saving cores with foreign code loaded is now
14 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
15 Timothy Moore's work for CMUCL)
16 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
17 parameters to * instead of NIL if no initform is supplied.
18 (reported by Johan Bockgaard)
19 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
20 correctly when FUNCALL appears as the car of the form. Note:
21 despite this FUNCALL forms are not currently subject to
22 compiler-macro expansion. (port of Raymond Toy's fix for the
23 same from CMUCL, reported by Johan Bockgaard)
24 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
25 (thanks for Teemu Kalvas)
26 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
27 case instead of terminating the iteration (thanks for Julian
28 Squires, Sean Champ and Raymond Toy)
29 * bug fix: incorrect expansion of defgeneric that caused a style
30 warning. (thanks for Zach Beane)
31 * bug fix: slot accessor effective method computation works properly
32 for classes with multiple non-standard applicable methods on
33 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
34 * on x86 compiler supports stack allocation of results of LIST and
35 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
36 CMUCL implementation by Gerd Moellmann)
37 * fixed some bugs revealed by Paul Dietz' test suite:
38 ** FORMAT strings with both the ~<~:;~> form of the justification
39 directive and pretty-printing directives cause an error.
41 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
42 * incompatible change: the internal functions
43 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
44 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
45 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
46 instead of the old functions.
47 * new feature: on platforms where "dladdr" is available foreign
48 function names now appear in backtraces. (based on Helmut Eller's
50 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
51 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
53 * documentation: ASDF manual is now distributed with SBCL.
54 * bug fix: SBCL can now load its contributed modules with REQUIRE
55 even if the system-provided entries have been removed from
56 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
58 * bug fix: backtraces involving undefined functions or assembly
59 routines are more informative. (thanks to Brian Downing)
60 * bug fix: mutually referent alien structures now work correctly.
61 (reported by Rick Taube)
62 * bug fix: structures defined by WITH-ALIEN can be referred to
63 within other definitions in the same WITH-ALIEN.
64 * bug fix: division operators (MOD, TRUNCATE and the like) with
65 constant zero divisors and integer dividends no longer generate
67 * bug fix: provide default methods for INPUT-STREAM-P and
68 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
69 * bug fix: improve the detection and resolution of MOP metacycles
70 (where effective method computation depends on the generic function
71 having its effective method computed). (reported by Bruno Haible)
72 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
73 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
74 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
75 * bug fix: pathnames with relative directory components can be
76 represented relative to default pathnames.
77 * optimization: in taking the GCD of bignums, reduce the two bignums
78 to approximately the same size (using Euclid's algorithm) before
79 applying the more sophisticated binary GCD. (thanks to Juho
81 * optimization: COUNT on bitvectors now operates word-at-a-time.
82 * optimization: ASH with a positive, but not necessarily constant,
83 (leftwards) shift, when in a modular context, is compiled to a
85 * fixed some bugs revealed by Paul Dietz' test suite:
86 ** FORMAT variable parameters ("~V<char>") are defaulted properly
87 if the corresponding argument is NIL.
88 ** FORMAT directives accepting a minpad parameter treat negative
90 ** literal commas as character parameters to FORMAT directives are
92 ** literal spaces directly after ~<Newline> directives within a
93 format-logical-block (~:< ~@:>) do not induce :FILL-style
95 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
96 as specified: it no longer includes conditional newlines).
97 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
99 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
100 :PER-LINE-PREFIX argument does not evaluate to a string.
102 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
103 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
104 the manual for details; for now, package locks can be disabled by
105 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
106 if no major problems are found then it is likely that they will be
107 compiled in unconditionally.
108 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
109 unsupported operators on all platforms. To load a shared library
110 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
111 object file, link it into a shared library outside of SBCL and
112 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
114 * bug workaround: ROOM T and the GC were not working together
115 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
116 reporting facility aren't preserved. That reporting has been
117 disabled, so now until and unless someone figures out how to make it
118 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
119 * minor incompatible change: as threatened around sbcl-0.8.0, the
120 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
121 The symbols are also exported from SB-PCL for backwards
122 compatibility, but more so than before SB-PCL should be treated as
123 an implementation-internal package.
124 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
126 * the SB-SPROF contrib now works on (most) non-x86 architectures.
127 It is known as of this release not to work on the Alpha, however.
128 * fixed bug #167: errors signalled due to illegal syntax in method
129 bodies are now more legible.
130 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
131 designators and can hence be used with TYPEP.
132 * fixed bug #333: CHECK-TYPE now ensures that the type error
133 signalled, if any, has the right object to be accessed by
134 TYPE-ERROR-DATUM. (reported by Tony Martinez)
135 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
136 5.1.2.3 for multiple-value place subforms. (reported by Kalle
138 * fixed bug #334: programmatic addition of slots using specialized
139 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
140 and :ALLOCATION :CLASS slots.
141 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
142 float exponents. (rereported by Peter Seibel)
143 * fixed a bug: #\Space (and other whitespace characters) are no
144 longer considered to be macro characters in standard syntax by
146 * fixed bug: initialization of condition class metaobjects no longer
147 causes an instance of the condition to be created. (reported by
149 * fixed bug: it is now possible to have slots such that
150 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
151 description is neither :INSTANCE nor :CLASS.
152 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
153 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
154 SLOT-VALUE-USING-CLASS), and no longer causes errors with
155 non-standard SLOT-DEFINITION-ALLOCATION values.
156 * fixed bugs: various race conditions handling exiting threads.
157 CL-PPCRE's thread torture test now passes.
158 * fixed arguably-a-bug: GC time proportional to number of threads in
159 system even when most of them are idle
160 * optimization: improved performance of BIT and SBIT on bit-vectors.
161 * .fasl file incompatibility: The fasl file version number has
162 been incremented because of changes associated with package locks.
164 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
165 * minor incompatible change: the system no longer provides
166 optimization hints (conditions which are TYPEP
167 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
168 mode; these hints are still emitted when higher SPEED optimization
170 * new contrib module: a sampling profiler (profiling by statistical
171 sampling, rather than by instrumenting functions) is available as
172 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
173 * the behaviour of the standard function ED is now customizeable by
174 third parties through a hook variable: see ED's documentation
175 string for information on the protocol.
176 * the compiler no longer emits efficiency notes for (FUNCALL X)
177 when the type of X is uncertain under default optimization
179 * fixed bug 276: mutating a binding of a specialized parameter to a
180 method to something that is not TYPEP the specializer is now
182 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
183 exhibit the required behaviour on the x86 platform. (thanks to
184 Peter van Eynde, Eric Marsden and Bruno Haible)
185 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
186 even for difficult arguments. (reported by Peter Graves)
187 * fixed bug 141a: the backquote printer now descends quoted
189 * fixed another bug in backquote printing: no more destructive
190 modification of the form's list structure. (reported by Brian
192 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
193 move between its address being taken and the call to
194 interrupt_thread, fixing a crashing race condition.
195 * the SB-POSIX contrib implementation has been adjusted so that it
196 no longer exhibits ridiculously poor performance when constructing
197 instances corresponding to C structs.
199 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
200 * minor incompatible change: the sb-grovel contrib now treats C
201 structures as alien (in the sense of SB-ALIEN) objects rather than
202 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
203 has implications for memory management of client code
204 (sb-grovel-returned objects must now be manually managed) and for
205 type safety (alien objects now have full types).
206 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
207 used to control emission of compiler diagnostics, rather than the
208 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
209 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
210 quality should be considered deprecated.
211 * install.sh now installs the user manual as well
212 * (not quite a new documentable feature, but worth considering in
213 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
214 beginnings of a semantically meaningful condition hierarchy is
215 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
217 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
218 designator as the defaults argument.
219 * fixed bug: Displaced arrays whose displaced-to array has become
220 too small now cause ARRAY-DIMENSION to signal an error, providing
221 for safer bounds-checking. (reported by Bruno Haible)
222 * fixed bug: DEFCLASS slot definitions with identical :READER and
223 :WRITER names now signal a reasonable error. (reported by Thomas
225 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
227 * fixed bug: Invalid slot specification errors now print correctly.
228 (thanks to Zach Beane)
229 * fixed bug 320: Shared to local slot value transfers in class
230 redefinitions now happen corrently from superclasses as
231 well. (reported by Bruno Haible)
232 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
234 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
235 improper lists correctly. (reported by Bruno Haible)
236 * fixed bug 313: source-transform for <fun-name> was erroneously
237 applied to a call of a value of a variable with name <fun-name>.
238 (reported by Antonio Menezes Leitao)
239 * fixed bug 307: The obsolete instance protocol ensures that
240 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
241 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
242 remove dead unknown-values globs from the middle of the stack.
243 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
244 easier to resume long computations after using *BREAK-ON-SIGNALS*
245 to diagnose and fix failures. (thanks to Nikodemus Siivola)
246 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
247 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
248 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
249 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
250 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
252 * fixed a (fairly theoretical) bug in string printing: if
253 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
254 does not have array-element-type equal to the most general string
256 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
257 function name. (thanks to Nikodemus Siivola)
258 * fixed bug: compile-time format string checker failed on
259 non-closed ~{. (reported by Thomas F Burdick)
260 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
261 don't warn on legal loop code involving more than one aggregate
262 boolean. (thanks to Nikodemus Siivola)
263 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
264 in keyword position to MAKE-PACKAGE was not being reported as
266 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
267 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
269 * fixed bugs 280 and 312: the checking for multiple definitions in a
270 file is less likely to become confused by uses of inline
272 * fixed bug: the #S reader macro performs the keyword coercion
273 specified for slot names. (reported by Kalle Niemitalo)
274 * fixed bug: lambda lists may contain symbols whose names start with
275 & but are not lambda-list-keywords; their occurrence triggers a
277 * fixed bug 321: define-method-combination argument lambda lists do
278 not cause invalid code to be generated when &optional or &aux
279 variables have default values. (reported by Bruno Haible)
280 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
281 have CLOS classes; this appears also to have fixed reported
282 instability in stack exhaustion detection.
283 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
284 MUFFLE-WARNING when no associated restart is present is now
286 * optimization: rearranged the expansion of various defining macros
287 so that each expands into only one top-level form in a
288 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
290 * optimization: used a previously little-used slot in symbols to
291 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
293 * fixed some bugs revealed by Paul Dietz' test suite:
294 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
296 ** Fixed an optimization bug related to inheritance of initforms
297 from local to shared slots.
298 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
299 ** CAST optimizer forgot to flush argument derived type.
300 ** print/read consistency on floats is now orders of magnitude
301 more likely. (thanks also to Bruno Haible for a similar report
303 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
304 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
305 for objects of type REAL. Make it so.
306 ** SXHASH is specified (infelicitously) to respect similarity,
307 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
308 it so. (thanks to Markus Ziegler)
309 ** on the Alpha, the compiler succeeds in compiling functions
310 returning a known number of arguments greater than 63.
311 ** fixed handling of invalid NIL arguments in keyword position
313 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
315 ** provided a readably-printable representation for RANDOM-STATE
317 ** ensured that pathnames, where they have a namestring, always
318 print using #P"..." syntax.
320 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
321 * Support for the forthcoming 2.0 version of the NetBSD kernel
322 running on x86 hardware has been added. (thanks to Perry
323 E. Metzger most immediately, and others for their past work)
324 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
325 current release version; SBCL's OpenBSD support had been broken
326 since about the time of OpenBSD's switch to ELF binary format.)
327 * [placeholder for DX summary]
328 ** user code with &REST lists declared dynamic-extent, under high
329 speed or space and low safety and debug optimization policy.
330 * The manual has been converted to Texinfo format and the debugger
331 chapter from the cmucl manual has been added.
332 * A facility has been added to extract documentation strings from
333 sbcl and store them as Texinfo-formatted snippets for inclusion in
334 the manual (via Texinfo's @include directive)
335 * bug fix: compiler emitted division in optimized DEREF. (thanks for
336 the test case to Dave Roberts)
337 * bug fix: multidimensional simple arrays loaded from FASLs had fill
338 pointers. (reported by Sean Ross)
339 * bug fix: PROFILE output is printed nicely even for large numerical
340 values. (thanks to Zach Beane)
341 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
342 greater than 32 handle EOF correctly.
343 * bug fix: on X86 an immediate argument of the IMUL instruction is
344 correctly printed in disassembly. (thanks to Lutz Euler)
345 * bug fix: class slots in redefined classes preserve their old
346 values. (thanks to Bruno Haible and Nikodemus Siivola)
347 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
348 succeeds. (reported by Marco Baringer)
349 * fixed some bugs revealed by Paul Dietz' test suite:
350 ** READ-SEQUENCE now works on ECHO-STREAMs.
351 ** RATIONALIZE works more according to its specification. (thanks
354 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
355 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
356 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
357 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
358 should matter to you only if you rebind the printer control
359 variables and then find you want different bindings in the
360 debugger than in the ordinary execution of your program.)
361 * The runtime build system has been tweaked to support building
362 (on SPARC/SunOS) using a C compiler which invokes Sun's own
363 assembler and linker. (thanks to Nikodemus Siivola)
364 * Unbound, undefined, undeclared variables now trigger full
365 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
366 more often programmer error than deliberate exploitation of undefined
368 * optimization: the hash algorithm for strings has changed to one
369 that is less vulnerable to spurious collisions. (thanks to Juho
371 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
372 needless bounds checking. (thanks to Juho Snellman)
373 * optimization: implemented multiplication as a modular
374 (UNSIGNED-BYTE 32) operation on the PPC backend.
