1 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2 * the system now has rudimentary external-format support; the
3 primary user-visible change at this time is that characters with
4 the high bit set (such as lower-case-e-acute) will print correctly
5 to a terminal in a UTF-8 environment.
6 * minor incompatible change: BASE-CHAR no longer names a class;
7 however, CHARACTER continues to do so, as required by ANSI.
8 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
9 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
10 has been moved to the SB-EXT package (temporarily re-exported from
12 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
13 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
15 * bug fix: Cyclic structures and unprintable objects in compiler
16 messages no longer cause errors. (reported by Bruno Haible)
17 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
18 and READ-FROM-STRING all now return a primary value of NIL if
19 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
20 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
21 (reported by Bruno Haible for CMUCL)
22 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
23 do not designate a condition. (reported by Bruno Haible for
25 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
26 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
27 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
28 permitting resolution in favour of any of the conflicting symbols.
29 (reported by Bruno Haible for CMUCL)
30 * FORMAT compile-time argument count checking has been enhanced.
31 (report from Bruno Haible for CMUCL)
32 * fixed some bugs revealed by Paul Dietz' test suite:
33 ** RENAME-PACKAGE allows all package designators as new package
36 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
37 * enhancement: saving cores with foreign code loaded is now
38 supported on x86/NetBSD and sparc/Linux in addition to the previously
40 * bug fix: on some platforms repeated installations caused multiple
41 copies of HTML documentation to be installed -- should not happen
42 any more. (reported by Stefan Scholl)
43 * bug fix: parsing self-recursive alien record types multiple times
44 no longer causes infinite recursion. (reported by Thomas F. Burdick,
45 original patch by Helmut Eller for CMUCL)
46 * bug fix: stack-exhaustion detection works now on NetBSD as well.
47 (thanks to Richard Kreuter)
48 * bug fix: defining classes whose accessors are methods on existing
49 generic functions in other (locked) packages no longer signals
50 bogus package lock violations. (reported by François-René Rideau)
51 * bug fix: special variables as DEFMETHOD parameters no longer have
52 associated bogus type declarations. (reported by David Wragg and
54 * bug fix: read-write consistency on streams of element-type
55 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
56 * bug fix: redefiniton of the only method of a generic function with
57 no DEFGENERIC no longer emits a full WARNING. In addition,
58 redefinition of generic functions with no DEFGENERIC to an
59 incompatible lambda list now signals an error. (thanks to Zach
61 * bug fix: DEFGENERIC now works even when there's a function of the
62 same name in an enclosing lexical environment. (thanks to Zach
64 * fixed compiler failure, caused by instrumenting code during
65 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
66 * optimization: added loop analysis and improved register allocation
67 * fixed some bugs revealed by Paul Dietz' test suite:
68 ** POSITION on displaced vectors with non-zero displacement
69 returns the right answer.
70 ** (SIMPLE-STRING) is a valid type specifier for sequence
72 ** *PRINT-LEVEL* handling for slotless structures is pedantically
74 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
75 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
77 ** The justification version of the FORMAT ~< directive treats
78 non-zero minpad parameter correctly.
80 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
81 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
82 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
83 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
84 the supported interface.
85 * new feature: Single-stepping of code compiled with DEBUG 2 or
86 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
87 * new feature: saving cores with foreign code loaded is now
88 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
89 Timothy Moore's work for CMUCL)
90 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
91 parameters to * instead of NIL if no initform is supplied.
92 (reported by Johan Bockgaard)
93 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
94 correctly when FUNCALL appears as the car of the form. Note:
95 despite this FUNCALL forms are not currently subject to
96 compiler-macro expansion. (port of Raymond Toy's fix for the
97 same from CMUCL, reported by Johan Bockgaard)
98 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
99 (thanks for Teemu Kalvas)
100 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
101 case instead of terminating the iteration (thanks for Julian
102 Squires, Sean Champ and Raymond Toy)
103 * bug fix: incorrect expansion of defgeneric that caused a style
104 warning. (thanks for Zach Beane)
105 * bug fix: slot accessor effective method computation works properly
106 for classes with multiple non-standard applicable methods on
107 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
108 * on x86 compiler supports stack allocation of results of LIST and
109 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
110 CMUCL implementation by Gerd Moellmann)
111 * fixed some bugs revealed by Paul Dietz' test suite:
112 ** FORMAT strings with both the ~<~:;~> form of the justification
113 directive and pretty-printing directives cause an error.
115 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
116 * incompatible change: the internal functions
117 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
118 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
119 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
120 instead of the old functions.
121 * new feature: on platforms where "dladdr" is available foreign
122 function names now appear in backtraces. (based on Helmut Eller's
124 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
125 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
127 * documentation: ASDF manual is now distributed with SBCL.
128 * bug fix: SBCL can now load its contributed modules with REQUIRE
129 even if the system-provided entries have been removed from
130 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
132 * bug fix: backtraces involving undefined functions or assembly
133 routines are more informative. (thanks to Brian Downing)
134 * bug fix: mutually referent alien structures now work correctly.
135 (reported by Rick Taube)
136 * bug fix: structures defined by WITH-ALIEN can be referred to
137 within other definitions in the same WITH-ALIEN.
138 * bug fix: division operators (MOD, TRUNCATE and the like) with
139 constant zero divisors and integer dividends no longer generate
141 * bug fix: provide default methods for INPUT-STREAM-P and
142 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
143 * bug fix: improve the detection and resolution of MOP metacycles
144 (where effective method computation depends on the generic function
145 having its effective method computed). (reported by Bruno Haible)
146 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
147 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
148 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
149 * bug fix: pathnames with relative directory components can be
150 represented relative to default pathnames.
151 * optimization: in taking the GCD of bignums, reduce the two bignums
152 to approximately the same size (using Euclid's algorithm) before
153 applying the more sophisticated binary GCD. (thanks to Juho
155 * optimization: COUNT on bitvectors now operates word-at-a-time.
156 * optimization: ASH with a positive, but not necessarily constant,
157 (leftwards) shift, when in a modular context, is compiled to a
159 * fixed some bugs revealed by Paul Dietz' test suite:
160 ** FORMAT variable parameters ("~V<char>") are defaulted properly
161 if the corresponding argument is NIL.
162 ** FORMAT directives accepting a minpad parameter treat negative
164 ** literal commas as character parameters to FORMAT directives are
166 ** literal spaces directly after ~<Newline> directives within a
167 format-logical-block (~:< ~@:>) do not induce :FILL-style
168 conditional newlines.
169 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
170 as specified: it no longer includes conditional newlines).
171 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
173 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
174 :PER-LINE-PREFIX argument does not evaluate to a string.
176 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
177 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
178 the manual for details; for now, package locks can be disabled by
179 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
180 if no major problems are found then it is likely that they will be
181 compiled in unconditionally.
182 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
183 unsupported operators on all platforms. To load a shared library
184 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
185 object file, link it into a shared library outside of SBCL and
186 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
188 * bug workaround: ROOM T and the GC were not working together
189 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
190 reporting facility aren't preserved. That reporting has been
191 disabled, so now until and unless someone figures out how to make it
192 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
193 * minor incompatible change: as threatened around sbcl-0.8.0, the
194 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
195 The symbols are also exported from SB-PCL for backwards
196 compatibility, but more so than before SB-PCL should be treated as
197 an implementation-internal package.
198 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
200 * the SB-SPROF contrib now works on (most) non-x86 architectures.
201 It is known as of this release not to work on the Alpha, however.
202 * fixed bug #167: errors signalled due to illegal syntax in method
203 bodies are now more legible.
204 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
205 designators and can hence be used with TYPEP.
206 * fixed bug #333: CHECK-TYPE now ensures that the type error
207 signalled, if any, has the right object to be accessed by
208 TYPE-ERROR-DATUM. (reported by Tony Martinez)
209 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
210 5.1.2.3 for multiple-value place subforms. (reported by Kalle
212 * fixed bug #334: programmatic addition of slots using specialized
213 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
214 and :ALLOCATION :CLASS slots.
215 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
216 float exponents. (rereported by Peter Seibel)
217 * fixed a bug: #\Space (and other whitespace characters) are no
218 longer considered to be macro characters in standard syntax by
220 * fixed bug: initialization of condition class metaobjects no longer
221 causes an instance of the condition to be created. (reported by
223 * fixed bug: it is now possible to have slots such that
224 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
225 description is neither :INSTANCE nor :CLASS.
226 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
227 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
228 SLOT-VALUE-USING-CLASS), and no longer causes errors with
229 non-standard SLOT-DEFINITION-ALLOCATION values.
230 * fixed bugs: various race conditions handling exiting threads.
231 CL-PPCRE's thread torture test now passes.
232 * fixed arguably-a-bug: GC time proportional to number of threads in
233 system even when most of them are idle
234 * optimization: improved performance of BIT and SBIT on bit-vectors.
235 * .fasl file incompatibility: The fasl file version number has
236 been incremented because of changes associated with package locks.
238 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
239 * minor incompatible change: the system no longer provides
240 optimization hints (conditions which are TYPEP
241 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
242 mode; these hints are still emitted when higher SPEED optimization
244 * new contrib module: a sampling profiler (profiling by statistical
245 sampling, rather than by instrumenting functions) is available as
246 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
247 * the behaviour of the standard function ED is now customizeable by
248 third parties through a hook variable: see ED's documentation
249 string for information on the protocol.
250 * the compiler no longer emits efficiency notes for (FUNCALL X)
251 when the type of X is uncertain under default optimization
253 * fixed bug 276: mutating a binding of a specialized parameter to a
254 method to something that is not TYPEP the specializer is now
256 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
257 exhibit the required behaviour on the x86 platform. (thanks to
258 Peter van Eynde, Eric Marsden and Bruno Haible)
259 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
260 even for difficult arguments. (reported by Peter Graves)
261 * fixed bug 141a: the backquote printer now descends quoted
263 * fixed another bug in backquote printing: no more destructive
264 modification of the form's list structure. (reported by Brian
266 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
267 move between its address being taken and the call to
268 interrupt_thread, fixing a crashing race condition.
269 * the SB-POSIX contrib implementation has been adjusted so that it
270 no longer exhibits ridiculously poor performance when constructing
271 instances corresponding to C structs.
273 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
274 * minor incompatible change: the sb-grovel contrib now treats C
275 structures as alien (in the sense of SB-ALIEN) objects rather than
276 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
277 has implications for memory management of client code
278 (sb-grovel-returned objects must now be manually managed) and for
279 type safety (alien objects now have full types).
280 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
281 used to control emission of compiler diagnostics, rather than the
282 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
283 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
284 quality should be considered deprecated.
285 * install.sh now installs the user manual as well
286 * (not quite a new documentable feature, but worth considering in
287 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
288 beginnings of a semantically meaningful condition hierarchy is
289 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
291 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
292 designator as the defaults argument.
293 * fixed bug: Displaced arrays whose displaced-to array has become
294 too small now cause ARRAY-DIMENSION to signal an error, providing
295 for safer bounds-checking. (reported by Bruno Haible)
296 * fixed bug: DEFCLASS slot definitions with identical :READER and
297 :WRITER names now signal a reasonable error. (reported by Thomas
299 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
301 * fixed bug: Invalid slot specification errors now print correctly.
302 (thanks to Zach Beane)
303 * fixed bug 320: Shared to local slot value transfers in class
304 redefinitions now happen corrently from superclasses as
305 well. (reported by Bruno Haible)
306 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
308 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
309 improper lists correctly. (reported by Bruno Haible)
310 * fixed bug 313: source-transform for <fun-name> was erroneously
311 applied to a call of a value of a variable with name <fun-name>.
312 (reported by Antonio Menezes Leitao)
313 * fixed bug 307: The obsolete instance protocol ensures that
314 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
315 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
316 remove dead unknown-values globs from the middle of the stack.
317 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
318 easier to resume long computations after using *BREAK-ON-SIGNALS*
319 to diagnose and fix failures. (thanks to Nikodemus Siivola)
320 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
321 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
322 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
323 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
324 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
326 * fixed a (fairly theoretical) bug in string printing: if
327 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
328 does not have array-element-type equal to the most general string
330 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
331 function name. (thanks to Nikodemus Siivola)
332 * fixed bug: compile-time format string checker failed on
333 non-closed ~{. (reported by Thomas F Burdick)
334 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
335 don't warn on legal loop code involving more than one aggregate
336 boolean. (thanks to Nikodemus Siivola)
337 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
338 in keyword position to MAKE-PACKAGE was not being reported as
340 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
341 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
343 * fixed bugs 280 and 312: the checking for multiple definitions in a
344 file is less likely to become confused by uses of inline
346 * fixed bug: the #S reader macro performs the keyword coercion
347 specified for slot names. (reported by Kalle Niemitalo)
348 * fixed bug: lambda lists may contain symbols whose names start with
349 & but are not lambda-list-keywords; their occurrence triggers a
351 * fixed bug 321: define-method-combination argument lambda lists do
352 not cause invalid code to be generated when &optional or &aux
353 variables have default values. (reported by Bruno Haible)
354 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
355 have CLOS classes; this appears also to have fixed reported
356 instability in stack exhaustion detection.
357 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
358 MUFFLE-WARNING when no associated restart is present is now
360 * optimization: rearranged the expansion of various defining macros
361 so that each expands into only one top-level form in a
362 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
364 * optimization: used a previously little-used slot in symbols to
365 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
367 * fixed some bugs revealed by Paul Dietz' test suite:
368 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
370 ** Fixed an optimization bug related to inheritance of initforms
371 from local to shared slots.
372 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
373 ** CAST optimizer forgot to flush argument derived type.
374 ** print/read consistency on floats is now orders of magnitude
375 more likely. (thanks also to Bruno Haible for a similar report
377 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
378 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
379 for objects of type REAL. Make it so.
