1 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2 * bug fix: read-write consistency on streams of element-type
3 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
4 * fixed some bugs revealed by Paul Dietz' test suite:
5 ** POSITION on displaced vectors with non-zero displacement
6 returns the right answer.
8 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
9 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
10 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
11 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
12 the supported interface.
13 * new feature: Single-stepping of code compiled with DEBUG 2 or
14 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
15 * new feature: saving cores with foreign code loaded is now
16 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
17 Timothy Moore's work for CMUCL)
18 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
19 parameters to * instead of NIL if no initform is supplied.
20 (reported by Johan Bockgaard)
21 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
22 correctly when FUNCALL appears as the car of the form. Note:
23 despite this FUNCALL forms are not currently subject to
24 compiler-macro expansion. (port of Raymond Toy's fix for the
25 same from CMUCL, reported by Johan Bockgaard)
26 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
27 (thanks for Teemu Kalvas)
28 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
29 case instead of terminating the iteration (thanks for Julian
30 Squires, Sean Champ and Raymond Toy)
31 * bug fix: incorrect expansion of defgeneric that caused a style
32 warning. (thanks for Zach Beane)
33 * bug fix: slot accessor effective method computation works properly
34 for classes with multiple non-standard applicable methods on
35 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
36 * on x86 compiler supports stack allocation of results of LIST and
37 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
38 CMUCL implementation by Gerd Moellmann)
39 * fixed some bugs revealed by Paul Dietz' test suite:
40 ** FORMAT strings with both the ~<~:;~> form of the justification
41 directive and pretty-printing directives cause an error.
43 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
44 * incompatible change: the internal functions
45 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
46 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
47 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
48 instead of the old functions.
49 * new feature: on platforms where "dladdr" is available foreign
50 function names now appear in backtraces. (based on Helmut Eller's
52 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
53 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
55 * documentation: ASDF manual is now distributed with SBCL.
56 * bug fix: SBCL can now load its contributed modules with REQUIRE
57 even if the system-provided entries have been removed from
58 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
60 * bug fix: backtraces involving undefined functions or assembly
61 routines are more informative. (thanks to Brian Downing)
62 * bug fix: mutually referent alien structures now work correctly.
63 (reported by Rick Taube)
64 * bug fix: structures defined by WITH-ALIEN can be referred to
65 within other definitions in the same WITH-ALIEN.
66 * bug fix: division operators (MOD, TRUNCATE and the like) with
67 constant zero divisors and integer dividends no longer generate
69 * bug fix: provide default methods for INPUT-STREAM-P and
70 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
71 * bug fix: improve the detection and resolution of MOP metacycles
72 (where effective method computation depends on the generic function
73 having its effective method computed). (reported by Bruno Haible)
74 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
75 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
76 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
77 * bug fix: pathnames with relative directory components can be
78 represented relative to default pathnames.
79 * optimization: in taking the GCD of bignums, reduce the two bignums
80 to approximately the same size (using Euclid's algorithm) before
81 applying the more sophisticated binary GCD. (thanks to Juho
83 * optimization: COUNT on bitvectors now operates word-at-a-time.
84 * optimization: ASH with a positive, but not necessarily constant,
85 (leftwards) shift, when in a modular context, is compiled to a
87 * fixed some bugs revealed by Paul Dietz' test suite:
88 ** FORMAT variable parameters ("~V<char>") are defaulted properly
89 if the corresponding argument is NIL.
90 ** FORMAT directives accepting a minpad parameter treat negative
92 ** literal commas as character parameters to FORMAT directives are
94 ** literal spaces directly after ~<Newline> directives within a
95 format-logical-block (~:< ~@:>) do not induce :FILL-style
97 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
98 as specified: it no longer includes conditional newlines).
99 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
101 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
102 :PER-LINE-PREFIX argument does not evaluate to a string.
104 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
105 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
106 the manual for details; for now, package locks can be disabled by
107 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
108 if no major problems are found then it is likely that they will be
109 compiled in unconditionally.
110 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
111 unsupported operators on all platforms. To load a shared library
112 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
113 object file, link it into a shared library outside of SBCL and
114 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
116 * bug workaround: ROOM T and the GC were not working together
117 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
118 reporting facility aren't preserved. That reporting has been
119 disabled, so now until and unless someone figures out how to make it
120 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
121 * minor incompatible change: as threatened around sbcl-0.8.0, the
122 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
123 The symbols are also exported from SB-PCL for backwards
124 compatibility, but more so than before SB-PCL should be treated as
125 an implementation-internal package.
126 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
128 * the SB-SPROF contrib now works on (most) non-x86 architectures.
129 It is known as of this release not to work on the Alpha, however.
130 * fixed bug #167: errors signalled due to illegal syntax in method
131 bodies are now more legible.
132 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
133 designators and can hence be used with TYPEP.
134 * fixed bug #333: CHECK-TYPE now ensures that the type error
135 signalled, if any, has the right object to be accessed by
136 TYPE-ERROR-DATUM. (reported by Tony Martinez)
137 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
138 5.1.2.3 for multiple-value place subforms. (reported by Kalle
140 * fixed bug #334: programmatic addition of slots using specialized
141 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
142 and :ALLOCATION :CLASS slots.
143 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
144 float exponents. (rereported by Peter Seibel)
145 * fixed a bug: #\Space (and other whitespace characters) are no
146 longer considered to be macro characters in standard syntax by
148 * fixed bug: initialization of condition class metaobjects no longer
149 causes an instance of the condition to be created. (reported by
151 * fixed bug: it is now possible to have slots such that
152 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
153 description is neither :INSTANCE nor :CLASS.
154 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
155 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
156 SLOT-VALUE-USING-CLASS), and no longer causes errors with
157 non-standard SLOT-DEFINITION-ALLOCATION values.
158 * fixed bugs: various race conditions handling exiting threads.
159 CL-PPCRE's thread torture test now passes.
160 * fixed arguably-a-bug: GC time proportional to number of threads in
161 system even when most of them are idle
162 * optimization: improved performance of BIT and SBIT on bit-vectors.
163 * .fasl file incompatibility: The fasl file version number has
164 been incremented because of changes associated with package locks.
166 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
167 * minor incompatible change: the system no longer provides
168 optimization hints (conditions which are TYPEP
169 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
170 mode; these hints are still emitted when higher SPEED optimization
172 * new contrib module: a sampling profiler (profiling by statistical
173 sampling, rather than by instrumenting functions) is available as
174 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
175 * the behaviour of the standard function ED is now customizeable by
176 third parties through a hook variable: see ED's documentation
177 string for information on the protocol.
178 * the compiler no longer emits efficiency notes for (FUNCALL X)
179 when the type of X is uncertain under default optimization
181 * fixed bug 276: mutating a binding of a specialized parameter to a
182 method to something that is not TYPEP the specializer is now
184 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
185 exhibit the required behaviour on the x86 platform. (thanks to
186 Peter van Eynde, Eric Marsden and Bruno Haible)
187 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
188 even for difficult arguments. (reported by Peter Graves)
189 * fixed bug 141a: the backquote printer now descends quoted
191 * fixed another bug in backquote printing: no more destructive
192 modification of the form's list structure. (reported by Brian
194 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
195 move between its address being taken and the call to
196 interrupt_thread, fixing a crashing race condition.
197 * the SB-POSIX contrib implementation has been adjusted so that it
198 no longer exhibits ridiculously poor performance when constructing
199 instances corresponding to C structs.
201 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
202 * minor incompatible change: the sb-grovel contrib now treats C
203 structures as alien (in the sense of SB-ALIEN) objects rather than
204 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
205 has implications for memory management of client code
206 (sb-grovel-returned objects must now be manually managed) and for
207 type safety (alien objects now have full types).
208 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
209 used to control emission of compiler diagnostics, rather than the
210 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
211 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
212 quality should be considered deprecated.
213 * install.sh now installs the user manual as well
214 * (not quite a new documentable feature, but worth considering in
215 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
216 beginnings of a semantically meaningful condition hierarchy is
217 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
219 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
220 designator as the defaults argument.
221 * fixed bug: Displaced arrays whose displaced-to array has become
222 too small now cause ARRAY-DIMENSION to signal an error, providing
223 for safer bounds-checking. (reported by Bruno Haible)
224 * fixed bug: DEFCLASS slot definitions with identical :READER and
225 :WRITER names now signal a reasonable error. (reported by Thomas
227 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
229 * fixed bug: Invalid slot specification errors now print correctly.
230 (thanks to Zach Beane)
231 * fixed bug 320: Shared to local slot value transfers in class
232 redefinitions now happen corrently from superclasses as
233 well. (reported by Bruno Haible)
234 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
236 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
237 improper lists correctly. (reported by Bruno Haible)
238 * fixed bug 313: source-transform for <fun-name> was erroneously
239 applied to a call of a value of a variable with name <fun-name>.
240 (reported by Antonio Menezes Leitao)
241 * fixed bug 307: The obsolete instance protocol ensures that
242 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
243 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
244 remove dead unknown-values globs from the middle of the stack.
245 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
246 easier to resume long computations after using *BREAK-ON-SIGNALS*
247 to diagnose and fix failures. (thanks to Nikodemus Siivola)
248 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
249 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
250 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
251 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
252 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
254 * fixed a (fairly theoretical) bug in string printing: if
255 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
256 does not have array-element-type equal to the most general string
258 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
259 function name. (thanks to Nikodemus Siivola)
260 * fixed bug: compile-time format string checker failed on
261 non-closed ~{. (reported by Thomas F Burdick)
262 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
263 don't warn on legal loop code involving more than one aggregate
264 boolean. (thanks to Nikodemus Siivola)
265 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
266 in keyword position to MAKE-PACKAGE was not being reported as
268 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
269 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
271 * fixed bugs 280 and 312: the checking for multiple definitions in a
272 file is less likely to become confused by uses of inline
274 * fixed bug: the #S reader macro performs the keyword coercion
275 specified for slot names. (reported by Kalle Niemitalo)
276 * fixed bug: lambda lists may contain symbols whose names start with
277 & but are not lambda-list-keywords; their occurrence triggers a
279 * fixed bug 321: define-method-combination argument lambda lists do
280 not cause invalid code to be generated when &optional or &aux
281 variables have default values. (reported by Bruno Haible)
282 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
283 have CLOS classes; this appears also to have fixed reported
284 instability in stack exhaustion detection.
285 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
286 MUFFLE-WARNING when no associated restart is present is now
288 * optimization: rearranged the expansion of various defining macros
289 so that each expands into only one top-level form in a
290 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
292 * optimization: used a previously little-used slot in symbols to
293 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
295 * fixed some bugs revealed by Paul Dietz' test suite:
296 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
298 ** Fixed an optimization bug related to inheritance of initforms
299 from local to shared slots.
300 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
301 ** CAST optimizer forgot to flush argument derived type.
302 ** print/read consistency on floats is now orders of magnitude
303 more likely. (thanks also to Bruno Haible for a similar report
305 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
306 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
307 for objects of type REAL. Make it so.
308 ** SXHASH is specified (infelicitously) to respect similarity,
309 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
310 it so. (thanks to Markus Ziegler)
311 ** on the Alpha, the compiler succeeds in compiling functions
312 returning a known number of arguments greater than 63.
313 ** fixed handling of invalid NIL arguments in keyword position
315 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
317 ** provided a readably-printable representation for RANDOM-STATE
319 ** ensured that pathnames, where they have a namestring, always
320 print using #P"..." syntax.
322 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
323 * Support for the forthcoming 2.0 version of the NetBSD kernel
324 running on x86 hardware has been added. (thanks to Perry
325 E. Metzger most immediately, and others for their past work)
326 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
327 current release version; SBCL's OpenBSD support had been broken
328 since about the time of OpenBSD's switch to ELF binary format.)
329 * [placeholder for DX summary]
330 ** user code with &REST lists declared dynamic-extent, under high
331 speed or space and low safety and debug optimization policy.
332 * The manual has been converted to Texinfo format and the debugger
333 chapter from the cmucl manual has been added.
334 * A facility has been added to extract documentation strings from
335 sbcl and store them as Texinfo-formatted snippets for inclusion in
336 the manual (via Texinfo's @include directive)
337 * bug fix: compiler emitted division in optimized DEREF. (thanks for
338 the test case to Dave Roberts)
339 * bug fix: multidimensional simple arrays loaded from FASLs had fill
340 pointers. (reported by Sean Ross)
341 * bug fix: PROFILE output is printed nicely even for large numerical
342 values. (thanks to Zach Beane)
343 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
344 greater than 32 handle EOF correctly.
345 * bug fix: on X86 an immediate argument of the IMUL instruction is
346 correctly printed in disassembly. (thanks to Lutz Euler)
347 * bug fix: class slots in redefined classes preserve their old
348 values. (thanks to Bruno Haible and Nikodemus Siivola)
349 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
350 succeeds. (reported by Marco Baringer)
351 * fixed some bugs revealed by Paul Dietz' test suite:
352 ** READ-SEQUENCE now works on ECHO-STREAMs.
353 ** RATIONALIZE works more according to its specification. (thanks
356 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
357 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
358 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
359 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
360 should matter to you only if you rebind the printer control
361 variables and then find you want different bindings in the
362 debugger than in the ordinary execution of your program.)