375 * fixed some bugs revealed by Paul Dietz' test suite:
376 ** ADJUST-ARRAY now copies the datum in a zero rank array if
378 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
379 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
380 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
382 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
384 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
386 ** ECHO-STREAMs no longer attempt to echo the end of file value to
387 their output stream on EOF from read.
388 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
389 have been read to end-of-file.
390 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
392 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
393 description of determination of which consecutive characters
395 ** Printing the "Space" character with escaping on now yields "#\\ ",
396 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
397 ** Reading floating-point numbers with *READ-BASE* set to a number
398 less than 10 works correctly.
399 ** Reading floating-point numbers with *READ-BASE* set to a number
400 more than 10 works correctly.
401 ** Printing with *PRINT-READABLY* targets the standard readtable, not
402 the readtable currently in effect.
404 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
405 * minor incompatible change: parsing of namestrings on a physical
406 (Unix) host has changed; numbers after the final #\. in a
407 namestring are no longer interpreted as a version field. This is
408 intented to be largely invisible to the user, except that the
409 meaning of the namestring "*.*.*" has changed: it now refers to a
410 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
411 should usually be replaced by
412 (make-pathname :name :wild :type :wild :version :wild)
413 with the added benefit that this is more likely to be portable.
414 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
415 option to OPEN now signals an error if the file being opened
416 exists; this may have an impact on existing code.
417 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
418 signals received, on the PowerPC platforms (both Linux and
419 Darwin). (thanks to Pierre Mai for pointing out the location of
421 * several fixes on OS X: The system now builds and runs cleanly on
422 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
423 (thanks to Brian Mastenbrook)
424 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
425 second-resolution: integer multiples of 1/3600 between -24 and 24.
426 (thanks to Vincent Arkesteijn)
427 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
428 type when called with 1 argument; PEEK-CHAR checked type of
429 PEEK-TYPE only after having read first character from a
430 stream. (reported by Peter Graves)
431 * bug fix: the garbage collector now has much better locality
432 behaviour, and in particular no longer treats all memory as being
433 exhausted when many small objects point to each other in a deeply
435 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
436 recognized as being TYPEP their class.
437 * bug fix: the PUSHNEW documentation string has been corrected.
438 (thanks to Vincent Arkesteijn)
439 * bug fix: defaulting of the value for the last of an atypically
440 large number of multiple values being bound was not being
441 performed correctly on the Alpha or PPC platforms
442 * optimization: implemented multiplication as a modular
443 (UNSIGNED-BYTE 32) operation on the x86 backend.
444 * optimization: SEARCH on simple-base-strings can now be open-coded.
445 (see also contrib/compiler-extras.lisp for inspiration for
446 teaching the compiler about the Boyer-Moore algorithm).
447 * value, returned by MAX (and MIN) called with several EQUALP, but
448 not EQL, arguments now does not depend on compiler settings.
449 * fixed some bugs revealed by Paul Dietz' test suite:
450 ** in stack analysis liveness information is propagated from
451 non-local entry points.
452 ** pathwise CAST removing failed when the CAST node did not start
454 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
455 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
457 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
458 host is already defined.
459 ** RENAME-FILE works on streams instead of signalling an internal
461 ** PEEK-CHAR uses the current readtable when determining whether
462 or not a character is whitespace.
463 ** MERGE-PATHNAMES handles the case when the pathname does not
464 specify a name while the default-pathname specifies a version.
465 ** Pathnames now stand a better chance of respecting print/read
467 ** Attempting to use standardized file system operators with a
468 pathname with invalid :DIRECTORY components signals a
470 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
472 ** DIRECTORY on logical pathnames is more correct.
473 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
474 signal a TYPE-ERROR if their argument is not a stream.
475 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
476 designator argument does not designate a stream.
477 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
478 examining the synonym.
479 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
481 ** OPEN and WITH-OPEN-STREAM allow opening streams with
482 element-type larger than ([UN]SIGNED-BYTE 32).
484 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
485 * When built with the :SB-FUTEX feature, threaded builds now take
486 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
487 for faster/more reliable mutex and condition variable support.
488 * Incompatible change (but one you probably shouldn't have been using
489 anyway): the interface and code for arbitrating between multiple
490 threads in the same user session has been redesigned.
491 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
492 error when called without an explicit environment argument.
493 (thanks to Nikodemus Siivola)
494 * bug fix: buffered :DIRECTION :IO streams are less likely to become
495 confused about their position. (thanks to Adam Warner and Gerd
497 * bug fix: Pretty printing backquoted forms with unquotations in the
498 argument list position of various code constructs such as LAMBDA
499 now works correctly. (reported by Paul Dietz)
500 * bug fix: Pretty printing unquotations no longer loses all
501 stream position information.
502 * optimization: performance of string output streams is now less
503 poor for multiple small sequence writes.
504 * optimization: performance of CSUBTYPEP in the presence of complex
505 expressions involving CONS and NOT many times has been improved.
506 (reported by Paul Dietz)
507 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
509 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
510 streams for the REPL, for the debugger, and so on. (thanks to
512 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
513 * fixed some bugs revealed by Paul Dietz' test suite:
514 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
515 is ignored if it is too small, rather than propagating through
516 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
517 ** extremely complex negations of CONS types were not being
518 sufficiently canonicalized, leading to inconsistencies in
520 ** VALUES tranformer lost derived type.
522 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
523 * fixed a bootstrapping bug: the build process no longer assumes
524 that the various BOOLE-related constants have the same value in
525 host and target lisps. (noted by Paul Dietz' test suite on an
526 SBCL binary built from CLISP)
527 * The system can now be dynamically linked on the MIPS platform,
528 which enables dynamic loading of foreign code from Lisp. (thanks
529 to Ralf Baechle for discussions on the MIPS ABI)
530 * The system now records debugging information for its own source
531 files in a filesystem-position-independent manner, relative to
532 the "SYS" logical host.
533 * fixed a compiler bug: MV-LET convertion did not check references
534 to the "max args" entry point. (reported by Brian Downing)
535 * tweaked disassembly notes to be less confident about proclaiming
536 some instruction as an LRA. (thanks to Brian Downing)
537 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
538 now each have their own history, command character, and other
539 characteristics. (thanks to David Lichteblau)
540 * fixed some bugs revealed by Paul Dietz' test suite:
541 ** compiler failure in compiling LOGAND expressions including a
543 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
544 shift greater than 32.
545 ** FLUSH-DEST did not mark blocks for type check regeneration.
546 ** HANDLER-CASE failed to accept declarations in handler clauses
547 in some circumstances.
549 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
550 * New code in contrib/sb-introspect (still probably not entirely
551 stable yet) provides some support for smart Lisp development
552 environments like SLIME.
553 * The conditions signalled for errors occurring when loading .fasl
554 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
555 in a way which should help ASDF recover gracefully.
556 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
557 stuff has been cleaned up. If you code contrib/ stuff, this might
558 affect you, and you can look at contrib/README, contrib/STANDARDS,
559 and/or the 0.8.4.27 diff to check.
560 * In full calls the compiler now does not generate checks for declared
561 argument types for all arguments.
562 * various threading fixes
563 ** and some experimental patches which didn't make it into
564 the main tree for this release, but which are shipped in
565 contrib/experimental-thread.patch as a possible fix for some
566 failures (deadlock, spinning...) in GC-intensive multithreaded
568 * fixed PPC build problem (source code incompatibility of different
569 library versions): added offsetof() hackery which attempts to divine
570 where glibc maintainers put uc_mcontext today
571 * fixed bug 282: compiler does not trust type assertions while passing
572 arguments to a full call.
573 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
574 &OPTIONAL VALUES type parameter.
575 * fix bug 214: algorithm for noting rejected templates is now more
576 similar to that of template seletion. (also reported by rydis on
578 * fixed bug 141b: printing backquoted information readably and prettily
579 inserts a space where necessary.
580 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
581 (CHARACTER 10) are now reported as errors, rather than propagated
582 as unknown types. (reported by piso on #lisp)
583 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
584 with values NIL and :ERROR. (thanks to Milan Zamazal)
585 * fixed bug 191c: CLOS now does proper keyword argument checking as
586 described in CLHS 7.6.5 and 7.6.5.1.
587 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
588 longer raise an error; further, using a list as a for-as-arithmetic
589 counter now raises a meaningful error.
590 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
591 now understood by sequence creation functions such as MAKE-SEQUENCE
593 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
594 a STRING-INPUT-STREAM.
595 * compiler enhancement: SIGNUM is now better able to derive the type
597 * type declarations inside WITH-SLOTS are checked. (reported by
599 * fixed some bugs revealed by Paul Dietz' test suite:
600 ** incorrect optimization of TRUNCATE for a positive first
601 argument and negative second.
602 ** compiler failure in let-convertion during flushing dead code.
603 ** compiler failure while deriving type of TRUNCATE on an
604 interval, containing 0.
605 ** ASH of a negative bignum by a negative bignum count now returns
607 ** intersection of CONS types now canonicalizes properly, fixing
608 inconsistencies in SUBTYPEP.
610 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
611 * incompatible change: The --disable-debugger command line
612 option now clobbers the debugger at a more fundamental
613 level, by redefining #'INVOKE-DEBUGGER instead of by
614 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
615 is specified by ANSI to ignore *DEBUGGER-HOOK* and
616 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
617 enter the debugger REPL and then suffer recursive errors
618 because *DEBUG-IO* is also messed up in --disable-debugger mode;
619 while under the new system, BREAK in --disable-debugger mode
620 terminates the system just as an unhandled error would.
621 * fixed compiler performance when processing loops with a step >1;
622 * bug fix: DOCUMENTATION now retrieves generic function
623 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
624 support has been systematized, and now supports the methods
625 specified by ANSI, along with a default method and a method for
626 slot documentation. (reported by Nathan Froyd)
627 * bug fix: effective methods associated with a generic function are
628 no longer cached over a change of that generic function's method
629 combination. (reported by Andreas Fuchs)
630 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
631 names a non-existent directory. (thanks to Andreas Fuchs)
632 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
633 the PPC platform, lead to stack corruption; this has been fixed.
634 (reported by Rainer Joswig)
635 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
636 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
637 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
638 arithmetic in general, is now much more reliable on the PPC
640 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
641 platform now returns the right answer.
642 * optimization: restored some effective method precomputation in
643 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
644 precomputation is now tunable.
645 * optimization: compiler-internal data structure use has been
646 reviewed, and changes have been made that should improve the
647 performance of the compiler by about 20%.
648 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
649 simple-base-strings and simple-bit-vectors is improved.
650 * optimization: the optimization of 32-bit logical and arithmetic
651 functions introduced in version 0.8.3 on the x86 has been
652 implemented on the mips, ppc and sparc platforms; an
653 implementation of the same facility, but for 64-bit arithmetic,
654 has been added for the alpha.
655 * microoptimization: the compiler is better able to make use of the
656 x86 LEA instruction for multiplication by constants.
657 * bug fix: in some situations compiler did not report usage of
658 generic arithmetic in (SPEED 3) policy.
659 * bug 145b fix: compiler used wrong type specifier while converting
660 MEMBER-types to numeric.
661 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
663 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
664 index argument now works properly on the Alpha platform.
665 * bug fix: floating point exception treatment on the Alpha platform
667 * bug fix: FILE-POSITION works much better on string input and
668 output streams. (thanks to Nikodemus Siivola)
669 * bug fix: many threading/garbage collection symptoms sorted.
670 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
671 might be pseudo-atomic.
672 * internal change: Stopping for GC is now done with signals not
673 ptrace. GC is now done in whichever thread wanted it, instead of
675 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
677 * bug fix: result form in DO is not contained in the implicit
679 * incompatible change: ICR structure is changed; the value part of
680 CONTINUATION is now called LVAR; corresponding functions are
681 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
682 * added type deriver for ISQRT (thanks to Robert E. Brown).
683 * bug fix: better support for loading from the command line when an
684 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
686 * fixed some bugs revealed by Paul Dietz' test suite:
687 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
688 ** ROUND and FROUND now give the right answer when given very
689 small float arguments.
690 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
692 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
693 ** (GCD 0 <negative-integer>) returned <negative-integer>.
694 ** LCM should return a non-negative integer.
695 ** PARSE-INTEGER returned the index of a terminator instead of the
696 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
697 ** PARSE-INTEGER returned an incorrect index being applied to a
699 ** LCM with two arguments of 0 returns 0 rather than signalling
701 ** unsigned addition of a 32-bit constant with the high bit set no
702 longer causes an internal compiler error on the x86.
703 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
704 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
705 with negative last argument.
706 ** byte specifiers with zero size and position no longer cause
707 an error during type derivation.
708 ** bignum multiplication on the Alpha platform now returns the
710 * porting: The system now builds on SuSE AMD64, although it still
711 generates a 32-bit binary.
712 * .fasl file incompatibility: The fasl file version number has
713 been incremented (because of the changes to internal compiler
714 data structures referred to above).
716 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
717 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
718 more accurately, on the Darwin kernel running on PowerPC hardware.
719 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
720 * Compiler code deletion notes now signal a condition of type
721 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
722 an associated MUFFLE-WARNING restart.
723 * The compiler now performs limited argument count validation of
724 constant format strings in FORMAT, and where appropriate in ERROR,
725 CERROR and WARN. (thanks to Gerd Moellmann)
726 * New ASDF-INSTALL contrib can be used for automatic download and
727 installation of third-party Lisp code from CCLAN or other sites
729 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
730 which forces another thread to execute a function supplied by the
732 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
733 now accept and act upon their :ELEMENT-TYPE keyword argument.