380 ** SXHASH is specified (infelicitously) to respect similarity,
381 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
382 it so. (thanks to Markus Ziegler)
383 ** on the Alpha, the compiler succeeds in compiling functions
384 returning a known number of arguments greater than 63.
385 ** fixed handling of invalid NIL arguments in keyword position
387 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
389 ** provided a readably-printable representation for RANDOM-STATE
391 ** ensured that pathnames, where they have a namestring, always
392 print using #P"..." syntax.
394 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
395 * Support for the forthcoming 2.0 version of the NetBSD kernel
396 running on x86 hardware has been added. (thanks to Perry
397 E. Metzger most immediately, and others for their past work)
398 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
399 current release version; SBCL's OpenBSD support had been broken
400 since about the time of OpenBSD's switch to ELF binary format.)
401 * [placeholder for DX summary]
402 ** user code with &REST lists declared dynamic-extent, under high
403 speed or space and low safety and debug optimization policy.
404 * The manual has been converted to Texinfo format and the debugger
405 chapter from the cmucl manual has been added.
406 * A facility has been added to extract documentation strings from
407 sbcl and store them as Texinfo-formatted snippets for inclusion in
408 the manual (via Texinfo's @include directive)
409 * bug fix: compiler emitted division in optimized DEREF. (thanks for
410 the test case to Dave Roberts)
411 * bug fix: multidimensional simple arrays loaded from FASLs had fill
412 pointers. (reported by Sean Ross)
413 * bug fix: PROFILE output is printed nicely even for large numerical
414 values. (thanks to Zach Beane)
415 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
416 greater than 32 handle EOF correctly.
417 * bug fix: on X86 an immediate argument of the IMUL instruction is
418 correctly printed in disassembly. (thanks to Lutz Euler)
419 * bug fix: class slots in redefined classes preserve their old
420 values. (thanks to Bruno Haible and Nikodemus Siivola)
421 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
422 succeeds. (reported by Marco Baringer)
423 * fixed some bugs revealed by Paul Dietz' test suite:
424 ** READ-SEQUENCE now works on ECHO-STREAMs.
425 ** RATIONALIZE works more according to its specification. (thanks
428 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
429 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
430 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
431 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
432 should matter to you only if you rebind the printer control
433 variables and then find you want different bindings in the
434 debugger than in the ordinary execution of your program.)
435 * The runtime build system has been tweaked to support building
436 (on SPARC/SunOS) using a C compiler which invokes Sun's own
437 assembler and linker. (thanks to Nikodemus Siivola)
438 * Unbound, undefined, undeclared variables now trigger full
439 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
440 more often programmer error than deliberate exploitation of undefined
442 * optimization: the hash algorithm for strings has changed to one
443 that is less vulnerable to spurious collisions. (thanks to Juho
445 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
446 needless bounds checking. (thanks to Juho Snellman)
447 * optimization: implemented multiplication as a modular
448 (UNSIGNED-BYTE 32) operation on the PPC backend.
449 * fixed some bugs revealed by Paul Dietz' test suite:
450 ** ADJUST-ARRAY now copies the datum in a zero rank array if
452 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
453 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
454 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
456 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
458 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
460 ** ECHO-STREAMs no longer attempt to echo the end of file value to
461 their output stream on EOF from read.
462 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
463 have been read to end-of-file.
464 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
466 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
467 description of determination of which consecutive characters
469 ** Printing the "Space" character with escaping on now yields "#\\ ",
470 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
471 ** Reading floating-point numbers with *READ-BASE* set to a number
472 less than 10 works correctly.
473 ** Reading floating-point numbers with *READ-BASE* set to a number
474 more than 10 works correctly.
475 ** Printing with *PRINT-READABLY* targets the standard readtable, not
476 the readtable currently in effect.
478 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
479 * minor incompatible change: parsing of namestrings on a physical
480 (Unix) host has changed; numbers after the final #\. in a
481 namestring are no longer interpreted as a version field. This is
482 intented to be largely invisible to the user, except that the
483 meaning of the namestring "*.*.*" has changed: it now refers to a
484 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
485 should usually be replaced by
486 (make-pathname :name :wild :type :wild :version :wild)
487 with the added benefit that this is more likely to be portable.
488 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
489 option to OPEN now signals an error if the file being opened
490 exists; this may have an impact on existing code.
491 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
492 signals received, on the PowerPC platforms (both Linux and
493 Darwin). (thanks to Pierre Mai for pointing out the location of
495 * several fixes on OS X: The system now builds and runs cleanly on
496 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
497 (thanks to Brian Mastenbrook)
498 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
499 second-resolution: integer multiples of 1/3600 between -24 and 24.
500 (thanks to Vincent Arkesteijn)
501 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
502 type when called with 1 argument; PEEK-CHAR checked type of
503 PEEK-TYPE only after having read first character from a
504 stream. (reported by Peter Graves)
505 * bug fix: the garbage collector now has much better locality
506 behaviour, and in particular no longer treats all memory as being
507 exhausted when many small objects point to each other in a deeply
509 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
510 recognized as being TYPEP their class.
511 * bug fix: the PUSHNEW documentation string has been corrected.
512 (thanks to Vincent Arkesteijn)
513 * bug fix: defaulting of the value for the last of an atypically
514 large number of multiple values being bound was not being
515 performed correctly on the Alpha or PPC platforms
516 * optimization: implemented multiplication as a modular
517 (UNSIGNED-BYTE 32) operation on the x86 backend.
518 * optimization: SEARCH on simple-base-strings can now be open-coded.
519 (see also contrib/compiler-extras.lisp for inspiration for
520 teaching the compiler about the Boyer-Moore algorithm).
521 * value, returned by MAX (and MIN) called with several EQUALP, but
522 not EQL, arguments now does not depend on compiler settings.
523 * fixed some bugs revealed by Paul Dietz' test suite:
524 ** in stack analysis liveness information is propagated from
525 non-local entry points.
526 ** pathwise CAST removing failed when the CAST node did not start
528 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
529 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
531 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
532 host is already defined.
533 ** RENAME-FILE works on streams instead of signalling an internal
535 ** PEEK-CHAR uses the current readtable when determining whether
536 or not a character is whitespace.
537 ** MERGE-PATHNAMES handles the case when the pathname does not
538 specify a name while the default-pathname specifies a version.
539 ** Pathnames now stand a better chance of respecting print/read
541 ** Attempting to use standardized file system operators with a
542 pathname with invalid :DIRECTORY components signals a
544 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
546 ** DIRECTORY on logical pathnames is more correct.
547 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
548 signal a TYPE-ERROR if their argument is not a stream.
549 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
550 designator argument does not designate a stream.
551 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
552 examining the synonym.
553 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
555 ** OPEN and WITH-OPEN-STREAM allow opening streams with
556 element-type larger than ([UN]SIGNED-BYTE 32).
558 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
559 * When built with the :SB-FUTEX feature, threaded builds now take
560 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
561 for faster/more reliable mutex and condition variable support.
562 * Incompatible change (but one you probably shouldn't have been using
563 anyway): the interface and code for arbitrating between multiple
564 threads in the same user session has been redesigned.
565 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
566 error when called without an explicit environment argument.
567 (thanks to Nikodemus Siivola)
568 * bug fix: buffered :DIRECTION :IO streams are less likely to become
569 confused about their position. (thanks to Adam Warner and Gerd
571 * bug fix: Pretty printing backquoted forms with unquotations in the
572 argument list position of various code constructs such as LAMBDA
573 now works correctly. (reported by Paul Dietz)
574 * bug fix: Pretty printing unquotations no longer loses all
575 stream position information.
576 * optimization: performance of string output streams is now less
577 poor for multiple small sequence writes.
578 * optimization: performance of CSUBTYPEP in the presence of complex
579 expressions involving CONS and NOT many times has been improved.
580 (reported by Paul Dietz)
581 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
583 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
584 streams for the REPL, for the debugger, and so on. (thanks to
586 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
587 * fixed some bugs revealed by Paul Dietz' test suite:
588 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
589 is ignored if it is too small, rather than propagating through
590 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
591 ** extremely complex negations of CONS types were not being
592 sufficiently canonicalized, leading to inconsistencies in
594 ** VALUES tranformer lost derived type.
596 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
597 * fixed a bootstrapping bug: the build process no longer assumes
598 that the various BOOLE-related constants have the same value in
599 host and target lisps. (noted by Paul Dietz' test suite on an
600 SBCL binary built from CLISP)
601 * The system can now be dynamically linked on the MIPS platform,
602 which enables dynamic loading of foreign code from Lisp. (thanks
603 to Ralf Baechle for discussions on the MIPS ABI)
604 * The system now records debugging information for its own source
605 files in a filesystem-position-independent manner, relative to
606 the "SYS" logical host.
607 * fixed a compiler bug: MV-LET convertion did not check references
608 to the "max args" entry point. (reported by Brian Downing)
609 * tweaked disassembly notes to be less confident about proclaiming
610 some instruction as an LRA. (thanks to Brian Downing)
611 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
612 now each have their own history, command character, and other
613 characteristics. (thanks to David Lichteblau)
614 * fixed some bugs revealed by Paul Dietz' test suite:
615 ** compiler failure in compiling LOGAND expressions including a
617 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
618 shift greater than 32.
619 ** FLUSH-DEST did not mark blocks for type check regeneration.
620 ** HANDLER-CASE failed to accept declarations in handler clauses
621 in some circumstances.
623 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
624 * New code in contrib/sb-introspect (still probably not entirely
625 stable yet) provides some support for smart Lisp development
626 environments like SLIME.
627 * The conditions signalled for errors occurring when loading .fasl
628 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
629 in a way which should help ASDF recover gracefully.
630 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
631 stuff has been cleaned up. If you code contrib/ stuff, this might
632 affect you, and you can look at contrib/README, contrib/STANDARDS,
633 and/or the 0.8.4.27 diff to check.
634 * In full calls the compiler now does not generate checks for declared
635 argument types for all arguments.
636 * various threading fixes
637 ** and some experimental patches which didn't make it into
638 the main tree for this release, but which are shipped in
639 contrib/experimental-thread.patch as a possible fix for some
640 failures (deadlock, spinning...) in GC-intensive multithreaded
642 * fixed PPC build problem (source code incompatibility of different
643 library versions): added offsetof() hackery which attempts to divine
644 where glibc maintainers put uc_mcontext today
645 * fixed bug 282: compiler does not trust type assertions while passing
646 arguments to a full call.
647 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
648 &OPTIONAL VALUES type parameter.
649 * fix bug 214: algorithm for noting rejected templates is now more
650 similar to that of template seletion. (also reported by rydis on
652 * fixed bug 141b: printing backquoted information readably and prettily
653 inserts a space where necessary.
654 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
655 (CHARACTER 10) are now reported as errors, rather than propagated
656 as unknown types. (reported by piso on #lisp)
657 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
658 with values NIL and :ERROR. (thanks to Milan Zamazal)
659 * fixed bug 191c: CLOS now does proper keyword argument checking as
660 described in CLHS 7.6.5 and 7.6.5.1.
661 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
662 longer raise an error; further, using a list as a for-as-arithmetic
663 counter now raises a meaningful error.
664 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
665 now understood by sequence creation functions such as MAKE-SEQUENCE
667 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
668 a STRING-INPUT-STREAM.
669 * compiler enhancement: SIGNUM is now better able to derive the type
671 * type declarations inside WITH-SLOTS are checked. (reported by
673 * fixed some bugs revealed by Paul Dietz' test suite:
674 ** incorrect optimization of TRUNCATE for a positive first
675 argument and negative second.
676 ** compiler failure in let-convertion during flushing dead code.
677 ** compiler failure while deriving type of TRUNCATE on an
678 interval, containing 0.
679 ** ASH of a negative bignum by a negative bignum count now returns
681 ** intersection of CONS types now canonicalizes properly, fixing
682 inconsistencies in SUBTYPEP.
684 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
685 * incompatible change: The --disable-debugger command line
686 option now clobbers the debugger at a more fundamental
687 level, by redefining #'INVOKE-DEBUGGER instead of by
688 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
689 is specified by ANSI to ignore *DEBUGGER-HOOK* and
690 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
691 enter the debugger REPL and then suffer recursive errors
692 because *DEBUG-IO* is also messed up in --disable-debugger mode;
693 while under the new system, BREAK in --disable-debugger mode
694 terminates the system just as an unhandled error would.
695 * fixed compiler performance when processing loops with a step >1;
696 * bug fix: DOCUMENTATION now retrieves generic function
697 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
698 support has been systematized, and now supports the methods
699 specified by ANSI, along with a default method and a method for
700 slot documentation. (reported by Nathan Froyd)
701 * bug fix: effective methods associated with a generic function are
702 no longer cached over a change of that generic function's method
703 combination. (reported by Andreas Fuchs)
704 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
705 names a non-existent directory. (thanks to Andreas Fuchs)
706 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
707 the PPC platform, lead to stack corruption; this has been fixed.
708 (reported by Rainer Joswig)
709 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
710 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
711 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
712 arithmetic in general, is now much more reliable on the PPC
714 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
715 platform now returns the right answer.
716 * optimization: restored some effective method precomputation in
717 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
718 precomputation is now tunable.
719 * optimization: compiler-internal data structure use has been
720 reviewed, and changes have been made that should improve the
721 performance of the compiler by about 20%.
722 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
723 simple-base-strings and simple-bit-vectors is improved.
724 * optimization: the optimization of 32-bit logical and arithmetic
725 functions introduced in version 0.8.3 on the x86 has been
726 implemented on the mips, ppc and sparc platforms; an
727 implementation of the same facility, but for 64-bit arithmetic,
728 has been added for the alpha.
729 * microoptimization: the compiler is better able to make use of the
730 x86 LEA instruction for multiplication by constants.