363 * The runtime build system has been tweaked to support building
364 (on SPARC/SunOS) using a C compiler which invokes Sun's own
365 assembler and linker. (thanks to Nikodemus Siivola)
366 * Unbound, undefined, undeclared variables now trigger full
367 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
368 more often programmer error than deliberate exploitation of undefined
370 * optimization: the hash algorithm for strings has changed to one
371 that is less vulnerable to spurious collisions. (thanks to Juho
373 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
374 needless bounds checking. (thanks to Juho Snellman)
375 * optimization: implemented multiplication as a modular
376 (UNSIGNED-BYTE 32) operation on the PPC backend.
377 * fixed some bugs revealed by Paul Dietz' test suite:
378 ** ADJUST-ARRAY now copies the datum in a zero rank array if
380 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
381 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
382 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
384 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
386 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
388 ** ECHO-STREAMs no longer attempt to echo the end of file value to
389 their output stream on EOF from read.
390 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
391 have been read to end-of-file.
392 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
394 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
395 description of determination of which consecutive characters
397 ** Printing the "Space" character with escaping on now yields "#\\ ",
398 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
399 ** Reading floating-point numbers with *READ-BASE* set to a number
400 less than 10 works correctly.
401 ** Reading floating-point numbers with *READ-BASE* set to a number
402 more than 10 works correctly.
403 ** Printing with *PRINT-READABLY* targets the standard readtable, not
404 the readtable currently in effect.
406 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
407 * minor incompatible change: parsing of namestrings on a physical
408 (Unix) host has changed; numbers after the final #\. in a
409 namestring are no longer interpreted as a version field. This is
410 intented to be largely invisible to the user, except that the
411 meaning of the namestring "*.*.*" has changed: it now refers to a
412 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
413 should usually be replaced by
414 (make-pathname :name :wild :type :wild :version :wild)
415 with the added benefit that this is more likely to be portable.
416 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
417 option to OPEN now signals an error if the file being opened
418 exists; this may have an impact on existing code.
419 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
420 signals received, on the PowerPC platforms (both Linux and
421 Darwin). (thanks to Pierre Mai for pointing out the location of
423 * several fixes on OS X: The system now builds and runs cleanly on
424 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
425 (thanks to Brian Mastenbrook)
426 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
427 second-resolution: integer multiples of 1/3600 between -24 and 24.
428 (thanks to Vincent Arkesteijn)
429 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
430 type when called with 1 argument; PEEK-CHAR checked type of
431 PEEK-TYPE only after having read first character from a
432 stream. (reported by Peter Graves)
433 * bug fix: the garbage collector now has much better locality
434 behaviour, and in particular no longer treats all memory as being
435 exhausted when many small objects point to each other in a deeply
437 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
438 recognized as being TYPEP their class.
439 * bug fix: the PUSHNEW documentation string has been corrected.
440 (thanks to Vincent Arkesteijn)
441 * bug fix: defaulting of the value for the last of an atypically
442 large number of multiple values being bound was not being
443 performed correctly on the Alpha or PPC platforms
444 * optimization: implemented multiplication as a modular
445 (UNSIGNED-BYTE 32) operation on the x86 backend.
446 * optimization: SEARCH on simple-base-strings can now be open-coded.
447 (see also contrib/compiler-extras.lisp for inspiration for
448 teaching the compiler about the Boyer-Moore algorithm).
449 * value, returned by MAX (and MIN) called with several EQUALP, but
450 not EQL, arguments now does not depend on compiler settings.
451 * fixed some bugs revealed by Paul Dietz' test suite:
452 ** in stack analysis liveness information is propagated from
453 non-local entry points.
454 ** pathwise CAST removing failed when the CAST node did not start
456 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
457 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
459 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
460 host is already defined.
461 ** RENAME-FILE works on streams instead of signalling an internal
463 ** PEEK-CHAR uses the current readtable when determining whether
464 or not a character is whitespace.
465 ** MERGE-PATHNAMES handles the case when the pathname does not
466 specify a name while the default-pathname specifies a version.
467 ** Pathnames now stand a better chance of respecting print/read
469 ** Attempting to use standardized file system operators with a
470 pathname with invalid :DIRECTORY components signals a
472 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
474 ** DIRECTORY on logical pathnames is more correct.
475 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
476 signal a TYPE-ERROR if their argument is not a stream.
477 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
478 designator argument does not designate a stream.
479 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
480 examining the synonym.
481 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
483 ** OPEN and WITH-OPEN-STREAM allow opening streams with
484 element-type larger than ([UN]SIGNED-BYTE 32).
486 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
487 * When built with the :SB-FUTEX feature, threaded builds now take
488 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
489 for faster/more reliable mutex and condition variable support.
490 * Incompatible change (but one you probably shouldn't have been using
491 anyway): the interface and code for arbitrating between multiple
492 threads in the same user session has been redesigned.
493 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
494 error when called without an explicit environment argument.
495 (thanks to Nikodemus Siivola)
496 * bug fix: buffered :DIRECTION :IO streams are less likely to become
497 confused about their position. (thanks to Adam Warner and Gerd
499 * bug fix: Pretty printing backquoted forms with unquotations in the
500 argument list position of various code constructs such as LAMBDA
501 now works correctly. (reported by Paul Dietz)
502 * bug fix: Pretty printing unquotations no longer loses all
503 stream position information.
504 * optimization: performance of string output streams is now less
505 poor for multiple small sequence writes.
506 * optimization: performance of CSUBTYPEP in the presence of complex
507 expressions involving CONS and NOT many times has been improved.
508 (reported by Paul Dietz)
509 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
511 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
512 streams for the REPL, for the debugger, and so on. (thanks to
514 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
515 * fixed some bugs revealed by Paul Dietz' test suite:
516 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
517 is ignored if it is too small, rather than propagating through
518 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
519 ** extremely complex negations of CONS types were not being
520 sufficiently canonicalized, leading to inconsistencies in
522 ** VALUES tranformer lost derived type.
524 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
525 * fixed a bootstrapping bug: the build process no longer assumes
526 that the various BOOLE-related constants have the same value in
527 host and target lisps. (noted by Paul Dietz' test suite on an
528 SBCL binary built from CLISP)
529 * The system can now be dynamically linked on the MIPS platform,
530 which enables dynamic loading of foreign code from Lisp. (thanks
531 to Ralf Baechle for discussions on the MIPS ABI)
532 * The system now records debugging information for its own source
533 files in a filesystem-position-independent manner, relative to
534 the "SYS" logical host.
535 * fixed a compiler bug: MV-LET convertion did not check references
536 to the "max args" entry point. (reported by Brian Downing)
537 * tweaked disassembly notes to be less confident about proclaiming
538 some instruction as an LRA. (thanks to Brian Downing)
539 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
540 now each have their own history, command character, and other
541 characteristics. (thanks to David Lichteblau)
542 * fixed some bugs revealed by Paul Dietz' test suite:
543 ** compiler failure in compiling LOGAND expressions including a
545 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
546 shift greater than 32.
547 ** FLUSH-DEST did not mark blocks for type check regeneration.
548 ** HANDLER-CASE failed to accept declarations in handler clauses
549 in some circumstances.
551 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
552 * New code in contrib/sb-introspect (still probably not entirely
553 stable yet) provides some support for smart Lisp development
554 environments like SLIME.
555 * The conditions signalled for errors occurring when loading .fasl
556 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
557 in a way which should help ASDF recover gracefully.
558 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
559 stuff has been cleaned up. If you code contrib/ stuff, this might
560 affect you, and you can look at contrib/README, contrib/STANDARDS,
561 and/or the 0.8.4.27 diff to check.
562 * In full calls the compiler now does not generate checks for declared
563 argument types for all arguments.
564 * various threading fixes
565 ** and some experimental patches which didn't make it into
566 the main tree for this release, but which are shipped in
567 contrib/experimental-thread.patch as a possible fix for some
568 failures (deadlock, spinning...) in GC-intensive multithreaded
570 * fixed PPC build problem (source code incompatibility of different
571 library versions): added offsetof() hackery which attempts to divine
572 where glibc maintainers put uc_mcontext today
573 * fixed bug 282: compiler does not trust type assertions while passing
574 arguments to a full call.
575 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
576 &OPTIONAL VALUES type parameter.
577 * fix bug 214: algorithm for noting rejected templates is now more
578 similar to that of template seletion. (also reported by rydis on
580 * fixed bug 141b: printing backquoted information readably and prettily
581 inserts a space where necessary.
582 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
583 (CHARACTER 10) are now reported as errors, rather than propagated
584 as unknown types. (reported by piso on #lisp)
585 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
586 with values NIL and :ERROR. (thanks to Milan Zamazal)
587 * fixed bug 191c: CLOS now does proper keyword argument checking as
588 described in CLHS 7.6.5 and 7.6.5.1.
589 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
590 longer raise an error; further, using a list as a for-as-arithmetic
591 counter now raises a meaningful error.
592 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
593 now understood by sequence creation functions such as MAKE-SEQUENCE
595 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
596 a STRING-INPUT-STREAM.
597 * compiler enhancement: SIGNUM is now better able to derive the type
599 * type declarations inside WITH-SLOTS are checked. (reported by
601 * fixed some bugs revealed by Paul Dietz' test suite:
602 ** incorrect optimization of TRUNCATE for a positive first
603 argument and negative second.
604 ** compiler failure in let-convertion during flushing dead code.
605 ** compiler failure while deriving type of TRUNCATE on an
606 interval, containing 0.
607 ** ASH of a negative bignum by a negative bignum count now returns
609 ** intersection of CONS types now canonicalizes properly, fixing
610 inconsistencies in SUBTYPEP.
612 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
613 * incompatible change: The --disable-debugger command line
614 option now clobbers the debugger at a more fundamental
615 level, by redefining #'INVOKE-DEBUGGER instead of by
616 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
617 is specified by ANSI to ignore *DEBUGGER-HOOK* and
618 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
619 enter the debugger REPL and then suffer recursive errors
620 because *DEBUG-IO* is also messed up in --disable-debugger mode;
621 while under the new system, BREAK in --disable-debugger mode
622 terminates the system just as an unhandled error would.
623 * fixed compiler performance when processing loops with a step >1;
624 * bug fix: DOCUMENTATION now retrieves generic function
625 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
626 support has been systematized, and now supports the methods
627 specified by ANSI, along with a default method and a method for
628 slot documentation. (reported by Nathan Froyd)
629 * bug fix: effective methods associated with a generic function are
630 no longer cached over a change of that generic function's method
631 combination. (reported by Andreas Fuchs)
632 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
633 names a non-existent directory. (thanks to Andreas Fuchs)
634 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
635 the PPC platform, lead to stack corruption; this has been fixed.
636 (reported by Rainer Joswig)
637 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
638 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
639 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
640 arithmetic in general, is now much more reliable on the PPC
642 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
643 platform now returns the right answer.
644 * optimization: restored some effective method precomputation in
645 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
646 precomputation is now tunable.
647 * optimization: compiler-internal data structure use has been
648 reviewed, and changes have been made that should improve the
649 performance of the compiler by about 20%.
650 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
651 simple-base-strings and simple-bit-vectors is improved.
652 * optimization: the optimization of 32-bit logical and arithmetic
653 functions introduced in version 0.8.3 on the x86 has been
654 implemented on the mips, ppc and sparc platforms; an
655 implementation of the same facility, but for 64-bit arithmetic,
656 has been added for the alpha.
657 * microoptimization: the compiler is better able to make use of the
658 x86 LEA instruction for multiplication by constants.
659 * bug fix: in some situations compiler did not report usage of
660 generic arithmetic in (SPEED 3) policy.
661 * bug 145b fix: compiler used wrong type specifier while converting
662 MEMBER-types to numeric.
663 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
665 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
666 index argument now works properly on the Alpha platform.
667 * bug fix: floating point exception treatment on the Alpha platform
669 * bug fix: FILE-POSITION works much better on string input and
670 output streams. (thanks to Nikodemus Siivola)
671 * bug fix: many threading/garbage collection symptoms sorted.
672 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
673 might be pseudo-atomic.
674 * internal change: Stopping for GC is now done with signals not
675 ptrace. GC is now done in whichever thread wanted it, instead of
677 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
679 * bug fix: result form in DO is not contained in the implicit
681 * incompatible change: ICR structure is changed; the value part of
682 CONTINUATION is now called LVAR; corresponding functions are
683 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
684 * added type deriver for ISQRT (thanks to Robert E. Brown).
685 * bug fix: better support for loading from the command line when an
686 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
688 * fixed some bugs revealed by Paul Dietz' test suite:
689 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
690 ** ROUND and FROUND now give the right answer when given very
691 small float arguments.
692 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
694 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
695 ** (GCD 0 <negative-integer>) returned <negative-integer>.
696 ** LCM should return a non-negative integer.
697 ** PARSE-INTEGER returned the index of a terminator instead of the
698 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
699 ** PARSE-INTEGER returned an incorrect index being applied to a
701 ** LCM with two arguments of 0 returns 0 rather than signalling
703 ** unsigned addition of a 32-bit constant with the high bit set no
704 longer causes an internal compiler error on the x86.
705 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
706 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
707 with negative last argument.
708 ** byte specifiers with zero size and position no longer cause
709 an error during type derivation.
710 ** bignum multiplication on the Alpha platform now returns the
712 * porting: The system now builds on SuSE AMD64, although it still
713 generates a 32-bit binary.
714 * .fasl file incompatibility: The fasl file version number has
715 been incremented (because of the changes to internal compiler
716 data structures referred to above).
718 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
719 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
720 more accurately, on the Darwin kernel running on PowerPC hardware.
721 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
722 * Compiler code deletion notes now signal a condition of type
723 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
724 an associated MUFFLE-WARNING restart.