734 (reported by Martin Atzmueller, Edi Weitz)
735 * bug fix: FILE-POSITION now accepts position designators up to
736 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
737 is the greater. (thanks to Patrik Nordebo)
738 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
739 Kalle Olavi Niemitalo)
740 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
741 printable. (reported by Eric Marsden)
742 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
743 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
744 * x86 bug fix in control stack exhaustion checking: now shows backtrace
745 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
746 (thanks to Stig Sandoe)
747 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
748 UNREAD-CHAR now correctly includes the unread character in the
749 target sequence. (thanks to Gerd Moellmann)
750 * bug fix in threaded builds: the system can now be suspended and
751 resumed by shell job control with minimal disruption.
752 * bug fixes in times and timezones >2038AD
753 * better handling of "where is GNU make?" problem in build scripts
754 (thanks to Nikodemus Siivola)
755 * new optimization: inside a named function any reference to a
756 function with the same name is considered to be a self-reference;
757 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
758 optimization quality.
759 * new optimization on x86: logical functions and + now have
760 optimized (UNSIGNED-BYTE 32) versions, which are automatically
761 used when the result is truncated to 32 bits.
762 * VALUES declaration is partially enabled.
763 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
764 * fixed some bugs revealed by Paul Dietz' test suite:
765 ** The system now obeys the constraint imposed by
766 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
767 types form a lattice under type intersection.
768 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
769 ** ASSOC now ignores NIL elements in an alist.
770 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
771 and (1+ MOST-POSITIVE-FIXNUM) answers.
772 ** The addition of a method with invalid qualifiers to a generic
773 function does not cause an error to be signalled immediately;
774 a warning is signalled, and the error is generated only on
775 calling the generic function.
776 * changed .fasl file version number, in part to add type codes for
777 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
778 obscure ANSI requirements
780 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
781 * fixed bug 148: failure to inline-expand a local function left
782 garbage, confusing the compiler.
783 * fixed bugs 3cd: structure slot readers perform type check if the
784 slot can have an invalid value (i.e. it is either not initialized
785 or can be written with a less specific slot writer).
786 * bug fix: the compiler now traps array references to elements off
787 the end of an array; previously, the bounds checking in some
788 circumstances could go off-by-one.
789 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
791 * type declarations for array element types now obey the description
792 on the CLHS page "Declaration TYPE", as per discussions on
793 sbcl-help around 2003-05-08. This means that a declaration
794 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
795 declaration, all references to BAR will be asserted or assumed
796 (with THE, so dependent on compiler policy) to involve objects of
797 type FOO. Note that no such declaration is implied in
798 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
799 * declared types of functions from the "Conditions"
800 chapter. (reported by Paul Dietz)
801 * bug fix: CERROR accepts a function as its first argument.
802 * bug fix: NTH an NTHCDR accept a bignum as index
803 arguments. (reported by Adam Warner)
804 * optimization: character compare routines now optimize comparing
805 against a constant character. (reported by Gilbert Baumann)
806 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
807 argument now works properly on the MIPS platform.
808 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
809 * repeated evaluation of the same DEFSTRUCT, a slot of which is
810 declared to have a functional type, does not cause an error
812 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
813 many :DEFAULT-INITARGS. (reported by Istvan Marko)
814 * fixed bug: if last continuation of a deleted block has a
815 destination, this destination should be deleted too. (reported by
817 * fixed a bug in the bootstrap process: the host compiler's values
818 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
819 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
820 test case from Patrik Nordebo)
821 * improved the ability of the disassembler on the PPC platform to
822 provide helpful disassembly notes.
823 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
824 the class in more cases than previously.
825 * bug fix: FILE-POSITION now understands :START and :END for
826 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
827 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
828 8) data. (thanks to David Lichteblau)
829 * bug fix: it is possible to add a method to a generic function
831 * bug fix: reader failed to signal END-OF-FILE inside an
832 object representation. (reported by Nikodemus Siivola)
833 * fixed some bugs revealed by Paul Dietz' test suite:
834 ** LAST and [N]BUTLAST should accept a bignum.
835 ** condition slot accessors are methods.
836 ** (VECTOR NIL) is a subtype of STRING.
838 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
839 * minor incompatible change: some nonsensical specialized lambda
840 lists (used in DEFMETHOD) which were previously ignored now signal
842 * minor incompatible change: the system is now aware of the types of
843 variables in the COMMON-LISP package, and will signal errors for
844 most violations of these type constraints (where previously they
845 were silently accepted).
846 * minor incompatible change: COMPILE-FILE now uses the freedom
847 afforded (ANSI 3.2.2.3) to use derived function types for
848 functions defined in the same file. This also permits the system
849 to warn on static type mismatches and function
850 redefinition. (Currently it does not work with high DEBUG level.)
851 * minor incompatible change: VALUES declaration is disabled.
852 * When issuing notes, the compiler now signals a condition of type
853 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
854 restart for use in user handlers. It is expected that the
855 COMPILER-NOTE condition will eventually become a condition
856 supertype to a hierarchy of note types, which will then be
857 handleable in a similar fashion. However, at the moment, no such
858 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
859 but it's an implementation detail, not a classification for the
861 * Changes in type checking closed the following bugs:
862 ** type checking of unused values (192b, 194d, 203);
863 ** template selection based on unsafe type assertions (192c, 236);
864 ** type checking in branches (194bc).
865 * A short form of VALUES type specifier has ANSI meaning (it has
866 increased the number of situations when SBCL cannot perform type
868 * fixed bug in DEFSTRUCT: once again, naming structure slots with
869 keywords or constants is permissible.
870 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
871 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
872 classes. (thanks to Antonio Martinez)
873 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
874 outputs FRESH-LINE or TERPRI, and no longer converts its stream
875 argument to a pretty-print stream. Instead, it leaves any such
876 operations to DESCRIBE-OBJECT methods.
877 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
879 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
880 lambda lists. (thanks to David Lichteblau)
881 * bug fix: Defining a generic function with a :METHOD-CLASS being a
882 subclass of STANDARD-METHOD no longer causes stack exhaustion.
883 (thanks to Gerd Moellmann)
884 * fixed bug 246: increased compilation speed of long
885 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
887 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
889 * DEFCONSTANT now throws a condition of type
890 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
891 constant to a non-EQL value; CONTINUE and ABORT restarts
892 respectively change and preserve the value.
893 * fixed bug 63: The code walker, part of the implementation of CLOS,
894 is now better at handling symbol macros.
895 * bug fix: There is no longer an internal implementation type named
896 CL:LENGTH. (reported by Raymond Toy)
897 * bug fix: In macro-like defining macros/special operators the
898 implicit block does not enclose the lambda list.
899 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
900 enclosed in parentheses) are not suitable as type specifiers, and
901 their use properly signals an error now.
902 * bug fix: An argument count mismatch for a type specifier in code
903 being compiled no longer causes an unhandled error at compile
904 time, but signals a compile-time warning.
905 * fixed simple vector readable printing
906 * bug fix: DESCRIBE takes more care over whether the class
907 precedence list slot of a class is bound before accessing it.
908 (reported by Markus Krummenacker)
909 * bug fix: FORMATTER can successfully compile pretty-printer format
910 strings which use variants of the ~* directive inside.
911 * bug fix: SEARCH now applies its TEST predicate to the elements of
912 the arguments in the correct order. (thanks to Wolfhard Buss)
913 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
914 in type inference. (thanks to Robert E. Brown)
915 * bug fix: Reading in symbols with an explicit package name of ""
916 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
917 * fixed some bugs revealed by Paul Dietz' test suite:
918 ** NIL is now allowed as a structure slot name.
919 ** Arbitrary numbers, not just REALs, are allowed in certain
920 circumstances in LOOP for-as-arithmetic clauses.
921 ** Multiple class redefinitions before slot access no longer
923 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
924 association between the name and a class.
925 ** Generic functions with non-standard method-combination and over
926 five methods all of which return constants no longer return NIL
927 after the first few invocations. (thanks to Gerd Moellmann)
928 ** CALL-NEXT-METHOD with no arguments now passes the original
929 values of the arguments, even in the presence of assignment.
930 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
932 ** DISASSEMBLE works with closures and funcallable instances.
933 ** ADD-METHOD now returns the generic function, not the new method.
934 ** FIND-METHOD signals an error if the lengths of the specializers
935 is incompatible with the generic function, even if the ERRORP
937 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
938 which its argument is a member.
939 ** DEFCLASS only redefines the class named by its class-name
940 argument if that name is the proper name of the class;
941 otherwise, it creates a new class.
942 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
943 of the UNBOUND-SLOT condition to the name of the slot.
944 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
945 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
946 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
947 SLOT-MAKUNBOUND in the specified fashion.
949 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
950 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
951 cross-compilation host. As a consequence, we can now bootstrap our
952 way up to SBCL starting with a bare gcc toolchain and human-readable
953 source code (first the source to CLISP, then the source to SBCL).
954 * A contributed module containing a partial implementation of the
955 simple-streams interface has been included. (thanks to Rudi
957 * A contributed module implementing the RFC1321 Message Digest
958 Algorithm, known as MD5, has been included.
959 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
960 no longer has any effect, as the code controlled by this feature
961 has been deleted. (As far as we know, no-one has ever built using
962 this feature, and its semantics were confused in any case).
963 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
964 work on conditions (as required by the ANSI specification),
965 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
966 expected behaviour on conditions. Users should note, however,
967 that such behaviour is not required by the ANSI specification,
968 and so use of this behaviour may render their code unportable.
969 * fixed some bugs revealed by Paul Dietz' test suite:
970 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
972 ** &ENVIRONMENT parameter in macro lambda list is bound first.
973 ** SXHASH on condition objects no longer returns NIL.
974 ** :ALLOCATION :CLASS slots are better treated; their values are
975 updated on class redefinition, and initforms inherited from
976 superclasses are applied.
977 ** REMOVE-METHOD returns its generic function argument even when
978 no method was removed.
979 ** SHARED-INITIALIZE now initializes the values of the requested
980 slots, including those with :ALLOCATION :CLASS.
981 ** ALLOCATE-INSTANCE now works on structure classes defined via
982 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
984 ** SLOT-EXISTS-P now works on conditions, as well as structures
986 ** MAKE-LOAD-FORM now has the required methods on
987 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
988 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
989 keyword, and now implements the SLOT-NAMES argument.
990 ** methods with &OPTIONAL arguments no longer allow too many
991 arguments to be passed in the call without error.
992 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
993 option is consistent with the required arguments of the generic
994 function lambda list.
995 * bug fix: REQUIRE accepts a string designator. (Thanks to
997 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
998 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
999 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1000 * bug fix: The compiler now checks for duplicated variables in macro
1002 * bug fix: SETQ on globals returns the correct value.
1003 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1004 not a COMPILER-ERROR (followed by some other strange error on
1005 choosing the CONTINUE restart).
1006 * bug fix: make.sh and friends are now more consistent in the way that
1007 they look for GNU "make".
1009 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1010 * experimental native threads support (on x86 Linux >=2.4 only).
1011 This is not compiled in by default: you need to add :SB-THREAD to
1012 the target features. See the "Beyond ANSI" chapter of the manual
1014 * fix for longstanding nonANSIism: The old distinction between
1015 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1016 The return value from CL:FIND-CLASS is now a CLOS class, and
1017 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1018 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1019 * An interface to the MetaObject Protocol, as described in Kiczales,
1020 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1021 MIT Press, 1991, is available from the SB-MOP package.
1022 * incompatible change: the SB-PCL package should now be considered
1023 a private implementation detail, and no longer a semi-private MOP
1025 * minor incompatible change: due to rearrangement for threads, the
1026 control stack and binding stack are now allocated at arbitrary
1027 addresses instead of being hardcoded per-port. Users affected by
1028 this probably have to be doing advanced things with shared
1029 libraries, and will know who they are.
1030 * minor incompatible change: Previously, all --eval forms used were
1031 processed with READ before any of them were processed with EVAL.
1032 Now each --eval form is processed with both READ and EVAL before
1033 the next --eval form is processed. (Thus package operations like
1034 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1035 work as the user might reasonably expect.)
1036 * minor incompatible change: *STANDARD-INPUT* is now only an
1037 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1039 * minor incompatible change: Y-OR-N-P is now character-oriented,
1040 not line oriented. Also, YES-OR-NO-P now works without errors.
1041 (thanks to Antonio Martinez)
1042 * sb-aclrepl module improvements: an integrated inspector, added
1043 repl features, and a bug fix to :trace command.
1044 * Known functions, which cannot be open coded by the backend, are
1045 considered to be able to check types of their arguments. (fixing
1046 a bug report by Nathan J. Froyd)
1047 * fixed a bug in computing method discriminating functions: It is
1048 now possible to define methods specialized on classes which have
1049 forward-referenced superclasses. (thanks to Gerd Moellmann)
1050 * fixed evaluation order in optional entries (reported by Gilbert
1052 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1053 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1054 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1055 required-by-AMOP NAME argument, as well as CLASS and
1056 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1057 * fixed bug 20: DEFMETHOD can define methods using names that are
1058 not the proper names of classes to designate class specializers.
1059 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1060 Unix file descriptors, instead of blowing up. (thanks to Antonio
1062 * Garbage collection refactoring: user-visible change is that a
1063 call to the GC function during WITHOUT-GCING will not do garbage
1064 collection until the end of the WITHOUT-GCING. If you were doing
1065 this you were probably losing anyway.