731 * bug fix: in some situations compiler did not report usage of
732 generic arithmetic in (SPEED 3) policy.
733 * bug 145b fix: compiler used wrong type specifier while converting
734 MEMBER-types to numeric.
735 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
737 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
738 index argument now works properly on the Alpha platform.
739 * bug fix: floating point exception treatment on the Alpha platform
741 * bug fix: FILE-POSITION works much better on string input and
742 output streams. (thanks to Nikodemus Siivola)
743 * bug fix: many threading/garbage collection symptoms sorted.
744 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
745 might be pseudo-atomic.
746 * internal change: Stopping for GC is now done with signals not
747 ptrace. GC is now done in whichever thread wanted it, instead of
749 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
751 * bug fix: result form in DO is not contained in the implicit
753 * incompatible change: ICR structure is changed; the value part of
754 CONTINUATION is now called LVAR; corresponding functions are
755 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
756 * added type deriver for ISQRT (thanks to Robert E. Brown).
757 * bug fix: better support for loading from the command line when an
758 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
760 * fixed some bugs revealed by Paul Dietz' test suite:
761 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
762 ** ROUND and FROUND now give the right answer when given very
763 small float arguments.
764 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
766 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
767 ** (GCD 0 <negative-integer>) returned <negative-integer>.
768 ** LCM should return a non-negative integer.
769 ** PARSE-INTEGER returned the index of a terminator instead of the
770 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
771 ** PARSE-INTEGER returned an incorrect index being applied to a
773 ** LCM with two arguments of 0 returns 0 rather than signalling
775 ** unsigned addition of a 32-bit constant with the high bit set no
776 longer causes an internal compiler error on the x86.
777 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
778 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
779 with negative last argument.
780 ** byte specifiers with zero size and position no longer cause
781 an error during type derivation.
782 ** bignum multiplication on the Alpha platform now returns the
784 * porting: The system now builds on SuSE AMD64, although it still
785 generates a 32-bit binary.
786 * .fasl file incompatibility: The fasl file version number has
787 been incremented (because of the changes to internal compiler
788 data structures referred to above).
790 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
791 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
792 more accurately, on the Darwin kernel running on PowerPC hardware.
793 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
794 * Compiler code deletion notes now signal a condition of type
795 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
796 an associated MUFFLE-WARNING restart.
797 * The compiler now performs limited argument count validation of
798 constant format strings in FORMAT, and where appropriate in ERROR,
799 CERROR and WARN. (thanks to Gerd Moellmann)
800 * New ASDF-INSTALL contrib can be used for automatic download and
801 installation of third-party Lisp code from CCLAN or other sites
803 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
804 which forces another thread to execute a function supplied by the
806 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
807 now accept and act upon their :ELEMENT-TYPE keyword argument.
808 (reported by Martin Atzmueller, Edi Weitz)
809 * bug fix: FILE-POSITION now accepts position designators up to
810 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
811 is the greater. (thanks to Patrik Nordebo)
812 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
813 Kalle Olavi Niemitalo)
814 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
815 printable. (reported by Eric Marsden)
816 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
817 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
818 * x86 bug fix in control stack exhaustion checking: now shows backtrace
819 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
820 (thanks to Stig Sandoe)
821 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
822 UNREAD-CHAR now correctly includes the unread character in the
823 target sequence. (thanks to Gerd Moellmann)
824 * bug fix in threaded builds: the system can now be suspended and
825 resumed by shell job control with minimal disruption.
826 * bug fixes in times and timezones >2038AD
827 * better handling of "where is GNU make?" problem in build scripts
828 (thanks to Nikodemus Siivola)
829 * new optimization: inside a named function any reference to a
830 function with the same name is considered to be a self-reference;
831 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
832 optimization quality.
833 * new optimization on x86: logical functions and + now have
834 optimized (UNSIGNED-BYTE 32) versions, which are automatically
835 used when the result is truncated to 32 bits.
836 * VALUES declaration is partially enabled.
837 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
838 * fixed some bugs revealed by Paul Dietz' test suite:
839 ** The system now obeys the constraint imposed by
840 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
841 types form a lattice under type intersection.
842 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
843 ** ASSOC now ignores NIL elements in an alist.
844 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
845 and (1+ MOST-POSITIVE-FIXNUM) answers.
846 ** The addition of a method with invalid qualifiers to a generic
847 function does not cause an error to be signalled immediately;
848 a warning is signalled, and the error is generated only on
849 calling the generic function.
850 * changed .fasl file version number, in part to add type codes for
851 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
852 obscure ANSI requirements
854 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
855 * fixed bug 148: failure to inline-expand a local function left
856 garbage, confusing the compiler.
857 * fixed bugs 3cd: structure slot readers perform type check if the
858 slot can have an invalid value (i.e. it is either not initialized
859 or can be written with a less specific slot writer).
860 * bug fix: the compiler now traps array references to elements off
861 the end of an array; previously, the bounds checking in some
862 circumstances could go off-by-one.
863 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
865 * type declarations for array element types now obey the description
866 on the CLHS page "Declaration TYPE", as per discussions on
867 sbcl-help around 2003-05-08. This means that a declaration
868 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
869 declaration, all references to BAR will be asserted or assumed
870 (with THE, so dependent on compiler policy) to involve objects of
871 type FOO. Note that no such declaration is implied in
872 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
873 * declared types of functions from the "Conditions"
874 chapter. (reported by Paul Dietz)
875 * bug fix: CERROR accepts a function as its first argument.
876 * bug fix: NTH an NTHCDR accept a bignum as index
877 arguments. (reported by Adam Warner)
878 * optimization: character compare routines now optimize comparing
879 against a constant character. (reported by Gilbert Baumann)
880 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
881 argument now works properly on the MIPS platform.
882 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
883 * repeated evaluation of the same DEFSTRUCT, a slot of which is
884 declared to have a functional type, does not cause an error
886 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
887 many :DEFAULT-INITARGS. (reported by Istvan Marko)
888 * fixed bug: if last continuation of a deleted block has a
889 destination, this destination should be deleted too. (reported by
891 * fixed a bug in the bootstrap process: the host compiler's values
892 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
893 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
894 test case from Patrik Nordebo)
895 * improved the ability of the disassembler on the PPC platform to
896 provide helpful disassembly notes.
897 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
898 the class in more cases than previously.
899 * bug fix: FILE-POSITION now understands :START and :END for
900 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
901 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
902 8) data. (thanks to David Lichteblau)
903 * bug fix: it is possible to add a method to a generic function
905 * bug fix: reader failed to signal END-OF-FILE inside an
906 object representation. (reported by Nikodemus Siivola)
907 * fixed some bugs revealed by Paul Dietz' test suite:
908 ** LAST and [N]BUTLAST should accept a bignum.
909 ** condition slot accessors are methods.
910 ** (VECTOR NIL) is a subtype of STRING.
912 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
913 * minor incompatible change: some nonsensical specialized lambda
914 lists (used in DEFMETHOD) which were previously ignored now signal
916 * minor incompatible change: the system is now aware of the types of
917 variables in the COMMON-LISP package, and will signal errors for
918 most violations of these type constraints (where previously they
919 were silently accepted).
920 * minor incompatible change: COMPILE-FILE now uses the freedom
921 afforded (ANSI 3.2.2.3) to use derived function types for
922 functions defined in the same file. This also permits the system
923 to warn on static type mismatches and function
924 redefinition. (Currently it does not work with high DEBUG level.)
925 * minor incompatible change: VALUES declaration is disabled.
926 * When issuing notes, the compiler now signals a condition of type
927 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
928 restart for use in user handlers. It is expected that the
929 COMPILER-NOTE condition will eventually become a condition
930 supertype to a hierarchy of note types, which will then be
931 handleable in a similar fashion. However, at the moment, no such
932 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
933 but it's an implementation detail, not a classification for the
935 * Changes in type checking closed the following bugs:
936 ** type checking of unused values (192b, 194d, 203);
937 ** template selection based on unsafe type assertions (192c, 236);
938 ** type checking in branches (194bc).
939 * A short form of VALUES type specifier has ANSI meaning (it has
940 increased the number of situations when SBCL cannot perform type
942 * fixed bug in DEFSTRUCT: once again, naming structure slots with
943 keywords or constants is permissible.
944 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
945 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
946 classes. (thanks to Antonio Martinez)
947 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
948 outputs FRESH-LINE or TERPRI, and no longer converts its stream
949 argument to a pretty-print stream. Instead, it leaves any such
950 operations to DESCRIBE-OBJECT methods.
951 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
953 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
954 lambda lists. (thanks to David Lichteblau)
955 * bug fix: Defining a generic function with a :METHOD-CLASS being a
956 subclass of STANDARD-METHOD no longer causes stack exhaustion.
957 (thanks to Gerd Moellmann)
958 * fixed bug 246: increased compilation speed of long
959 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
961 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
963 * DEFCONSTANT now throws a condition of type
964 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
965 constant to a non-EQL value; CONTINUE and ABORT restarts
966 respectively change and preserve the value.
967 * fixed bug 63: The code walker, part of the implementation of CLOS,
968 is now better at handling symbol macros.
969 * bug fix: There is no longer an internal implementation type named
970 CL:LENGTH. (reported by Raymond Toy)
971 * bug fix: In macro-like defining macros/special operators the
972 implicit block does not enclose the lambda list.
973 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
974 enclosed in parentheses) are not suitable as type specifiers, and
975 their use properly signals an error now.
976 * bug fix: An argument count mismatch for a type specifier in code
977 being compiled no longer causes an unhandled error at compile
978 time, but signals a compile-time warning.
979 * fixed simple vector readable printing
980 * bug fix: DESCRIBE takes more care over whether the class
981 precedence list slot of a class is bound before accessing it.
982 (reported by Markus Krummenacker)
983 * bug fix: FORMATTER can successfully compile pretty-printer format
984 strings which use variants of the ~* directive inside.
985 * bug fix: SEARCH now applies its TEST predicate to the elements of
986 the arguments in the correct order. (thanks to Wolfhard Buss)
987 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
988 in type inference. (thanks to Robert E. Brown)
989 * bug fix: Reading in symbols with an explicit package name of ""
990 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
991 * fixed some bugs revealed by Paul Dietz' test suite:
992 ** NIL is now allowed as a structure slot name.
993 ** Arbitrary numbers, not just REALs, are allowed in certain
994 circumstances in LOOP for-as-arithmetic clauses.
995 ** Multiple class redefinitions before slot access no longer
997 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
998 association between the name and a class.
999 ** Generic functions with non-standard method-combination and over
1000 five methods all of which return constants no longer return NIL
1001 after the first few invocations. (thanks to Gerd Moellmann)
1002 ** CALL-NEXT-METHOD with no arguments now passes the original
1003 values of the arguments, even in the presence of assignment.
1004 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1006 ** DISASSEMBLE works with closures and funcallable instances.
1007 ** ADD-METHOD now returns the generic function, not the new method.
1008 ** FIND-METHOD signals an error if the lengths of the specializers
1009 is incompatible with the generic function, even if the ERRORP
1011 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1012 which its argument is a member.
1013 ** DEFCLASS only redefines the class named by its class-name
1014 argument if that name is the proper name of the class;
1015 otherwise, it creates a new class.
1016 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1017 of the UNBOUND-SLOT condition to the name of the slot.
1018 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1019 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1020 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1021 SLOT-MAKUNBOUND in the specified fashion.
1023 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1024 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1025 cross-compilation host. As a consequence, we can now bootstrap our
1026 way up to SBCL starting with a bare gcc toolchain and human-readable
1027 source code (first the source to CLISP, then the source to SBCL).
1028 * A contributed module containing a partial implementation of the
1029 simple-streams interface has been included. (thanks to Rudi
1031 * A contributed module implementing the RFC1321 Message Digest
1032 Algorithm, known as MD5, has been included.
1033 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1034 no longer has any effect, as the code controlled by this feature
1035 has been deleted. (As far as we know, no-one has ever built using
1036 this feature, and its semantics were confused in any case).
1037 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1038 work on conditions (as required by the ANSI specification),
1039 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1040 expected behaviour on conditions. Users should note, however,
1041 that such behaviour is not required by the ANSI specification,
1042 and so use of this behaviour may render their code unportable.
1043 * fixed some bugs revealed by Paul Dietz' test suite:
1044 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1046 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1047 ** SXHASH on condition objects no longer returns NIL.
1048 ** :ALLOCATION :CLASS slots are better treated; their values are
1049 updated on class redefinition, and initforms inherited from
1050 superclasses are applied.
1051 ** REMOVE-METHOD returns its generic function argument even when
1052 no method was removed.
1053 ** SHARED-INITIALIZE now initializes the values of the requested
1054 slots, including those with :ALLOCATION :CLASS.
1055 ** ALLOCATE-INSTANCE now works on structure classes defined via
1056 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1058 ** SLOT-EXISTS-P now works on conditions, as well as structures
1060 ** MAKE-LOAD-FORM now has the required methods on
1061 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1062 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1063 keyword, and now implements the SLOT-NAMES argument.
1064 ** methods with &OPTIONAL arguments no longer allow too many
1065 arguments to be passed in the call without error.
1066 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1067 option is consistent with the required arguments of the generic
1068 function lambda list.
1069 * bug fix: REQUIRE accepts a string designator. (Thanks to
1071 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1072 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1073 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1074 * bug fix: The compiler now checks for duplicated variables in macro
1076 * bug fix: SETQ on globals returns the correct value.
1077 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1078 not a COMPILER-ERROR (followed by some other strange error on
1079 choosing the CONTINUE restart).
1080 * bug fix: make.sh and friends are now more consistent in the way that
1081 they look for GNU "make".
1083 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1084 * experimental native threads support (on x86 Linux >=2.4 only).