725 * The compiler now performs limited argument count validation of
726 constant format strings in FORMAT, and where appropriate in ERROR,
727 CERROR and WARN. (thanks to Gerd Moellmann)
728 * New ASDF-INSTALL contrib can be used for automatic download and
729 installation of third-party Lisp code from CCLAN or other sites
731 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
732 which forces another thread to execute a function supplied by the
734 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
735 now accept and act upon their :ELEMENT-TYPE keyword argument.
736 (reported by Martin Atzmueller, Edi Weitz)
737 * bug fix: FILE-POSITION now accepts position designators up to
738 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
739 is the greater. (thanks to Patrik Nordebo)
740 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
741 Kalle Olavi Niemitalo)
742 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
743 printable. (reported by Eric Marsden)
744 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
745 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
746 * x86 bug fix in control stack exhaustion checking: now shows backtrace
747 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
748 (thanks to Stig Sandoe)
749 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
750 UNREAD-CHAR now correctly includes the unread character in the
751 target sequence. (thanks to Gerd Moellmann)
752 * bug fix in threaded builds: the system can now be suspended and
753 resumed by shell job control with minimal disruption.
754 * bug fixes in times and timezones >2038AD
755 * better handling of "where is GNU make?" problem in build scripts
756 (thanks to Nikodemus Siivola)
757 * new optimization: inside a named function any reference to a
758 function with the same name is considered to be a self-reference;
759 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
760 optimization quality.
761 * new optimization on x86: logical functions and + now have
762 optimized (UNSIGNED-BYTE 32) versions, which are automatically
763 used when the result is truncated to 32 bits.
764 * VALUES declaration is partially enabled.
765 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
766 * fixed some bugs revealed by Paul Dietz' test suite:
767 ** The system now obeys the constraint imposed by
768 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
769 types form a lattice under type intersection.
770 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
771 ** ASSOC now ignores NIL elements in an alist.
772 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
773 and (1+ MOST-POSITIVE-FIXNUM) answers.
774 ** The addition of a method with invalid qualifiers to a generic
775 function does not cause an error to be signalled immediately;
776 a warning is signalled, and the error is generated only on
777 calling the generic function.
778 * changed .fasl file version number, in part to add type codes for
779 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
780 obscure ANSI requirements
782 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
783 * fixed bug 148: failure to inline-expand a local function left
784 garbage, confusing the compiler.
785 * fixed bugs 3cd: structure slot readers perform type check if the
786 slot can have an invalid value (i.e. it is either not initialized
787 or can be written with a less specific slot writer).
788 * bug fix: the compiler now traps array references to elements off
789 the end of an array; previously, the bounds checking in some
790 circumstances could go off-by-one.
791 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
793 * type declarations for array element types now obey the description
794 on the CLHS page "Declaration TYPE", as per discussions on
795 sbcl-help around 2003-05-08. This means that a declaration
796 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
797 declaration, all references to BAR will be asserted or assumed
798 (with THE, so dependent on compiler policy) to involve objects of
799 type FOO. Note that no such declaration is implied in
800 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
801 * declared types of functions from the "Conditions"
802 chapter. (reported by Paul Dietz)
803 * bug fix: CERROR accepts a function as its first argument.
804 * bug fix: NTH an NTHCDR accept a bignum as index
805 arguments. (reported by Adam Warner)
806 * optimization: character compare routines now optimize comparing
807 against a constant character. (reported by Gilbert Baumann)
808 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
809 argument now works properly on the MIPS platform.
810 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
811 * repeated evaluation of the same DEFSTRUCT, a slot of which is
812 declared to have a functional type, does not cause an error
814 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
815 many :DEFAULT-INITARGS. (reported by Istvan Marko)
816 * fixed bug: if last continuation of a deleted block has a
817 destination, this destination should be deleted too. (reported by
819 * fixed a bug in the bootstrap process: the host compiler's values
820 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
821 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
822 test case from Patrik Nordebo)
823 * improved the ability of the disassembler on the PPC platform to
824 provide helpful disassembly notes.
825 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
826 the class in more cases than previously.
827 * bug fix: FILE-POSITION now understands :START and :END for
828 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
829 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
830 8) data. (thanks to David Lichteblau)
831 * bug fix: it is possible to add a method to a generic function
833 * bug fix: reader failed to signal END-OF-FILE inside an
834 object representation. (reported by Nikodemus Siivola)
835 * fixed some bugs revealed by Paul Dietz' test suite:
836 ** LAST and [N]BUTLAST should accept a bignum.
837 ** condition slot accessors are methods.
838 ** (VECTOR NIL) is a subtype of STRING.
840 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
841 * minor incompatible change: some nonsensical specialized lambda
842 lists (used in DEFMETHOD) which were previously ignored now signal
844 * minor incompatible change: the system is now aware of the types of
845 variables in the COMMON-LISP package, and will signal errors for
846 most violations of these type constraints (where previously they
847 were silently accepted).
848 * minor incompatible change: COMPILE-FILE now uses the freedom
849 afforded (ANSI 3.2.2.3) to use derived function types for
850 functions defined in the same file. This also permits the system
851 to warn on static type mismatches and function
852 redefinition. (Currently it does not work with high DEBUG level.)
853 * minor incompatible change: VALUES declaration is disabled.
854 * When issuing notes, the compiler now signals a condition of type
855 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
856 restart for use in user handlers. It is expected that the
857 COMPILER-NOTE condition will eventually become a condition
858 supertype to a hierarchy of note types, which will then be
859 handleable in a similar fashion. However, at the moment, no such
860 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
861 but it's an implementation detail, not a classification for the
863 * Changes in type checking closed the following bugs:
864 ** type checking of unused values (192b, 194d, 203);
865 ** template selection based on unsafe type assertions (192c, 236);
866 ** type checking in branches (194bc).
867 * A short form of VALUES type specifier has ANSI meaning (it has
868 increased the number of situations when SBCL cannot perform type
870 * fixed bug in DEFSTRUCT: once again, naming structure slots with
871 keywords or constants is permissible.
872 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
873 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
874 classes. (thanks to Antonio Martinez)
875 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
876 outputs FRESH-LINE or TERPRI, and no longer converts its stream
877 argument to a pretty-print stream. Instead, it leaves any such
878 operations to DESCRIBE-OBJECT methods.
879 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
881 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
882 lambda lists. (thanks to David Lichteblau)
883 * bug fix: Defining a generic function with a :METHOD-CLASS being a
884 subclass of STANDARD-METHOD no longer causes stack exhaustion.
885 (thanks to Gerd Moellmann)
886 * fixed bug 246: increased compilation speed of long
887 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
889 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
891 * DEFCONSTANT now throws a condition of type
892 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
893 constant to a non-EQL value; CONTINUE and ABORT restarts
894 respectively change and preserve the value.
895 * fixed bug 63: The code walker, part of the implementation of CLOS,
896 is now better at handling symbol macros.
897 * bug fix: There is no longer an internal implementation type named
898 CL:LENGTH. (reported by Raymond Toy)
899 * bug fix: In macro-like defining macros/special operators the
900 implicit block does not enclose the lambda list.
901 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
902 enclosed in parentheses) are not suitable as type specifiers, and
903 their use properly signals an error now.
904 * bug fix: An argument count mismatch for a type specifier in code
905 being compiled no longer causes an unhandled error at compile
906 time, but signals a compile-time warning.
907 * fixed simple vector readable printing
908 * bug fix: DESCRIBE takes more care over whether the class
909 precedence list slot of a class is bound before accessing it.
910 (reported by Markus Krummenacker)
911 * bug fix: FORMATTER can successfully compile pretty-printer format
912 strings which use variants of the ~* directive inside.
913 * bug fix: SEARCH now applies its TEST predicate to the elements of
914 the arguments in the correct order. (thanks to Wolfhard Buss)
915 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
916 in type inference. (thanks to Robert E. Brown)
917 * bug fix: Reading in symbols with an explicit package name of ""
918 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
919 * fixed some bugs revealed by Paul Dietz' test suite:
920 ** NIL is now allowed as a structure slot name.
921 ** Arbitrary numbers, not just REALs, are allowed in certain
922 circumstances in LOOP for-as-arithmetic clauses.
923 ** Multiple class redefinitions before slot access no longer
925 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
926 association between the name and a class.
927 ** Generic functions with non-standard method-combination and over
928 five methods all of which return constants no longer return NIL
929 after the first few invocations. (thanks to Gerd Moellmann)
930 ** CALL-NEXT-METHOD with no arguments now passes the original
931 values of the arguments, even in the presence of assignment.
932 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
934 ** DISASSEMBLE works with closures and funcallable instances.
935 ** ADD-METHOD now returns the generic function, not the new method.
936 ** FIND-METHOD signals an error if the lengths of the specializers
937 is incompatible with the generic function, even if the ERRORP
939 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
940 which its argument is a member.
941 ** DEFCLASS only redefines the class named by its class-name
942 argument if that name is the proper name of the class;
943 otherwise, it creates a new class.
944 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
945 of the UNBOUND-SLOT condition to the name of the slot.
946 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
947 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
948 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
949 SLOT-MAKUNBOUND in the specified fashion.
951 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
952 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
953 cross-compilation host. As a consequence, we can now bootstrap our
954 way up to SBCL starting with a bare gcc toolchain and human-readable
955 source code (first the source to CLISP, then the source to SBCL).
956 * A contributed module containing a partial implementation of the
957 simple-streams interface has been included. (thanks to Rudi
959 * A contributed module implementing the RFC1321 Message Digest
960 Algorithm, known as MD5, has been included.
961 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
962 no longer has any effect, as the code controlled by this feature
963 has been deleted. (As far as we know, no-one has ever built using
964 this feature, and its semantics were confused in any case).
965 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
966 work on conditions (as required by the ANSI specification),
967 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
968 expected behaviour on conditions. Users should note, however,
969 that such behaviour is not required by the ANSI specification,
970 and so use of this behaviour may render their code unportable.
971 * fixed some bugs revealed by Paul Dietz' test suite:
972 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
974 ** &ENVIRONMENT parameter in macro lambda list is bound first.
975 ** SXHASH on condition objects no longer returns NIL.
976 ** :ALLOCATION :CLASS slots are better treated; their values are
977 updated on class redefinition, and initforms inherited from
978 superclasses are applied.
979 ** REMOVE-METHOD returns its generic function argument even when
980 no method was removed.
981 ** SHARED-INITIALIZE now initializes the values of the requested
982 slots, including those with :ALLOCATION :CLASS.
983 ** ALLOCATE-INSTANCE now works on structure classes defined via
984 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
986 ** SLOT-EXISTS-P now works on conditions, as well as structures
988 ** MAKE-LOAD-FORM now has the required methods on
989 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
990 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
991 keyword, and now implements the SLOT-NAMES argument.
992 ** methods with &OPTIONAL arguments no longer allow too many
993 arguments to be passed in the call without error.
994 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
995 option is consistent with the required arguments of the generic
996 function lambda list.
997 * bug fix: REQUIRE accepts a string designator. (Thanks to
999 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1000 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1001 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1002 * bug fix: The compiler now checks for duplicated variables in macro
1004 * bug fix: SETQ on globals returns the correct value.
1005 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1006 not a COMPILER-ERROR (followed by some other strange error on
1007 choosing the CONTINUE restart).
1008 * bug fix: make.sh and friends are now more consistent in the way that
1009 they look for GNU "make".
1011 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1012 * experimental native threads support (on x86 Linux >=2.4 only).
1013 This is not compiled in by default: you need to add :SB-THREAD to
1014 the target features. See the "Beyond ANSI" chapter of the manual
1016 * fix for longstanding nonANSIism: The old distinction between
1017 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1018 The return value from CL:FIND-CLASS is now a CLOS class, and
1019 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1020 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1021 * An interface to the MetaObject Protocol, as described in Kiczales,
1022 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1023 MIT Press, 1991, is available from the SB-MOP package.
1024 * incompatible change: the SB-PCL package should now be considered
1025 a private implementation detail, and no longer a semi-private MOP
1027 * minor incompatible change: due to rearrangement for threads, the
1028 control stack and binding stack are now allocated at arbitrary
1029 addresses instead of being hardcoded per-port. Users affected by
1030 this probably have to be doing advanced things with shared
1031 libraries, and will know who they are.
1032 * minor incompatible change: Previously, all --eval forms used were
1033 processed with READ before any of them were processed with EVAL.
1034 Now each --eval form is processed with both READ and EVAL before
1035 the next --eval form is processed. (Thus package operations like
1036 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1037 work as the user might reasonably expect.)
1038 * minor incompatible change: *STANDARD-INPUT* is now only an
1039 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1041 * minor incompatible change: Y-OR-N-P is now character-oriented,
1042 not line oriented. Also, YES-OR-NO-P now works without errors.
1043 (thanks to Antonio Martinez)
1044 * sb-aclrepl module improvements: an integrated inspector, added
1045 repl features, and a bug fix to :trace command.
1046 * Known functions, which cannot be open coded by the backend, are
1047 considered to be able to check types of their arguments. (fixing
1048 a bug report by Nathan J. Froyd)
1049 * fixed a bug in computing method discriminating functions: It is
1050 now possible to define methods specialized on classes which have
1051 forward-referenced superclasses. (thanks to Gerd Moellmann)
1052 * fixed evaluation order in optional entries (reported by Gilbert
1054 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1055 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1056 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1057 required-by-AMOP NAME argument, as well as CLASS and
1058 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1059 * fixed bug 20: DEFMETHOD can define methods using names that are
1060 not the proper names of classes to designate class specializers.