1066 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1067 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1068 TYPEP the latter but not the former.
1069 * The compiler issues a full WARNING for calls to undefined functions
1070 with names from the CL package.
1071 * MAP-INTO for a vector destination is open coded. (reported by
1072 Brian Downing on c.l.l)
1073 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1074 documentation string.
1075 * fixed some bugs revealed by Paul Dietz' test suite:
1076 ** COPY-ALIST now signals an error if its argument is a dotted
1078 ** Condition slots are now accessed more correctly in the presence
1079 of multiple initargs for a given slot.
1080 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1081 exclude restarts of the same name associated with a different
1083 ** DEFCLASS of forward-referenced classes with another
1084 forward-referenced class in the superclasses list no longer
1086 ** Condition slots are now initialized once each, not multiple
1087 times. (thanks to Gerd Moellmann)
1088 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1090 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1091 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1092 arguments contain duplicated elements.
1093 ** RESTART-CASE understands local macros.
1094 ** RESTART-CASE associates exactly its own restarts with a condition.
1095 ** ENDP in safe mode checks its argument to be of type LIST.
1096 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1097 in question is unbound.
1098 ** Optimized MAKE-INSTANCE functions no longer cause internal
1099 assertion failures in the presence of duplicate initargs.
1100 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1101 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1102 propagated even in the ctor optimized implementation of
1104 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1106 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1107 * a better implementation of SXHASH on (simple) bit vectors,
1108 measured both in execution speed and in distribution of results
1109 over the positive fixnums, has been installed. Likewise, a better
1110 implementation of EQUAL for simple bit vectors is now available.
1111 * fixed CEILING optimization for a divisor of form 2^k.
1112 * fixed bug 240 (emitting extra style warnings "using the lexical
1113 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1114 by Antonio Martinez)
1115 * fixed SXHASH, giving different results for NIL depending on type
1116 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1117 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1118 variable. (found by Rolf Wester)
1119 * SBCL does not ignore type declarations for special
1120 variables. (reported by rif on c.l.l 2003-03-05)
1121 * some bug fixes in contrib/sb-aclrepl/
1122 * fixed some bugs revealed by Paul Dietz' test suite:
1123 ** a bug in the CONS type specifier, whereby the CAR and CDR
1124 types got intertwined, has been fixed;
1125 ** the type system is now able to reason about the interaction
1126 between INTEGER and RATIO types more completely;
1127 ** APPEND, [N]REVERSE and NRECONC check that those their
1128 arguments, which must be proper lists, are really so;
1129 ** An array specialized to be unable to hold elements has been
1130 implemented, as required -- yes, really -- by ANSI;
1131 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1132 on malformed property lists;
1134 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1135 * incompatible packaging change: in line with Unix convention,
1136 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1137 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1138 * REQUIRE and PROVIDE are now optionally capable of doing something
1139 useful. See the documentation string for REQUIRE.
1140 * infrastructure for a managed SBCL contrib system: contributed
1141 modules in this release include:
1142 ** the ASDF system definition facility;
1143 ** an interface to the BSD Sockets API;
1144 ** an ACL-like convenience interface to the repl;
1145 (thanks to Kevin Rosenberg)
1146 ** an implementation of ROTATE-BYTE, with efficient implementation
1148 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1149 gives a closer approximation to the right answer than previously.
1150 (thanks to Raymond Toy)
1151 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1152 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1153 optional environment arguments, as required by ANSI.
1154 * fixed bugs in other functions taking environment objects, allowing
1155 calls with an explicit NIL environment argument to be compiled
1157 * fixed bug 228: primary return values from
1158 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1159 COMPILE or FUNCTION.
1160 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1161 :TYPE will no longer signal errors on innocuous objects.
1162 * fixed bug 231b: SETQ is better at respecting type declarations in
1163 the lexical environment.
1164 * fixed a bug in DEFCLASS: classes named by symbols with no or
1165 unprintable packages can now be defined.
1166 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1167 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1168 * fixed compiler failure related to checking types of functions.
1169 (reported by Robert E. Brown)
1170 * the compiler is now much more consistent in its error-checking
1171 treatment of bounding index arguments to sequence functions: in
1172 (SAFETY 3) code, errors will be signalled in almost all cases if
1173 invalid sequence bounding indices are passed to functions defined
1174 by ANSI to operate on sequences.
1175 * fixed a bug in the build procedure: documentation of SBCL-specific
1176 packages is now preserved and available in the final Lisp image.
1177 * lifted FDEFINITION lookup out of loops in the implementation of
1178 many list operations. (thanks to Robert E. Brown)
1179 * fixed a bug in the reader: the #n# reader macro now works for
1180 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1181 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1182 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1183 E. Brown and Tony Martinez respectively)
1184 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1185 SLOT-BOUNDP: in particular, it is now able to optimize them much
1186 better, and is now not vulnerable to having packages renamed.
1187 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1188 always called when a slot is not present in an instance. (thanks
1190 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1191 longer signal an error on unreading a character following EOF on
1192 the previous constituent stream. (thanks to Tony Martinez)
1193 * fixed some bugs revealed by Paul Dietz' test suite:
1194 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1195 not just nonnegative fixnums;
1196 ** the logical bit-array operators such as BIT-AND now accept an
1197 explicit NIL for their "opt-arg" argument (to indicate a
1198 freshly-consed result bit-array);
1199 ** ELT now signals an error on an invalid sequence index in safe
1201 ** the type system is now cleverer about negations of numeric
1202 types, and consequently understands the BIGNUM and RATIO types
1204 ** the type system is now cleverer about the interaction between
1205 INTEGER and RATIO types: while bugs still remain, many more
1206 cases are accurately computed;
1207 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1208 if it is in the last clause;
1209 ** CONSTANTLY now correctly returns a side-effect-free function in
1211 ** DECLARE is no longer treated as a special-operator; in
1212 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1213 * incremented fasl file version number due to the change in the
1214 DEFSTRUCT-SLOT-DESCRIPTION structure.
1216 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1217 * minor incompatible change: code processed by the "interpreter" or
1218 EVAL now has a compilation optimization policy of (DEBUG 2)
1219 (changed from (DEBUG 1)) to improve debuggability of interactive
1220 development, and to allow the use of the debug RETURN command in
1222 * an experimental implementation of the RETURN command for the
1223 debugger has been included. (thanks to Frederik Kuivinen)
1224 * fixed bug 62: constraints were not propagated into a loop.
1225 * fixed bug in embedded calls of SORT (reported and investigated by
1227 * fixed some bugs revealed by Paul Dietz' test suite:
1228 ** printing and reading of arrays with some dimensions having
1229 length 0 (thanks to Gerd Moellmann);
1230 ** BOA constructor with &AUX argument without a default value does
1231 not cause a type error;
1232 ** CONSTANTP now returns true for all self-evaluating objects.
1234 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1235 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1236 accessors that are related by inheritance, as specified in the
1237 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1238 Valtteri Vuorikoski)
1239 * The compiler is now able to inline functions that were defined in
1240 a complex lexical environment (e.g. inside a MACROLET).
1241 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1243 * Support for the upcoming FreeBSD-5.0 release has been included.
1244 (thanks to Dag-Erling Smorgrav)
1245 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1246 effect when it is not in a toplevel context.
1247 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1248 stand a better chance of being correct. (thanks to Gerd
1250 * fixed bug in COERCE, which now signals an error on coercing a
1251 rational to a bounded real type which excludes the expected
1253 * The compiler is now able to derive types more accurately from the
1254 COERCE and COMPILE functions.
1255 * fixed bug 223: functional binding is considered to be constant
1256 only for symbols in the CL package.
1257 * fixed bug 231: SETQ did not check the type of a variable being set
1258 (reported by Robert E. Brown)
1259 * A new optimization for MAKE-INSTANCE has been included, fixing
1260 various bugs (including relating to :ALLOCATION :CLASS slots and
1261 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1263 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1264 ** As required by ANSI, LOOP now disallows anonymous collection
1265 clauses such as COLLECT I in conjunction with aggregate boolean
1266 clauses such as THEREIS (= I 1);
1267 ** LOOP now signals an error when any variable is reused in the
1268 same loop (including the potentially useful construct analogous
1269 to WITH A = 1 WITH A = (1+ A);
1270 ** IT is only a special loop symbol within the first clause of a
1271 conditional loop clause;
1272 ** LOOP with a typed iteration variable over a hashtable now
1273 signals a type error iff it should.
1274 * fixed some other bugs revealed by Paul Dietz' test suite:
1275 ** FILE-STREAM now names the class previously known as FD-STREAM;
1276 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1277 argument) no longer signals an error;
1278 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1279 of the slot symbol, rather than using the current package
1280 ((:CONC-NAME "") continues to intern the slot's name in the
1282 * incremented fasl file version number, because of the incompatible
1283 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1284 of the new implementation of DEFINE-COMPILER-MACRO.
1286 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1287 * Support for building SBCL for MIPS platforms running in
1288 little-endian mode has now been checked in, and basic
1289 functionality on said platforms verified.
1290 * minor incompatible change: PCL now records the pathname of a file
1291 in which methods and the like are defined, rather than its
1293 * minor incompatible change: TRUENAME now considers the truename of
1294 a file naming a directory to be the pathname with :DIRECTORY
1295 component indicating that directory.
1296 * minor incompatible change: a NAMED clause in the extended form of
1297 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1298 reason for the previous behaviour is unclear.
1299 * more systematization and improvement of CLOS and MOP conformance
1300 in PCL (thanks to Gerd Moellman and Pierre Mai):
1301 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1303 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1304 primary methods with no specializers;
1305 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1307 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1308 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1309 FORWARD-REFERENCED-CLASSes; error reporting on
1310 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1312 ** SXHASH on CLOS instances now uses a slot internal to the
1313 instance to return different numbers on distinct instances,
1314 while preserving the same return value through invocations of
1316 ** DEFMETHOD signals errors when methods with longer incongruent
1317 lambda lists are added to generic functions;
1318 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1319 CLASS, as specified in AMOP;
1320 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1321 on the order returned by the primary method for classes of
1322 class STANDARD-CLASS;
1323 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1324 * fixed some bugs shown by Paul Dietz' test suite:
1325 ** DOLIST puts its body in TAGBODY;
1326 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1328 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1329 value producing form;
1330 ** if more variables are given to PROGV than values, extra
1331 variables are bound and made to have no value;
1332 ** NSUBSTITUTE on list arguments gets the right answer with
1334 ** ELT signals an error of type TYPE-ERROR when the index argument
1335 is not a valid sequence index;
1336 ** LOOP signals (at macroexpansion time) an error of type
1337 PROGRAM-ERROR when duplicate variable names are found;
1338 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1339 ** FUNCALL of special-operators now cause an error of type
1341 ** PSETQ now works as required in the presence of side-effecting
1342 symbol-macro places;
1343 ** NCONC accepts any object as its last argument;
1344 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1346 ** loop-for-as-package does not require a package to be explicitely
1348 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1350 * fixed bug 166: compiler preserves "there is a way to go"
1351 invariant when deleting code.
1352 * fixed bug 172: macro lambda lists with required arguments after
1353 &REST arguments now cause an error to be signalled. (thanks to
1355 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1356 bug. (thanks to Matthew Danish)
1357 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1359 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1360 arguments are lexically rebound. (thanks to Gerd Moellmann and
1362 * fixed bug 194: error messages are now more informative when there
1363 is no primary method applicable in a call to a generic
1364 function. (thanks to Gerd Moellmann)
1365 * fixed bug in command line argument checking (thanks to Julian
1367 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1368 (thanks to Matthew Danish)
1369 * incremented fasl file version number, because of the
1370 SXHASH-related changes in the layout of CLOS data structures
1372 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1373 * minor incompatible change: The runtime (the Unix executable named
1374 "sbcl") is now much pickier about the .core files it will load.
1375 Essentially it now requires .core files to descend from the same
1376 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1377 as the runtime does. (The intent is to prevent the crashes which
1378 can occur, and which can even be reported as mysterious failures,
1379 when people patch the sources or change the build parameters
1380 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1381 sbcl and .core files.)
1382 * fixed bug: VALUES-LIST is no longer optimized away.
1383 * fixed bug 142: The FFI conversion of C string values to Lisp
1384 string values no longer conses excessively. (thanks to Nathan
1385 Froyd porting Raymond Toy's fix to CMU CL)
1386 * began to systematize and improve MOP conformance in PCL (thanks to
1387 Nathan Froyd, Gerd Moellman and Pierre Mai):
1388 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1390 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1391 ** FINALIZE-INHERITANCE is now called on class finalization;
1392 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1393 argument precedence order.
1394 * fixed bug 202: The compiler no longer fails on functions whose
1395 derived types contradict their declared type.
1396 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1397 so it can be non-toplevel.
1398 * The fasl file version number has changed (because of the new
1399 implementation of DEFMACRO).
1400 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1401 safe code, that any length requirement by their type-specifier
1402 argument is valid. The exceptions to this are described in bug
1404 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1405 functions now check the types of their inputs as required by ANSI.
1406 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1407 introduced symbol is DECLAREd to be SPECIAL.