1085 This is not compiled in by default: you need to add :SB-THREAD to
1086 the target features. See the "Beyond ANSI" chapter of the manual
1088 * fix for longstanding nonANSIism: The old distinction between
1089 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1090 The return value from CL:FIND-CLASS is now a CLOS class, and
1091 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1092 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1093 * An interface to the MetaObject Protocol, as described in Kiczales,
1094 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1095 MIT Press, 1991, is available from the SB-MOP package.
1096 * incompatible change: the SB-PCL package should now be considered
1097 a private implementation detail, and no longer a semi-private MOP
1099 * minor incompatible change: due to rearrangement for threads, the
1100 control stack and binding stack are now allocated at arbitrary
1101 addresses instead of being hardcoded per-port. Users affected by
1102 this probably have to be doing advanced things with shared
1103 libraries, and will know who they are.
1104 * minor incompatible change: Previously, all --eval forms used were
1105 processed with READ before any of them were processed with EVAL.
1106 Now each --eval form is processed with both READ and EVAL before
1107 the next --eval form is processed. (Thus package operations like
1108 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1109 work as the user might reasonably expect.)
1110 * minor incompatible change: *STANDARD-INPUT* is now only an
1111 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1113 * minor incompatible change: Y-OR-N-P is now character-oriented,
1114 not line oriented. Also, YES-OR-NO-P now works without errors.
1115 (thanks to Antonio Martinez)
1116 * sb-aclrepl module improvements: an integrated inspector, added
1117 repl features, and a bug fix to :trace command.
1118 * Known functions, which cannot be open coded by the backend, are
1119 considered to be able to check types of their arguments. (fixing
1120 a bug report by Nathan J. Froyd)
1121 * fixed a bug in computing method discriminating functions: It is
1122 now possible to define methods specialized on classes which have
1123 forward-referenced superclasses. (thanks to Gerd Moellmann)
1124 * fixed evaluation order in optional entries (reported by Gilbert
1126 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1127 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1128 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1129 required-by-AMOP NAME argument, as well as CLASS and
1130 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1131 * fixed bug 20: DEFMETHOD can define methods using names that are
1132 not the proper names of classes to designate class specializers.
1133 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1134 Unix file descriptors, instead of blowing up. (thanks to Antonio
1136 * Garbage collection refactoring: user-visible change is that a
1137 call to the GC function during WITHOUT-GCING will not do garbage
1138 collection until the end of the WITHOUT-GCING. If you were doing
1139 this you were probably losing anyway.
1140 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1141 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1142 TYPEP the latter but not the former.
1143 * The compiler issues a full WARNING for calls to undefined functions
1144 with names from the CL package.
1145 * MAP-INTO for a vector destination is open coded. (reported by
1146 Brian Downing on c.l.l)
1147 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1148 documentation string.
1149 * fixed some bugs revealed by Paul Dietz' test suite:
1150 ** COPY-ALIST now signals an error if its argument is a dotted
1152 ** Condition slots are now accessed more correctly in the presence
1153 of multiple initargs for a given slot.
1154 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1155 exclude restarts of the same name associated with a different
1157 ** DEFCLASS of forward-referenced classes with another
1158 forward-referenced class in the superclasses list no longer
1160 ** Condition slots are now initialized once each, not multiple
1161 times. (thanks to Gerd Moellmann)
1162 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1164 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1165 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1166 arguments contain duplicated elements.
1167 ** RESTART-CASE understands local macros.
1168 ** RESTART-CASE associates exactly its own restarts with a condition.
1169 ** ENDP in safe mode checks its argument to be of type LIST.
1170 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1171 in question is unbound.
1172 ** Optimized MAKE-INSTANCE functions no longer cause internal
1173 assertion failures in the presence of duplicate initargs.
1174 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1175 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1176 propagated even in the ctor optimized implementation of
1178 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1180 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1181 * a better implementation of SXHASH on (simple) bit vectors,
1182 measured both in execution speed and in distribution of results
1183 over the positive fixnums, has been installed. Likewise, a better
1184 implementation of EQUAL for simple bit vectors is now available.
1185 * fixed CEILING optimization for a divisor of form 2^k.
1186 * fixed bug 240 (emitting extra style warnings "using the lexical
1187 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1188 by Antonio Martinez)
1189 * fixed SXHASH, giving different results for NIL depending on type
1190 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1191 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1192 variable. (found by Rolf Wester)
1193 * SBCL does not ignore type declarations for special
1194 variables. (reported by rif on c.l.l 2003-03-05)
1195 * some bug fixes in contrib/sb-aclrepl/
1196 * fixed some bugs revealed by Paul Dietz' test suite:
1197 ** a bug in the CONS type specifier, whereby the CAR and CDR
1198 types got intertwined, has been fixed;
1199 ** the type system is now able to reason about the interaction
1200 between INTEGER and RATIO types more completely;
1201 ** APPEND, [N]REVERSE and NRECONC check that those their
1202 arguments, which must be proper lists, are really so;
1203 ** An array specialized to be unable to hold elements has been
1204 implemented, as required -- yes, really -- by ANSI;
1205 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1206 on malformed property lists;
1208 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1209 * incompatible packaging change: in line with Unix convention,
1210 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1211 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1212 * REQUIRE and PROVIDE are now optionally capable of doing something
1213 useful. See the documentation string for REQUIRE.
1214 * infrastructure for a managed SBCL contrib system: contributed
1215 modules in this release include:
1216 ** the ASDF system definition facility;
1217 ** an interface to the BSD Sockets API;
1218 ** an ACL-like convenience interface to the repl;
1219 (thanks to Kevin Rosenberg)
1220 ** an implementation of ROTATE-BYTE, with efficient implementation
1222 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1223 gives a closer approximation to the right answer than previously.
1224 (thanks to Raymond Toy)
1225 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1226 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1227 optional environment arguments, as required by ANSI.
1228 * fixed bugs in other functions taking environment objects, allowing
1229 calls with an explicit NIL environment argument to be compiled
1231 * fixed bug 228: primary return values from
1232 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1233 COMPILE or FUNCTION.
1234 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1235 :TYPE will no longer signal errors on innocuous objects.
1236 * fixed bug 231b: SETQ is better at respecting type declarations in
1237 the lexical environment.
1238 * fixed a bug in DEFCLASS: classes named by symbols with no or
1239 unprintable packages can now be defined.
1240 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1241 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1242 * fixed compiler failure related to checking types of functions.
1243 (reported by Robert E. Brown)
1244 * the compiler is now much more consistent in its error-checking
1245 treatment of bounding index arguments to sequence functions: in
1246 (SAFETY 3) code, errors will be signalled in almost all cases if
1247 invalid sequence bounding indices are passed to functions defined
1248 by ANSI to operate on sequences.
1249 * fixed a bug in the build procedure: documentation of SBCL-specific
1250 packages is now preserved and available in the final Lisp image.
1251 * lifted FDEFINITION lookup out of loops in the implementation of
1252 many list operations. (thanks to Robert E. Brown)
1253 * fixed a bug in the reader: the #n# reader macro now works for
1254 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1255 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1256 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1257 E. Brown and Tony Martinez respectively)
1258 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1259 SLOT-BOUNDP: in particular, it is now able to optimize them much
1260 better, and is now not vulnerable to having packages renamed.
1261 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1262 always called when a slot is not present in an instance. (thanks
1264 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1265 longer signal an error on unreading a character following EOF on
1266 the previous constituent stream. (thanks to Tony Martinez)
1267 * fixed some bugs revealed by Paul Dietz' test suite:
1268 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1269 not just nonnegative fixnums;
1270 ** the logical bit-array operators such as BIT-AND now accept an
1271 explicit NIL for their "opt-arg" argument (to indicate a
1272 freshly-consed result bit-array);
1273 ** ELT now signals an error on an invalid sequence index in safe
1275 ** the type system is now cleverer about negations of numeric
1276 types, and consequently understands the BIGNUM and RATIO types
1278 ** the type system is now cleverer about the interaction between
1279 INTEGER and RATIO types: while bugs still remain, many more
1280 cases are accurately computed;
1281 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1282 if it is in the last clause;
1283 ** CONSTANTLY now correctly returns a side-effect-free function in
1285 ** DECLARE is no longer treated as a special-operator; in
1286 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1287 * incremented fasl file version number due to the change in the
1288 DEFSTRUCT-SLOT-DESCRIPTION structure.
1290 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1291 * minor incompatible change: code processed by the "interpreter" or
1292 EVAL now has a compilation optimization policy of (DEBUG 2)
1293 (changed from (DEBUG 1)) to improve debuggability of interactive
1294 development, and to allow the use of the debug RETURN command in
1296 * an experimental implementation of the RETURN command for the
1297 debugger has been included. (thanks to Frederik Kuivinen)
1298 * fixed bug 62: constraints were not propagated into a loop.
1299 * fixed bug in embedded calls of SORT (reported and investigated by
1301 * fixed some bugs revealed by Paul Dietz' test suite:
1302 ** printing and reading of arrays with some dimensions having
1303 length 0 (thanks to Gerd Moellmann);
1304 ** BOA constructor with &AUX argument without a default value does
1305 not cause a type error;
1306 ** CONSTANTP now returns true for all self-evaluating objects.
1308 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1309 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1310 accessors that are related by inheritance, as specified in the
1311 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1312 Valtteri Vuorikoski)
1313 * The compiler is now able to inline functions that were defined in
1314 a complex lexical environment (e.g. inside a MACROLET).
1315 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1317 * Support for the upcoming FreeBSD-5.0 release has been included.
1318 (thanks to Dag-Erling Smorgrav)
1319 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1320 effect when it is not in a toplevel context.
1321 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1322 stand a better chance of being correct. (thanks to Gerd
1324 * fixed bug in COERCE, which now signals an error on coercing a
1325 rational to a bounded real type which excludes the expected
1327 * The compiler is now able to derive types more accurately from the
1328 COERCE and COMPILE functions.
1329 * fixed bug 223: functional binding is considered to be constant
1330 only for symbols in the CL package.
1331 * fixed bug 231: SETQ did not check the type of a variable being set
1332 (reported by Robert E. Brown)
1333 * A new optimization for MAKE-INSTANCE has been included, fixing
1334 various bugs (including relating to :ALLOCATION :CLASS slots and
1335 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1337 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1338 ** As required by ANSI, LOOP now disallows anonymous collection
1339 clauses such as COLLECT I in conjunction with aggregate boolean
1340 clauses such as THEREIS (= I 1);
1341 ** LOOP now signals an error when any variable is reused in the
1342 same loop (including the potentially useful construct analogous
1343 to WITH A = 1 WITH A = (1+ A);
1344 ** IT is only a special loop symbol within the first clause of a
1345 conditional loop clause;
1346 ** LOOP with a typed iteration variable over a hashtable now
1347 signals a type error iff it should.
1348 * fixed some other bugs revealed by Paul Dietz' test suite:
1349 ** FILE-STREAM now names the class previously known as FD-STREAM;
1350 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1351 argument) no longer signals an error;
1352 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1353 of the slot symbol, rather than using the current package
1354 ((:CONC-NAME "") continues to intern the slot's name in the
1356 * incremented fasl file version number, because of the incompatible
1357 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1358 of the new implementation of DEFINE-COMPILER-MACRO.
1360 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1361 * Support for building SBCL for MIPS platforms running in
1362 little-endian mode has now been checked in, and basic
1363 functionality on said platforms verified.
1364 * minor incompatible change: PCL now records the pathname of a file
1365 in which methods and the like are defined, rather than its
1367 * minor incompatible change: TRUENAME now considers the truename of
1368 a file naming a directory to be the pathname with :DIRECTORY
1369 component indicating that directory.
1370 * minor incompatible change: a NAMED clause in the extended form of
1371 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1372 reason for the previous behaviour is unclear.
1373 * more systematization and improvement of CLOS and MOP conformance
1374 in PCL (thanks to Gerd Moellman and Pierre Mai):
1375 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1377 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1378 primary methods with no specializers;
1379 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1381 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1382 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1383 FORWARD-REFERENCED-CLASSes; error reporting on
1384 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1386 ** SXHASH on CLOS instances now uses a slot internal to the
1387 instance to return different numbers on distinct instances,
1388 while preserving the same return value through invocations of
1390 ** DEFMETHOD signals errors when methods with longer incongruent
1391 lambda lists are added to generic functions;
1392 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1393 CLASS, as specified in AMOP;
1394 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1395 on the order returned by the primary method for classes of
1396 class STANDARD-CLASS;
1397 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1398 * fixed some bugs shown by Paul Dietz' test suite:
1399 ** DOLIST puts its body in TAGBODY;
1400 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1402 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1403 value producing form;
1404 ** if more variables are given to PROGV than values, extra
1405 variables are bound and made to have no value;
1406 ** NSUBSTITUTE on list arguments gets the right answer with
1408 ** ELT signals an error of type TYPE-ERROR when the index argument
1409 is not a valid sequence index;
1410 ** LOOP signals (at macroexpansion time) an error of type
1411 PROGRAM-ERROR when duplicate variable names are found;
1412 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1413 ** FUNCALL of special-operators now cause an error of type
1415 ** PSETQ now works as required in the presence of side-effecting
1416 symbol-macro places;
1417 ** NCONC accepts any object as its last argument;
1418 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1420 ** loop-for-as-package does not require a package to be explicitely
1422 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1424 * fixed bug 166: compiler preserves "there is a way to go"
1425 invariant when deleting code.