1061 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1062 Unix file descriptors, instead of blowing up. (thanks to Antonio
1064 * Garbage collection refactoring: user-visible change is that a
1065 call to the GC function during WITHOUT-GCING will not do garbage
1066 collection until the end of the WITHOUT-GCING. If you were doing
1067 this you were probably losing anyway.
1068 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1069 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1070 TYPEP the latter but not the former.
1071 * The compiler issues a full WARNING for calls to undefined functions
1072 with names from the CL package.
1073 * MAP-INTO for a vector destination is open coded. (reported by
1074 Brian Downing on c.l.l)
1075 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1076 documentation string.
1077 * fixed some bugs revealed by Paul Dietz' test suite:
1078 ** COPY-ALIST now signals an error if its argument is a dotted
1080 ** Condition slots are now accessed more correctly in the presence
1081 of multiple initargs for a given slot.
1082 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1083 exclude restarts of the same name associated with a different
1085 ** DEFCLASS of forward-referenced classes with another
1086 forward-referenced class in the superclasses list no longer
1088 ** Condition slots are now initialized once each, not multiple
1089 times. (thanks to Gerd Moellmann)
1090 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1092 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1093 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1094 arguments contain duplicated elements.
1095 ** RESTART-CASE understands local macros.
1096 ** RESTART-CASE associates exactly its own restarts with a condition.
1097 ** ENDP in safe mode checks its argument to be of type LIST.
1098 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1099 in question is unbound.
1100 ** Optimized MAKE-INSTANCE functions no longer cause internal
1101 assertion failures in the presence of duplicate initargs.
1102 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1103 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1104 propagated even in the ctor optimized implementation of
1106 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1108 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1109 * a better implementation of SXHASH on (simple) bit vectors,
1110 measured both in execution speed and in distribution of results
1111 over the positive fixnums, has been installed. Likewise, a better
1112 implementation of EQUAL for simple bit vectors is now available.
1113 * fixed CEILING optimization for a divisor of form 2^k.
1114 * fixed bug 240 (emitting extra style warnings "using the lexical
1115 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1116 by Antonio Martinez)
1117 * fixed SXHASH, giving different results for NIL depending on type
1118 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1119 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1120 variable. (found by Rolf Wester)
1121 * SBCL does not ignore type declarations for special
1122 variables. (reported by rif on c.l.l 2003-03-05)
1123 * some bug fixes in contrib/sb-aclrepl/
1124 * fixed some bugs revealed by Paul Dietz' test suite:
1125 ** a bug in the CONS type specifier, whereby the CAR and CDR
1126 types got intertwined, has been fixed;
1127 ** the type system is now able to reason about the interaction
1128 between INTEGER and RATIO types more completely;
1129 ** APPEND, [N]REVERSE and NRECONC check that those their
1130 arguments, which must be proper lists, are really so;
1131 ** An array specialized to be unable to hold elements has been
1132 implemented, as required -- yes, really -- by ANSI;
1133 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1134 on malformed property lists;
1136 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1137 * incompatible packaging change: in line with Unix convention,
1138 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1139 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1140 * REQUIRE and PROVIDE are now optionally capable of doing something
1141 useful. See the documentation string for REQUIRE.
1142 * infrastructure for a managed SBCL contrib system: contributed
1143 modules in this release include:
1144 ** the ASDF system definition facility;
1145 ** an interface to the BSD Sockets API;
1146 ** an ACL-like convenience interface to the repl;
1147 (thanks to Kevin Rosenberg)
1148 ** an implementation of ROTATE-BYTE, with efficient implementation
1150 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1151 gives a closer approximation to the right answer than previously.
1152 (thanks to Raymond Toy)
1153 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1154 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1155 optional environment arguments, as required by ANSI.
1156 * fixed bugs in other functions taking environment objects, allowing
1157 calls with an explicit NIL environment argument to be compiled
1159 * fixed bug 228: primary return values from
1160 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1161 COMPILE or FUNCTION.
1162 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1163 :TYPE will no longer signal errors on innocuous objects.
1164 * fixed bug 231b: SETQ is better at respecting type declarations in
1165 the lexical environment.
1166 * fixed a bug in DEFCLASS: classes named by symbols with no or
1167 unprintable packages can now be defined.
1168 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1169 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1170 * fixed compiler failure related to checking types of functions.
1171 (reported by Robert E. Brown)
1172 * the compiler is now much more consistent in its error-checking
1173 treatment of bounding index arguments to sequence functions: in
1174 (SAFETY 3) code, errors will be signalled in almost all cases if
1175 invalid sequence bounding indices are passed to functions defined
1176 by ANSI to operate on sequences.
1177 * fixed a bug in the build procedure: documentation of SBCL-specific
1178 packages is now preserved and available in the final Lisp image.
1179 * lifted FDEFINITION lookup out of loops in the implementation of
1180 many list operations. (thanks to Robert E. Brown)
1181 * fixed a bug in the reader: the #n# reader macro now works for
1182 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1183 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1184 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1185 E. Brown and Tony Martinez respectively)
1186 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1187 SLOT-BOUNDP: in particular, it is now able to optimize them much
1188 better, and is now not vulnerable to having packages renamed.
1189 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1190 always called when a slot is not present in an instance. (thanks
1192 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1193 longer signal an error on unreading a character following EOF on
1194 the previous constituent stream. (thanks to Tony Martinez)
1195 * fixed some bugs revealed by Paul Dietz' test suite:
1196 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1197 not just nonnegative fixnums;
1198 ** the logical bit-array operators such as BIT-AND now accept an
1199 explicit NIL for their "opt-arg" argument (to indicate a
1200 freshly-consed result bit-array);
1201 ** ELT now signals an error on an invalid sequence index in safe
1203 ** the type system is now cleverer about negations of numeric
1204 types, and consequently understands the BIGNUM and RATIO types
1206 ** the type system is now cleverer about the interaction between
1207 INTEGER and RATIO types: while bugs still remain, many more
1208 cases are accurately computed;
1209 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1210 if it is in the last clause;
1211 ** CONSTANTLY now correctly returns a side-effect-free function in
1213 ** DECLARE is no longer treated as a special-operator; in
1214 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1215 * incremented fasl file version number due to the change in the
1216 DEFSTRUCT-SLOT-DESCRIPTION structure.
1218 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1219 * minor incompatible change: code processed by the "interpreter" or
1220 EVAL now has a compilation optimization policy of (DEBUG 2)
1221 (changed from (DEBUG 1)) to improve debuggability of interactive
1222 development, and to allow the use of the debug RETURN command in
1224 * an experimental implementation of the RETURN command for the
1225 debugger has been included. (thanks to Frederik Kuivinen)
1226 * fixed bug 62: constraints were not propagated into a loop.
1227 * fixed bug in embedded calls of SORT (reported and investigated by
1229 * fixed some bugs revealed by Paul Dietz' test suite:
1230 ** printing and reading of arrays with some dimensions having
1231 length 0 (thanks to Gerd Moellmann);
1232 ** BOA constructor with &AUX argument without a default value does
1233 not cause a type error;
1234 ** CONSTANTP now returns true for all self-evaluating objects.
1236 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1237 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1238 accessors that are related by inheritance, as specified in the
1239 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1240 Valtteri Vuorikoski)
1241 * The compiler is now able to inline functions that were defined in
1242 a complex lexical environment (e.g. inside a MACROLET).
1243 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1245 * Support for the upcoming FreeBSD-5.0 release has been included.
1246 (thanks to Dag-Erling Smorgrav)
1247 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1248 effect when it is not in a toplevel context.
1249 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1250 stand a better chance of being correct. (thanks to Gerd
1252 * fixed bug in COERCE, which now signals an error on coercing a
1253 rational to a bounded real type which excludes the expected
1255 * The compiler is now able to derive types more accurately from the
1256 COERCE and COMPILE functions.
1257 * fixed bug 223: functional binding is considered to be constant
1258 only for symbols in the CL package.
1259 * fixed bug 231: SETQ did not check the type of a variable being set
1260 (reported by Robert E. Brown)
1261 * A new optimization for MAKE-INSTANCE has been included, fixing
1262 various bugs (including relating to :ALLOCATION :CLASS slots and
1263 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1265 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1266 ** As required by ANSI, LOOP now disallows anonymous collection
1267 clauses such as COLLECT I in conjunction with aggregate boolean
1268 clauses such as THEREIS (= I 1);
1269 ** LOOP now signals an error when any variable is reused in the
1270 same loop (including the potentially useful construct analogous
1271 to WITH A = 1 WITH A = (1+ A);
1272 ** IT is only a special loop symbol within the first clause of a
1273 conditional loop clause;
1274 ** LOOP with a typed iteration variable over a hashtable now
1275 signals a type error iff it should.
1276 * fixed some other bugs revealed by Paul Dietz' test suite:
1277 ** FILE-STREAM now names the class previously known as FD-STREAM;
1278 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1279 argument) no longer signals an error;
1280 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1281 of the slot symbol, rather than using the current package
1282 ((:CONC-NAME "") continues to intern the slot's name in the
1284 * incremented fasl file version number, because of the incompatible
1285 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1286 of the new implementation of DEFINE-COMPILER-MACRO.
1288 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1289 * Support for building SBCL for MIPS platforms running in
1290 little-endian mode has now been checked in, and basic
1291 functionality on said platforms verified.
1292 * minor incompatible change: PCL now records the pathname of a file
1293 in which methods and the like are defined, rather than its
1295 * minor incompatible change: TRUENAME now considers the truename of
1296 a file naming a directory to be the pathname with :DIRECTORY
1297 component indicating that directory.
1298 * minor incompatible change: a NAMED clause in the extended form of
1299 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1300 reason for the previous behaviour is unclear.
1301 * more systematization and improvement of CLOS and MOP conformance
1302 in PCL (thanks to Gerd Moellman and Pierre Mai):
1303 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1305 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1306 primary methods with no specializers;
1307 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1309 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1310 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1311 FORWARD-REFERENCED-CLASSes; error reporting on
1312 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1314 ** SXHASH on CLOS instances now uses a slot internal to the
1315 instance to return different numbers on distinct instances,
1316 while preserving the same return value through invocations of
1318 ** DEFMETHOD signals errors when methods with longer incongruent
1319 lambda lists are added to generic functions;
1320 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1321 CLASS, as specified in AMOP;
1322 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1323 on the order returned by the primary method for classes of
1324 class STANDARD-CLASS;
1325 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1326 * fixed some bugs shown by Paul Dietz' test suite:
1327 ** DOLIST puts its body in TAGBODY;
1328 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1330 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1331 value producing form;
1332 ** if more variables are given to PROGV than values, extra
1333 variables are bound and made to have no value;
1334 ** NSUBSTITUTE on list arguments gets the right answer with
1336 ** ELT signals an error of type TYPE-ERROR when the index argument
1337 is not a valid sequence index;
1338 ** LOOP signals (at macroexpansion time) an error of type
1339 PROGRAM-ERROR when duplicate variable names are found;
1340 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1341 ** FUNCALL of special-operators now cause an error of type
1343 ** PSETQ now works as required in the presence of side-effecting
1344 symbol-macro places;
1345 ** NCONC accepts any object as its last argument;
1346 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1348 ** loop-for-as-package does not require a package to be explicitely
1350 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1352 * fixed bug 166: compiler preserves "there is a way to go"
1353 invariant when deleting code.
1354 * fixed bug 172: macro lambda lists with required arguments after
1355 &REST arguments now cause an error to be signalled. (thanks to
1357 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1358 bug. (thanks to Matthew Danish)
1359 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1361 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1362 arguments are lexically rebound. (thanks to Gerd Moellmann and
1364 * fixed bug 194: error messages are now more informative when there
1365 is no primary method applicable in a call to a generic
1366 function. (thanks to Gerd Moellmann)
1367 * fixed bug in command line argument checking (thanks to Julian
1369 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1370 (thanks to Matthew Danish)
1371 * incremented fasl file version number, because of the
1372 SXHASH-related changes in the layout of CLOS data structures
1374 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1375 * minor incompatible change: The runtime (the Unix executable named
1376 "sbcl") is now much pickier about the .core files it will load.
1377 Essentially it now requires .core files to descend from the same
1378 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1379 as the runtime does. (The intent is to prevent the crashes which
1380 can occur, and which can even be reported as mysterious failures,
1381 when people patch the sources or change the build parameters
1382 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1383 sbcl and .core files.)
1384 * fixed bug: VALUES-LIST is no longer optimized away.
1385 * fixed bug 142: The FFI conversion of C string values to Lisp
1386 string values no longer conses excessively. (thanks to Nathan
1387 Froyd porting Raymond Toy's fix to CMU CL)
1388 * began to systematize and improve MOP conformance in PCL (thanks to
1389 Nathan Froyd, Gerd Moellman and Pierre Mai):
1390 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1392 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1393 ** FINALIZE-INHERITANCE is now called on class finalization;
1394 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1395 argument precedence order.
1396 * fixed bug 202: The compiler no longer fails on functions whose
1397 derived types contradict their declared type.
1398 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1399 so it can be non-toplevel.
1400 * The fasl file version number has changed (because of the new
1401 implementation of DEFMACRO).
1402 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1403 safe code, that any length requirement by their type-specifier
1404 argument is valid. The exceptions to this are described in bug
1406 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1407 functions now check the types of their inputs as required by ANSI.
1408 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1409 introduced symbol is DECLAREd to be SPECIAL.