1408 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1409 * fixed bug: :COUNT argument to sequence functions may be negative
1410 * fixed bug: body of DO-SYMBOLS may contain declarations
1411 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1412 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1413 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1414 symbol macro only once
1415 * fixed printing of call frame when argument list is unavailable
1416 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1417 * compiler no longer signals WARNING on unknown keyword
1420 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1421 * A beta-quality port to the mips architecture running Linux,
1422 based on the old CMUCL backend, has been made. It has been tested
1423 on a big-endian kernel, and works sufficiently well to be able to
1424 rebuild itself; it has not been tested in little-endian mode.
1425 * fixed an inconsistency between gencgc.c and purify.c which made
1426 dumping/loading .core files unreliable
1427 * fixed bug 120a: The compiler now deals correctly with IFs where
1428 the consequent is the same as the alternative, instead of
1429 misderiving the return type. (thanks to Alexey Dejneka)
1430 * fixed bug 113: Logical pathnames are now dumpable (the logical
1431 host is resolved at load-time, throwing an error if it is not
1433 * fixed bug 174: FORMAT's error message is slightly clearer when a
1434 non-printing character is used in a format directive.
1435 * fixed several bugs in compiler checking of type declarations, i.e.
1436 violations of the Python "declarations are assertions" principle
1437 (thanks to Alexey Dejneka)
1438 * fixed several bugs in PCL's error checking (thanks to Gerd
1440 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1442 * fixed bug in compilation of functions as first class values
1443 (thanks to Antonio Martinez-Shotton)
1444 * The compiler's handling TYPE-ERRORs which it can prove will
1445 inevitably happen at runtime has been cleaned up and corrected
1446 in several ways. (thanks to Alexey Dejneka)
1447 * improved argument type checking for various basic arithmetic
1448 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1449 lavished on them in the past that they can be compiled in many
1450 ways in different special cases
1451 * fixed bug 181: compiler checks validity of user supplied type
1453 * cleaned up code flushing in optimization: Function calls which
1454 should signal errors for safety purposes (e.g. which ANSI says
1455 should signal errors when their arguments are of incorrect type)
1456 are no longer optimized away.
1457 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1458 * incremented fasl file version number, because changes in the
1459 implementation of sequence functions like COERCE caused
1460 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1461 in old inline expansions) to become undefined. (Actually these
1462 changes were later undone, so we might very well be binary
1463 compatible with 0.7.7 after all, but leaving the version number
1464 incremented seemed like the simplest and most conservative
1467 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1468 * An alpha-quality port to the parisc architecture running Linux,
1469 based on the old CMUCL backend, has been made. This, even more so
1470 than the other backends, should be considered still a work in
1471 progress; known problems include that the Linux kernel in 64-bit
1472 mode does not propagate the correct sigcontext structure to
1473 userspace, and consequently SBCL on a parisc64 kernel will not
1475 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1476 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1477 suppressed.) Also now that the compiler is looking at declarations
1478 in the environment, it checks optimization declarations as well,
1479 and suppresses inlining when (> DEBUG SPEED).
1480 * More fixes have been made to treatment of floating point exception
1481 treatment and other Unix signals. In particular, floating point
1482 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1483 * The detection and handling of control stack exhaustion (infinite
1484 or very deeply nested recursion) has changed. Stack exhaustion
1485 detection is now done by write-protecting pages at the OS level
1486 and applies at all optimization settings; when found, a
1487 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1488 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1489 be caught using IGNORE-ERRORS.
1490 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1491 correctly when transforming recursive LABELS functions to LETs.
1492 (thanks to Alexey Dejneka)
1493 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1494 that are names of constants or global variables.
1495 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1496 alien routines with docstrings.
1497 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1498 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1500 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1501 (thanks to Eric Marsden)
1502 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1503 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1504 * Bug fix: The SPARC backend can now compile functions involving
1505 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1506 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1507 to a functions expecting a PATHNAME-DESIGNATOR.
1508 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1509 lambda lists. (thanks to Alexey Dejneka)
1510 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1511 functions so that EXPORTing the name of the class doesn't cause
1512 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1513 to Antonio Martinez for reporting this)
1514 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1515 constant-folded without failing an assertion. (thanks to Einar
1516 Floystad Dorum for reporting this)
1517 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1518 types are now dealt with more correctly by the compiler.
1519 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1520 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1521 behaviour analogous to RENAME-FILE. This puts its behaviour more
1522 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1524 * The fasl file version number has changed again. (because of the
1525 bug fix involving the names of PCL MAKE-INSTANCE functions)
1527 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1528 * bug fix: Floating point exceptions are treated much more
1529 consistently on the x86/Linux and PPC/Linux platforms.
1530 * Array initialization with :INITIAL-ELEMENT is now much faster for
1531 cases when the compiler cannot open code the array creation, but
1532 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1533 array accesses have also seen a speed increase.
1534 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1535 specified. (This was at the root of some bad interactions between
1536 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1538 * bug fix: Internal error arguments for undefined functions are now
1539 computed correctly on the PPC/Linux platform.
1540 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1541 Raymond Toy's patch for CMU CL)
1542 * Support for the Solaris 9 operating environment has been included
1543 (thanks to Daniel Merritt)
1544 * A very ugly but hopefully complete draft of the missing FFI chapter
1545 of the manual has been created by reformatting the corresponding
1546 CMU CL manual chapter into (currently very ugly and incoherent)
1547 DocBook and bringing it up to date for SBCL behavior. Thus, the
1548 manual is now essentially complete, at least by my extreme
1549 once-and-only-once standards, whereby it's acceptable to refer to
1550 the doc strings of SB-EXT functions as the primary documentation.
1551 * The fasl file version number has changed again, due to cleanup of
1552 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1553 is no longer a static symbol.)
1555 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1556 * SBCL now builds with OpenMCL (version 0.12) as the
1557 cross-compilation host; also, more progress has been made toward
1558 bootstrapping under CLISP.
1559 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1561 * bug 158 fixed: The compiler can now deal with integer loop
1562 increments different from 1; fixing this turned out also to fix
1564 * bug 169 fixed: no more bogus warnings about using lexical bindings
1565 despite the presence of perfectly good SPECIAL declarations (thanks
1566 to David Lichteblau)
1567 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1568 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1569 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1571 * bug 180 fixed: Method combination specifications no longer ignore
1572 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1573 * bug fix: Structure type predicate functions now check their argument
1574 count as they should.
1575 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1576 correctly. (thanks to Pierre Mai)
1577 * minor incompatible change: The --noprogrammer option is deprecated
1578 in favor of the new --disable-debugger option, which is very similar.
1579 (The major difference is that it takes effect at a slightly different
1580 time at startup, causing handling of errors in --sysinit and
1581 --userinit files will be affected differently.) The
1582 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1583 been added to allow this functionality to be controlled from ordinary
1584 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1585 maintainers, who might want to run non-interactive scripts to
1586 build SBCL cores which will later be used interactively.)
1587 * minor incompatible change: The LOAD function no longer, when given
1588 a wild pathname to load, loads all files matching that pathname.
1589 Instead, an error of type FILE-ERROR is signalled.
1591 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1592 * bug 147 fixed: The compiler preserves its block link/count
1593 invariants more correctly now so that it doesn't crash. (thanks
1595 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1597 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1598 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1599 no-value-for-this-character as NIL (as specified by ANSI).
1600 * HOST-NAMESTRING on physical pathnames now returns a string that is
1601 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1602 (thanks to Christophe Rhodes)
1603 * The Alpha port handles icache flushing more correctly. (thanks to
1605 * More progress has been made toward bootstrapping under CLISP. (thanks
1606 to Christophe Rhodes)
1607 * The fasl file format has changed again, because dynamic loading
1608 on OpenBSD (which has non-ELF object files) motivated some cleanups
1609 in the way that foreign symbols are transformed and passed around.
1610 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1611 is no longer treated as whitespace by the reader, but instead as
1612 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1613 |A
\7fB|, instead of A as it used to.
1615 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1616 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1617 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1618 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1619 this as of 0.7.1.45, but was left out of the previous news section
1620 (thanks to Dan Barlow)
1621 * SBCL now runs on the Solaris operating system on SPARC architectures
1622 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1623 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1624 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1625 * SPARC backend cleanups, allowing builds of cores optimized for V8
1626 and V9 SPARCS, and also emission of code targeted to a particular
1627 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1629 * SBCL is closer to bootstrapping under CLISP, thanks to various
1630 fixes by Christophe Rhodes.
1631 * The fasl file format has changed again, to allow the compiler's
1632 INFO database to support symbol macros.
1633 * The user manual (in doc/) is formatted into HTML more nicely.
1634 (thanks to coreythomas)
1635 * The system is smarter about SUBTYPEP relationships, especially
1636 those involving NOT types (including types such as ATOM which are
1637 represented internally using NOT types). Thus SUBTYPEP is less
1638 likely to return (VALUES NIL NIL) in general, and in particular
1639 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1641 * The fasl file format has changed again, because the internal
1642 representation of types now includes a new slot to support the new
1643 SUBTYPEP-of-NOT-types logic.
1644 * (not a change in the main branch of SBCL, but a related prototype
1645 which can hopefully be merged into the main branch of SBCL in the
1646 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1647 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1648 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1649 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1650 dispatch macro character combinations. (thanks to Alexey Dejneka)
1651 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1652 that superficially look like logical namestrings correctly.
1653 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1655 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1656 default supplied-p) for &optional and &key arguments. (thanks to
1659 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1660 * incompatible change: The compiler is now less aggressive about
1661 tail call optimization, doing it only when (> SPACE DEBUG) or
1662 (> SPEED DEBUG). (This is an incompatible change because there are
1663 programs which relied on the old CMU-CL-style behavior to optimize
1664 away their unbounded recursion which will now die of stack overflow.)
1665 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1666 for non-GENCGC systems has been increased to 20M (since that
1667 seems much closer to the likely performance optimum for modern
1668 systems than the old 4M value was)
1669 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1670 and *DEBUG-PRINT-LEVEL*
1671 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1672 of CMU CL's support for SPARC, and various endianness and other
1673 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1674 * new syntactic sugar for the Unix command line: --load foo.bar is now
1675 an alternate notation for --eval '(load "foo.bar")'.
1677 ** The system now detects stack overflow and handles it gracefully,
1678 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1679 optimization settings. (This is a good thing in general, and
1680 its introduction in this version should be particularly timely
1681 for anyone whose code fails because of suppression of tail
1683 ** The system now hunts for the C variable "environ" in a more
1684 devious way, to avoid segfaults when the C library version differs
1685 between compile time and run time. (thanks to Christophe Rhodes)
1686 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1687 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1688 ** The compiler no longer issues bogus style warnings for undefined
1689 classes in the same source file as the DEFCLASSes which defined
1690 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1692 ** fixes in CONDITION class precedence list for undefined function
1693 errors (thanks to Alexei Dejneka)
1694 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1695 correctly. (thanks to Dan Barlow)
1696 ** portability fixes aiming at bootstrapping under CLISP (thanks
1697 to Dave McDonald and Christophe Rhodes)
1698 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1699 ** fixes in type translation and and type inference (thanks to
1701 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1702 ** various fixes in the new ports (thanks to Dan Barlow)
1703 * several changes related to debugging:
1704 ** suppression of tail recursion, as noted above
1705 ** stack overflow detection, as noted above
1706 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1707 is now the default. (For some time encapsulation has been more
1708 reliable than the breakpoint-based :ENCAPSULATE NIL
1709 implementation, at least on X86 systems; and I just noticed that
1710 encapsulation also seems closer to the spirit of the ANSI
1713 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1715 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1716 up properly again. (There was a packaging bug in 0.7.0 which
1717 left their definitions in SB-SYS::LOAD-FOREIGN and
1718 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1719 vital for most things which interface to C-level interfaces,
1720 like extensions working with sockets or databases or
1721 Perl-compatible regexes or whatever, and the need to fix
1722 this bug is the main reason that 0.7.1 was released so
1724 ** DEFGENERIC is now choosier about the methods it redefines, so that
1725 reLOADing a previously-LOADed file containing DEFGENERICs does
1726 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1727 cycle now works as it should. (thanks to Alexey Dejneka)
1728 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1729 fixed by Christophe Rhodes. (He actually submitted this patch
1730 months ago, and I delayed until after 0.7.0.)
1731 ** Bug 111 (internal compiler confusion about runtime checks on
1732 FUNCTION types) was fixed by Alexey Dejneka.
1733 * Some internal cleanups (getting rid of variables which aren't
1734 needed now that the byte interpreter is gone) caused the fasl
1735 file format number to change again.
1737 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1738 * major incompatible change: The default fasl file extension, i.e. the
1739 default extension for files produced by COMPILE-FILE, has changed
1740 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1742 ** There are many changes in the implementation of the compiler.
1743 SBCL is now essentially a compiler-only implementation of ANSI
1744 Common Lisp. EVAL still "interprets" a few special cases, but
1745 almost all the interesting cases are handled by creating
1746 a LAMBDA expression, calling COMPILE on it, then calling
1747 FUNCALL on the result.
1748 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1749 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1750 Since the code is newer, there might still be some new bugs
1751 (though not as many as before Martin Atzmueller's fixes:-). But
1752 the new code is substantially simpler and clearer, and hopefully
1753 any remaining bugs will be simpler, less fundamental, and more
1754 fixable then the bugs in the old code.
1755 ** The revised compiler is still a little unsteady on its feet.
1757 *** The debugging information it produces (particularly the names
1758 of FUNCTION objects) is sometimes much less useful than what
1759 the old compiler produced.
1760 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1761 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1762 within a MACROLET) has been temporarily weakened.