1426 * fixed bug 172: macro lambda lists with required arguments after
1427 &REST arguments now cause an error to be signalled. (thanks to
1429 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1430 bug. (thanks to Matthew Danish)
1431 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1433 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1434 arguments are lexically rebound. (thanks to Gerd Moellmann and
1436 * fixed bug 194: error messages are now more informative when there
1437 is no primary method applicable in a call to a generic
1438 function. (thanks to Gerd Moellmann)
1439 * fixed bug in command line argument checking (thanks to Julian
1441 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1442 (thanks to Matthew Danish)
1443 * incremented fasl file version number, because of the
1444 SXHASH-related changes in the layout of CLOS data structures
1446 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1447 * minor incompatible change: The runtime (the Unix executable named
1448 "sbcl") is now much pickier about the .core files it will load.
1449 Essentially it now requires .core files to descend from the same
1450 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1451 as the runtime does. (The intent is to prevent the crashes which
1452 can occur, and which can even be reported as mysterious failures,
1453 when people patch the sources or change the build parameters
1454 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1455 sbcl and .core files.)
1456 * fixed bug: VALUES-LIST is no longer optimized away.
1457 * fixed bug 142: The FFI conversion of C string values to Lisp
1458 string values no longer conses excessively. (thanks to Nathan
1459 Froyd porting Raymond Toy's fix to CMU CL)
1460 * began to systematize and improve MOP conformance in PCL (thanks to
1461 Nathan Froyd, Gerd Moellman and Pierre Mai):
1462 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1464 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1465 ** FINALIZE-INHERITANCE is now called on class finalization;
1466 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1467 argument precedence order.
1468 * fixed bug 202: The compiler no longer fails on functions whose
1469 derived types contradict their declared type.
1470 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1471 so it can be non-toplevel.
1472 * The fasl file version number has changed (because of the new
1473 implementation of DEFMACRO).
1474 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1475 safe code, that any length requirement by their type-specifier
1476 argument is valid. The exceptions to this are described in bug
1478 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1479 functions now check the types of their inputs as required by ANSI.
1480 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1481 introduced symbol is DECLAREd to be SPECIAL.
1482 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1483 * fixed bug: :COUNT argument to sequence functions may be negative
1484 * fixed bug: body of DO-SYMBOLS may contain declarations
1485 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1486 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1487 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1488 symbol macro only once
1489 * fixed printing of call frame when argument list is unavailable
1490 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1491 * compiler no longer signals WARNING on unknown keyword
1494 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1495 * A beta-quality port to the mips architecture running Linux,
1496 based on the old CMUCL backend, has been made. It has been tested
1497 on a big-endian kernel, and works sufficiently well to be able to
1498 rebuild itself; it has not been tested in little-endian mode.
1499 * fixed an inconsistency between gencgc.c and purify.c which made
1500 dumping/loading .core files unreliable
1501 * fixed bug 120a: The compiler now deals correctly with IFs where
1502 the consequent is the same as the alternative, instead of
1503 misderiving the return type. (thanks to Alexey Dejneka)
1504 * fixed bug 113: Logical pathnames are now dumpable (the logical
1505 host is resolved at load-time, throwing an error if it is not
1507 * fixed bug 174: FORMAT's error message is slightly clearer when a
1508 non-printing character is used in a format directive.
1509 * fixed several bugs in compiler checking of type declarations, i.e.
1510 violations of the Python "declarations are assertions" principle
1511 (thanks to Alexey Dejneka)
1512 * fixed several bugs in PCL's error checking (thanks to Gerd
1514 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1516 * fixed bug in compilation of functions as first class values
1517 (thanks to Antonio Martinez-Shotton)
1518 * The compiler's handling TYPE-ERRORs which it can prove will
1519 inevitably happen at runtime has been cleaned up and corrected
1520 in several ways. (thanks to Alexey Dejneka)
1521 * improved argument type checking for various basic arithmetic
1522 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1523 lavished on them in the past that they can be compiled in many
1524 ways in different special cases
1525 * fixed bug 181: compiler checks validity of user supplied type
1527 * cleaned up code flushing in optimization: Function calls which
1528 should signal errors for safety purposes (e.g. which ANSI says
1529 should signal errors when their arguments are of incorrect type)
1530 are no longer optimized away.
1531 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1532 * incremented fasl file version number, because changes in the
1533 implementation of sequence functions like COERCE caused
1534 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1535 in old inline expansions) to become undefined. (Actually these
1536 changes were later undone, so we might very well be binary
1537 compatible with 0.7.7 after all, but leaving the version number
1538 incremented seemed like the simplest and most conservative
1541 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1542 * An alpha-quality port to the parisc architecture running Linux,
1543 based on the old CMUCL backend, has been made. This, even more so
1544 than the other backends, should be considered still a work in
1545 progress; known problems include that the Linux kernel in 64-bit
1546 mode does not propagate the correct sigcontext structure to
1547 userspace, and consequently SBCL on a parisc64 kernel will not
1549 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1550 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1551 suppressed.) Also now that the compiler is looking at declarations
1552 in the environment, it checks optimization declarations as well,
1553 and suppresses inlining when (> DEBUG SPEED).
1554 * More fixes have been made to treatment of floating point exception
1555 treatment and other Unix signals. In particular, floating point
1556 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1557 * The detection and handling of control stack exhaustion (infinite
1558 or very deeply nested recursion) has changed. Stack exhaustion
1559 detection is now done by write-protecting pages at the OS level
1560 and applies at all optimization settings; when found, a
1561 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1562 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1563 be caught using IGNORE-ERRORS.
1564 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1565 correctly when transforming recursive LABELS functions to LETs.
1566 (thanks to Alexey Dejneka)
1567 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1568 that are names of constants or global variables.
1569 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1570 alien routines with docstrings.
1571 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1572 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1574 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1575 (thanks to Eric Marsden)
1576 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1577 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1578 * Bug fix: The SPARC backend can now compile functions involving
1579 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1580 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1581 to a functions expecting a PATHNAME-DESIGNATOR.
1582 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1583 lambda lists. (thanks to Alexey Dejneka)
1584 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1585 functions so that EXPORTing the name of the class doesn't cause
1586 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1587 to Antonio Martinez for reporting this)
1588 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1589 constant-folded without failing an assertion. (thanks to Einar
1590 Floystad Dorum for reporting this)
1591 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1592 types are now dealt with more correctly by the compiler.
1593 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1594 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1595 behaviour analogous to RENAME-FILE. This puts its behaviour more
1596 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1598 * The fasl file version number has changed again. (because of the
1599 bug fix involving the names of PCL MAKE-INSTANCE functions)
1601 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1602 * bug fix: Floating point exceptions are treated much more
1603 consistently on the x86/Linux and PPC/Linux platforms.
1604 * Array initialization with :INITIAL-ELEMENT is now much faster for
1605 cases when the compiler cannot open code the array creation, but
1606 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1607 array accesses have also seen a speed increase.
1608 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1609 specified. (This was at the root of some bad interactions between
1610 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1612 * bug fix: Internal error arguments for undefined functions are now
1613 computed correctly on the PPC/Linux platform.
1614 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1615 Raymond Toy's patch for CMU CL)
1616 * Support for the Solaris 9 operating environment has been included
1617 (thanks to Daniel Merritt)
1618 * A very ugly but hopefully complete draft of the missing FFI chapter
1619 of the manual has been created by reformatting the corresponding
1620 CMU CL manual chapter into (currently very ugly and incoherent)
1621 DocBook and bringing it up to date for SBCL behavior. Thus, the
1622 manual is now essentially complete, at least by my extreme
1623 once-and-only-once standards, whereby it's acceptable to refer to
1624 the doc strings of SB-EXT functions as the primary documentation.
1625 * The fasl file version number has changed again, due to cleanup of
1626 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1627 is no longer a static symbol.)
1629 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1630 * SBCL now builds with OpenMCL (version 0.12) as the
1631 cross-compilation host; also, more progress has been made toward
1632 bootstrapping under CLISP.
1633 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1635 * bug 158 fixed: The compiler can now deal with integer loop
1636 increments different from 1; fixing this turned out also to fix
1638 * bug 169 fixed: no more bogus warnings about using lexical bindings
1639 despite the presence of perfectly good SPECIAL declarations (thanks
1640 to David Lichteblau)
1641 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1642 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1643 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1645 * bug 180 fixed: Method combination specifications no longer ignore
1646 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1647 * bug fix: Structure type predicate functions now check their argument
1648 count as they should.
1649 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1650 correctly. (thanks to Pierre Mai)
1651 * minor incompatible change: The --noprogrammer option is deprecated
1652 in favor of the new --disable-debugger option, which is very similar.
1653 (The major difference is that it takes effect at a slightly different
1654 time at startup, causing handling of errors in --sysinit and
1655 --userinit files will be affected differently.) The
1656 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1657 been added to allow this functionality to be controlled from ordinary
1658 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1659 maintainers, who might want to run non-interactive scripts to
1660 build SBCL cores which will later be used interactively.)
1661 * minor incompatible change: The LOAD function no longer, when given
1662 a wild pathname to load, loads all files matching that pathname.
1663 Instead, an error of type FILE-ERROR is signalled.
1665 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1666 * bug 147 fixed: The compiler preserves its block link/count
1667 invariants more correctly now so that it doesn't crash. (thanks
1669 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1671 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1672 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1673 no-value-for-this-character as NIL (as specified by ANSI).
1674 * HOST-NAMESTRING on physical pathnames now returns a string that is
1675 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1676 (thanks to Christophe Rhodes)
1677 * The Alpha port handles icache flushing more correctly. (thanks to
1679 * More progress has been made toward bootstrapping under CLISP. (thanks
1680 to Christophe Rhodes)
1681 * The fasl file format has changed again, because dynamic loading
1682 on OpenBSD (which has non-ELF object files) motivated some cleanups
1683 in the way that foreign symbols are transformed and passed around.
1684 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1685 is no longer treated as whitespace by the reader, but instead as
1686 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1687 |A
\7fB|, instead of A as it used to.
1689 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1690 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1691 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1692 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1693 this as of 0.7.1.45, but was left out of the previous news section
1694 (thanks to Dan Barlow)
1695 * SBCL now runs on the Solaris operating system on SPARC architectures
1696 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1697 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1698 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1699 * SPARC backend cleanups, allowing builds of cores optimized for V8
1700 and V9 SPARCS, and also emission of code targeted to a particular
1701 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1703 * SBCL is closer to bootstrapping under CLISP, thanks to various
1704 fixes by Christophe Rhodes.
1705 * The fasl file format has changed again, to allow the compiler's
1706 INFO database to support symbol macros.
1707 * The user manual (in doc/) is formatted into HTML more nicely.
1708 (thanks to coreythomas)
1709 * The system is smarter about SUBTYPEP relationships, especially
1710 those involving NOT types (including types such as ATOM which are
1711 represented internally using NOT types). Thus SUBTYPEP is less
1712 likely to return (VALUES NIL NIL) in general, and in particular
1713 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1715 * The fasl file format has changed again, because the internal
1716 representation of types now includes a new slot to support the new
1717 SUBTYPEP-of-NOT-types logic.
1718 * (not a change in the main branch of SBCL, but a related prototype
1719 which can hopefully be merged into the main branch of SBCL in the
1720 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1721 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1722 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1723 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1724 dispatch macro character combinations. (thanks to Alexey Dejneka)
1725 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1726 that superficially look like logical namestrings correctly.
1727 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1729 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1730 default supplied-p) for &optional and &key arguments. (thanks to
1733 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1734 * incompatible change: The compiler is now less aggressive about
1735 tail call optimization, doing it only when (> SPACE DEBUG) or
1736 (> SPEED DEBUG). (This is an incompatible change because there are
1737 programs which relied on the old CMU-CL-style behavior to optimize
1738 away their unbounded recursion which will now die of stack overflow.)
1739 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1740 for non-GENCGC systems has been increased to 20M (since that
1741 seems much closer to the likely performance optimum for modern
1742 systems than the old 4M value was)
1743 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1744 and *DEBUG-PRINT-LEVEL*
1745 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1746 of CMU CL's support for SPARC, and various endianness and other
1747 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1748 * new syntactic sugar for the Unix command line: --load foo.bar is now
1749 an alternate notation for --eval '(load "foo.bar")'.
1751 ** The system now detects stack overflow and handles it gracefully,
1752 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1753 optimization settings. (This is a good thing in general, and
1754 its introduction in this version should be particularly timely
1755 for anyone whose code fails because of suppression of tail
1757 ** The system now hunts for the C variable "environ" in a more
1758 devious way, to avoid segfaults when the C library version differs
1759 between compile time and run time. (thanks to Christophe Rhodes)
1760 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1761 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1762 ** The compiler no longer issues bogus style warnings for undefined
1763 classes in the same source file as the DEFCLASSes which defined
1764 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1766 ** fixes in CONDITION class precedence list for undefined function
1767 errors (thanks to Alexei Dejneka)
1768 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1769 correctly. (thanks to Dan Barlow)
1770 ** portability fixes aiming at bootstrapping under CLISP (thanks
1771 to Dave McDonald and Christophe Rhodes)
1772 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1773 ** fixes in type translation and and type inference (thanks to
1775 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1776 ** various fixes in the new ports (thanks to Dan Barlow)
1777 * several changes related to debugging:
1778 ** suppression of tail recursion, as noted above
1779 ** stack overflow detection, as noted above
1780 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1781 is now the default. (For some time encapsulation has been more
1782 reliable than the breakpoint-based :ENCAPSULATE NIL
1783 implementation, at least on X86 systems; and I just noticed that
1784 encapsulation also seems closer to the spirit of the ANSI
1787 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1789 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1790 up properly again. (There was a packaging bug in 0.7.0 which
1791 left their definitions in SB-SYS::LOAD-FOREIGN and
1792 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1793 vital for most things which interface to C-level interfaces,
1794 like extensions working with sockets or databases or
1795 Perl-compatible regexes or whatever, and the need to fix
1796 this bug is the main reason that 0.7.1 was released so
1798 ** DEFGENERIC is now choosier about the methods it redefines, so that
1799 reLOADing a previously-LOADed file containing DEFGENERICs does
1800 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1801 cycle now works as it should. (thanks to Alexey Dejneka)
1802 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1803 fixed by Christophe Rhodes. (He actually submitted this patch
1804 months ago, and I delayed until after 0.7.0.)