1410 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1411 * fixed bug: :COUNT argument to sequence functions may be negative
1412 * fixed bug: body of DO-SYMBOLS may contain declarations
1413 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1414 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1415 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1416 symbol macro only once
1417 * fixed printing of call frame when argument list is unavailable
1418 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1419 * compiler no longer signals WARNING on unknown keyword
1422 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1423 * A beta-quality port to the mips architecture running Linux,
1424 based on the old CMUCL backend, has been made. It has been tested
1425 on a big-endian kernel, and works sufficiently well to be able to
1426 rebuild itself; it has not been tested in little-endian mode.
1427 * fixed an inconsistency between gencgc.c and purify.c which made
1428 dumping/loading .core files unreliable
1429 * fixed bug 120a: The compiler now deals correctly with IFs where
1430 the consequent is the same as the alternative, instead of
1431 misderiving the return type. (thanks to Alexey Dejneka)
1432 * fixed bug 113: Logical pathnames are now dumpable (the logical
1433 host is resolved at load-time, throwing an error if it is not
1435 * fixed bug 174: FORMAT's error message is slightly clearer when a
1436 non-printing character is used in a format directive.
1437 * fixed several bugs in compiler checking of type declarations, i.e.
1438 violations of the Python "declarations are assertions" principle
1439 (thanks to Alexey Dejneka)
1440 * fixed several bugs in PCL's error checking (thanks to Gerd
1442 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1444 * fixed bug in compilation of functions as first class values
1445 (thanks to Antonio Martinez-Shotton)
1446 * The compiler's handling TYPE-ERRORs which it can prove will
1447 inevitably happen at runtime has been cleaned up and corrected
1448 in several ways. (thanks to Alexey Dejneka)
1449 * improved argument type checking for various basic arithmetic
1450 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1451 lavished on them in the past that they can be compiled in many
1452 ways in different special cases
1453 * fixed bug 181: compiler checks validity of user supplied type
1455 * cleaned up code flushing in optimization: Function calls which
1456 should signal errors for safety purposes (e.g. which ANSI says
1457 should signal errors when their arguments are of incorrect type)
1458 are no longer optimized away.
1459 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1460 * incremented fasl file version number, because changes in the
1461 implementation of sequence functions like COERCE caused
1462 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1463 in old inline expansions) to become undefined. (Actually these
1464 changes were later undone, so we might very well be binary
1465 compatible with 0.7.7 after all, but leaving the version number
1466 incremented seemed like the simplest and most conservative
1469 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1470 * An alpha-quality port to the parisc architecture running Linux,
1471 based on the old CMUCL backend, has been made. This, even more so
1472 than the other backends, should be considered still a work in
1473 progress; known problems include that the Linux kernel in 64-bit
1474 mode does not propagate the correct sigcontext structure to
1475 userspace, and consequently SBCL on a parisc64 kernel will not
1477 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1478 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1479 suppressed.) Also now that the compiler is looking at declarations
1480 in the environment, it checks optimization declarations as well,
1481 and suppresses inlining when (> DEBUG SPEED).
1482 * More fixes have been made to treatment of floating point exception
1483 treatment and other Unix signals. In particular, floating point
1484 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1485 * The detection and handling of control stack exhaustion (infinite
1486 or very deeply nested recursion) has changed. Stack exhaustion
1487 detection is now done by write-protecting pages at the OS level
1488 and applies at all optimization settings; when found, a
1489 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1490 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1491 be caught using IGNORE-ERRORS.
1492 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1493 correctly when transforming recursive LABELS functions to LETs.
1494 (thanks to Alexey Dejneka)
1495 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1496 that are names of constants or global variables.
1497 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1498 alien routines with docstrings.
1499 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1500 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1502 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1503 (thanks to Eric Marsden)
1504 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1505 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1506 * Bug fix: The SPARC backend can now compile functions involving
1507 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1508 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1509 to a functions expecting a PATHNAME-DESIGNATOR.
1510 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1511 lambda lists. (thanks to Alexey Dejneka)
1512 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1513 functions so that EXPORTing the name of the class doesn't cause
1514 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1515 to Antonio Martinez for reporting this)
1516 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1517 constant-folded without failing an assertion. (thanks to Einar
1518 Floystad Dorum for reporting this)
1519 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1520 types are now dealt with more correctly by the compiler.
1521 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1522 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1523 behaviour analogous to RENAME-FILE. This puts its behaviour more
1524 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1526 * The fasl file version number has changed again. (because of the
1527 bug fix involving the names of PCL MAKE-INSTANCE functions)
1529 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1530 * bug fix: Floating point exceptions are treated much more
1531 consistently on the x86/Linux and PPC/Linux platforms.
1532 * Array initialization with :INITIAL-ELEMENT is now much faster for
1533 cases when the compiler cannot open code the array creation, but
1534 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1535 array accesses have also seen a speed increase.
1536 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1537 specified. (This was at the root of some bad interactions between
1538 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1540 * bug fix: Internal error arguments for undefined functions are now
1541 computed correctly on the PPC/Linux platform.
1542 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1543 Raymond Toy's patch for CMU CL)
1544 * Support for the Solaris 9 operating environment has been included
1545 (thanks to Daniel Merritt)
1546 * A very ugly but hopefully complete draft of the missing FFI chapter
1547 of the manual has been created by reformatting the corresponding
1548 CMU CL manual chapter into (currently very ugly and incoherent)
1549 DocBook and bringing it up to date for SBCL behavior. Thus, the
1550 manual is now essentially complete, at least by my extreme
1551 once-and-only-once standards, whereby it's acceptable to refer to
1552 the doc strings of SB-EXT functions as the primary documentation.
1553 * The fasl file version number has changed again, due to cleanup of
1554 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1555 is no longer a static symbol.)
1557 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1558 * SBCL now builds with OpenMCL (version 0.12) as the
1559 cross-compilation host; also, more progress has been made toward
1560 bootstrapping under CLISP.
1561 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1563 * bug 158 fixed: The compiler can now deal with integer loop
1564 increments different from 1; fixing this turned out also to fix
1566 * bug 169 fixed: no more bogus warnings about using lexical bindings
1567 despite the presence of perfectly good SPECIAL declarations (thanks
1568 to David Lichteblau)
1569 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1570 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1571 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1573 * bug 180 fixed: Method combination specifications no longer ignore
1574 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1575 * bug fix: Structure type predicate functions now check their argument
1576 count as they should.
1577 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1578 correctly. (thanks to Pierre Mai)
1579 * minor incompatible change: The --noprogrammer option is deprecated
1580 in favor of the new --disable-debugger option, which is very similar.
1581 (The major difference is that it takes effect at a slightly different
1582 time at startup, causing handling of errors in --sysinit and
1583 --userinit files will be affected differently.) The
1584 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1585 been added to allow this functionality to be controlled from ordinary
1586 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1587 maintainers, who might want to run non-interactive scripts to
1588 build SBCL cores which will later be used interactively.)
1589 * minor incompatible change: The LOAD function no longer, when given
1590 a wild pathname to load, loads all files matching that pathname.
1591 Instead, an error of type FILE-ERROR is signalled.
1593 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1594 * bug 147 fixed: The compiler preserves its block link/count
1595 invariants more correctly now so that it doesn't crash. (thanks
1597 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1599 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1600 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1601 no-value-for-this-character as NIL (as specified by ANSI).
1602 * HOST-NAMESTRING on physical pathnames now returns a string that is
1603 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1604 (thanks to Christophe Rhodes)
1605 * The Alpha port handles icache flushing more correctly. (thanks to
1607 * More progress has been made toward bootstrapping under CLISP. (thanks
1608 to Christophe Rhodes)
1609 * The fasl file format has changed again, because dynamic loading
1610 on OpenBSD (which has non-ELF object files) motivated some cleanups
1611 in the way that foreign symbols are transformed and passed around.
1612 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1613 is no longer treated as whitespace by the reader, but instead as
1614 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1615 |A
\7fB|, instead of A as it used to.
1617 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1618 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1619 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1620 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1621 this as of 0.7.1.45, but was left out of the previous news section
1622 (thanks to Dan Barlow)
1623 * SBCL now runs on the Solaris operating system on SPARC architectures
1624 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1625 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1626 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1627 * SPARC backend cleanups, allowing builds of cores optimized for V8
1628 and V9 SPARCS, and also emission of code targeted to a particular
1629 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1631 * SBCL is closer to bootstrapping under CLISP, thanks to various
1632 fixes by Christophe Rhodes.
1633 * The fasl file format has changed again, to allow the compiler's
1634 INFO database to support symbol macros.
1635 * The user manual (in doc/) is formatted into HTML more nicely.
1636 (thanks to coreythomas)
1637 * The system is smarter about SUBTYPEP relationships, especially
1638 those involving NOT types (including types such as ATOM which are
1639 represented internally using NOT types). Thus SUBTYPEP is less
1640 likely to return (VALUES NIL NIL) in general, and in particular
1641 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1643 * The fasl file format has changed again, because the internal
1644 representation of types now includes a new slot to support the new
1645 SUBTYPEP-of-NOT-types logic.
1646 * (not a change in the main branch of SBCL, but a related prototype
1647 which can hopefully be merged into the main branch of SBCL in the
1648 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1649 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1650 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1651 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1652 dispatch macro character combinations. (thanks to Alexey Dejneka)
1653 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1654 that superficially look like logical namestrings correctly.
1655 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1657 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1658 default supplied-p) for &optional and &key arguments. (thanks to
1661 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1662 * incompatible change: The compiler is now less aggressive about
1663 tail call optimization, doing it only when (> SPACE DEBUG) or
1664 (> SPEED DEBUG). (This is an incompatible change because there are
1665 programs which relied on the old CMU-CL-style behavior to optimize
1666 away their unbounded recursion which will now die of stack overflow.)
1667 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1668 for non-GENCGC systems has been increased to 20M (since that
1669 seems much closer to the likely performance optimum for modern
1670 systems than the old 4M value was)
1671 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1672 and *DEBUG-PRINT-LEVEL*
1673 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1674 of CMU CL's support for SPARC, and various endianness and other
1675 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1676 * new syntactic sugar for the Unix command line: --load foo.bar is now
1677 an alternate notation for --eval '(load "foo.bar")'.
1679 ** The system now detects stack overflow and handles it gracefully,
1680 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1681 optimization settings. (This is a good thing in general, and
1682 its introduction in this version should be particularly timely
1683 for anyone whose code fails because of suppression of tail
1685 ** The system now hunts for the C variable "environ" in a more
1686 devious way, to avoid segfaults when the C library version differs
1687 between compile time and run time. (thanks to Christophe Rhodes)
1688 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1689 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1690 ** The compiler no longer issues bogus style warnings for undefined
1691 classes in the same source file as the DEFCLASSes which defined
1692 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1694 ** fixes in CONDITION class precedence list for undefined function
1695 errors (thanks to Alexei Dejneka)
1696 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1697 correctly. (thanks to Dan Barlow)
1698 ** portability fixes aiming at bootstrapping under CLISP (thanks
1699 to Dave McDonald and Christophe Rhodes)
1700 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1701 ** fixes in type translation and and type inference (thanks to
1703 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1704 ** various fixes in the new ports (thanks to Dan Barlow)
1705 * several changes related to debugging:
1706 ** suppression of tail recursion, as noted above
1707 ** stack overflow detection, as noted above
1708 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1709 is now the default. (For some time encapsulation has been more
1710 reliable than the breakpoint-based :ENCAPSULATE NIL
1711 implementation, at least on X86 systems; and I just noticed that
1712 encapsulation also seems closer to the spirit of the ANSI
1715 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1717 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1718 up properly again. (There was a packaging bug in 0.7.0 which
1719 left their definitions in SB-SYS::LOAD-FOREIGN and
1720 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1721 vital for most things which interface to C-level interfaces,
1722 like extensions working with sockets or databases or
1723 Perl-compatible regexes or whatever, and the need to fix
1724 this bug is the main reason that 0.7.1 was released so
1726 ** DEFGENERIC is now choosier about the methods it redefines, so that
1727 reLOADing a previously-LOADed file containing DEFGENERICs does
1728 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1729 cycle now works as it should. (thanks to Alexey Dejneka)
1730 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1731 fixed by Christophe Rhodes. (He actually submitted this patch
1732 months ago, and I delayed until after 0.7.0.)
1733 ** Bug 111 (internal compiler confusion about runtime checks on
1734 FUNCTION types) was fixed by Alexey Dejneka.
1735 * Some internal cleanups (getting rid of variables which aren't
1736 needed now that the byte interpreter is gone) caused the fasl
1737 file format number to change again.
1739 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1740 * major incompatible change: The default fasl file extension, i.e. the
1741 default extension for files produced by COMPILE-FILE, has changed
1742 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1744 ** There are many changes in the implementation of the compiler.
1745 SBCL is now essentially a compiler-only implementation of ANSI
1746 Common Lisp. EVAL still "interprets" a few special cases, but
1747 almost all the interesting cases are handled by creating
1748 a LAMBDA expression, calling COMPILE on it, then calling
1749 FUNCALL on the result.
1750 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1751 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1752 Since the code is newer, there might still be some new bugs
1753 (though not as many as before Martin Atzmueller's fixes:-). But
1754 the new code is substantially simpler and clearer, and hopefully
1755 any remaining bugs will be simpler, less fundamental, and more
1756 fixable then the bugs in the old code.
1757 ** The revised compiler is still a little unsteady on its feet.
1759 *** The debugging information it produces (particularly the names
1760 of FUNCTION objects) is sometimes much less useful than what
1761 the old compiler produced.