1763 ** There are new compiler optimizations for various functions:
1764 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1765 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1766 *** the math functions TRUNCATE, FLOOR, and CEILING
1767 *** the function-of-all-trades COERCE
1768 Mostly these should be transparent, but there's one
1769 potentially-annoying problem (bug 117): when the compiler
1770 inline-expands a function and does type analysis on the result,
1771 it can create control paths which have type mismatches, and
1772 when it can't prove that those control paths aren't taken,
1773 it will issue WARNINGs about the type mismatches. This is
1774 a particular problem in practice for the new sequence functions.
1775 It's not clear how this should be fixed, and for now, a
1776 workaround is given in the entry for 117 in the BUGS file.
1777 ** (Because of the interaction between the two previous items --
1778 occasional inlining problems and new inline expansions -- some
1779 of the new sequence function optimizations won't really kick in
1780 completely until debugging information, and then inlining, are
1781 straightened out in some future version.)
1782 * minor incompatible changes:
1783 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1784 DIRECTORY no longer implicitly promotes NIL slots of its
1785 pathname argument to :WILD. In particular, when you ask for the
1786 contents of a directory (which you used to be able to do without
1787 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1788 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1789 ** changes in behavior that ANSI explicitly defines to be
1790 implementation dependent:
1791 *** The new compiler-only implementation still conforms with ANSI,
1792 but acts a little different than before. Besides the obvious
1793 changes in performance tradeoffs (that the cost per form passed
1794 to EVAL has gone up, and the cost per form executed by EVAL
1795 has gone down), the behavior of the system changes a little
1796 because there are no longer any interpreted function objects.
1797 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1798 e.g. doing COMPILE on the output of interactive DEFUN is
1800 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1802 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1803 has changed from (:CL) to NIL.
1804 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1805 e.g. #\Newline and #\Tab, don't have names specified in the
1806 ANSI Common Lisp standard, is now based on their ASCII symbolic
1807 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1808 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1809 are no longer used for output.
1810 ** changes in internal implementation constants:
1811 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1812 4 million. (If your application spends a lot of time GCing and
1813 you have a lot of RAM, you might want to experiment with
1814 increasing it even more.)
1815 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1816 However, almost all old code should still continue to work without
1817 immediate update, as SB-C-CALL is now a (deprecated) nickname
1819 ** Old operator names in the style DEF-FOO are now deprecated in
1820 favor of new corresponding names DEFINE-FOO, for consistency with
1821 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1822 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1823 internal symbols, but a few supported extensions like
1824 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1825 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1826 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1827 etc. as you get deeper into recursive calls to the debugger
1828 command loop, instead of the old "5]", "5]]", "5]]]"
1829 sequence. (I was motivated to do this when squabbles between
1830 ILISP and SBCL left me very deeply nested in the debugger. In the
1831 short term, this change will probably provoke more ILISP/SBCL
1832 squabbles, but hopefully it will be an improvement in the long run.)
1833 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1834 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1835 function, since declaiming return types involving aliens is
1836 (1) annoyingly messy to do by hand and (2) vital to efficient
1837 compilation of code which calls such functions.
1838 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1839 longer reexported by the SB-EXT package. They're solely useful
1840 for alien code, so it seems more logical that you should get
1841 them from the SB-ALIEN package, not in SB-EXT.
1842 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1843 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1844 features. Instead, the code that they used to control is always
1845 built into the system.
1846 * many other bug fixes
1847 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1848 advantage of the new EVAL-WHEN stuff and to clean them up in
1849 general, and they are now more ANSI-compliant in a number of
1850 ways. Martin Atzmueller is responsible for a lot of this.
1851 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1853 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1854 *** correct ERROR type for various file operations
1855 *** some fixes for Lisp streams
1856 *** DEFMETHOD syntax checking
1857 *** changing old weird representation of debug information as
1858 strings (which, among their other deficiencies, don't transform
1859 correctly when you rename packages, and don't change their
1860 print representation when you change things like *PACKAGE*
1861 and *PRINT-LENGTH*) to symbols and lists of symbols
1862 He also made several improvements and fixed several bugs in DESCRIBE.
1863 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1865 *** misbehavior of WRITE-STRING/WRITE-LINE
1866 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1867 and several other LOOP problems as well
1868 *** DIRECTORY when similar filenames are present
1869 *** DEFGENERIC with :METHOD options
1870 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1871 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1872 *** argument ordering in FIND with :TEST option
1873 *** mishandled package designator argument in APROPOS-LIST
1874 *** various problems in the backquote readmacro
1875 *** a bug in APROPOS
1876 *** probably some others that I'm not describing very well here,
1877 since the CVS log documents them by reference to sbcl-devel
1878 messages, and the SourceForge archives aren't working well.:-(
1879 ** Dan Barlow improved the Alpha port (and is making progress on the
1880 PPC port, for those of you who think different).
1881 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
1882 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
1883 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
1884 continues to work on the SPARC port (for those of you in a position
1885 to look down upon our little PC-compatible boxes from a great height).
1886 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
1887 and uses that copy, rather than the current dynamic value, when
1888 it's trying to decide whether to truncate output. Thus e.g.
1889 (let ((*print-lines* 50))
1890 (pprint-logical-block (stream nil)
1892 (let ((*print-lines* 8))
1893 (print (aref possiblybigthings i) stream)))))
1894 should now truncate the logical block only at 50 lines, instead of
1895 often truncating it at 8 lines, as it did before.
1896 * The doc/cmucl/ directory, containing old CMU CL documentation
1897 from the time of the fork, is no longer part of the base system.
1898 SourceForge has shut down its anonymous FTP service, and with it
1899 my original plan for distributing the old CMU CL documentation
1900 there. For now, if you need these files you can download an old
1901 SBCL source release and extract them from it.
1902 * The fasl file version number changed again, for dozens of reasons,
1903 some of which are apparent above.
1905 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
1906 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
1907 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
1908 patch, so that DISASSEMBLE output is much nicer.
1909 * The code in the SB-PROFILE package now seems reasonably stable.
1910 I still haven't decided what the final interface should look like
1911 (I'd like PROFILE to interact cleanly with TRACE, since both
1912 facilities use function encapsulation) but if you have a need
1913 for profiling now, you can probably use it successfully with
1914 the current CMU-CL-style interface.
1915 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
1916 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
1917 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
1918 link now returns the dangling link itself, and for similar
1919 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
1920 link itself. (In these cases the old code signalled an error and
1921 looped endlessly, respectively.) Thus, DIRECTORY now works even
1922 in the presence of dangling and cyclic symbolic links.
1923 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
1924 is now a supported extension again, since the consensus on
1925 sbcl-devel was that it can be useful for ordinary development
1926 work, not just for debugging SBCL itself.
1927 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
1928 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
1929 that currently-defined functions might be redefined later with
1930 different return types.
1931 * Hash tables can be printed readably, as inspired by CMU CL code
1932 of Eric Marsden and SBCL code of Martin Atzmueller.
1933 * better error handling in CLOS method combination, thanks to
1934 Martin Atzmueller porting Pierre Mai's CMU CL patches
1935 * more overflow fixes for >16Mbyte I/O buffers
1936 * A bug in READ has been fixed, so that now a single Ctrl-D
1937 character suffices to cause end-of-file on character streams.
1938 In particular, now you only need one Ctrl-D at the command
1939 line (not two) to exit SBCL.
1940 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
1942 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
1943 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
1944 does the right thing.
1945 * The implementation of some type tests, especially for CONDITION
1946 types, is now tidier and maybe faster, due to CMU CL code
1947 originally by Douglas Crosher, ported by Martin Atzmueller.
1948 * Some math functions have been fixed, and there are new
1949 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
1950 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
1951 * (There are also some new optimizers in contrib/*-extras.lisp. Those
1952 aren't built into sbcl-0.6.13, but are a sneak preview of what's
1953 likely to be built into sbcl-0.7.0.)
1954 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
1955 CL, ported to SBCL by Martin Atzmueller)
1956 * DESCRIBE now gives more information in some cases. (Pierre Mai's
1957 patch to CMU CL, ported to SBCL by Martin Atzmueller)
1958 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
1959 * There's a new slam.sh hack to shorten the edit/compile/debug
1960 cycle for low-level changes to SBCL itself, and a new
1961 :SB-AFTER-XC-CORE target feature to control the generation of
1962 the after-xc.core file needed by slam.sh.
1963 * minor incompatible change: The ENTRY-POINTS &KEY argument to
1964 COMPILE-FILE is no longer supported, so that now every function
1965 gets an entry point, so that block compilation looks a little
1966 more like the plain vanilla ANSI section 3.2.2.3 scheme.
1967 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
1968 returns the number of bytes consed since the system started,
1969 rather than the number consed since the first time the function
1970 was called. (The new definition parallels ANSI functions like
1971 CL:GET-INTERNAL-RUN-TIME.)
1972 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
1973 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
1974 supported. Now DIRECTORY always does the abstract Common-Lisp-y
1975 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
1976 * Fasl file version numbers are now independent of the target CPU,
1977 since historically most system changes which required version
1978 number changes have affected all CPUs equally. Similarly,
1979 the byte fasl file version is now equal to the ordinary
1982 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
1983 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
1984 is no longer recognized. I apologize for this, because it was
1985 listed in SB-EXT as a supported extension, but I found that
1986 its existing behavior was poorly specified, as well as incorrectly
1987 specified, and it looked like too much of a mess to straighten it
1988 out. I have enough on my hands trying to get ANSI stuff to work..
1989 * many patches ported from CMU CL by Martin Atzmueller, with
1990 half a dozen bug fixes in pretty-printing and the debugger, and
1991 half a dozen others elsewhere
1992 * fixed bug 13: Floating point infinities are now supported again.
1993 They might still be a little bit flaky, but thanks to bug reports
1994 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
1995 as flaky as they were.
1996 * The --noprogrammer command line option is now supported. (Its
1997 behavior is slightly different in detail from what the old man
1998 page claimed it would do, but it's still appropriate under the
1999 same circumstances that the man page talks about.)
2000 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2001 are now supported, and enabled by default. Thus, the compiler can
2002 handle many floating point and complex operations much less
2003 inefficiently. (Thus e.g. you can implement a complex FFT
2005 * The compiler now detects type mismatches between DECLAIM FTYPE
2006 and DEFUN better, and implements CHECK-TYPE more correctly, and
2007 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2008 thanks to patches from Martin Atzmueller.
2009 * various fixes to make the cross-compiler more portable to
2010 ANSI-conforming-but-different cross-compilation hosts (notably
2011 Lispworks for Windows, following bug reports from Arthur Lemmens)
2012 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2013 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2014 fixed thanks to Pierre Mai's CMU CL patches.
2015 * fixes to keep the system from overflowing internal counters when
2016 it tries to use i/o buffers larger than 16M bytes
2017 * fixed bug 45a: Various internal functions required to support
2018 complex special functions have been merged from CMU CL sources.
2019 (When I was first setting up SBCL, I misunderstood a compile-time
2020 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2021 * improved support for type intersection and union, fixing bug 12
2022 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2023 more obscure bugs as well
2024 * some steps toward byte-compiling non-performance-critical
2025 parts of the system, courtesy of patches from Martin Atzmueller
2026 * Christophe Rhodes has made some debian packages of sbcl at
2027 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2028 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2029 stable, but are nonetheless usable. When he's ready, I'd be happy
2030 to add them to the SourceForge "File Releases" section. (And if
2031 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2032 * new fasl file format version number (because of changes in
2033 internal representation of (OR ..) types to accommodate the new
2034 support for (AND ..) types, among other things)
2036 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2037 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2039 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2041 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2042 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2043 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2044 * DESCRIBE now works on structure objects again.
2045 * Most function call argument type mismatches are now handled as
2046 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2047 know whether the function will be redefined before the call is
2048 executed. (The compiler could flag local calls with full WARNINGs,
2049 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2050 it doesn't keep track of enough information to know whether calls
2051 are local in this sense.)
2052 * Compiler output is now more verbose, with messages truncated
2053 later than before. (There should be some supported way for users
2054 to override the default verbosity, but I haven't decided how to
2055 provide it yet, so this behavior is still controlled by the internal
2056 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2057 src/compiler/ir1util.lisp.)
2058 * Fasl file format version numbers have increased again, because
2059 support for the Gray streams extension changes the layout of the
2060 system's STREAM objects.
2061 * The Gray subclassable streams extension now works, thanks to a
2062 patch from Martin Atzmueller.
2063 * The full LOAD-FOREIGN extension (not just the primitive
2064 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2065 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2066 but like most other programs, it defaults to copying the Unix
2067 environment from the original process instead of starting the
2068 new process in an empty environment.
2069 * Extensions which manipulate the Unix environment now support
2070 an :ENVIRONMENT keyword option which doesn't smash case or
2071 do other bad things. The CMU-CL-style :ENV option is retained
2072 for porting convenience.
2073 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2074 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2076 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2078 * A patch from Martin Atzmueller seems to have solved the SIGINT
2079 problem, and as far as we know, signal-handling now works cleanly.
2080 (If you find any new bugs, please report them!)
2081 * The system no longer defaults Lisp source file names to types
2082 ".l", ".cl", or ".lsp", but only to ".lisp".
2083 * The compiler no longer uses special default file extensions for
2084 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2085 seems to expect a single default extension for all compiled code,
2086 and there's no compelling reason to try to stretch the standard
2087 to allow two different extensions.) Instead, byte-compiled files
2088 default to the same extension as native-compiled files.
2089 * Fasl file format version numbers have increased again, because
2090 a rearrangement of internal implementation packages made some
2091 dumped symbols in old fasl files unreadable in new cores.
2092 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2093 many fewer weird special cases.
2094 * Bug #17 (differing COMPILE-FILE behavior between logical and
2095 physical pathnames) has been fixed, and some related misbehavior too,
2096 thanks to a patch from Martin Atzmueller.
2097 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2098 by Tim Moore, ported to SBCL by Martin Atzmueller.
2099 * Martin Atzmueller fixed several filesystem-related problems,
2100 including bug #36, in part by porting CMU CL patches, which were
2101 written in part by Paul Werkowski.
2102 * More compiler warnings in src/runtime/ are gone, thanks to
2103 more patches from Martin Atzmueller.
2104 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2107 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2109 * DESCRIBE now works on CONDITION objects.
2110 * The debugger now handles errors which arise when trying to print
2111 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2113 * The build system now uses an additional file, customize-target-features.lisp,
2114 to allow local modifications to the target *FEATURES* list. (The point of
2115 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2116 debugging features enabled, without having to worry about propagating it
2117 into everyone's system when I do a "cvs update".) When no
2118 customize-target-features.lisp file exists, the target *FEATURES* list
2119 should be constructed the same way as before.
2120 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2121 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2122 warn about behavior which is undefined under the ANSI spec. Note: This
2123 is specified by ANSI, but it's not very popular with programmers.
2124 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2125 macro in the SBCL sources for an example of a workaround which you
2126 might use to make portable ANSI-standard code which does what you want.
2127 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2128 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2129 when it's used as a non-toplevel form. (This is required in order
2130 to implement the DEFCONSTANT-EQX macro.)
2131 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2132 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2133 funny things, due to bugs in the implementation of EVAL-WHEN
2134 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2135 non-toplevel DEFCONSTANTs are uncommon.)
2136 * The core file version number and fasl file version number have been
2137 incremented, because the old noncompliant DEFCONSTANT behavior involved
2138 calling functions which no longer exist, and because I also took the
2139 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2140 * fixed bug 1 (error handling before read-eval-print loop starts), and
2141 redid debugger restarts and related debugger commands somewhat while
2143 ** The QUIT debugger command is gone, since it did something
2144 rather different than the SB-EXT:QUIT command, and since it never
2145 worked properly outside the main toplevel read/eval/print loop.
2146 Invoking the new TOPLEVEL restart provides the same functionality.
2147 ** The GO debugger command is also gone, since you can just invoke
2148 the CONTINUE restart directly instead.
2149 ** The TOP debugger command is also gone, since it's redundant with the
2150 FRAME 0 command, and since it interfered with abbreviations for the
2152 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2154 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2155 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2156 some porting work by Martin Atzmueller)
2157 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2158 with a boatload of patches which he ported to SBCL. Now that those
2160 ** The system tries to make sure that its low-priority messages
2161 are prefixed by semicolons, to help people who like to use
2162 syntax highlighting in their ILISP buffer. (This patch
2163 was originally due to Raymond Toy.)
2164 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2165 patches originally written by Raymond Toy.
2166 ** The compiler understands coercion between single-value and
2167 multiple-VALUES type expressions better, getting rid of some very
2168 weird behavior, thanks to patches originally by Robert MacLachlan
2169 and Douglas Crosher.
2170 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2171 lambda lists, thanks to a patch originally by Pierre Mai.
2172 ** The system no longer bogusly warns about "abbreviated type
2174 ** The compiler gets less confused by inlining and RETURN-FROM,
2175 thanks to some patches originally by Tim Moore.
2176 ** The system no longer hangs when dumping circular lists to fasl
2177 files, thanks to a patch originally from Douglas Crosher.
2178 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2179 undefined function error.
2180 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2181 in structure slot initforms) for now, documented workaround instead:-|
2182 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2183 * fixed bug 5: added stubs for various Gray stream functions called
2184 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2185 installed, at least appropriate type errors are generated
2186 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2187 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2188 consistently in DEFMETHOD forms.
2189 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2190 it doesn't seem to affect SBCL after all
2191 * The C runtime system now builds with better optimization and many
2192 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2194 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2196 * The system is now under CVS at SourceForge (instead of the
2197 CVS repository on my home machine).
2198 * The new signal handling code has been tweaked to treat register
2199 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2200 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2201 downstream, e.g. in debug-int.lisp, has implicit dependencies
2202 on the unsignedness of integer representation of machine words,
2203 and that caused the system to bomb out with infinite regress
2204 when trying to recover from type errors involving signed values,
2205 e.g. (BUTLAST '(1 2 3) -1).)
2206 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2207 (This was one of the bugs Peter Van Eynde reported back in July.)
2208 * The system now uses code inspired by Colin Walters' O(N)
2209 implementation of MAP (from the cmucl-imp@cons.org mailing
2210 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2211 inline the MAP operation, and there is more than one
2212 sequence argument to the MAP call (so that it can't just
2213 do ETYPECASE once and for all based on the type of the
2214 single sequence argument). (The old non-inline implementation
2215 of the general M-argument sequence-of-length-N case required
2216 O(M*N*N) time when any of the sequence arguments were LISTs.)
2217 * The QUIT :UNIX-CODE keyword argument has been renamed to
2218 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2219 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2221 * Raymond Wiker's patches to port dynamic loading from Linux to
2222 FreeBSD have been added.
2223 * The BUGS file is now more nearly up to date, thanks in large part
2224 to Martin Atzmueller's review of it.
2225 * The debugger now flushes standard output streams before it begins
2226 its output ("debugger invoked" and so forth).
2227 * The core version number and fasl file version number have both
2228 been incremented, because of incompatible changes in the layout
2230 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2231 used to not be called for a saved Lisp image.)
2232 * Martin Atzmueller's version of a patch to fix a compiler crash,
2233 as posted on sbcl-devel 13 September 2000, has been installed.
2234 * Instead of installing Martin Atzmueller's patch for the
2235 compiler transform for SUBSEQ, I deleted the compiler transform,
2236 and transforms for some similar consing operations.
2237 * A bug in signal handling which kept TRACE from working on OpenBSD
2239 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2240 away by constant folding
2241 * The system now defines its address space constants in one place
2242 (in the Lisp sources), and propagates them automatically elsewhere
2243 (through GENESIS and the sbcl.h file). Therefore, patching the
2244 address map is less unnecessarily tedious and error-prone. The
2245 Lisp names of address space constants have also been systematized.
2246 * CVS tags like dollar-Header-dollar have been removed from
2247 the sources, because they have never saved me trouble and
2248 they've been source of trouble working with patches and other
2249 diff-related operations.
2250 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2251 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2253 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2255 * The system has been ported to OpenBSD.
2256 * The system now compiles with a simple "sh make.sh" on the systems
2257 that it's supported on. I.e., now you no longer need to tweak
2258 text in the target-features.lisp-expr and symlinks in src/runtime/
2259 by hand, the make.sh takes care of it for you.
2260 * The system is no longer so grossly inefficient when compiling code
2261 involving vectors implemented as general (not simple) vectors (VECTOR T),
2262 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2263 which dares to use the various sequence functions on non-simple
2264 vectors, takes less of a performance hit.
2265 * There is now a primitive type predicate VECTOR-T-P
2266 to test for the (VECTOR T) type, so that e.g.
2267 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2268 can now be compiled with some semblance of efficiency. (The old code
2269 turned the type declaration into a full call to %TYPEP at runtime!)
2270 * AREF on (VECTOR T) is still not fast, since it's still compiled
2271 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2272 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2273 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2274 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2275 without making full calls to SUBTYPEP at runtime.
2276 (Some analogous efficiency issues for non-simple vectors specialized to
2277 element types other than T, or for non-simple multidimensional arrays,
2278 have not been addressed. They could almost certainly be handled the
2279 same way if anyone is motivated to do so.)
2280 * The changes in array handling break binary compatibility, so
2281 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2282 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2283 instead of (VALUES T T).
2284 * By following the instructions that Dan Barlow posted to sbcl-devel
2285 on 2 July 2000, I was able to enable primitive dynamic object
2286 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2287 functionality is not implemented (since it calls ld to resolve
2288 library references automatically, requiring RUN-PROGRAM for its
2289 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2290 try to resolve library references) is now supported.
2291 * The system now flushes the standard output streams when it terminates,
2292 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2293 them at several other probably-convenient times, e.g. in each pass of
2294 the toplevel read-eval-print loop, and after evaluating a form given
2295 as an "--eval" command-line option. (These changes were motivated by a
2296 discussion of stream flushing issues on cmucl-imp in August 2000.)
2297 * The source transform for TYPEP of array types no longer assumes
2298 that an array whose element type is a not-yet-defined type
2299 is implemented as an array of T, but instead punts, so that the
2300 type will be interpreted at runtime.
2301 * There is now some support for cross-compiling in make.sh: each of
2302 the phases of make.sh has its own script. (This should be transparent
2303 to people doing ordinary, non-cross-compile builds.)
2304 * Since my laptop doesn't have hundreds of megabytes of memory like
2305 my desktop machine, I became more motivated to do some items on
2306 my to-do list in order to reduce the size of the system a little:
2307 ** Arrange for various needed-only-at-cold-init things to be
2308 uninterned after cold init. To support this, those things have
2309 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2310 symbols with such names are now uninterned after cold init).
2311 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2312 fasl files for cold load.
2313 ** Remove the old compiler structure pooling code (which used to
2314 be conditional on the target feature :SB-ALLOC) completely.
2315 ** Redo the representation of some data in cold init to be more compact.
2316 (I also looked into supporting byte compiled code at bootstrap time,
2317 which would probably reduce the size of the system a lot, but that
2318 looked too complicated, so I punted for now.)
2319 * The maximum signal nesting depth in the src/runtime/ support code has
2320 been reduced from 4096 to 256. (I don't know any reason for the very
2321 large old value. If the new smaller value turns out to break something,
2322 I'll probably just bump it back up.)
2323 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2325 * Many, many bugs reported by Peter Van Eynde have been added to
2326 the BUGS list; some have even been fixed.
2327 * While enabling dynamic object file loading, I tried to make the
2328 code easier to understand, renaming various functions and variables
2329 with less ambiguous names, and changing some function calling
2330 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2331 * While trying to figure out how to do the OpenBSD port, I tried to
2332 clean up some of the code in src/runtime/. In particular, I dropped
2333 support for non-POSIX signal handling, added various comments,
2334 tweaked the code to reduce the number of compilation warnings, and
2335 renamed some files to increase consistency.
2336 * To support the new automatic configuration functionality in make.sh,
2337 the source file target-features.lisp-expr has been replaced with the
2338 source file base-target-features.lisp-expr and the machine-generated
2339 file local-target-features.lisp-expr.
2340 * fixed a stupid quoting error in make.sh so that using CMU CL
2341 "lisp -batch" as cross-compilation host works again
2343 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2345 * DESCRIBE no longer tries to call itself recursively to describe
2346 bound/fbound values, so that it no longer fails on symbols which are
2347 bound to themselves (like keywords, T, and NIL).
2348 * DESCRIBE now works on generic functions.
2349 * The printer now prints less-screwed-up representations of closures
2350 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2351 * A private symbol is used instead of the :EMPTY keyword previously
2352 used to mark empty slots in hash tables. Thus
2353 (DEFVAR *HT* (MAKE-HASH-TABLE))
2354 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2355 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2356 now does what ANSI says that it should. (You can still get
2357 similar noncompliant behavior if bang on the hash table
2358 implementation with all the symbols you get back from
2359 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2360 This breaks binary compatibility, since tests for equality to
2361 :EMPTY are wired into things like the macroexpansion of
2362 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2364 * There's now a minimal placeholder implementation for CL:STEP,
2365 as required by ANSI.
2366 * An obscure bug in the interaction of the normal compiler, the byte
2367 compiler, inlining, and structure predicates has been patched
2368 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2369 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2371 * Missing ordinary arguments in a macro call are now detected even
2372 when the macro lambda list contains &KEY or &REST.
2373 * The debugger no longer complains about encountering the top of the
2374 stack when you type "FRAME 0" to explicitly instruct it to go to
2375 the top of the stack. And it now prints the frame you request even
2376 if it's the current frame (instead of saying "You are here.").
2377 * As specified by ANSI, the system now always prints keywords
2378 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2380 * The default initial SIZE of HASH-TABLEs is now smaller.
2381 * Type information from CLOS class dispatch is now propagated
2382 into DEFMETHOD bodies, so that e.g.
2383 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2385 is now basically equivalent to
2386 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2387 (DECLARE (TYPE SINGLE-FLOAT X))
2389 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2390 operation, without having to do run-time type dispatch.
2391 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2392 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2393 * Since I don't seem to be making big file reorganizations very often
2394 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2395 is overflowing my ability to conveniently back them up), I've finally
2396 checked the system into CVS. (The CVS repository is on my home system,
2397 not at SourceForge -- putting it on SourceForge might come later.)
2398 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2399 high-level GC-NOTIFY-FOO functions send their output. (There's
2400 still very little control of where low-level verbose GC functions
2401 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2402 less than it used to -- the GC-NOTIFY-FOO functions are now under
2403 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2404 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2405 in only one place in the source code, and propagates it automatically
2406 everywhere that it's needed. Thus e.g. when I bump the version from
2407 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2408 * The C source files now include boilerplate legalese and documentation
2409 at the head of each file (just as the Lisp source files already did).