1805 ** Bug 111 (internal compiler confusion about runtime checks on
1806 FUNCTION types) was fixed by Alexey Dejneka.
1807 * Some internal cleanups (getting rid of variables which aren't
1808 needed now that the byte interpreter is gone) caused the fasl
1809 file format number to change again.
1811 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1812 * major incompatible change: The default fasl file extension, i.e. the
1813 default extension for files produced by COMPILE-FILE, has changed
1814 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1816 ** There are many changes in the implementation of the compiler.
1817 SBCL is now essentially a compiler-only implementation of ANSI
1818 Common Lisp. EVAL still "interprets" a few special cases, but
1819 almost all the interesting cases are handled by creating
1820 a LAMBDA expression, calling COMPILE on it, then calling
1821 FUNCALL on the result.
1822 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1823 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1824 Since the code is newer, there might still be some new bugs
1825 (though not as many as before Martin Atzmueller's fixes:-). But
1826 the new code is substantially simpler and clearer, and hopefully
1827 any remaining bugs will be simpler, less fundamental, and more
1828 fixable then the bugs in the old code.
1829 ** The revised compiler is still a little unsteady on its feet.
1831 *** The debugging information it produces (particularly the names
1832 of FUNCTION objects) is sometimes much less useful than what
1833 the old compiler produced.
1834 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1835 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1836 within a MACROLET) has been temporarily weakened.
1837 ** There are new compiler optimizations for various functions:
1838 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1839 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1840 *** the math functions TRUNCATE, FLOOR, and CEILING
1841 *** the function-of-all-trades COERCE
1842 Mostly these should be transparent, but there's one
1843 potentially-annoying problem (bug 117): when the compiler
1844 inline-expands a function and does type analysis on the result,
1845 it can create control paths which have type mismatches, and
1846 when it can't prove that those control paths aren't taken,
1847 it will issue WARNINGs about the type mismatches. This is
1848 a particular problem in practice for the new sequence functions.
1849 It's not clear how this should be fixed, and for now, a
1850 workaround is given in the entry for 117 in the BUGS file.
1851 ** (Because of the interaction between the two previous items --
1852 occasional inlining problems and new inline expansions -- some
1853 of the new sequence function optimizations won't really kick in
1854 completely until debugging information, and then inlining, are
1855 straightened out in some future version.)
1856 * minor incompatible changes:
1857 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1858 DIRECTORY no longer implicitly promotes NIL slots of its
1859 pathname argument to :WILD. In particular, when you ask for the
1860 contents of a directory (which you used to be able to do without
1861 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1862 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1863 ** changes in behavior that ANSI explicitly defines to be
1864 implementation dependent:
1865 *** The new compiler-only implementation still conforms with ANSI,
1866 but acts a little different than before. Besides the obvious
1867 changes in performance tradeoffs (that the cost per form passed
1868 to EVAL has gone up, and the cost per form executed by EVAL
1869 has gone down), the behavior of the system changes a little
1870 because there are no longer any interpreted function objects.
1871 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1872 e.g. doing COMPILE on the output of interactive DEFUN is
1874 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1876 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1877 has changed from (:CL) to NIL.
1878 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1879 e.g. #\Newline and #\Tab, don't have names specified in the
1880 ANSI Common Lisp standard, is now based on their ASCII symbolic
1881 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1882 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1883 are no longer used for output.
1884 ** changes in internal implementation constants:
1885 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1886 4 million. (If your application spends a lot of time GCing and
1887 you have a lot of RAM, you might want to experiment with
1888 increasing it even more.)
1889 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1890 However, almost all old code should still continue to work without
1891 immediate update, as SB-C-CALL is now a (deprecated) nickname
1893 ** Old operator names in the style DEF-FOO are now deprecated in
1894 favor of new corresponding names DEFINE-FOO, for consistency with
1895 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1896 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1897 internal symbols, but a few supported extensions like
1898 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1899 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1900 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1901 etc. as you get deeper into recursive calls to the debugger
1902 command loop, instead of the old "5]", "5]]", "5]]]"
1903 sequence. (I was motivated to do this when squabbles between
1904 ILISP and SBCL left me very deeply nested in the debugger. In the
1905 short term, this change will probably provoke more ILISP/SBCL
1906 squabbles, but hopefully it will be an improvement in the long run.)
1907 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1908 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1909 function, since declaiming return types involving aliens is
1910 (1) annoyingly messy to do by hand and (2) vital to efficient
1911 compilation of code which calls such functions.
1912 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1913 longer reexported by the SB-EXT package. They're solely useful
1914 for alien code, so it seems more logical that you should get
1915 them from the SB-ALIEN package, not in SB-EXT.
1916 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1917 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1918 features. Instead, the code that they used to control is always
1919 built into the system.
1920 * many other bug fixes
1921 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1922 advantage of the new EVAL-WHEN stuff and to clean them up in
1923 general, and they are now more ANSI-compliant in a number of
1924 ways. Martin Atzmueller is responsible for a lot of this.
1925 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1927 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1928 *** correct ERROR type for various file operations
1929 *** some fixes for Lisp streams
1930 *** DEFMETHOD syntax checking
1931 *** changing old weird representation of debug information as
1932 strings (which, among their other deficiencies, don't transform
1933 correctly when you rename packages, and don't change their
1934 print representation when you change things like *PACKAGE*
1935 and *PRINT-LENGTH*) to symbols and lists of symbols
1936 He also made several improvements and fixed several bugs in DESCRIBE.
1937 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1939 *** misbehavior of WRITE-STRING/WRITE-LINE
1940 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1941 and several other LOOP problems as well
1942 *** DIRECTORY when similar filenames are present
1943 *** DEFGENERIC with :METHOD options
1944 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1945 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1946 *** argument ordering in FIND with :TEST option
1947 *** mishandled package designator argument in APROPOS-LIST
1948 *** various problems in the backquote readmacro
1949 *** a bug in APROPOS
1950 *** probably some others that I'm not describing very well here,
1951 since the CVS log documents them by reference to sbcl-devel
1952 messages, and the SourceForge archives aren't working well.:-(
1953 ** Dan Barlow improved the Alpha port (and is making progress on the
1954 PPC port, for those of you who think different).
1955 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
1956 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
1957 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
1958 continues to work on the SPARC port (for those of you in a position
1959 to look down upon our little PC-compatible boxes from a great height).
1960 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
1961 and uses that copy, rather than the current dynamic value, when
1962 it's trying to decide whether to truncate output. Thus e.g.
1963 (let ((*print-lines* 50))
1964 (pprint-logical-block (stream nil)
1966 (let ((*print-lines* 8))
1967 (print (aref possiblybigthings i) stream)))))
1968 should now truncate the logical block only at 50 lines, instead of
1969 often truncating it at 8 lines, as it did before.
1970 * The doc/cmucl/ directory, containing old CMU CL documentation
1971 from the time of the fork, is no longer part of the base system.
1972 SourceForge has shut down its anonymous FTP service, and with it
1973 my original plan for distributing the old CMU CL documentation
1974 there. For now, if you need these files you can download an old
1975 SBCL source release and extract them from it.
1976 * The fasl file version number changed again, for dozens of reasons,
1977 some of which are apparent above.
1979 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
1980 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
1981 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
1982 patch, so that DISASSEMBLE output is much nicer.
1983 * The code in the SB-PROFILE package now seems reasonably stable.
1984 I still haven't decided what the final interface should look like
1985 (I'd like PROFILE to interact cleanly with TRACE, since both
1986 facilities use function encapsulation) but if you have a need
1987 for profiling now, you can probably use it successfully with
1988 the current CMU-CL-style interface.
1989 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
1990 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
1991 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
1992 link now returns the dangling link itself, and for similar
1993 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
1994 link itself. (In these cases the old code signalled an error and
1995 looped endlessly, respectively.) Thus, DIRECTORY now works even
1996 in the presence of dangling and cyclic symbolic links.
1997 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
1998 is now a supported extension again, since the consensus on
1999 sbcl-devel was that it can be useful for ordinary development
2000 work, not just for debugging SBCL itself.
2001 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2002 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2003 that currently-defined functions might be redefined later with
2004 different return types.
2005 * Hash tables can be printed readably, as inspired by CMU CL code
2006 of Eric Marsden and SBCL code of Martin Atzmueller.
2007 * better error handling in CLOS method combination, thanks to
2008 Martin Atzmueller porting Pierre Mai's CMU CL patches
2009 * more overflow fixes for >16Mbyte I/O buffers
2010 * A bug in READ has been fixed, so that now a single Ctrl-D
2011 character suffices to cause end-of-file on character streams.
2012 In particular, now you only need one Ctrl-D at the command
2013 line (not two) to exit SBCL.
2014 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2016 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2017 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2018 does the right thing.
2019 * The implementation of some type tests, especially for CONDITION
2020 types, is now tidier and maybe faster, due to CMU CL code
2021 originally by Douglas Crosher, ported by Martin Atzmueller.
2022 * Some math functions have been fixed, and there are new
2023 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2024 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2025 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2026 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2027 likely to be built into sbcl-0.7.0.)
2028 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2029 CL, ported to SBCL by Martin Atzmueller)
2030 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2031 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2032 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2033 * There's a new slam.sh hack to shorten the edit/compile/debug
2034 cycle for low-level changes to SBCL itself, and a new
2035 :SB-AFTER-XC-CORE target feature to control the generation of
2036 the after-xc.core file needed by slam.sh.
2037 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2038 COMPILE-FILE is no longer supported, so that now every function
2039 gets an entry point, so that block compilation looks a little
2040 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2041 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2042 returns the number of bytes consed since the system started,
2043 rather than the number consed since the first time the function
2044 was called. (The new definition parallels ANSI functions like
2045 CL:GET-INTERNAL-RUN-TIME.)
2046 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2047 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2048 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2049 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2050 * Fasl file version numbers are now independent of the target CPU,
2051 since historically most system changes which required version
2052 number changes have affected all CPUs equally. Similarly,
2053 the byte fasl file version is now equal to the ordinary
2056 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2057 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2058 is no longer recognized. I apologize for this, because it was
2059 listed in SB-EXT as a supported extension, but I found that
2060 its existing behavior was poorly specified, as well as incorrectly
2061 specified, and it looked like too much of a mess to straighten it
2062 out. I have enough on my hands trying to get ANSI stuff to work..
2063 * many patches ported from CMU CL by Martin Atzmueller, with
2064 half a dozen bug fixes in pretty-printing and the debugger, and
2065 half a dozen others elsewhere
2066 * fixed bug 13: Floating point infinities are now supported again.
2067 They might still be a little bit flaky, but thanks to bug reports
2068 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2069 as flaky as they were.
2070 * The --noprogrammer command line option is now supported. (Its
2071 behavior is slightly different in detail from what the old man
2072 page claimed it would do, but it's still appropriate under the
2073 same circumstances that the man page talks about.)
2074 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2075 are now supported, and enabled by default. Thus, the compiler can
2076 handle many floating point and complex operations much less
2077 inefficiently. (Thus e.g. you can implement a complex FFT
2079 * The compiler now detects type mismatches between DECLAIM FTYPE
2080 and DEFUN better, and implements CHECK-TYPE more correctly, and
2081 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2082 thanks to patches from Martin Atzmueller.
2083 * various fixes to make the cross-compiler more portable to
2084 ANSI-conforming-but-different cross-compilation hosts (notably
2085 Lispworks for Windows, following bug reports from Arthur Lemmens)
2086 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2087 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2088 fixed thanks to Pierre Mai's CMU CL patches.
2089 * fixes to keep the system from overflowing internal counters when
2090 it tries to use i/o buffers larger than 16M bytes
2091 * fixed bug 45a: Various internal functions required to support
2092 complex special functions have been merged from CMU CL sources.
2093 (When I was first setting up SBCL, I misunderstood a compile-time
2094 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2095 * improved support for type intersection and union, fixing bug 12
2096 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2097 more obscure bugs as well
2098 * some steps toward byte-compiling non-performance-critical
2099 parts of the system, courtesy of patches from Martin Atzmueller
2100 * Christophe Rhodes has made some debian packages of sbcl at
2101 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2102 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2103 stable, but are nonetheless usable. When he's ready, I'd be happy
2104 to add them to the SourceForge "File Releases" section. (And if
2105 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2106 * new fasl file format version number (because of changes in
2107 internal representation of (OR ..) types to accommodate the new
2108 support for (AND ..) types, among other things)
2110 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2111 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2113 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2115 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2116 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2117 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2118 * DESCRIBE now works on structure objects again.
2119 * Most function call argument type mismatches are now handled as
2120 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2121 know whether the function will be redefined before the call is
2122 executed. (The compiler could flag local calls with full WARNINGs,
2123 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2124 it doesn't keep track of enough information to know whether calls
2125 are local in this sense.)
2126 * Compiler output is now more verbose, with messages truncated
2127 later than before. (There should be some supported way for users
2128 to override the default verbosity, but I haven't decided how to
2129 provide it yet, so this behavior is still controlled by the internal
2130 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2131 src/compiler/ir1util.lisp.)
2132 * Fasl file format version numbers have increased again, because
2133 support for the Gray streams extension changes the layout of the
2134 system's STREAM objects.
2135 * The Gray subclassable streams extension now works, thanks to a
2136 patch from Martin Atzmueller.
2137 * The full LOAD-FOREIGN extension (not just the primitive
2138 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2139 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2140 but like most other programs, it defaults to copying the Unix
2141 environment from the original process instead of starting the
2142 new process in an empty environment.