1762 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1763 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1764 within a MACROLET) has been temporarily weakened.
1765 ** There are new compiler optimizations for various functions:
1766 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1767 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1768 *** the math functions TRUNCATE, FLOOR, and CEILING
1769 *** the function-of-all-trades COERCE
1770 Mostly these should be transparent, but there's one
1771 potentially-annoying problem (bug 117): when the compiler
1772 inline-expands a function and does type analysis on the result,
1773 it can create control paths which have type mismatches, and
1774 when it can't prove that those control paths aren't taken,
1775 it will issue WARNINGs about the type mismatches. This is
1776 a particular problem in practice for the new sequence functions.
1777 It's not clear how this should be fixed, and for now, a
1778 workaround is given in the entry for 117 in the BUGS file.
1779 ** (Because of the interaction between the two previous items --
1780 occasional inlining problems and new inline expansions -- some
1781 of the new sequence function optimizations won't really kick in
1782 completely until debugging information, and then inlining, are
1783 straightened out in some future version.)
1784 * minor incompatible changes:
1785 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1786 DIRECTORY no longer implicitly promotes NIL slots of its
1787 pathname argument to :WILD. In particular, when you ask for the
1788 contents of a directory (which you used to be able to do without
1789 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1790 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1791 ** changes in behavior that ANSI explicitly defines to be
1792 implementation dependent:
1793 *** The new compiler-only implementation still conforms with ANSI,
1794 but acts a little different than before. Besides the obvious
1795 changes in performance tradeoffs (that the cost per form passed
1796 to EVAL has gone up, and the cost per form executed by EVAL
1797 has gone down), the behavior of the system changes a little
1798 because there are no longer any interpreted function objects.
1799 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1800 e.g. doing COMPILE on the output of interactive DEFUN is
1802 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1804 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1805 has changed from (:CL) to NIL.
1806 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1807 e.g. #\Newline and #\Tab, don't have names specified in the
1808 ANSI Common Lisp standard, is now based on their ASCII symbolic
1809 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1810 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1811 are no longer used for output.
1812 ** changes in internal implementation constants:
1813 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1814 4 million. (If your application spends a lot of time GCing and
1815 you have a lot of RAM, you might want to experiment with
1816 increasing it even more.)
1817 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1818 However, almost all old code should still continue to work without
1819 immediate update, as SB-C-CALL is now a (deprecated) nickname
1821 ** Old operator names in the style DEF-FOO are now deprecated in
1822 favor of new corresponding names DEFINE-FOO, for consistency with
1823 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1824 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1825 internal symbols, but a few supported extensions like
1826 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1827 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1828 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1829 etc. as you get deeper into recursive calls to the debugger
1830 command loop, instead of the old "5]", "5]]", "5]]]"
1831 sequence. (I was motivated to do this when squabbles between
1832 ILISP and SBCL left me very deeply nested in the debugger. In the
1833 short term, this change will probably provoke more ILISP/SBCL
1834 squabbles, but hopefully it will be an improvement in the long run.)
1835 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1836 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1837 function, since declaiming return types involving aliens is
1838 (1) annoyingly messy to do by hand and (2) vital to efficient
1839 compilation of code which calls such functions.
1840 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1841 longer reexported by the SB-EXT package. They're solely useful
1842 for alien code, so it seems more logical that you should get
1843 them from the SB-ALIEN package, not in SB-EXT.
1844 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1845 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1846 features. Instead, the code that they used to control is always
1847 built into the system.
1848 * many other bug fixes
1849 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1850 advantage of the new EVAL-WHEN stuff and to clean them up in
1851 general, and they are now more ANSI-compliant in a number of
1852 ways. Martin Atzmueller is responsible for a lot of this.
1853 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1855 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1856 *** correct ERROR type for various file operations
1857 *** some fixes for Lisp streams
1858 *** DEFMETHOD syntax checking
1859 *** changing old weird representation of debug information as
1860 strings (which, among their other deficiencies, don't transform
1861 correctly when you rename packages, and don't change their
1862 print representation when you change things like *PACKAGE*
1863 and *PRINT-LENGTH*) to symbols and lists of symbols
1864 He also made several improvements and fixed several bugs in DESCRIBE.
1865 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1867 *** misbehavior of WRITE-STRING/WRITE-LINE
1868 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1869 and several other LOOP problems as well
1870 *** DIRECTORY when similar filenames are present
1871 *** DEFGENERIC with :METHOD options
1872 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1873 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1874 *** argument ordering in FIND with :TEST option
1875 *** mishandled package designator argument in APROPOS-LIST
1876 *** various problems in the backquote readmacro
1877 *** a bug in APROPOS
1878 *** probably some others that I'm not describing very well here,
1879 since the CVS log documents them by reference to sbcl-devel
1880 messages, and the SourceForge archives aren't working well.:-(
1881 ** Dan Barlow improved the Alpha port (and is making progress on the
1882 PPC port, for those of you who think different).
1883 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
1884 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
1885 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
1886 continues to work on the SPARC port (for those of you in a position
1887 to look down upon our little PC-compatible boxes from a great height).
1888 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
1889 and uses that copy, rather than the current dynamic value, when
1890 it's trying to decide whether to truncate output. Thus e.g.
1891 (let ((*print-lines* 50))
1892 (pprint-logical-block (stream nil)
1894 (let ((*print-lines* 8))
1895 (print (aref possiblybigthings i) stream)))))
1896 should now truncate the logical block only at 50 lines, instead of
1897 often truncating it at 8 lines, as it did before.
1898 * The doc/cmucl/ directory, containing old CMU CL documentation
1899 from the time of the fork, is no longer part of the base system.
1900 SourceForge has shut down its anonymous FTP service, and with it
1901 my original plan for distributing the old CMU CL documentation
1902 there. For now, if you need these files you can download an old
1903 SBCL source release and extract them from it.
1904 * The fasl file version number changed again, for dozens of reasons,
1905 some of which are apparent above.
1907 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
1908 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
1909 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
1910 patch, so that DISASSEMBLE output is much nicer.
1911 * The code in the SB-PROFILE package now seems reasonably stable.
1912 I still haven't decided what the final interface should look like
1913 (I'd like PROFILE to interact cleanly with TRACE, since both
1914 facilities use function encapsulation) but if you have a need
1915 for profiling now, you can probably use it successfully with
1916 the current CMU-CL-style interface.
1917 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
1918 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
1919 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
1920 link now returns the dangling link itself, and for similar
1921 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
1922 link itself. (In these cases the old code signalled an error and
1923 looped endlessly, respectively.) Thus, DIRECTORY now works even
1924 in the presence of dangling and cyclic symbolic links.
1925 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
1926 is now a supported extension again, since the consensus on
1927 sbcl-devel was that it can be useful for ordinary development
1928 work, not just for debugging SBCL itself.
1929 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
1930 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
1931 that currently-defined functions might be redefined later with
1932 different return types.
1933 * Hash tables can be printed readably, as inspired by CMU CL code
1934 of Eric Marsden and SBCL code of Martin Atzmueller.
1935 * better error handling in CLOS method combination, thanks to
1936 Martin Atzmueller porting Pierre Mai's CMU CL patches
1937 * more overflow fixes for >16Mbyte I/O buffers
1938 * A bug in READ has been fixed, so that now a single Ctrl-D
1939 character suffices to cause end-of-file on character streams.
1940 In particular, now you only need one Ctrl-D at the command
1941 line (not two) to exit SBCL.
1942 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
1944 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
1945 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
1946 does the right thing.
1947 * The implementation of some type tests, especially for CONDITION
1948 types, is now tidier and maybe faster, due to CMU CL code
1949 originally by Douglas Crosher, ported by Martin Atzmueller.
1950 * Some math functions have been fixed, and there are new
1951 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
1952 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
1953 * (There are also some new optimizers in contrib/*-extras.lisp. Those
1954 aren't built into sbcl-0.6.13, but are a sneak preview of what's
1955 likely to be built into sbcl-0.7.0.)
1956 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
1957 CL, ported to SBCL by Martin Atzmueller)
1958 * DESCRIBE now gives more information in some cases. (Pierre Mai's
1959 patch to CMU CL, ported to SBCL by Martin Atzmueller)
1960 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
1961 * There's a new slam.sh hack to shorten the edit/compile/debug
1962 cycle for low-level changes to SBCL itself, and a new
1963 :SB-AFTER-XC-CORE target feature to control the generation of
1964 the after-xc.core file needed by slam.sh.
1965 * minor incompatible change: The ENTRY-POINTS &KEY argument to
1966 COMPILE-FILE is no longer supported, so that now every function
1967 gets an entry point, so that block compilation looks a little
1968 more like the plain vanilla ANSI section 3.2.2.3 scheme.
1969 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
1970 returns the number of bytes consed since the system started,
1971 rather than the number consed since the first time the function
1972 was called. (The new definition parallels ANSI functions like
1973 CL:GET-INTERNAL-RUN-TIME.)
1974 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
1975 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
1976 supported. Now DIRECTORY always does the abstract Common-Lisp-y
1977 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
1978 * Fasl file version numbers are now independent of the target CPU,
1979 since historically most system changes which required version
1980 number changes have affected all CPUs equally. Similarly,
1981 the byte fasl file version is now equal to the ordinary
1984 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
1985 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
1986 is no longer recognized. I apologize for this, because it was
1987 listed in SB-EXT as a supported extension, but I found that
1988 its existing behavior was poorly specified, as well as incorrectly
1989 specified, and it looked like too much of a mess to straighten it
1990 out. I have enough on my hands trying to get ANSI stuff to work..
1991 * many patches ported from CMU CL by Martin Atzmueller, with
1992 half a dozen bug fixes in pretty-printing and the debugger, and
1993 half a dozen others elsewhere
1994 * fixed bug 13: Floating point infinities are now supported again.
1995 They might still be a little bit flaky, but thanks to bug reports
1996 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
1997 as flaky as they were.
1998 * The --noprogrammer command line option is now supported. (Its
1999 behavior is slightly different in detail from what the old man
2000 page claimed it would do, but it's still appropriate under the
2001 same circumstances that the man page talks about.)
2002 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2003 are now supported, and enabled by default. Thus, the compiler can
2004 handle many floating point and complex operations much less
2005 inefficiently. (Thus e.g. you can implement a complex FFT
2007 * The compiler now detects type mismatches between DECLAIM FTYPE
2008 and DEFUN better, and implements CHECK-TYPE more correctly, and
2009 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2010 thanks to patches from Martin Atzmueller.
2011 * various fixes to make the cross-compiler more portable to
2012 ANSI-conforming-but-different cross-compilation hosts (notably
2013 Lispworks for Windows, following bug reports from Arthur Lemmens)
2014 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2015 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2016 fixed thanks to Pierre Mai's CMU CL patches.
2017 * fixes to keep the system from overflowing internal counters when
2018 it tries to use i/o buffers larger than 16M bytes
2019 * fixed bug 45a: Various internal functions required to support
2020 complex special functions have been merged from CMU CL sources.
2021 (When I was first setting up SBCL, I misunderstood a compile-time
2022 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2023 * improved support for type intersection and union, fixing bug 12
2024 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2025 more obscure bugs as well
2026 * some steps toward byte-compiling non-performance-critical
2027 parts of the system, courtesy of patches from Martin Atzmueller
2028 * Christophe Rhodes has made some debian packages of sbcl at
2029 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2030 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2031 stable, but are nonetheless usable. When he's ready, I'd be happy
2032 to add them to the SourceForge "File Releases" section. (And if
2033 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2034 * new fasl file format version number (because of changes in
2035 internal representation of (OR ..) types to accommodate the new
2036 support for (AND ..) types, among other things)
2038 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2039 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2041 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2043 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2044 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2045 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2046 * DESCRIBE now works on structure objects again.
2047 * Most function call argument type mismatches are now handled as
2048 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2049 know whether the function will be redefined before the call is
2050 executed. (The compiler could flag local calls with full WARNINGs,
2051 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2052 it doesn't keep track of enough information to know whether calls
2053 are local in this sense.)
2054 * Compiler output is now more verbose, with messages truncated
2055 later than before. (There should be some supported way for users
2056 to override the default verbosity, but I haven't decided how to
2057 provide it yet, so this behavior is still controlled by the internal
2058 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2059 src/compiler/ir1util.lisp.)
2060 * Fasl file format version numbers have increased again, because
2061 support for the Gray streams extension changes the layout of the
2062 system's STREAM objects.
2063 * The Gray subclassable streams extension now works, thanks to a
2064 patch from Martin Atzmueller.
2065 * The full LOAD-FOREIGN extension (not just the primitive
2066 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2067 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2068 but like most other programs, it defaults to copying the Unix
2069 environment from the original process instead of starting the
2070 new process in an empty environment.
2071 * Extensions which manipulate the Unix environment now support
2072 an :ENVIRONMENT keyword option which doesn't smash case or
2073 do other bad things. The CMU-CL-style :ENV option is retained
2074 for porting convenience.
2075 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2076 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2078 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2080 * A patch from Martin Atzmueller seems to have solved the SIGINT
2081 problem, and as far as we know, signal-handling now works cleanly.
2082 (If you find any new bugs, please report them!)
2083 * The system no longer defaults Lisp source file names to types
2084 ".l", ".cl", or ".lsp", but only to ".lisp".
2085 * The compiler no longer uses special default file extensions for
2086 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2087 seems to expect a single default extension for all compiled code,
2088 and there's no compelling reason to try to stretch the standard
2089 to allow two different extensions.) Instead, byte-compiled files
2090 default to the same extension as native-compiled files.