2410 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2411 to his page will be replaced with a link to his new CLiki service.
2413 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2415 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2416 * The build process now looks for GNU make under the default name "gmake",
2417 instead of "make" as it used to. If GNU make is not available as "gmake"
2418 on your system, you can change this default behavior by setting the
2419 GNUMAKE environment variable.
2420 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2421 can build without error under CMU CL.
2423 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2425 * There is now a partial SBCL user manual (with some new text and some
2426 text cribbed from the CMU CL manual).
2427 * The beginnings of a profiler have been added (starting with the
2428 CMU CL profiler and simplifying and cleaning up). Eventually the
2429 main interface should be through the TRACE macro, but for now,
2430 it's still accessed through vaguely CMU-CL-style functions and macros
2431 exported from the package SB-PROFILE.
2432 * Some problems left over from porting CMU CL to the new
2433 cross-compilation bootstrap process have been cleaned up:
2434 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2435 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2436 part when getting rid of the compiler *BACKEND* stuff.)
2437 ** The value of *NULL-TYPE* was screwed up, because it was
2438 being initialized before the type system knew the final
2439 definition of the 'NULL type. This screwed up several key
2440 optimizations in the compiler, causing inefficiency in all sorts
2441 of places. (I found it because I wanted to understand why
2442 GET-INTERNAL-RUN-TIME was consing.)
2443 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2444 PROCLAIM FTYPE information. Unfortunately this broke binary
2445 compatibility again, since now the forms output by DEFGENERIC
2446 to refer to functions which didn't exist in 0.6.3.
2447 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2448 can use the new (as of 0.6.3) transform for SOME into MAP into
2450 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2451 values to (INTEGER 0 1000000), so that the time code will no longer
2452 occasionally get blown up by Linux returning 1000000 microseconds
2453 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2454 its output conform to the ANSI spec. (Alas, this makes its output
2455 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2456 * A full call to MAP NIL with a single sequence argument no longer conses.
2457 * fixes to problems pointed out by Martin Atzmueller:
2458 * The manual page no longer talks about multiprocessing as though
2459 it were currently supported.
2460 * The ILISP support patches have been removed from the distribution,
2461 because as of version 5.10.1, ILISP now supports SBCL without us
2462 having to maintain patches.
2463 * added a modified version of Raymond Toy's recent CMU CL patch for
2464 EQUALP comparison of HASH-TABLE
2466 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2468 * The system still can't cross-compile itself with
2469 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2470 entails), but at least it can compile more of itself that way
2471 than it used to be able to, and various buglets which were uncovered
2472 by trying to cross-compile itself that way have now been fixed.
2473 * This release breaks binary compatibility again. This time
2474 at least I've incremented the FASL file format version to 2, so that the
2475 problem can be detected reliably instead of just causing weird errors.
2476 * various new style warnings:
2477 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2478 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2479 ** using the lexical binding of a variable named in the *FOO* style
2480 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2481 as specified by ANSI.
2482 * *RANDOM-STATE* is no longer automatically initialized from
2483 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2484 default behavior of the system is to repeat its behavior every time
2485 it's run. If you'd like to change this behavior, you can always
2486 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2487 old convention there was no comparably easy way to get the system to
2488 repeat its behavior every time it was run.
2489 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2490 FTYPE declarations has been removed, in favor of their ANSI
2491 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2492 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2493 the types of their sequence arguments can be determined at compile time.
2494 This is done through a new open code expansion for MAP which eliminates
2495 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2496 argument types can be determined at compile time.
2497 * The optimizer now transforms COERCE into an identity operation when it
2498 can prove that the coerced object is already of the correct type. (This
2499 can be a win for machine generated code, including the output of other
2500 optimization transforms, such as the MAP transform above.)
2501 * Credit information has been moved from source file headers into CREDITS.
2502 * Source file headers have been made more standard.
2503 * The CASE macro now compiles without complaining even when it has
2506 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2508 * (Note that the way that the PCL macroexpansions were rewritten
2509 to accommodate the change in DEFGENERIC below breaks binary
2510 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2511 not run under sbcl-0.6.2. Once we get out of alpha releases,
2512 i.e. hit release 1.0.0, we'll probably try to maintain binary
2513 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2514 and sbcl-1.4.4. Until then, however, it might be fairly common
2515 for maintenance releases to break binary compatibility.)
2516 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2518 * The reporting of unhandled signals has been changed to print some
2519 explanatory text as well as the report form. (Previously only
2520 the report form was printed.)
2521 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2522 it defines, so that the compiler no longer issues undefined function
2523 warnings for compiled-but-not-yet-loaded generic functions.
2524 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2525 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2526 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2527 section of the ANSI spec are supported.
2528 * The "" nickname for the "KEYWORD" package has been removed.
2529 The reader still handles symbol tokens which begin with a package marker
2530 as keywords, but it doesn't expose its mechanism for doing so in the
2531 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2532 * The system now issues STYLE-WARNINGs for contradictory TYPE
2533 proclamations. (Warnings for contradictory FTYPE proclamations would
2534 be nice too, but those can't be done usefully unless the type system
2535 is made smarter about FUNCTION types.)
2536 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2537 systematized, so that "*target-*.lisp is supposed to exist only on the
2538 target and imply that there's a related file which exists on the
2539 host, and *host-*.lisp is supposed to exist only on the host and imply
2540 that there's a related file which exists on the target. This involves a
2541 lot of renaming. Hopefully the acute confusion caused by the renaming
2542 will be justified by the reduction in chronic confusion..
2543 ** runtime-type.lisp -> early-target-type.lisp
2544 ** target-type.lisp -> late-target-type.lisp
2545 ** early-host-format.lisp -> early-format.lisp
2546 ** late-host-format.lisp -> late-format.lisp
2547 ** host-error.lisp -> misc-error.lisp
2548 ** early-error.lisp -> early-target-error.lisp
2549 ** late-error.lisp -> late-target-error.lisp
2550 ** host-defboot.lisp -> early-defboot.lisp
2551 ** code/misc.lisp -> code/target-misc.lisp
2552 ** code/host-misc.lisp -> code/misc.lisp
2553 ** code/numbers.lisp -> code/target-numbers.lisp
2554 ** code/early-numbers.lisp -> numbers.lisp
2555 ** early-host-type.lisp -> early-type.lisp
2556 ** late-host-type.lisp -> late-type.lisp
2557 ** host-typep.lisp -> typep.lisp
2558 ** load.lisp -> target-load.lisp
2559 ** host-load.lisp -> load.lisp
2560 ** host-disassem.lisp -> disassem.lisp
2561 ** host-insts.lisp -> insts.lisp
2562 ** byte-comp.lisp -> target-byte-comp.lisp
2563 ** host-byte-comp.lisp -> byte-comp.lisp
2564 ** host-signal.lisp -> signal.lisp
2565 ** host-defstruct.lisp -> defstruct.lisp
2566 ** late-target-type.lisp -> deftypes-for-target.lisp
2567 Furthermore, several other previously target-only files foo.lisp (e.g.
2568 hash-table.lisp and random.lisp) have been split into a target-and-host
2569 foo.lisp file and a target-only target-foo.lisp file, with their key type
2570 definitions in the target-and-host part, so that the cross-compiler will
2571 know more about target types.
2572 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2573 gone away. In their place are various *BACKEND-FOO* variables
2574 corresponding to the slots of the old structure.
2575 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2576 into the target SBCL has been fixed.
2577 * The chill.lisp system for loading cold code into a running SBCL
2579 * Support for the CMU CL "scavenger hook" extension has been removed.
2580 (It was undocumented and unused in the CMU CL sources that SBCL was
2581 derived from, and stale in sbcl-0.6.1.)
2582 * Various errors in the cross-compiler type system were detected
2583 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2584 (enabling various consistency checks). Many of them were fixed,
2585 but some hard problems remain, so the compiler is back to
2586 running without *TYPE-SYSTEM-INITIALIZED* for now.
2587 * As part of the cross-compiler type system cleanup, I implemented
2588 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2589 * I have started adding UNCROSS calls throughout the type system
2590 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2591 on cross-compiler input files will be able to go away, and various
2593 * CONSTANTP now returns true for quoted forms (as explicitly required
2596 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2598 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2599 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2600 (MAKE-LIST -1). In the longer term, it ought to become true
2601 that declarations are assertions even at SAFETY 1. For now, it's not
2602 quite true even at SAFETY 3, but it's at least more nearly true..
2603 (Note that this change seems to increases the size of the system by
2604 O(5%) and to decrease the speed of the compiler by 20% or more.)
2605 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2606 for the problem of printing dozens of lines of distracting information
2607 about low-level system machinery as part of the top stack frame
2608 on entry to the debugger when an undefined function was called.
2609 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2610 is not reset to COMMON-LISP-USER.
2611 * Compilation of stuff related to dyncount.lisp has been made conditional
2612 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2613 smaller. The various dyncount-related symbols have been moved into
2614 a new "SB-DYNCOUNT" package.
2615 * tty-inspect.lisp has been renamed to inspect.lisp.
2616 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2617 feature has gone away. (When we eventually port to other flavors of
2618 libc and/or Unix, we'll try to make the differences between flavors
2619 invisible at the user level.)
2620 * Various other *FEATURES* tags, and/or their associated conditionals,
2621 have been removed if obsolescent, or given better documentation, or
2622 sometimes given more-mnemonic names.
2624 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2626 * tidied up "make.sh" script
2627 * tidied up system directory structure
2628 * better "clean.sh" behavior
2629 * added doc/FOR-CMUCL-DEVELOPERS
2630 * many many small tweaks to output format, e.g. removing possibly-confusing
2631 trailing #\. character in DESCRIBE-INSTANCE
2632 * (EQUALP #\A 'A) no longer signals an error.
2633 * new hashing code, including EQUALP hashing
2634 * tidied up Lisp initialization and toplevel
2635 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2636 * command line argument processing
2637 * added POSIX-GETENV function to deal with Unix-ish environment variables
2638 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2639 terminating SBCL on EOF
2640 * non-verbose GC by default
2641 * There is no more "sbcl" shell script; the sbcl file is now the C
2642 runtime executable (just like CMU CL).
2643 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2645 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2646 symbol and package information as Lisp native symbol and package objects
2647 instead of strings naming symbols and strings naming packages. This way,
2648 whenever packages are renamed (as in warm init), debug information is
2649 transformed along with everything else.
2650 * tweaked the optimization policy declarations which control the building
2651 of SBCL itself. Now, among other things, the system no longer saves
2652 source location debugging information. (This helps two problems at once
2653 by reducing SBCL size and by keeping SBCL from trying to look for its
2654 sources -- which may not exist -- when reporting errors.)
2655 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2656 debugging and testing purposes
2657 * cleaned up printing, making the printer call PRINT-OBJECT for
2658 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2659 methods, giving nearly-ANSI behavior
2660 * converted almost all special variables to use *FOO* naming convention
2661 * deleted PARSE-TIME functionality, since it can be done portably
2662 * moved some files out of cold init into warm init
2663 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2665 * regularized formatting of source files
2666 * added an install.sh script
2667 * fixed ridiculous memory usage of cross-compiler by making
2668 compiler/alloc.lisp not try to do pooling unless it can hook
2669 itself into the GC of the cross-compilation host. Now the system
2670 builds nicely on my old laptop.
2671 * added :SB-ALLOC in target-features.lisp-expr
2672 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2673 was not implemented)
2674 * re-did condition handling and note reporting in the compiler. Notes
2675 are no longer handled by signalling conditions. Style warnings
2676 and warnings are handled more correctly and reported in such a way
2677 that it's easy to find one or the other in your output (so that you
2678 can e.g. figure out which of many problems caused COMPILE-FILE to
2680 * changed the severity of several compiler warnings from full WARNING
2681 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2682 compiler note reporting so that it doesn't use the condition system
2683 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2684 COMPILE-FILE command)
2685 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2686 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2687 no longer supported, since their implementation was deeply intertwingled
2688 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2689 * removed bogus "support" for compiler macros named (SETF FOO), and
2690 removed the compiler macro for SETF INFO (but only after making a fool
2691 of myself on the cmucl-imp mailing list by posting a bogus patch for
2692 DEFINE-COMPILER-MACRO..)
2693 * Compiled files containing forms which have side effects on the Lisp
2694 reader (such as DEFPACKAGE forms) are now handled more correctly.
2695 (Compiler queuing of top level lambdas has been suppressed by setting
2696 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2697 * deleted various currently-unused source files, e.g. gengc.lisp. They
2698 may be added back at some point e.g. when porting to other architectures,
2699 but until they are it's distracting to distribute them and to try to
2701 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2702 there were so many of them they're just distractions, and UNCROSS is
2703 known to be able to handle the current sources
2704 * moved PROFILE functionality into TRACE, so that it will be clear
2705 how the wrapping and unwrapping of functions when you profile them
2706 interacts with the wrapping and unwrapping of functions when you
2707 trace them. (Actually, the functionality isn't there yet, but at least
2708 the interface specification is there. Hopefully, the functionality will
2709 arrive with some future maintenance release.)
2710 * removed host-oops.lisp
2711 * changed signature of QUIT function to allow UNIX-CODE argument
2712 * fixed READ-SEQUENCE bug
2713 * tweaked verbose GC output so that it looks more like the progress
2714 output that ANSI specifies for functions like LOAD
2715 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2716 * added <http://sbcl.sourceforge.com> to the banner information printed by