2143 * Extensions which manipulate the Unix environment now support
2144 an :ENVIRONMENT keyword option which doesn't smash case or
2145 do other bad things. The CMU-CL-style :ENV option is retained
2146 for porting convenience.
2147 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2148 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2150 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2152 * A patch from Martin Atzmueller seems to have solved the SIGINT
2153 problem, and as far as we know, signal-handling now works cleanly.
2154 (If you find any new bugs, please report them!)
2155 * The system no longer defaults Lisp source file names to types
2156 ".l", ".cl", or ".lsp", but only to ".lisp".
2157 * The compiler no longer uses special default file extensions for
2158 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2159 seems to expect a single default extension for all compiled code,
2160 and there's no compelling reason to try to stretch the standard
2161 to allow two different extensions.) Instead, byte-compiled files
2162 default to the same extension as native-compiled files.
2163 * Fasl file format version numbers have increased again, because
2164 a rearrangement of internal implementation packages made some
2165 dumped symbols in old fasl files unreadable in new cores.
2166 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2167 many fewer weird special cases.
2168 * Bug #17 (differing COMPILE-FILE behavior between logical and
2169 physical pathnames) has been fixed, and some related misbehavior too,
2170 thanks to a patch from Martin Atzmueller.
2171 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2172 by Tim Moore, ported to SBCL by Martin Atzmueller.
2173 * Martin Atzmueller fixed several filesystem-related problems,
2174 including bug #36, in part by porting CMU CL patches, which were
2175 written in part by Paul Werkowski.
2176 * More compiler warnings in src/runtime/ are gone, thanks to
2177 more patches from Martin Atzmueller.
2178 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2181 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2183 * DESCRIBE now works on CONDITION objects.
2184 * The debugger now handles errors which arise when trying to print
2185 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2187 * The build system now uses an additional file, customize-target-features.lisp,
2188 to allow local modifications to the target *FEATURES* list. (The point of
2189 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2190 debugging features enabled, without having to worry about propagating it
2191 into everyone's system when I do a "cvs update".) When no
2192 customize-target-features.lisp file exists, the target *FEATURES* list
2193 should be constructed the same way as before.
2194 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2195 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2196 warn about behavior which is undefined under the ANSI spec. Note: This
2197 is specified by ANSI, but it's not very popular with programmers.
2198 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2199 macro in the SBCL sources for an example of a workaround which you
2200 might use to make portable ANSI-standard code which does what you want.
2201 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2202 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2203 when it's used as a non-toplevel form. (This is required in order
2204 to implement the DEFCONSTANT-EQX macro.)
2205 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2206 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2207 funny things, due to bugs in the implementation of EVAL-WHEN
2208 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2209 non-toplevel DEFCONSTANTs are uncommon.)
2210 * The core file version number and fasl file version number have been
2211 incremented, because the old noncompliant DEFCONSTANT behavior involved
2212 calling functions which no longer exist, and because I also took the
2213 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2214 * fixed bug 1 (error handling before read-eval-print loop starts), and
2215 redid debugger restarts and related debugger commands somewhat while
2217 ** The QUIT debugger command is gone, since it did something
2218 rather different than the SB-EXT:QUIT command, and since it never
2219 worked properly outside the main toplevel read/eval/print loop.
2220 Invoking the new TOPLEVEL restart provides the same functionality.
2221 ** The GO debugger command is also gone, since you can just invoke
2222 the CONTINUE restart directly instead.
2223 ** The TOP debugger command is also gone, since it's redundant with the
2224 FRAME 0 command, and since it interfered with abbreviations for the
2226 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2228 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2229 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2230 some porting work by Martin Atzmueller)
2231 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2232 with a boatload of patches which he ported to SBCL. Now that those
2234 ** The system tries to make sure that its low-priority messages
2235 are prefixed by semicolons, to help people who like to use
2236 syntax highlighting in their ILISP buffer. (This patch
2237 was originally due to Raymond Toy.)
2238 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2239 patches originally written by Raymond Toy.
2240 ** The compiler understands coercion between single-value and
2241 multiple-VALUES type expressions better, getting rid of some very
2242 weird behavior, thanks to patches originally by Robert MacLachlan
2243 and Douglas Crosher.
2244 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2245 lambda lists, thanks to a patch originally by Pierre Mai.
2246 ** The system no longer bogusly warns about "abbreviated type
2248 ** The compiler gets less confused by inlining and RETURN-FROM,
2249 thanks to some patches originally by Tim Moore.
2250 ** The system no longer hangs when dumping circular lists to fasl
2251 files, thanks to a patch originally from Douglas Crosher.
2252 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2253 undefined function error.
2254 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2255 in structure slot initforms) for now, documented workaround instead:-|
2256 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2257 * fixed bug 5: added stubs for various Gray stream functions called
2258 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2259 installed, at least appropriate type errors are generated
2260 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2261 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2262 consistently in DEFMETHOD forms.
2263 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2264 it doesn't seem to affect SBCL after all
2265 * The C runtime system now builds with better optimization and many
2266 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2268 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2270 * The system is now under CVS at SourceForge (instead of the
2271 CVS repository on my home machine).
2272 * The new signal handling code has been tweaked to treat register
2273 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2274 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2275 downstream, e.g. in debug-int.lisp, has implicit dependencies
2276 on the unsignedness of integer representation of machine words,
2277 and that caused the system to bomb out with infinite regress
2278 when trying to recover from type errors involving signed values,
2279 e.g. (BUTLAST '(1 2 3) -1).)
2280 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2281 (This was one of the bugs Peter Van Eynde reported back in July.)
2282 * The system now uses code inspired by Colin Walters' O(N)
2283 implementation of MAP (from the cmucl-imp@cons.org mailing
2284 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2285 inline the MAP operation, and there is more than one
2286 sequence argument to the MAP call (so that it can't just
2287 do ETYPECASE once and for all based on the type of the
2288 single sequence argument). (The old non-inline implementation
2289 of the general M-argument sequence-of-length-N case required
2290 O(M*N*N) time when any of the sequence arguments were LISTs.)
2291 * The QUIT :UNIX-CODE keyword argument has been renamed to
2292 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2293 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2295 * Raymond Wiker's patches to port dynamic loading from Linux to
2296 FreeBSD have been added.
2297 * The BUGS file is now more nearly up to date, thanks in large part
2298 to Martin Atzmueller's review of it.
2299 * The debugger now flushes standard output streams before it begins
2300 its output ("debugger invoked" and so forth).
2301 * The core version number and fasl file version number have both
2302 been incremented, because of incompatible changes in the layout
2304 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2305 used to not be called for a saved Lisp image.)
2306 * Martin Atzmueller's version of a patch to fix a compiler crash,
2307 as posted on sbcl-devel 13 September 2000, has been installed.
2308 * Instead of installing Martin Atzmueller's patch for the
2309 compiler transform for SUBSEQ, I deleted the compiler transform,
2310 and transforms for some similar consing operations.
2311 * A bug in signal handling which kept TRACE from working on OpenBSD
2313 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2314 away by constant folding
2315 * The system now defines its address space constants in one place
2316 (in the Lisp sources), and propagates them automatically elsewhere
2317 (through GENESIS and the sbcl.h file). Therefore, patching the
2318 address map is less unnecessarily tedious and error-prone. The
2319 Lisp names of address space constants have also been systematized.
2320 * CVS tags like dollar-Header-dollar have been removed from
2321 the sources, because they have never saved me trouble and
2322 they've been source of trouble working with patches and other
2323 diff-related operations.
2324 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2325 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2327 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2329 * The system has been ported to OpenBSD.
2330 * The system now compiles with a simple "sh make.sh" on the systems
2331 that it's supported on. I.e., now you no longer need to tweak
2332 text in the target-features.lisp-expr and symlinks in src/runtime/
2333 by hand, the make.sh takes care of it for you.
2334 * The system is no longer so grossly inefficient when compiling code
2335 involving vectors implemented as general (not simple) vectors (VECTOR T),
2336 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2337 which dares to use the various sequence functions on non-simple
2338 vectors, takes less of a performance hit.
2339 * There is now a primitive type predicate VECTOR-T-P
2340 to test for the (VECTOR T) type, so that e.g.
2341 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2342 can now be compiled with some semblance of efficiency. (The old code
2343 turned the type declaration into a full call to %TYPEP at runtime!)
2344 * AREF on (VECTOR T) is still not fast, since it's still compiled
2345 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2346 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2347 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2348 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2349 without making full calls to SUBTYPEP at runtime.
2350 (Some analogous efficiency issues for non-simple vectors specialized to
2351 element types other than T, or for non-simple multidimensional arrays,
2352 have not been addressed. They could almost certainly be handled the
2353 same way if anyone is motivated to do so.)
2354 * The changes in array handling break binary compatibility, so
2355 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2356 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2357 instead of (VALUES T T).
2358 * By following the instructions that Dan Barlow posted to sbcl-devel
2359 on 2 July 2000, I was able to enable primitive dynamic object
2360 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2361 functionality is not implemented (since it calls ld to resolve
2362 library references automatically, requiring RUN-PROGRAM for its
2363 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2364 try to resolve library references) is now supported.
2365 * The system now flushes the standard output streams when it terminates,
2366 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2367 them at several other probably-convenient times, e.g. in each pass of
2368 the toplevel read-eval-print loop, and after evaluating a form given
2369 as an "--eval" command-line option. (These changes were motivated by a
2370 discussion of stream flushing issues on cmucl-imp in August 2000.)
2371 * The source transform for TYPEP of array types no longer assumes
2372 that an array whose element type is a not-yet-defined type
2373 is implemented as an array of T, but instead punts, so that the
2374 type will be interpreted at runtime.
2375 * There is now some support for cross-compiling in make.sh: each of
2376 the phases of make.sh has its own script. (This should be transparent
2377 to people doing ordinary, non-cross-compile builds.)
2378 * Since my laptop doesn't have hundreds of megabytes of memory like
2379 my desktop machine, I became more motivated to do some items on
2380 my to-do list in order to reduce the size of the system a little:
2381 ** Arrange for various needed-only-at-cold-init things to be
2382 uninterned after cold init. To support this, those things have
2383 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2384 symbols with such names are now uninterned after cold init).
2385 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2386 fasl files for cold load.
2387 ** Remove the old compiler structure pooling code (which used to
2388 be conditional on the target feature :SB-ALLOC) completely.
2389 ** Redo the representation of some data in cold init to be more compact.
2390 (I also looked into supporting byte compiled code at bootstrap time,
2391 which would probably reduce the size of the system a lot, but that
2392 looked too complicated, so I punted for now.)
2393 * The maximum signal nesting depth in the src/runtime/ support code has
2394 been reduced from 4096 to 256. (I don't know any reason for the very
2395 large old value. If the new smaller value turns out to break something,
2396 I'll probably just bump it back up.)
2397 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2399 * Many, many bugs reported by Peter Van Eynde have been added to
2400 the BUGS list; some have even been fixed.
2401 * While enabling dynamic object file loading, I tried to make the
2402 code easier to understand, renaming various functions and variables
2403 with less ambiguous names, and changing some function calling
2404 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2405 * While trying to figure out how to do the OpenBSD port, I tried to
2406 clean up some of the code in src/runtime/. In particular, I dropped
2407 support for non-POSIX signal handling, added various comments,
2408 tweaked the code to reduce the number of compilation warnings, and
2409 renamed some files to increase consistency.
2410 * To support the new automatic configuration functionality in make.sh,
2411 the source file target-features.lisp-expr has been replaced with the
2412 source file base-target-features.lisp-expr and the machine-generated
2413 file local-target-features.lisp-expr.
2414 * fixed a stupid quoting error in make.sh so that using CMU CL
2415 "lisp -batch" as cross-compilation host works again
2417 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2419 * DESCRIBE no longer tries to call itself recursively to describe
2420 bound/fbound values, so that it no longer fails on symbols which are
2421 bound to themselves (like keywords, T, and NIL).
2422 * DESCRIBE now works on generic functions.
2423 * The printer now prints less-screwed-up representations of closures
2424 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2425 * A private symbol is used instead of the :EMPTY keyword previously
2426 used to mark empty slots in hash tables. Thus
2427 (DEFVAR *HT* (MAKE-HASH-TABLE))
2428 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2429 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2430 now does what ANSI says that it should. (You can still get
2431 similar noncompliant behavior if bang on the hash table
2432 implementation with all the symbols you get back from
2433 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2434 This breaks binary compatibility, since tests for equality to
2435 :EMPTY are wired into things like the macroexpansion of
2436 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2438 * There's now a minimal placeholder implementation for CL:STEP,
2439 as required by ANSI.
2440 * An obscure bug in the interaction of the normal compiler, the byte
2441 compiler, inlining, and structure predicates has been patched
2442 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2443 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2445 * Missing ordinary arguments in a macro call are now detected even
2446 when the macro lambda list contains &KEY or &REST.
2447 * The debugger no longer complains about encountering the top of the
2448 stack when you type "FRAME 0" to explicitly instruct it to go to
2449 the top of the stack. And it now prints the frame you request even
2450 if it's the current frame (instead of saying "You are here.").
2451 * As specified by ANSI, the system now always prints keywords
2452 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2454 * The default initial SIZE of HASH-TABLEs is now smaller.
2455 * Type information from CLOS class dispatch is now propagated
2456 into DEFMETHOD bodies, so that e.g.
2457 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2459 is now basically equivalent to
2460 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2461 (DECLARE (TYPE SINGLE-FLOAT X))
2463 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2464 operation, without having to do run-time type dispatch.
2465 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2466 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2467 * Since I don't seem to be making big file reorganizations very often
2468 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2469 is overflowing my ability to conveniently back them up), I've finally
2470 checked the system into CVS. (The CVS repository is on my home system,
2471 not at SourceForge -- putting it on SourceForge might come later.)