2091 * Fasl file format version numbers have increased again, because
2092 a rearrangement of internal implementation packages made some
2093 dumped symbols in old fasl files unreadable in new cores.
2094 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2095 many fewer weird special cases.
2096 * Bug #17 (differing COMPILE-FILE behavior between logical and
2097 physical pathnames) has been fixed, and some related misbehavior too,
2098 thanks to a patch from Martin Atzmueller.
2099 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2100 by Tim Moore, ported to SBCL by Martin Atzmueller.
2101 * Martin Atzmueller fixed several filesystem-related problems,
2102 including bug #36, in part by porting CMU CL patches, which were
2103 written in part by Paul Werkowski.
2104 * More compiler warnings in src/runtime/ are gone, thanks to
2105 more patches from Martin Atzmueller.
2106 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2109 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2111 * DESCRIBE now works on CONDITION objects.
2112 * The debugger now handles errors which arise when trying to print
2113 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2115 * The build system now uses an additional file, customize-target-features.lisp,
2116 to allow local modifications to the target *FEATURES* list. (The point of
2117 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2118 debugging features enabled, without having to worry about propagating it
2119 into everyone's system when I do a "cvs update".) When no
2120 customize-target-features.lisp file exists, the target *FEATURES* list
2121 should be constructed the same way as before.
2122 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2123 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2124 warn about behavior which is undefined under the ANSI spec. Note: This
2125 is specified by ANSI, but it's not very popular with programmers.
2126 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2127 macro in the SBCL sources for an example of a workaround which you
2128 might use to make portable ANSI-standard code which does what you want.
2129 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2130 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2131 when it's used as a non-toplevel form. (This is required in order
2132 to implement the DEFCONSTANT-EQX macro.)
2133 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2134 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2135 funny things, due to bugs in the implementation of EVAL-WHEN
2136 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2137 non-toplevel DEFCONSTANTs are uncommon.)
2138 * The core file version number and fasl file version number have been
2139 incremented, because the old noncompliant DEFCONSTANT behavior involved
2140 calling functions which no longer exist, and because I also took the
2141 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2142 * fixed bug 1 (error handling before read-eval-print loop starts), and
2143 redid debugger restarts and related debugger commands somewhat while
2145 ** The QUIT debugger command is gone, since it did something
2146 rather different than the SB-EXT:QUIT command, and since it never
2147 worked properly outside the main toplevel read/eval/print loop.
2148 Invoking the new TOPLEVEL restart provides the same functionality.
2149 ** The GO debugger command is also gone, since you can just invoke
2150 the CONTINUE restart directly instead.
2151 ** The TOP debugger command is also gone, since it's redundant with the
2152 FRAME 0 command, and since it interfered with abbreviations for the
2154 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2156 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2157 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2158 some porting work by Martin Atzmueller)
2159 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2160 with a boatload of patches which he ported to SBCL. Now that those
2162 ** The system tries to make sure that its low-priority messages
2163 are prefixed by semicolons, to help people who like to use
2164 syntax highlighting in their ILISP buffer. (This patch
2165 was originally due to Raymond Toy.)
2166 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2167 patches originally written by Raymond Toy.
2168 ** The compiler understands coercion between single-value and
2169 multiple-VALUES type expressions better, getting rid of some very
2170 weird behavior, thanks to patches originally by Robert MacLachlan
2171 and Douglas Crosher.
2172 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2173 lambda lists, thanks to a patch originally by Pierre Mai.
2174 ** The system no longer bogusly warns about "abbreviated type
2176 ** The compiler gets less confused by inlining and RETURN-FROM,
2177 thanks to some patches originally by Tim Moore.
2178 ** The system no longer hangs when dumping circular lists to fasl
2179 files, thanks to a patch originally from Douglas Crosher.
2180 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2181 undefined function error.
2182 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2183 in structure slot initforms) for now, documented workaround instead:-|
2184 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2185 * fixed bug 5: added stubs for various Gray stream functions called
2186 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2187 installed, at least appropriate type errors are generated
2188 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2189 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2190 consistently in DEFMETHOD forms.
2191 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2192 it doesn't seem to affect SBCL after all
2193 * The C runtime system now builds with better optimization and many
2194 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2196 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2198 * The system is now under CVS at SourceForge (instead of the
2199 CVS repository on my home machine).
2200 * The new signal handling code has been tweaked to treat register
2201 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2202 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2203 downstream, e.g. in debug-int.lisp, has implicit dependencies
2204 on the unsignedness of integer representation of machine words,
2205 and that caused the system to bomb out with infinite regress
2206 when trying to recover from type errors involving signed values,
2207 e.g. (BUTLAST '(1 2 3) -1).)
2208 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2209 (This was one of the bugs Peter Van Eynde reported back in July.)
2210 * The system now uses code inspired by Colin Walters' O(N)
2211 implementation of MAP (from the cmucl-imp@cons.org mailing
2212 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2213 inline the MAP operation, and there is more than one
2214 sequence argument to the MAP call (so that it can't just
2215 do ETYPECASE once and for all based on the type of the
2216 single sequence argument). (The old non-inline implementation
2217 of the general M-argument sequence-of-length-N case required
2218 O(M*N*N) time when any of the sequence arguments were LISTs.)
2219 * The QUIT :UNIX-CODE keyword argument has been renamed to
2220 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2221 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2223 * Raymond Wiker's patches to port dynamic loading from Linux to
2224 FreeBSD have been added.
2225 * The BUGS file is now more nearly up to date, thanks in large part
2226 to Martin Atzmueller's review of it.
2227 * The debugger now flushes standard output streams before it begins
2228 its output ("debugger invoked" and so forth).
2229 * The core version number and fasl file version number have both
2230 been incremented, because of incompatible changes in the layout
2232 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2233 used to not be called for a saved Lisp image.)
2234 * Martin Atzmueller's version of a patch to fix a compiler crash,
2235 as posted on sbcl-devel 13 September 2000, has been installed.
2236 * Instead of installing Martin Atzmueller's patch for the
2237 compiler transform for SUBSEQ, I deleted the compiler transform,
2238 and transforms for some similar consing operations.
2239 * A bug in signal handling which kept TRACE from working on OpenBSD
2241 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2242 away by constant folding
2243 * The system now defines its address space constants in one place
2244 (in the Lisp sources), and propagates them automatically elsewhere
2245 (through GENESIS and the sbcl.h file). Therefore, patching the
2246 address map is less unnecessarily tedious and error-prone. The
2247 Lisp names of address space constants have also been systematized.
2248 * CVS tags like dollar-Header-dollar have been removed from
2249 the sources, because they have never saved me trouble and
2250 they've been source of trouble working with patches and other
2251 diff-related operations.
2252 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2253 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2255 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2257 * The system has been ported to OpenBSD.
2258 * The system now compiles with a simple "sh make.sh" on the systems
2259 that it's supported on. I.e., now you no longer need to tweak
2260 text in the target-features.lisp-expr and symlinks in src/runtime/
2261 by hand, the make.sh takes care of it for you.
2262 * The system is no longer so grossly inefficient when compiling code
2263 involving vectors implemented as general (not simple) vectors (VECTOR T),
2264 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2265 which dares to use the various sequence functions on non-simple
2266 vectors, takes less of a performance hit.
2267 * There is now a primitive type predicate VECTOR-T-P
2268 to test for the (VECTOR T) type, so that e.g.
2269 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2270 can now be compiled with some semblance of efficiency. (The old code
2271 turned the type declaration into a full call to %TYPEP at runtime!)
2272 * AREF on (VECTOR T) is still not fast, since it's still compiled
2273 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2274 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2275 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2276 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2277 without making full calls to SUBTYPEP at runtime.
2278 (Some analogous efficiency issues for non-simple vectors specialized to
2279 element types other than T, or for non-simple multidimensional arrays,
2280 have not been addressed. They could almost certainly be handled the
2281 same way if anyone is motivated to do so.)
2282 * The changes in array handling break binary compatibility, so
2283 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2284 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2285 instead of (VALUES T T).
2286 * By following the instructions that Dan Barlow posted to sbcl-devel
2287 on 2 July 2000, I was able to enable primitive dynamic object
2288 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2289 functionality is not implemented (since it calls ld to resolve
2290 library references automatically, requiring RUN-PROGRAM for its
2291 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2292 try to resolve library references) is now supported.
2293 * The system now flushes the standard output streams when it terminates,
2294 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2295 them at several other probably-convenient times, e.g. in each pass of
2296 the toplevel read-eval-print loop, and after evaluating a form given
2297 as an "--eval" command-line option. (These changes were motivated by a
2298 discussion of stream flushing issues on cmucl-imp in August 2000.)
2299 * The source transform for TYPEP of array types no longer assumes
2300 that an array whose element type is a not-yet-defined type
2301 is implemented as an array of T, but instead punts, so that the
2302 type will be interpreted at runtime.
2303 * There is now some support for cross-compiling in make.sh: each of
2304 the phases of make.sh has its own script. (This should be transparent
2305 to people doing ordinary, non-cross-compile builds.)
2306 * Since my laptop doesn't have hundreds of megabytes of memory like
2307 my desktop machine, I became more motivated to do some items on
2308 my to-do list in order to reduce the size of the system a little:
2309 ** Arrange for various needed-only-at-cold-init things to be
2310 uninterned after cold init. To support this, those things have
2311 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2312 symbols with such names are now uninterned after cold init).
2313 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2314 fasl files for cold load.
2315 ** Remove the old compiler structure pooling code (which used to
2316 be conditional on the target feature :SB-ALLOC) completely.
2317 ** Redo the representation of some data in cold init to be more compact.
2318 (I also looked into supporting byte compiled code at bootstrap time,
2319 which would probably reduce the size of the system a lot, but that
2320 looked too complicated, so I punted for now.)
2321 * The maximum signal nesting depth in the src/runtime/ support code has
2322 been reduced from 4096 to 256. (I don't know any reason for the very
2323 large old value. If the new smaller value turns out to break something,
2324 I'll probably just bump it back up.)
2325 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2327 * Many, many bugs reported by Peter Van Eynde have been added to
2328 the BUGS list; some have even been fixed.
2329 * While enabling dynamic object file loading, I tried to make the
2330 code easier to understand, renaming various functions and variables
2331 with less ambiguous names, and changing some function calling
2332 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2333 * While trying to figure out how to do the OpenBSD port, I tried to
2334 clean up some of the code in src/runtime/. In particular, I dropped
2335 support for non-POSIX signal handling, added various comments,
2336 tweaked the code to reduce the number of compilation warnings, and
2337 renamed some files to increase consistency.
2338 * To support the new automatic configuration functionality in make.sh,
2339 the source file target-features.lisp-expr has been replaced with the
2340 source file base-target-features.lisp-expr and the machine-generated
2341 file local-target-features.lisp-expr.
2342 * fixed a stupid quoting error in make.sh so that using CMU CL
2343 "lisp -batch" as cross-compilation host works again
2345 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2347 * DESCRIBE no longer tries to call itself recursively to describe
2348 bound/fbound values, so that it no longer fails on symbols which are
2349 bound to themselves (like keywords, T, and NIL).
2350 * DESCRIBE now works on generic functions.
2351 * The printer now prints less-screwed-up representations of closures
2352 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2353 * A private symbol is used instead of the :EMPTY keyword previously
2354 used to mark empty slots in hash tables. Thus
2355 (DEFVAR *HT* (MAKE-HASH-TABLE))
2356 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2357 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2358 now does what ANSI says that it should. (You can still get
2359 similar noncompliant behavior if bang on the hash table
2360 implementation with all the symbols you get back from
2361 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2362 This breaks binary compatibility, since tests for equality to
2363 :EMPTY are wired into things like the macroexpansion of
2364 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2366 * There's now a minimal placeholder implementation for CL:STEP,
2367 as required by ANSI.
2368 * An obscure bug in the interaction of the normal compiler, the byte
2369 compiler, inlining, and structure predicates has been patched
2370 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2371 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2373 * Missing ordinary arguments in a macro call are now detected even
2374 when the macro lambda list contains &KEY or &REST.
2375 * The debugger no longer complains about encountering the top of the
2376 stack when you type "FRAME 0" to explicitly instruct it to go to
2377 the top of the stack. And it now prints the frame you request even
2378 if it's the current frame (instead of saying "You are here.").
2379 * As specified by ANSI, the system now always prints keywords
2380 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2382 * The default initial SIZE of HASH-TABLEs is now smaller.
2383 * Type information from CLOS class dispatch is now propagated
2384 into DEFMETHOD bodies, so that e.g.
2385 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2387 is now basically equivalent to
2388 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2389 (DECLARE (TYPE SINGLE-FLOAT X))
2391 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2392 operation, without having to do run-time type dispatch.
2393 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2394 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2395 * Since I don't seem to be making big file reorganizations very often
2396 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2397 is overflowing my ability to conveniently back them up), I've finally
2398 checked the system into CVS. (The CVS repository is on my home system,
2399 not at SourceForge -- putting it on SourceForge might come later.)
2400 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2401 high-level GC-NOTIFY-FOO functions send their output. (There's
2402 still very little control of where low-level verbose GC functions
2403 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2404 less than it used to -- the GC-NOTIFY-FOO functions are now under
2405 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2406 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2407 in only one place in the source code, and propagates it automatically
2408 everywhere that it's needed. Thus e.g. when I bump the version from
2409 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2410 * The C source files now include boilerplate legalese and documentation
2411 at the head of each file (just as the Lisp source files already did).
2412 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2413 to his page will be replaced with a link to his new CLiki service.
2415 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2417 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2418 * The build process now looks for GNU make under the default name "gmake",
2419 instead of "make" as it used to. If GNU make is not available as "gmake"
2420 on your system, you can change this default behavior by setting the
2421 GNUMAKE environment variable.
2422 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2423 can build without error under CMU CL.
2425 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2427 * There is now a partial SBCL user manual (with some new text and some
2428 text cribbed from the CMU CL manual).
2429 * The beginnings of a profiler have been added (starting with the
2430 CMU CL profiler and simplifying and cleaning up). Eventually the
2431 main interface should be through the TRACE macro, but for now,
2432 it's still accessed through vaguely CMU-CL-style functions and macros
2433 exported from the package SB-PROFILE.
2434 * Some problems left over from porting CMU CL to the new
2435 cross-compilation bootstrap process have been cleaned up:
2436 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2437 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2438 part when getting rid of the compiler *BACKEND* stuff.)
2439 ** The value of *NULL-TYPE* was screwed up, because it was
2440 being initialized before the type system knew the final
2441 definition of the 'NULL type. This screwed up several key
2442 optimizations in the compiler, causing inefficiency in all sorts
2443 of places. (I found it because I wanted to understand why
2444 GET-INTERNAL-RUN-TIME was consing.)
2445 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2446 PROCLAIM FTYPE information. Unfortunately this broke binary
2447 compatibility again, since now the forms output by DEFGENERIC
2448 to refer to functions which didn't exist in 0.6.3.
2449 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2450 can use the new (as of 0.6.3) transform for SOME into MAP into
2452 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2453 values to (INTEGER 0 1000000), so that the time code will no longer
2454 occasionally get blown up by Linux returning 1000000 microseconds
2455 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2456 its output conform to the ANSI spec. (Alas, this makes its output
2457 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2458 * A full call to MAP NIL with a single sequence argument no longer conses.
2459 * fixes to problems pointed out by Martin Atzmueller:
2460 * The manual page no longer talks about multiprocessing as though
2461 it were currently supported.
2462 * The ILISP support patches have been removed from the distribution,
2463 because as of version 5.10.1, ILISP now supports SBCL without us
2464 having to maintain patches.
2465 * added a modified version of Raymond Toy's recent CMU CL patch for
2466 EQUALP comparison of HASH-TABLE
2468 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2470 * The system still can't cross-compile itself with
2471 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2472 entails), but at least it can compile more of itself that way
2473 than it used to be able to, and various buglets which were uncovered
2474 by trying to cross-compile itself that way have now been fixed.
2475 * This release breaks binary compatibility again. This time
2476 at least I've incremented the FASL file format version to 2, so that the
2477 problem can be detected reliably instead of just causing weird errors.
2478 * various new style warnings:
2479 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2480 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2481 ** using the lexical binding of a variable named in the *FOO* style
2482 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2483 as specified by ANSI.
2484 * *RANDOM-STATE* is no longer automatically initialized from
2485 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2486 default behavior of the system is to repeat its behavior every time
2487 it's run. If you'd like to change this behavior, you can always
2488 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2489 old convention there was no comparably easy way to get the system to
2490 repeat its behavior every time it was run.
2491 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2492 FTYPE declarations has been removed, in favor of their ANSI
2493 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2494 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2495 the types of their sequence arguments can be determined at compile time.
2496 This is done through a new open code expansion for MAP which eliminates
2497 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2498 argument types can be determined at compile time.
2499 * The optimizer now transforms COERCE into an identity operation when it
2500 can prove that the coerced object is already of the correct type. (This
2501 can be a win for machine generated code, including the output of other
2502 optimization transforms, such as the MAP transform above.)
2503 * Credit information has been moved from source file headers into CREDITS.
2504 * Source file headers have been made more standard.
2505 * The CASE macro now compiles without complaining even when it has
2508 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2510 * (Note that the way that the PCL macroexpansions were rewritten
2511 to accommodate the change in DEFGENERIC below breaks binary
2512 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2513 not run under sbcl-0.6.2. Once we get out of alpha releases,
2514 i.e. hit release 1.0.0, we'll probably try to maintain binary
2515 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2516 and sbcl-1.4.4. Until then, however, it might be fairly common
2517 for maintenance releases to break binary compatibility.)
2518 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2520 * The reporting of unhandled signals has been changed to print some
2521 explanatory text as well as the report form. (Previously only
2522 the report form was printed.)
2523 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2524 it defines, so that the compiler no longer issues undefined function
2525 warnings for compiled-but-not-yet-loaded generic functions.
2526 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2527 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2528 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2529 section of the ANSI spec are supported.
2530 * The "" nickname for the "KEYWORD" package has been removed.
2531 The reader still handles symbol tokens which begin with a package marker
2532 as keywords, but it doesn't expose its mechanism for doing so in the
2533 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2534 * The system now issues STYLE-WARNINGs for contradictory TYPE
2535 proclamations. (Warnings for contradictory FTYPE proclamations would
2536 be nice too, but those can't be done usefully unless the type system
2537 is made smarter about FUNCTION types.)
2538 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2539 systematized, so that "*target-*.lisp is supposed to exist only on the
2540 target and imply that there's a related file which exists on the
2541 host, and *host-*.lisp is supposed to exist only on the host and imply
2542 that there's a related file which exists on the target. This involves a
2543 lot of renaming. Hopefully the acute confusion caused by the renaming
2544 will be justified by the reduction in chronic confusion..
2545 ** runtime-type.lisp -> early-target-type.lisp
2546 ** target-type.lisp -> late-target-type.lisp
2547 ** early-host-format.lisp -> early-format.lisp
2548 ** late-host-format.lisp -> late-format.lisp
2549 ** host-error.lisp -> misc-error.lisp
2550 ** early-error.lisp -> early-target-error.lisp
2551 ** late-error.lisp -> late-target-error.lisp
2552 ** host-defboot.lisp -> early-defboot.lisp
2553 ** code/misc.lisp -> code/target-misc.lisp
2554 ** code/host-misc.lisp -> code/misc.lisp
2555 ** code/numbers.lisp -> code/target-numbers.lisp
2556 ** code/early-numbers.lisp -> numbers.lisp
2557 ** early-host-type.lisp -> early-type.lisp
2558 ** late-host-type.lisp -> late-type.lisp
2559 ** host-typep.lisp -> typep.lisp
2560 ** load.lisp -> target-load.lisp
2561 ** host-load.lisp -> load.lisp
2562 ** host-disassem.lisp -> disassem.lisp
2563 ** host-insts.lisp -> insts.lisp
2564 ** byte-comp.lisp -> target-byte-comp.lisp
2565 ** host-byte-comp.lisp -> byte-comp.lisp
2566 ** host-signal.lisp -> signal.lisp
2567 ** host-defstruct.lisp -> defstruct.lisp
2568 ** late-target-type.lisp -> deftypes-for-target.lisp
2569 Furthermore, several other previously target-only files foo.lisp (e.g.
2570 hash-table.lisp and random.lisp) have been split into a target-and-host
2571 foo.lisp file and a target-only target-foo.lisp file, with their key type
2572 definitions in the target-and-host part, so that the cross-compiler will
2573 know more about target types.
2574 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2575 gone away. In their place are various *BACKEND-FOO* variables
2576 corresponding to the slots of the old structure.
2577 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2578 into the target SBCL has been fixed.
2579 * The chill.lisp system for loading cold code into a running SBCL
2581 * Support for the CMU CL "scavenger hook" extension has been removed.
2582 (It was undocumented and unused in the CMU CL sources that SBCL was
2583 derived from, and stale in sbcl-0.6.1.)
2584 * Various errors in the cross-compiler type system were detected
2585 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2586 (enabling various consistency checks). Many of them were fixed,
2587 but some hard problems remain, so the compiler is back to
2588 running without *TYPE-SYSTEM-INITIALIZED* for now.
2589 * As part of the cross-compiler type system cleanup, I implemented
2590 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2591 * I have started adding UNCROSS calls throughout the type system
2592 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2593 on cross-compiler input files will be able to go away, and various
2595 * CONSTANTP now returns true for quoted forms (as explicitly required
2598 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2600 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2601 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2602 (MAKE-LIST -1). In the longer term, it ought to become true
2603 that declarations are assertions even at SAFETY 1. For now, it's not
2604 quite true even at SAFETY 3, but it's at least more nearly true..
2605 (Note that this change seems to increases the size of the system by
2606 O(5%) and to decrease the speed of the compiler by 20% or more.)
2607 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2608 for the problem of printing dozens of lines of distracting information
2609 about low-level system machinery as part of the top stack frame
2610 on entry to the debugger when an undefined function was called.
2611 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2612 is not reset to COMMON-LISP-USER.
2613 * Compilation of stuff related to dyncount.lisp has been made conditional
2614 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2615 smaller. The various dyncount-related symbols have been moved into
2616 a new "SB-DYNCOUNT" package.
2617 * tty-inspect.lisp has been renamed to inspect.lisp.
2618 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2619 feature has gone away. (When we eventually port to other flavors of
2620 libc and/or Unix, we'll try to make the differences between flavors
2621 invisible at the user level.)
2622 * Various other *FEATURES* tags, and/or their associated conditionals,
2623 have been removed if obsolescent, or given better documentation, or
2624 sometimes given more-mnemonic names.
2626 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2628 * tidied up "make.sh" script
2629 * tidied up system directory structure
2630 * better "clean.sh" behavior
2631 * added doc/FOR-CMUCL-DEVELOPERS
2632 * many many small tweaks to output format, e.g. removing possibly-confusing
2633 trailing #\. character in DESCRIBE-INSTANCE
2634 * (EQUALP #\A 'A) no longer signals an error.
2635 * new hashing code, including EQUALP hashing
2636 * tidied up Lisp initialization and toplevel
2637 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2638 * command line argument processing
2639 * added POSIX-GETENV function to deal with Unix-ish environment variables
2640 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2641 terminating SBCL on EOF
2642 * non-verbose GC by default
2643 * There is no more "sbcl" shell script; the sbcl file is now the C
2644 runtime executable (just like CMU CL).
2645 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2647 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2648 symbol and package information as Lisp native symbol and package objects
2649 instead of strings naming symbols and strings naming packages. This way,
2650 whenever packages are renamed (as in warm init), debug information is
2651 transformed along with everything else.
2652 * tweaked the optimization policy declarations which control the building
2653 of SBCL itself. Now, among other things, the system no longer saves
2654 source location debugging information. (This helps two problems at once
2655 by reducing SBCL size and by keeping SBCL from trying to look for its
2656 sources -- which may not exist -- when reporting errors.)
2657 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2658 debugging and testing purposes
2659 * cleaned up printing, making the printer call PRINT-OBJECT for
2660 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2661 methods, giving nearly-ANSI behavior
2662 * converted almost all special variables to use *FOO* naming convention
2663 * deleted PARSE-TIME functionality, since it can be done portably
2664 * moved some files out of cold init into warm init
2665 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2667 * regularized formatting of source files
2668 * added an install.sh script
2669 * fixed ridiculous memory usage of cross-compiler by making
2670 compiler/alloc.lisp not try to do pooling unless it can hook
2671 itself into the GC of the cross-compilation host. Now the system
2672 builds nicely on my old laptop.
2673 * added :SB-ALLOC in target-features.lisp-expr
2674 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2675 was not implemented)
2676 * re-did condition handling and note reporting in the compiler. Notes
2677 are no longer handled by signalling conditions. Style warnings
2678 and warnings are handled more correctly and reported in such a way
2679 that it's easy to find one or the other in your output (so that you
2680 can e.g. figure out which of many problems caused COMPILE-FILE to
2682 * changed the severity of several compiler warnings from full WARNING
2683 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2684 compiler note reporting so that it doesn't use the condition system
2685 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2686 COMPILE-FILE command)
2687 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2688 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2689 no longer supported, since their implementation was deeply intertwingled
2690 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2691 * removed bogus "support" for compiler macros named (SETF FOO), and
2692 removed the compiler macro for SETF INFO (but only after making a fool
2693 of myself on the cmucl-imp mailing list by posting a bogus patch for
2694 DEFINE-COMPILER-MACRO..)
2695 * Compiled files containing forms which have side effects on the Lisp
2696 reader (such as DEFPACKAGE forms) are now handled more correctly.
2697 (Compiler queuing of top level lambdas has been suppressed by setting
2698 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2699 * deleted various currently-unused source files, e.g. gengc.lisp. They
2700 may be added back at some point e.g. when porting to other architectures,
2701 but until they are it's distracting to distribute them and to try to
2703 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2704 there were so many of them they're just distractions, and UNCROSS is
2705 known to be able to handle the current sources
2706 * moved PROFILE functionality into TRACE, so that it will be clear
2707 how the wrapping and unwrapping of functions when you profile them
2708 interacts with the wrapping and unwrapping of functions when you
2709 trace them. (Actually, the functionality isn't there yet, but at least
2710 the interface specification is there. Hopefully, the functionality will
2711 arrive with some future maintenance release.)
2712 * removed host-oops.lisp
2713 * changed signature of QUIT function to allow UNIX-CODE argument
2714 * fixed READ-SEQUENCE bug
2715 * tweaked verbose GC output so that it looks more like the progress
2716 output that ANSI specifies for functions like LOAD
2717 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2718 * added <http://sbcl.sourceforge.com> to the banner information printed by