2472 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2473 high-level GC-NOTIFY-FOO functions send their output. (There's
2474 still very little control of where low-level verbose GC functions
2475 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2476 less than it used to -- the GC-NOTIFY-FOO functions are now under
2477 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2478 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2479 in only one place in the source code, and propagates it automatically
2480 everywhere that it's needed. Thus e.g. when I bump the version from
2481 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2482 * The C source files now include boilerplate legalese and documentation
2483 at the head of each file (just as the Lisp source files already did).
2484 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2485 to his page will be replaced with a link to his new CLiki service.
2487 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2489 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2490 * The build process now looks for GNU make under the default name "gmake",
2491 instead of "make" as it used to. If GNU make is not available as "gmake"
2492 on your system, you can change this default behavior by setting the
2493 GNUMAKE environment variable.
2494 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2495 can build without error under CMU CL.
2497 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2499 * There is now a partial SBCL user manual (with some new text and some
2500 text cribbed from the CMU CL manual).
2501 * The beginnings of a profiler have been added (starting with the
2502 CMU CL profiler and simplifying and cleaning up). Eventually the
2503 main interface should be through the TRACE macro, but for now,
2504 it's still accessed through vaguely CMU-CL-style functions and macros
2505 exported from the package SB-PROFILE.
2506 * Some problems left over from porting CMU CL to the new
2507 cross-compilation bootstrap process have been cleaned up:
2508 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2509 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2510 part when getting rid of the compiler *BACKEND* stuff.)
2511 ** The value of *NULL-TYPE* was screwed up, because it was
2512 being initialized before the type system knew the final
2513 definition of the 'NULL type. This screwed up several key
2514 optimizations in the compiler, causing inefficiency in all sorts
2515 of places. (I found it because I wanted to understand why
2516 GET-INTERNAL-RUN-TIME was consing.)
2517 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2518 PROCLAIM FTYPE information. Unfortunately this broke binary
2519 compatibility again, since now the forms output by DEFGENERIC
2520 to refer to functions which didn't exist in 0.6.3.
2521 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2522 can use the new (as of 0.6.3) transform for SOME into MAP into
2524 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2525 values to (INTEGER 0 1000000), so that the time code will no longer
2526 occasionally get blown up by Linux returning 1000000 microseconds
2527 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2528 its output conform to the ANSI spec. (Alas, this makes its output
2529 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2530 * A full call to MAP NIL with a single sequence argument no longer conses.
2531 * fixes to problems pointed out by Martin Atzmueller:
2532 * The manual page no longer talks about multiprocessing as though
2533 it were currently supported.
2534 * The ILISP support patches have been removed from the distribution,
2535 because as of version 5.10.1, ILISP now supports SBCL without us
2536 having to maintain patches.
2537 * added a modified version of Raymond Toy's recent CMU CL patch for
2538 EQUALP comparison of HASH-TABLE
2540 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2542 * The system still can't cross-compile itself with
2543 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2544 entails), but at least it can compile more of itself that way
2545 than it used to be able to, and various buglets which were uncovered
2546 by trying to cross-compile itself that way have now been fixed.
2547 * This release breaks binary compatibility again. This time
2548 at least I've incremented the FASL file format version to 2, so that the
2549 problem can be detected reliably instead of just causing weird errors.
2550 * various new style warnings:
2551 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2552 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2553 ** using the lexical binding of a variable named in the *FOO* style
2554 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2555 as specified by ANSI.
2556 * *RANDOM-STATE* is no longer automatically initialized from
2557 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2558 default behavior of the system is to repeat its behavior every time
2559 it's run. If you'd like to change this behavior, you can always
2560 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2561 old convention there was no comparably easy way to get the system to
2562 repeat its behavior every time it was run.
2563 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2564 FTYPE declarations has been removed, in favor of their ANSI
2565 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2566 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2567 the types of their sequence arguments can be determined at compile time.
2568 This is done through a new open code expansion for MAP which eliminates
2569 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2570 argument types can be determined at compile time.
2571 * The optimizer now transforms COERCE into an identity operation when it
2572 can prove that the coerced object is already of the correct type. (This
2573 can be a win for machine generated code, including the output of other
2574 optimization transforms, such as the MAP transform above.)
2575 * Credit information has been moved from source file headers into CREDITS.
2576 * Source file headers have been made more standard.
2577 * The CASE macro now compiles without complaining even when it has
2580 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2582 * (Note that the way that the PCL macroexpansions were rewritten
2583 to accommodate the change in DEFGENERIC below breaks binary
2584 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2585 not run under sbcl-0.6.2. Once we get out of alpha releases,
2586 i.e. hit release 1.0.0, we'll probably try to maintain binary
2587 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2588 and sbcl-1.4.4. Until then, however, it might be fairly common
2589 for maintenance releases to break binary compatibility.)
2590 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2592 * The reporting of unhandled signals has been changed to print some
2593 explanatory text as well as the report form. (Previously only
2594 the report form was printed.)
2595 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2596 it defines, so that the compiler no longer issues undefined function
2597 warnings for compiled-but-not-yet-loaded generic functions.
2598 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2599 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2600 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2601 section of the ANSI spec are supported.
2602 * The "" nickname for the "KEYWORD" package has been removed.
2603 The reader still handles symbol tokens which begin with a package marker
2604 as keywords, but it doesn't expose its mechanism for doing so in the
2605 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2606 * The system now issues STYLE-WARNINGs for contradictory TYPE
2607 proclamations. (Warnings for contradictory FTYPE proclamations would
2608 be nice too, but those can't be done usefully unless the type system
2609 is made smarter about FUNCTION types.)
2610 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2611 systematized, so that "*target-*.lisp is supposed to exist only on the
2612 target and imply that there's a related file which exists on the
2613 host, and *host-*.lisp is supposed to exist only on the host and imply
2614 that there's a related file which exists on the target. This involves a
2615 lot of renaming. Hopefully the acute confusion caused by the renaming
2616 will be justified by the reduction in chronic confusion..
2617 ** runtime-type.lisp -> early-target-type.lisp
2618 ** target-type.lisp -> late-target-type.lisp
2619 ** early-host-format.lisp -> early-format.lisp
2620 ** late-host-format.lisp -> late-format.lisp
2621 ** host-error.lisp -> misc-error.lisp
2622 ** early-error.lisp -> early-target-error.lisp
2623 ** late-error.lisp -> late-target-error.lisp
2624 ** host-defboot.lisp -> early-defboot.lisp
2625 ** code/misc.lisp -> code/target-misc.lisp
2626 ** code/host-misc.lisp -> code/misc.lisp
2627 ** code/numbers.lisp -> code/target-numbers.lisp
2628 ** code/early-numbers.lisp -> numbers.lisp
2629 ** early-host-type.lisp -> early-type.lisp
2630 ** late-host-type.lisp -> late-type.lisp
2631 ** host-typep.lisp -> typep.lisp
2632 ** load.lisp -> target-load.lisp
2633 ** host-load.lisp -> load.lisp
2634 ** host-disassem.lisp -> disassem.lisp
2635 ** host-insts.lisp -> insts.lisp
2636 ** byte-comp.lisp -> target-byte-comp.lisp
2637 ** host-byte-comp.lisp -> byte-comp.lisp
2638 ** host-signal.lisp -> signal.lisp
2639 ** host-defstruct.lisp -> defstruct.lisp
2640 ** late-target-type.lisp -> deftypes-for-target.lisp
2641 Furthermore, several other previously target-only files foo.lisp (e.g.
2642 hash-table.lisp and random.lisp) have been split into a target-and-host
2643 foo.lisp file and a target-only target-foo.lisp file, with their key type
2644 definitions in the target-and-host part, so that the cross-compiler will
2645 know more about target types.
2646 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2647 gone away. In their place are various *BACKEND-FOO* variables
2648 corresponding to the slots of the old structure.
2649 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2650 into the target SBCL has been fixed.
2651 * The chill.lisp system for loading cold code into a running SBCL
2653 * Support for the CMU CL "scavenger hook" extension has been removed.
2654 (It was undocumented and unused in the CMU CL sources that SBCL was
2655 derived from, and stale in sbcl-0.6.1.)
2656 * Various errors in the cross-compiler type system were detected
2657 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2658 (enabling various consistency checks). Many of them were fixed,
2659 but some hard problems remain, so the compiler is back to
2660 running without *TYPE-SYSTEM-INITIALIZED* for now.
2661 * As part of the cross-compiler type system cleanup, I implemented
2662 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2663 * I have started adding UNCROSS calls throughout the type system
2664 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2665 on cross-compiler input files will be able to go away, and various
2667 * CONSTANTP now returns true for quoted forms (as explicitly required
2670 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2672 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2673 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2674 (MAKE-LIST -1). In the longer term, it ought to become true
2675 that declarations are assertions even at SAFETY 1. For now, it's not
2676 quite true even at SAFETY 3, but it's at least more nearly true..
2677 (Note that this change seems to increases the size of the system by
2678 O(5%) and to decrease the speed of the compiler by 20% or more.)
2679 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2680 for the problem of printing dozens of lines of distracting information
2681 about low-level system machinery as part of the top stack frame
2682 on entry to the debugger when an undefined function was called.
2683 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2684 is not reset to COMMON-LISP-USER.
2685 * Compilation of stuff related to dyncount.lisp has been made conditional
2686 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2687 smaller. The various dyncount-related symbols have been moved into
2688 a new "SB-DYNCOUNT" package.
2689 * tty-inspect.lisp has been renamed to inspect.lisp.
2690 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2691 feature has gone away. (When we eventually port to other flavors of
2692 libc and/or Unix, we'll try to make the differences between flavors
2693 invisible at the user level.)
2694 * Various other *FEATURES* tags, and/or their associated conditionals,
2695 have been removed if obsolescent, or given better documentation, or
2696 sometimes given more-mnemonic names.
2698 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2700 * tidied up "make.sh" script
2701 * tidied up system directory structure
2702 * better "clean.sh" behavior
2703 * added doc/FOR-CMUCL-DEVELOPERS
2704 * many many small tweaks to output format, e.g. removing possibly-confusing
2705 trailing #\. character in DESCRIBE-INSTANCE
2706 * (EQUALP #\A 'A) no longer signals an error.
2707 * new hashing code, including EQUALP hashing
2708 * tidied up Lisp initialization and toplevel
2709 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2710 * command line argument processing
2711 * added POSIX-GETENV function to deal with Unix-ish environment variables
2712 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2713 terminating SBCL on EOF
2714 * non-verbose GC by default
2715 * There is no more "sbcl" shell script; the sbcl file is now the C
2716 runtime executable (just like CMU CL).
2717 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2719 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2720 symbol and package information as Lisp native symbol and package objects
2721 instead of strings naming symbols and strings naming packages. This way,
2722 whenever packages are renamed (as in warm init), debug information is
2723 transformed along with everything else.
2724 * tweaked the optimization policy declarations which control the building
2725 of SBCL itself. Now, among other things, the system no longer saves
2726 source location debugging information. (This helps two problems at once
2727 by reducing SBCL size and by keeping SBCL from trying to look for its
2728 sources -- which may not exist -- when reporting errors.)
2729 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2730 debugging and testing purposes
2731 * cleaned up printing, making the printer call PRINT-OBJECT for
2732 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2733 methods, giving nearly-ANSI behavior
2734 * converted almost all special variables to use *FOO* naming convention
2735 * deleted PARSE-TIME functionality, since it can be done portably
2736 * moved some files out of cold init into warm init
2737 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2739 * regularized formatting of source files
2740 * added an install.sh script
2741 * fixed ridiculous memory usage of cross-compiler by making
2742 compiler/alloc.lisp not try to do pooling unless it can hook
2743 itself into the GC of the cross-compilation host. Now the system
2744 builds nicely on my old laptop.
2745 * added :SB-ALLOC in target-features.lisp-expr
2746 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2747 was not implemented)
2748 * re-did condition handling and note reporting in the compiler. Notes
2749 are no longer handled by signalling conditions. Style warnings
2750 and warnings are handled more correctly and reported in such a way
2751 that it's easy to find one or the other in your output (so that you
2752 can e.g. figure out which of many problems caused COMPILE-FILE to
2754 * changed the severity of several compiler warnings from full WARNING
2755 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2756 compiler note reporting so that it doesn't use the condition system
2757 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2758 COMPILE-FILE command)
2759 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2760 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2761 no longer supported, since their implementation was deeply intertwingled
2762 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2763 * removed bogus "support" for compiler macros named (SETF FOO), and
2764 removed the compiler macro for SETF INFO (but only after making a fool
2765 of myself on the cmucl-imp mailing list by posting a bogus patch for
2766 DEFINE-COMPILER-MACRO..)
2767 * Compiled files containing forms which have side effects on the Lisp
2768 reader (such as DEFPACKAGE forms) are now handled more correctly.
2769 (Compiler queuing of top level lambdas has been suppressed by setting
2770 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2771 * deleted various currently-unused source files, e.g. gengc.lisp. They
2772 may be added back at some point e.g. when porting to other architectures,
2773 but until they are it's distracting to distribute them and to try to
2775 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2776 there were so many of them they're just distractions, and UNCROSS is
2777 known to be able to handle the current sources
2778 * moved PROFILE functionality into TRACE, so that it will be clear
2779 how the wrapping and unwrapping of functions when you profile them
2780 interacts with the wrapping and unwrapping of functions when you
2781 trace them. (Actually, the functionality isn't there yet, but at least
2782 the interface specification is there. Hopefully, the functionality will
2783 arrive with some future maintenance release.)
2784 * removed host-oops.lisp
2785 * changed signature of QUIT function to allow UNIX-CODE argument
2786 * fixed READ-SEQUENCE bug
2787 * tweaked verbose GC output so that it looks more like the progress
2788 output that ANSI specifies for functions like LOAD
2789 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2790 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob