1 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2 * minor incompatible change: BASE-CHAR no longer names a class;
3 however, CHARACTER continues to do so, as required by ANSI.
4 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
5 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
6 has been moved to the SB-EXT package (temporarily re-exported from
8 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
9 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
11 * bug fix: Cyclic structures and unprintable objects in compiler
12 messages no longer cause errors. (reported by Bruno Haible)
13 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
14 and READ-FROM-STRING all now return a primary value of NIL if
15 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
16 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
17 (reported by Bruno Haible for CMUCL)
18 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
19 do not designate a condition. (reported by Bruno Haible for
21 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
22 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
23 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
24 permitting resolution in favour of any of the conflicting symbols.
25 (reported by Bruno Haible for CMUCL)
26 * fixed some bugs revealed by Paul Dietz' test suite:
27 ** RENAME-PACKAGE allows all package designators as new package
30 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
31 * enhancement: saving cores with foreign code loaded is now
32 supported on x86/NetBSD and sparc/Linux in addition to the previously
34 * bug fix: on some platforms repeated installations caused multiple
35 copies of HTML documentation to be installed -- should not happen
36 any more. (reported by Stefan Scholl)
37 * bug fix: parsing self-recursive alien record types multiple times
38 no longer causes infinite recursion. (reported by Thomas F. Burdick,
39 original patch by Helmut Eller for CMUCL)
40 * bug fix: stack-exhaustion detection works now on NetBSD as well.
41 (thanks to Richard Kreuter)
42 * bug fix: defining classes whose accessors are methods on existing
43 generic functions in other (locked) packages no longer signals
44 bogus package lock violations. (reported by François-René Rideau)
45 * bug fix: special variables as DEFMETHOD parameters no longer have
46 associated bogus type declarations. (reported by David Wragg and
48 * bug fix: read-write consistency on streams of element-type
49 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
50 * bug fix: redefiniton of the only method of a generic function with
51 no DEFGENERIC no longer emits a full WARNING. In addition,
52 redefinition of generic functions with no DEFGENERIC to an
53 incompatible lambda list now signals an error. (thanks to Zach
55 * bug fix: DEFGENERIC now works even when there's a function of the
56 same name in an enclosing lexical environment. (thanks to Zach
58 * fixed compiler failure, caused by instrumenting code during
59 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
60 * optimization: added loop analysis and improved register allocation
61 * fixed some bugs revealed by Paul Dietz' test suite:
62 ** POSITION on displaced vectors with non-zero displacement
63 returns the right answer.
64 ** (SIMPLE-STRING) is a valid type specifier for sequence
66 ** *PRINT-LEVEL* handling for slotless structures is pedantically
68 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
69 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
71 ** The justification version of the FORMAT ~< directive treats
72 non-zero minpad parameter correctly.
74 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
75 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
76 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
77 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
78 the supported interface.
79 * new feature: Single-stepping of code compiled with DEBUG 2 or
80 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
81 * new feature: saving cores with foreign code loaded is now
82 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
83 Timothy Moore's work for CMUCL)
84 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
85 parameters to * instead of NIL if no initform is supplied.
86 (reported by Johan Bockgaard)
87 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
88 correctly when FUNCALL appears as the car of the form. Note:
89 despite this FUNCALL forms are not currently subject to
90 compiler-macro expansion. (port of Raymond Toy's fix for the
91 same from CMUCL, reported by Johan Bockgaard)
92 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
93 (thanks for Teemu Kalvas)
94 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
95 case instead of terminating the iteration (thanks for Julian
96 Squires, Sean Champ and Raymond Toy)
97 * bug fix: incorrect expansion of defgeneric that caused a style
98 warning. (thanks for Zach Beane)
99 * bug fix: slot accessor effective method computation works properly
100 for classes with multiple non-standard applicable methods on
101 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
102 * on x86 compiler supports stack allocation of results of LIST and
103 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
104 CMUCL implementation by Gerd Moellmann)
105 * fixed some bugs revealed by Paul Dietz' test suite:
106 ** FORMAT strings with both the ~<~:;~> form of the justification
107 directive and pretty-printing directives cause an error.
109 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
110 * incompatible change: the internal functions
111 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
112 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
113 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
114 instead of the old functions.
115 * new feature: on platforms where "dladdr" is available foreign
116 function names now appear in backtraces. (based on Helmut Eller's
118 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
119 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
121 * documentation: ASDF manual is now distributed with SBCL.
122 * bug fix: SBCL can now load its contributed modules with REQUIRE
123 even if the system-provided entries have been removed from
124 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
126 * bug fix: backtraces involving undefined functions or assembly
127 routines are more informative. (thanks to Brian Downing)
128 * bug fix: mutually referent alien structures now work correctly.
129 (reported by Rick Taube)
130 * bug fix: structures defined by WITH-ALIEN can be referred to
131 within other definitions in the same WITH-ALIEN.
132 * bug fix: division operators (MOD, TRUNCATE and the like) with
133 constant zero divisors and integer dividends no longer generate
135 * bug fix: provide default methods for INPUT-STREAM-P and
136 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
137 * bug fix: improve the detection and resolution of MOP metacycles
138 (where effective method computation depends on the generic function
139 having its effective method computed). (reported by Bruno Haible)
140 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
141 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
142 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
143 * bug fix: pathnames with relative directory components can be
144 represented relative to default pathnames.
145 * optimization: in taking the GCD of bignums, reduce the two bignums
146 to approximately the same size (using Euclid's algorithm) before
147 applying the more sophisticated binary GCD. (thanks to Juho
149 * optimization: COUNT on bitvectors now operates word-at-a-time.
150 * optimization: ASH with a positive, but not necessarily constant,
151 (leftwards) shift, when in a modular context, is compiled to a
153 * fixed some bugs revealed by Paul Dietz' test suite:
154 ** FORMAT variable parameters ("~V<char>") are defaulted properly
155 if the corresponding argument is NIL.
156 ** FORMAT directives accepting a minpad parameter treat negative
158 ** literal commas as character parameters to FORMAT directives are
160 ** literal spaces directly after ~<Newline> directives within a
161 format-logical-block (~:< ~@:>) do not induce :FILL-style
162 conditional newlines.
163 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
164 as specified: it no longer includes conditional newlines).
165 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
167 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
168 :PER-LINE-PREFIX argument does not evaluate to a string.
170 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
171 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
172 the manual for details; for now, package locks can be disabled by
173 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
174 if no major problems are found then it is likely that they will be
175 compiled in unconditionally.
176 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
177 unsupported operators on all platforms. To load a shared library
178 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
179 object file, link it into a shared library outside of SBCL and
180 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
182 * bug workaround: ROOM T and the GC were not working together
183 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
184 reporting facility aren't preserved. That reporting has been
185 disabled, so now until and unless someone figures out how to make it
186 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
187 * minor incompatible change: as threatened around sbcl-0.8.0, the
188 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
189 The symbols are also exported from SB-PCL for backwards
190 compatibility, but more so than before SB-PCL should be treated as
191 an implementation-internal package.
192 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
194 * the SB-SPROF contrib now works on (most) non-x86 architectures.
195 It is known as of this release not to work on the Alpha, however.
196 * fixed bug #167: errors signalled due to illegal syntax in method
197 bodies are now more legible.
198 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
199 designators and can hence be used with TYPEP.
200 * fixed bug #333: CHECK-TYPE now ensures that the type error
201 signalled, if any, has the right object to be accessed by
202 TYPE-ERROR-DATUM. (reported by Tony Martinez)
203 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
204 5.1.2.3 for multiple-value place subforms. (reported by Kalle
206 * fixed bug #334: programmatic addition of slots using specialized
207 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
208 and :ALLOCATION :CLASS slots.
209 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
210 float exponents. (rereported by Peter Seibel)
211 * fixed a bug: #\Space (and other whitespace characters) are no
212 longer considered to be macro characters in standard syntax by
214 * fixed bug: initialization of condition class metaobjects no longer
215 causes an instance of the condition to be created. (reported by
217 * fixed bug: it is now possible to have slots such that
218 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
219 description is neither :INSTANCE nor :CLASS.
220 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
221 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
222 SLOT-VALUE-USING-CLASS), and no longer causes errors with
223 non-standard SLOT-DEFINITION-ALLOCATION values.
224 * fixed bugs: various race conditions handling exiting threads.
225 CL-PPCRE's thread torture test now passes.
226 * fixed arguably-a-bug: GC time proportional to number of threads in
227 system even when most of them are idle
228 * optimization: improved performance of BIT and SBIT on bit-vectors.
229 * .fasl file incompatibility: The fasl file version number has
230 been incremented because of changes associated with package locks.
232 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
233 * minor incompatible change: the system no longer provides
234 optimization hints (conditions which are TYPEP
235 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
236 mode; these hints are still emitted when higher SPEED optimization
238 * new contrib module: a sampling profiler (profiling by statistical
239 sampling, rather than by instrumenting functions) is available as
240 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
241 * the behaviour of the standard function ED is now customizeable by
242 third parties through a hook variable: see ED's documentation
243 string for information on the protocol.
244 * the compiler no longer emits efficiency notes for (FUNCALL X)
245 when the type of X is uncertain under default optimization
247 * fixed bug 276: mutating a binding of a specialized parameter to a
248 method to something that is not TYPEP the specializer is now
250 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
251 exhibit the required behaviour on the x86 platform. (thanks to
252 Peter van Eynde, Eric Marsden and Bruno Haible)
253 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
254 even for difficult arguments. (reported by Peter Graves)
255 * fixed bug 141a: the backquote printer now descends quoted
257 * fixed another bug in backquote printing: no more destructive
258 modification of the form's list structure. (reported by Brian
260 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
261 move between its address being taken and the call to
262 interrupt_thread, fixing a crashing race condition.
263 * the SB-POSIX contrib implementation has been adjusted so that it
264 no longer exhibits ridiculously poor performance when constructing
265 instances corresponding to C structs.
267 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
268 * minor incompatible change: the sb-grovel contrib now treats C
269 structures as alien (in the sense of SB-ALIEN) objects rather than
270 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
271 has implications for memory management of client code
272 (sb-grovel-returned objects must now be manually managed) and for
273 type safety (alien objects now have full types).
274 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
275 used to control emission of compiler diagnostics, rather than the
276 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
277 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
278 quality should be considered deprecated.
279 * install.sh now installs the user manual as well
280 * (not quite a new documentable feature, but worth considering in
281 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
282 beginnings of a semantically meaningful condition hierarchy is
283 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
285 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
286 designator as the defaults argument.
287 * fixed bug: Displaced arrays whose displaced-to array has become
288 too small now cause ARRAY-DIMENSION to signal an error, providing
289 for safer bounds-checking. (reported by Bruno Haible)
290 * fixed bug: DEFCLASS slot definitions with identical :READER and
291 :WRITER names now signal a reasonable error. (reported by Thomas
293 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
295 * fixed bug: Invalid slot specification errors now print correctly.
296 (thanks to Zach Beane)
297 * fixed bug 320: Shared to local slot value transfers in class
298 redefinitions now happen corrently from superclasses as
299 well. (reported by Bruno Haible)
300 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
302 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
303 improper lists correctly. (reported by Bruno Haible)
304 * fixed bug 313: source-transform for <fun-name> was erroneously
305 applied to a call of a value of a variable with name <fun-name>.
306 (reported by Antonio Menezes Leitao)
307 * fixed bug 307: The obsolete instance protocol ensures that
308 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
309 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
310 remove dead unknown-values globs from the middle of the stack.
311 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
312 easier to resume long computations after using *BREAK-ON-SIGNALS*
313 to diagnose and fix failures. (thanks to Nikodemus Siivola)
314 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
315 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
316 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
317 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
318 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
320 * fixed a (fairly theoretical) bug in string printing: if
321 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
322 does not have array-element-type equal to the most general string
324 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
325 function name. (thanks to Nikodemus Siivola)
326 * fixed bug: compile-time format string checker failed on
327 non-closed ~{. (reported by Thomas F Burdick)
328 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
329 don't warn on legal loop code involving more than one aggregate
330 boolean. (thanks to Nikodemus Siivola)
331 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
332 in keyword position to MAKE-PACKAGE was not being reported as
334 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
335 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
337 * fixed bugs 280 and 312: the checking for multiple definitions in a
338 file is less likely to become confused by uses of inline
340 * fixed bug: the #S reader macro performs the keyword coercion
341 specified for slot names. (reported by Kalle Niemitalo)
342 * fixed bug: lambda lists may contain symbols whose names start with
343 & but are not lambda-list-keywords; their occurrence triggers a
345 * fixed bug 321: define-method-combination argument lambda lists do
346 not cause invalid code to be generated when &optional or &aux
347 variables have default values. (reported by Bruno Haible)
348 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
349 have CLOS classes; this appears also to have fixed reported
350 instability in stack exhaustion detection.
351 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
352 MUFFLE-WARNING when no associated restart is present is now
354 * optimization: rearranged the expansion of various defining macros
355 so that each expands into only one top-level form in a
356 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
358 * optimization: used a previously little-used slot in symbols to
359 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
361 * fixed some bugs revealed by Paul Dietz' test suite:
362 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
364 ** Fixed an optimization bug related to inheritance of initforms
365 from local to shared slots.
366 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
367 ** CAST optimizer forgot to flush argument derived type.
368 ** print/read consistency on floats is now orders of magnitude
369 more likely. (thanks also to Bruno Haible for a similar report
371 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
372 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
373 for objects of type REAL. Make it so.
374 ** SXHASH is specified (infelicitously) to respect similarity,
375 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
376 it so. (thanks to Markus Ziegler)
377 ** on the Alpha, the compiler succeeds in compiling functions
378 returning a known number of arguments greater than 63.
379 ** fixed handling of invalid NIL arguments in keyword position
381 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
383 ** provided a readably-printable representation for RANDOM-STATE
385 ** ensured that pathnames, where they have a namestring, always
386 print using #P"..." syntax.
388 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
389 * Support for the forthcoming 2.0 version of the NetBSD kernel
390 running on x86 hardware has been added. (thanks to Perry
391 E. Metzger most immediately, and others for their past work)
392 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
393 current release version; SBCL's OpenBSD support had been broken
394 since about the time of OpenBSD's switch to ELF binary format.)
395 * [placeholder for DX summary]
396 ** user code with &REST lists declared dynamic-extent, under high
397 speed or space and low safety and debug optimization policy.
398 * The manual has been converted to Texinfo format and the debugger
399 chapter from the cmucl manual has been added.
400 * A facility has been added to extract documentation strings from
401 sbcl and store them as Texinfo-formatted snippets for inclusion in
402 the manual (via Texinfo's @include directive)
403 * bug fix: compiler emitted division in optimized DEREF. (thanks for
404 the test case to Dave Roberts)
405 * bug fix: multidimensional simple arrays loaded from FASLs had fill
406 pointers. (reported by Sean Ross)
407 * bug fix: PROFILE output is printed nicely even for large numerical
408 values. (thanks to Zach Beane)
409 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
410 greater than 32 handle EOF correctly.
411 * bug fix: on X86 an immediate argument of the IMUL instruction is
412 correctly printed in disassembly. (thanks to Lutz Euler)
413 * bug fix: class slots in redefined classes preserve their old
414 values. (thanks to Bruno Haible and Nikodemus Siivola)
415 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
416 succeeds. (reported by Marco Baringer)
417 * fixed some bugs revealed by Paul Dietz' test suite:
418 ** READ-SEQUENCE now works on ECHO-STREAMs.
419 ** RATIONALIZE works more according to its specification. (thanks
422 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
423 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
424 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
425 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
426 should matter to you only if you rebind the printer control
427 variables and then find you want different bindings in the
428 debugger than in the ordinary execution of your program.)
429 * The runtime build system has been tweaked to support building
430 (on SPARC/SunOS) using a C compiler which invokes Sun's own
431 assembler and linker. (thanks to Nikodemus Siivola)
432 * Unbound, undefined, undeclared variables now trigger full
433 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
434 more often programmer error than deliberate exploitation of undefined
436 * optimization: the hash algorithm for strings has changed to one
437 that is less vulnerable to spurious collisions. (thanks to Juho
439 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
440 needless bounds checking. (thanks to Juho Snellman)
441 * optimization: implemented multiplication as a modular
442 (UNSIGNED-BYTE 32) operation on the PPC backend.
443 * fixed some bugs revealed by Paul Dietz' test suite:
444 ** ADJUST-ARRAY now copies the datum in a zero rank array if
446 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
447 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
448 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
450 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
452 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
454 ** ECHO-STREAMs no longer attempt to echo the end of file value to
455 their output stream on EOF from read.
456 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
457 have been read to end-of-file.
458 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
460 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
461 description of determination of which consecutive characters
463 ** Printing the "Space" character with escaping on now yields "#\\ ",
464 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
465 ** Reading floating-point numbers with *READ-BASE* set to a number
466 less than 10 works correctly.
467 ** Reading floating-point numbers with *READ-BASE* set to a number
468 more than 10 works correctly.
469 ** Printing with *PRINT-READABLY* targets the standard readtable, not
470 the readtable currently in effect.
472 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
473 * minor incompatible change: parsing of namestrings on a physical
474 (Unix) host has changed; numbers after the final #\. in a
475 namestring are no longer interpreted as a version field. This is
476 intented to be largely invisible to the user, except that the
477 meaning of the namestring "*.*.*" has changed: it now refers to a
478 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
479 should usually be replaced by
480 (make-pathname :name :wild :type :wild :version :wild)
481 with the added benefit that this is more likely to be portable.
482 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
483 option to OPEN now signals an error if the file being opened
484 exists; this may have an impact on existing code.
485 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
486 signals received, on the PowerPC platforms (both Linux and
487 Darwin). (thanks to Pierre Mai for pointing out the location of
489 * several fixes on OS X: The system now builds and runs cleanly on
490 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
491 (thanks to Brian Mastenbrook)
492 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
493 second-resolution: integer multiples of 1/3600 between -24 and 24.
494 (thanks to Vincent Arkesteijn)
495 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
496 type when called with 1 argument; PEEK-CHAR checked type of
497 PEEK-TYPE only after having read first character from a
498 stream. (reported by Peter Graves)
499 * bug fix: the garbage collector now has much better locality
500 behaviour, and in particular no longer treats all memory as being
501 exhausted when many small objects point to each other in a deeply
503 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
504 recognized as being TYPEP their class.
505 * bug fix: the PUSHNEW documentation string has been corrected.
506 (thanks to Vincent Arkesteijn)
507 * bug fix: defaulting of the value for the last of an atypically
508 large number of multiple values being bound was not being
509 performed correctly on the Alpha or PPC platforms
510 * optimization: implemented multiplication as a modular
511 (UNSIGNED-BYTE 32) operation on the x86 backend.
512 * optimization: SEARCH on simple-base-strings can now be open-coded.
513 (see also contrib/compiler-extras.lisp for inspiration for
514 teaching the compiler about the Boyer-Moore algorithm).
515 * value, returned by MAX (and MIN) called with several EQUALP, but
516 not EQL, arguments now does not depend on compiler settings.
517 * fixed some bugs revealed by Paul Dietz' test suite:
518 ** in stack analysis liveness information is propagated from
519 non-local entry points.
520 ** pathwise CAST removing failed when the CAST node did not start
522 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
523 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
525 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
526 host is already defined.
527 ** RENAME-FILE works on streams instead of signalling an internal
529 ** PEEK-CHAR uses the current readtable when determining whether
530 or not a character is whitespace.
531 ** MERGE-PATHNAMES handles the case when the pathname does not
532 specify a name while the default-pathname specifies a version.
533 ** Pathnames now stand a better chance of respecting print/read
535 ** Attempting to use standardized file system operators with a
536 pathname with invalid :DIRECTORY components signals a
538 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
540 ** DIRECTORY on logical pathnames is more correct.
541 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
542 signal a TYPE-ERROR if their argument is not a stream.
543 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
544 designator argument does not designate a stream.
545 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
546 examining the synonym.
547 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
549 ** OPEN and WITH-OPEN-STREAM allow opening streams with
550 element-type larger than ([UN]SIGNED-BYTE 32).
552 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
553 * When built with the :SB-FUTEX feature, threaded builds now take
554 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
555 for faster/more reliable mutex and condition variable support.
556 * Incompatible change (but one you probably shouldn't have been using
557 anyway): the interface and code for arbitrating between multiple
558 threads in the same user session has been redesigned.
559 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
560 error when called without an explicit environment argument.
561 (thanks to Nikodemus Siivola)
562 * bug fix: buffered :DIRECTION :IO streams are less likely to become
563 confused about their position. (thanks to Adam Warner and Gerd
565 * bug fix: Pretty printing backquoted forms with unquotations in the
566 argument list position of various code constructs such as LAMBDA
567 now works correctly. (reported by Paul Dietz)
568 * bug fix: Pretty printing unquotations no longer loses all
569 stream position information.
570 * optimization: performance of string output streams is now less
571 poor for multiple small sequence writes.
572 * optimization: performance of CSUBTYPEP in the presence of complex
573 expressions involving CONS and NOT many times has been improved.
574 (reported by Paul Dietz)
575 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
577 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
578 streams for the REPL, for the debugger, and so on. (thanks to
580 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
581 * fixed some bugs revealed by Paul Dietz' test suite:
582 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
583 is ignored if it is too small, rather than propagating through
584 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
585 ** extremely complex negations of CONS types were not being
586 sufficiently canonicalized, leading to inconsistencies in
588 ** VALUES tranformer lost derived type.
590 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
591 * fixed a bootstrapping bug: the build process no longer assumes
592 that the various BOOLE-related constants have the same value in
593 host and target lisps. (noted by Paul Dietz' test suite on an
594 SBCL binary built from CLISP)
595 * The system can now be dynamically linked on the MIPS platform,
596 which enables dynamic loading of foreign code from Lisp. (thanks
597 to Ralf Baechle for discussions on the MIPS ABI)
598 * The system now records debugging information for its own source
599 files in a filesystem-position-independent manner, relative to
600 the "SYS" logical host.
601 * fixed a compiler bug: MV-LET convertion did not check references
602 to the "max args" entry point. (reported by Brian Downing)
603 * tweaked disassembly notes to be less confident about proclaiming
604 some instruction as an LRA. (thanks to Brian Downing)
605 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
606 now each have their own history, command character, and other
607 characteristics. (thanks to David Lichteblau)
608 * fixed some bugs revealed by Paul Dietz' test suite:
609 ** compiler failure in compiling LOGAND expressions including a
611 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
612 shift greater than 32.
613 ** FLUSH-DEST did not mark blocks for type check regeneration.
614 ** HANDLER-CASE failed to accept declarations in handler clauses
615 in some circumstances.
617 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
618 * New code in contrib/sb-introspect (still probably not entirely
619 stable yet) provides some support for smart Lisp development
620 environments like SLIME.
621 * The conditions signalled for errors occurring when loading .fasl
622 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
623 in a way which should help ASDF recover gracefully.
624 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
625 stuff has been cleaned up. If you code contrib/ stuff, this might
626 affect you, and you can look at contrib/README, contrib/STANDARDS,
627 and/or the 0.8.4.27 diff to check.
628 * In full calls the compiler now does not generate checks for declared
629 argument types for all arguments.
630 * various threading fixes
631 ** and some experimental patches which didn't make it into
632 the main tree for this release, but which are shipped in
633 contrib/experimental-thread.patch as a possible fix for some
634 failures (deadlock, spinning...) in GC-intensive multithreaded
636 * fixed PPC build problem (source code incompatibility of different
637 library versions): added offsetof() hackery which attempts to divine
638 where glibc maintainers put uc_mcontext today
639 * fixed bug 282: compiler does not trust type assertions while passing
640 arguments to a full call.
641 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
642 &OPTIONAL VALUES type parameter.
643 * fix bug 214: algorithm for noting rejected templates is now more
644 similar to that of template seletion. (also reported by rydis on
646 * fixed bug 141b: printing backquoted information readably and prettily
647 inserts a space where necessary.
648 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
649 (CHARACTER 10) are now reported as errors, rather than propagated
650 as unknown types. (reported by piso on #lisp)
651 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
652 with values NIL and :ERROR. (thanks to Milan Zamazal)
653 * fixed bug 191c: CLOS now does proper keyword argument checking as
654 described in CLHS 7.6.5 and 7.6.5.1.
655 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
656 longer raise an error; further, using a list as a for-as-arithmetic
657 counter now raises a meaningful error.
658 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
659 now understood by sequence creation functions such as MAKE-SEQUENCE
661 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
662 a STRING-INPUT-STREAM.
663 * compiler enhancement: SIGNUM is now better able to derive the type
665 * type declarations inside WITH-SLOTS are checked. (reported by
667 * fixed some bugs revealed by Paul Dietz' test suite:
668 ** incorrect optimization of TRUNCATE for a positive first
669 argument and negative second.
670 ** compiler failure in let-convertion during flushing dead code.
671 ** compiler failure while deriving type of TRUNCATE on an
672 interval, containing 0.
673 ** ASH of a negative bignum by a negative bignum count now returns
675 ** intersection of CONS types now canonicalizes properly, fixing
676 inconsistencies in SUBTYPEP.
678 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
679 * incompatible change: The --disable-debugger command line
680 option now clobbers the debugger at a more fundamental
681 level, by redefining #'INVOKE-DEBUGGER instead of by
682 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
683 is specified by ANSI to ignore *DEBUGGER-HOOK* and
684 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
685 enter the debugger REPL and then suffer recursive errors
686 because *DEBUG-IO* is also messed up in --disable-debugger mode;
687 while under the new system, BREAK in --disable-debugger mode
688 terminates the system just as an unhandled error would.
689 * fixed compiler performance when processing loops with a step >1;
690 * bug fix: DOCUMENTATION now retrieves generic function
691 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
692 support has been systematized, and now supports the methods
693 specified by ANSI, along with a default method and a method for
694 slot documentation. (reported by Nathan Froyd)
695 * bug fix: effective methods associated with a generic function are
696 no longer cached over a change of that generic function's method
697 combination. (reported by Andreas Fuchs)
698 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
699 names a non-existent directory. (thanks to Andreas Fuchs)
700 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
701 the PPC platform, lead to stack corruption; this has been fixed.
702 (reported by Rainer Joswig)
703 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
704 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
705 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
706 arithmetic in general, is now much more reliable on the PPC
708 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
709 platform now returns the right answer.
710 * optimization: restored some effective method precomputation in
711 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
712 precomputation is now tunable.
713 * optimization: compiler-internal data structure use has been
714 reviewed, and changes have been made that should improve the
715 performance of the compiler by about 20%.
716 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
717 simple-base-strings and simple-bit-vectors is improved.
718 * optimization: the optimization of 32-bit logical and arithmetic
719 functions introduced in version 0.8.3 on the x86 has been
720 implemented on the mips, ppc and sparc platforms; an
721 implementation of the same facility, but for 64-bit arithmetic,
722 has been added for the alpha.
723 * microoptimization: the compiler is better able to make use of the
724 x86 LEA instruction for multiplication by constants.
725 * bug fix: in some situations compiler did not report usage of
726 generic arithmetic in (SPEED 3) policy.
727 * bug 145b fix: compiler used wrong type specifier while converting
728 MEMBER-types to numeric.
729 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
731 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
732 index argument now works properly on the Alpha platform.
733 * bug fix: floating point exception treatment on the Alpha platform
735 * bug fix: FILE-POSITION works much better on string input and
736 output streams. (thanks to Nikodemus Siivola)
737 * bug fix: many threading/garbage collection symptoms sorted.
738 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
739 might be pseudo-atomic.
740 * internal change: Stopping for GC is now done with signals not
741 ptrace. GC is now done in whichever thread wanted it, instead of
743 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
745 * bug fix: result form in DO is not contained in the implicit
747 * incompatible change: ICR structure is changed; the value part of
748 CONTINUATION is now called LVAR; corresponding functions are
749 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
750 * added type deriver for ISQRT (thanks to Robert E. Brown).
751 * bug fix: better support for loading from the command line when an
752 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
754 * fixed some bugs revealed by Paul Dietz' test suite:
755 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
756 ** ROUND and FROUND now give the right answer when given very
757 small float arguments.
758 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
760 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
761 ** (GCD 0 <negative-integer>) returned <negative-integer>.
762 ** LCM should return a non-negative integer.
763 ** PARSE-INTEGER returned the index of a terminator instead of the
764 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
765 ** PARSE-INTEGER returned an incorrect index being applied to a
767 ** LCM with two arguments of 0 returns 0 rather than signalling
769 ** unsigned addition of a 32-bit constant with the high bit set no
770 longer causes an internal compiler error on the x86.
771 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
772 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
773 with negative last argument.
774 ** byte specifiers with zero size and position no longer cause
775 an error during type derivation.
776 ** bignum multiplication on the Alpha platform now returns the
778 * porting: The system now builds on SuSE AMD64, although it still
779 generates a 32-bit binary.
780 * .fasl file incompatibility: The fasl file version number has
781 been incremented (because of the changes to internal compiler
782 data structures referred to above).
784 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
785 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
786 more accurately, on the Darwin kernel running on PowerPC hardware.
787 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
788 * Compiler code deletion notes now signal a condition of type
789 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
790 an associated MUFFLE-WARNING restart.
791 * The compiler now performs limited argument count validation of
792 constant format strings in FORMAT, and where appropriate in ERROR,
793 CERROR and WARN. (thanks to Gerd Moellmann)
794 * New ASDF-INSTALL contrib can be used for automatic download and
795 installation of third-party Lisp code from CCLAN or other sites
797 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
798 which forces another thread to execute a function supplied by the
800 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
801 now accept and act upon their :ELEMENT-TYPE keyword argument.
802 (reported by Martin Atzmueller, Edi Weitz)
803 * bug fix: FILE-POSITION now accepts position designators up to
804 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
805 is the greater. (thanks to Patrik Nordebo)
806 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
807 Kalle Olavi Niemitalo)
808 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
809 printable. (reported by Eric Marsden)
810 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
811 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
812 * x86 bug fix in control stack exhaustion checking: now shows backtrace
813 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
814 (thanks to Stig Sandoe)
815 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
816 UNREAD-CHAR now correctly includes the unread character in the
817 target sequence. (thanks to Gerd Moellmann)
818 * bug fix in threaded builds: the system can now be suspended and
819 resumed by shell job control with minimal disruption.
820 * bug fixes in times and timezones >2038AD
821 * better handling of "where is GNU make?" problem in build scripts
822 (thanks to Nikodemus Siivola)
823 * new optimization: inside a named function any reference to a
824 function with the same name is considered to be a self-reference;
825 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
826 optimization quality.
827 * new optimization on x86: logical functions and + now have
828 optimized (UNSIGNED-BYTE 32) versions, which are automatically
829 used when the result is truncated to 32 bits.
830 * VALUES declaration is partially enabled.
831 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
832 * fixed some bugs revealed by Paul Dietz' test suite:
833 ** The system now obeys the constraint imposed by
834 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
835 types form a lattice under type intersection.
836 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
837 ** ASSOC now ignores NIL elements in an alist.
838 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
839 and (1+ MOST-POSITIVE-FIXNUM) answers.
840 ** The addition of a method with invalid qualifiers to a generic
841 function does not cause an error to be signalled immediately;
842 a warning is signalled, and the error is generated only on
843 calling the generic function.
844 * changed .fasl file version number, in part to add type codes for
845 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
846 obscure ANSI requirements
848 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
849 * fixed bug 148: failure to inline-expand a local function left
850 garbage, confusing the compiler.
851 * fixed bugs 3cd: structure slot readers perform type check if the
852 slot can have an invalid value (i.e. it is either not initialized
853 or can be written with a less specific slot writer).
854 * bug fix: the compiler now traps array references to elements off
855 the end of an array; previously, the bounds checking in some
856 circumstances could go off-by-one.
857 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
859 * type declarations for array element types now obey the description
860 on the CLHS page "Declaration TYPE", as per discussions on
861 sbcl-help around 2003-05-08. This means that a declaration
862 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
863 declaration, all references to BAR will be asserted or assumed
864 (with THE, so dependent on compiler policy) to involve objects of
865 type FOO. Note that no such declaration is implied in
866 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
867 * declared types of functions from the "Conditions"
868 chapter. (reported by Paul Dietz)
869 * bug fix: CERROR accepts a function as its first argument.
870 * bug fix: NTH an NTHCDR accept a bignum as index
871 arguments. (reported by Adam Warner)
872 * optimization: character compare routines now optimize comparing
873 against a constant character. (reported by Gilbert Baumann)
874 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
875 argument now works properly on the MIPS platform.
876 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
877 * repeated evaluation of the same DEFSTRUCT, a slot of which is
878 declared to have a functional type, does not cause an error
880 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
881 many :DEFAULT-INITARGS. (reported by Istvan Marko)
882 * fixed bug: if last continuation of a deleted block has a
883 destination, this destination should be deleted too. (reported by
885 * fixed a bug in the bootstrap process: the host compiler's values
886 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
887 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
888 test case from Patrik Nordebo)
889 * improved the ability of the disassembler on the PPC platform to
890 provide helpful disassembly notes.
891 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
892 the class in more cases than previously.
893 * bug fix: FILE-POSITION now understands :START and :END for
894 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
895 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
896 8) data. (thanks to David Lichteblau)
897 * bug fix: it is possible to add a method to a generic function
899 * bug fix: reader failed to signal END-OF-FILE inside an
900 object representation. (reported by Nikodemus Siivola)
901 * fixed some bugs revealed by Paul Dietz' test suite:
902 ** LAST and [N]BUTLAST should accept a bignum.
903 ** condition slot accessors are methods.
904 ** (VECTOR NIL) is a subtype of STRING.
906 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
907 * minor incompatible change: some nonsensical specialized lambda
908 lists (used in DEFMETHOD) which were previously ignored now signal
910 * minor incompatible change: the system is now aware of the types of
911 variables in the COMMON-LISP package, and will signal errors for
912 most violations of these type constraints (where previously they
913 were silently accepted).
914 * minor incompatible change: COMPILE-FILE now uses the freedom
915 afforded (ANSI 3.2.2.3) to use derived function types for
916 functions defined in the same file. This also permits the system
917 to warn on static type mismatches and function
918 redefinition. (Currently it does not work with high DEBUG level.)
919 * minor incompatible change: VALUES declaration is disabled.
920 * When issuing notes, the compiler now signals a condition of type
921 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
922 restart for use in user handlers. It is expected that the
923 COMPILER-NOTE condition will eventually become a condition
924 supertype to a hierarchy of note types, which will then be
925 handleable in a similar fashion. However, at the moment, no such
926 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
927 but it's an implementation detail, not a classification for the
929 * Changes in type checking closed the following bugs:
930 ** type checking of unused values (192b, 194d, 203);
931 ** template selection based on unsafe type assertions (192c, 236);
932 ** type checking in branches (194bc).
933 * A short form of VALUES type specifier has ANSI meaning (it has
934 increased the number of situations when SBCL cannot perform type
936 * fixed bug in DEFSTRUCT: once again, naming structure slots with
937 keywords or constants is permissible.
938 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
939 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
940 classes. (thanks to Antonio Martinez)
941 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
942 outputs FRESH-LINE or TERPRI, and no longer converts its stream
943 argument to a pretty-print stream. Instead, it leaves any such
944 operations to DESCRIBE-OBJECT methods.
945 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
947 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
948 lambda lists. (thanks to David Lichteblau)
949 * bug fix: Defining a generic function with a :METHOD-CLASS being a
950 subclass of STANDARD-METHOD no longer causes stack exhaustion.
951 (thanks to Gerd Moellmann)
952 * fixed bug 246: increased compilation speed of long
953 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
955 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
957 * DEFCONSTANT now throws a condition of type
958 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
959 constant to a non-EQL value; CONTINUE and ABORT restarts
960 respectively change and preserve the value.
961 * fixed bug 63: The code walker, part of the implementation of CLOS,
962 is now better at handling symbol macros.
963 * bug fix: There is no longer an internal implementation type named
964 CL:LENGTH. (reported by Raymond Toy)
965 * bug fix: In macro-like defining macros/special operators the
966 implicit block does not enclose the lambda list.
967 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
968 enclosed in parentheses) are not suitable as type specifiers, and
969 their use properly signals an error now.
970 * bug fix: An argument count mismatch for a type specifier in code
971 being compiled no longer causes an unhandled error at compile
972 time, but signals a compile-time warning.
973 * fixed simple vector readable printing
974 * bug fix: DESCRIBE takes more care over whether the class
975 precedence list slot of a class is bound before accessing it.
976 (reported by Markus Krummenacker)
977 * bug fix: FORMATTER can successfully compile pretty-printer format
978 strings which use variants of the ~* directive inside.
979 * bug fix: SEARCH now applies its TEST predicate to the elements of
980 the arguments in the correct order. (thanks to Wolfhard Buss)
981 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
982 in type inference. (thanks to Robert E. Brown)
983 * bug fix: Reading in symbols with an explicit package name of ""
984 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
985 * fixed some bugs revealed by Paul Dietz' test suite:
986 ** NIL is now allowed as a structure slot name.
987 ** Arbitrary numbers, not just REALs, are allowed in certain
988 circumstances in LOOP for-as-arithmetic clauses.
989 ** Multiple class redefinitions before slot access no longer
991 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
992 association between the name and a class.
993 ** Generic functions with non-standard method-combination and over
994 five methods all of which return constants no longer return NIL
995 after the first few invocations. (thanks to Gerd Moellmann)
996 ** CALL-NEXT-METHOD with no arguments now passes the original
997 values of the arguments, even in the presence of assignment.
998 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1000 ** DISASSEMBLE works with closures and funcallable instances.
1001 ** ADD-METHOD now returns the generic function, not the new method.
1002 ** FIND-METHOD signals an error if the lengths of the specializers
1003 is incompatible with the generic function, even if the ERRORP
1005 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1006 which its argument is a member.
1007 ** DEFCLASS only redefines the class named by its class-name
1008 argument if that name is the proper name of the class;
1009 otherwise, it creates a new class.
1010 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1011 of the UNBOUND-SLOT condition to the name of the slot.
1012 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1013 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1014 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1015 SLOT-MAKUNBOUND in the specified fashion.
1017 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1018 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1019 cross-compilation host. As a consequence, we can now bootstrap our
1020 way up to SBCL starting with a bare gcc toolchain and human-readable
1021 source code (first the source to CLISP, then the source to SBCL).
1022 * A contributed module containing a partial implementation of the
1023 simple-streams interface has been included. (thanks to Rudi
1025 * A contributed module implementing the RFC1321 Message Digest
1026 Algorithm, known as MD5, has been included.
1027 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1028 no longer has any effect, as the code controlled by this feature
1029 has been deleted. (As far as we know, no-one has ever built using
1030 this feature, and its semantics were confused in any case).
1031 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1032 work on conditions (as required by the ANSI specification),
1033 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1034 expected behaviour on conditions. Users should note, however,
1035 that such behaviour is not required by the ANSI specification,
1036 and so use of this behaviour may render their code unportable.
1037 * fixed some bugs revealed by Paul Dietz' test suite:
1038 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1040 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1041 ** SXHASH on condition objects no longer returns NIL.
1042 ** :ALLOCATION :CLASS slots are better treated; their values are
1043 updated on class redefinition, and initforms inherited from
1044 superclasses are applied.
1045 ** REMOVE-METHOD returns its generic function argument even when
1046 no method was removed.
1047 ** SHARED-INITIALIZE now initializes the values of the requested
1048 slots, including those with :ALLOCATION :CLASS.
1049 ** ALLOCATE-INSTANCE now works on structure classes defined via
1050 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1052 ** SLOT-EXISTS-P now works on conditions, as well as structures
1054 ** MAKE-LOAD-FORM now has the required methods on
1055 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1056 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1057 keyword, and now implements the SLOT-NAMES argument.
1058 ** methods with &OPTIONAL arguments no longer allow too many
1059 arguments to be passed in the call without error.
1060 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1061 option is consistent with the required arguments of the generic
1062 function lambda list.
1063 * bug fix: REQUIRE accepts a string designator. (Thanks to
1065 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1066 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1067 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1068 * bug fix: The compiler now checks for duplicated variables in macro
1070 * bug fix: SETQ on globals returns the correct value.
1071 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1072 not a COMPILER-ERROR (followed by some other strange error on
1073 choosing the CONTINUE restart).
1074 * bug fix: make.sh and friends are now more consistent in the way that
1075 they look for GNU "make".
1077 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1078 * experimental native threads support (on x86 Linux >=2.4 only).
1079 This is not compiled in by default: you need to add :SB-THREAD to
1080 the target features. See the "Beyond ANSI" chapter of the manual
1082 * fix for longstanding nonANSIism: The old distinction between
1083 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1084 The return value from CL:FIND-CLASS is now a CLOS class, and
1085 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1086 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1087 * An interface to the MetaObject Protocol, as described in Kiczales,
1088 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1089 MIT Press, 1991, is available from the SB-MOP package.
1090 * incompatible change: the SB-PCL package should now be considered
1091 a private implementation detail, and no longer a semi-private MOP
1093 * minor incompatible change: due to rearrangement for threads, the
1094 control stack and binding stack are now allocated at arbitrary
1095 addresses instead of being hardcoded per-port. Users affected by
1096 this probably have to be doing advanced things with shared
1097 libraries, and will know who they are.
1098 * minor incompatible change: Previously, all --eval forms used were
1099 processed with READ before any of them were processed with EVAL.
1100 Now each --eval form is processed with both READ and EVAL before
1101 the next --eval form is processed. (Thus package operations like
1102 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1103 work as the user might reasonably expect.)
1104 * minor incompatible change: *STANDARD-INPUT* is now only an
1105 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1107 * minor incompatible change: Y-OR-N-P is now character-oriented,
1108 not line oriented. Also, YES-OR-NO-P now works without errors.
1109 (thanks to Antonio Martinez)
1110 * sb-aclrepl module improvements: an integrated inspector, added
1111 repl features, and a bug fix to :trace command.
1112 * Known functions, which cannot be open coded by the backend, are
1113 considered to be able to check types of their arguments. (fixing
1114 a bug report by Nathan J. Froyd)
1115 * fixed a bug in computing method discriminating functions: It is
1116 now possible to define methods specialized on classes which have
1117 forward-referenced superclasses. (thanks to Gerd Moellmann)
1118 * fixed evaluation order in optional entries (reported by Gilbert
1120 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1121 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1122 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1123 required-by-AMOP NAME argument, as well as CLASS and
1124 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1125 * fixed bug 20: DEFMETHOD can define methods using names that are
1126 not the proper names of classes to designate class specializers.
1127 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1128 Unix file descriptors, instead of blowing up. (thanks to Antonio
1130 * Garbage collection refactoring: user-visible change is that a
1131 call to the GC function during WITHOUT-GCING will not do garbage
1132 collection until the end of the WITHOUT-GCING. If you were doing
1133 this you were probably losing anyway.
1134 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1135 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1136 TYPEP the latter but not the former.
1137 * The compiler issues a full WARNING for calls to undefined functions
1138 with names from the CL package.
1139 * MAP-INTO for a vector destination is open coded. (reported by
1140 Brian Downing on c.l.l)
1141 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1142 documentation string.
1143 * fixed some bugs revealed by Paul Dietz' test suite:
1144 ** COPY-ALIST now signals an error if its argument is a dotted
1146 ** Condition slots are now accessed more correctly in the presence
1147 of multiple initargs for a given slot.
1148 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1149 exclude restarts of the same name associated with a different
1151 ** DEFCLASS of forward-referenced classes with another
1152 forward-referenced class in the superclasses list no longer
1154 ** Condition slots are now initialized once each, not multiple
1155 times. (thanks to Gerd Moellmann)
1156 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1158 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1159 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1160 arguments contain duplicated elements.
1161 ** RESTART-CASE understands local macros.
1162 ** RESTART-CASE associates exactly its own restarts with a condition.
1163 ** ENDP in safe mode checks its argument to be of type LIST.
1164 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1165 in question is unbound.
1166 ** Optimized MAKE-INSTANCE functions no longer cause internal
1167 assertion failures in the presence of duplicate initargs.
1168 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1169 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1170 propagated even in the ctor optimized implementation of
1172 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1174 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1175 * a better implementation of SXHASH on (simple) bit vectors,
1176 measured both in execution speed and in distribution of results
1177 over the positive fixnums, has been installed. Likewise, a better
1178 implementation of EQUAL for simple bit vectors is now available.
1179 * fixed CEILING optimization for a divisor of form 2^k.
1180 * fixed bug 240 (emitting extra style warnings "using the lexical
1181 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1182 by Antonio Martinez)
1183 * fixed SXHASH, giving different results for NIL depending on type
1184 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1185 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1186 variable. (found by Rolf Wester)
1187 * SBCL does not ignore type declarations for special
1188 variables. (reported by rif on c.l.l 2003-03-05)
1189 * some bug fixes in contrib/sb-aclrepl/
1190 * fixed some bugs revealed by Paul Dietz' test suite:
1191 ** a bug in the CONS type specifier, whereby the CAR and CDR
1192 types got intertwined, has been fixed;
1193 ** the type system is now able to reason about the interaction
1194 between INTEGER and RATIO types more completely;
1195 ** APPEND, [N]REVERSE and NRECONC check that those their
1196 arguments, which must be proper lists, are really so;
1197 ** An array specialized to be unable to hold elements has been
1198 implemented, as required -- yes, really -- by ANSI;
1199 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1200 on malformed property lists;
1202 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1203 * incompatible packaging change: in line with Unix convention,
1204 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1205 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1206 * REQUIRE and PROVIDE are now optionally capable of doing something
1207 useful. See the documentation string for REQUIRE.
1208 * infrastructure for a managed SBCL contrib system: contributed
1209 modules in this release include:
1210 ** the ASDF system definition facility;
1211 ** an interface to the BSD Sockets API;
1212 ** an ACL-like convenience interface to the repl;
1213 (thanks to Kevin Rosenberg)
1214 ** an implementation of ROTATE-BYTE, with efficient implementation
1216 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1217 gives a closer approximation to the right answer than previously.
1218 (thanks to Raymond Toy)
1219 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1220 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1221 optional environment arguments, as required by ANSI.
1222 * fixed bugs in other functions taking environment objects, allowing
1223 calls with an explicit NIL environment argument to be compiled
1225 * fixed bug 228: primary return values from
1226 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1227 COMPILE or FUNCTION.
1228 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1229 :TYPE will no longer signal errors on innocuous objects.
1230 * fixed bug 231b: SETQ is better at respecting type declarations in
1231 the lexical environment.
1232 * fixed a bug in DEFCLASS: classes named by symbols with no or
1233 unprintable packages can now be defined.
1234 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1235 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1236 * fixed compiler failure related to checking types of functions.
1237 (reported by Robert E. Brown)
1238 * the compiler is now much more consistent in its error-checking
1239 treatment of bounding index arguments to sequence functions: in
1240 (SAFETY 3) code, errors will be signalled in almost all cases if
1241 invalid sequence bounding indices are passed to functions defined
1242 by ANSI to operate on sequences.
1243 * fixed a bug in the build procedure: documentation of SBCL-specific
1244 packages is now preserved and available in the final Lisp image.
1245 * lifted FDEFINITION lookup out of loops in the implementation of
1246 many list operations. (thanks to Robert E. Brown)
1247 * fixed a bug in the reader: the #n# reader macro now works for
1248 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1249 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1250 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1251 E. Brown and Tony Martinez respectively)
1252 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1253 SLOT-BOUNDP: in particular, it is now able to optimize them much
1254 better, and is now not vulnerable to having packages renamed.
1255 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1256 always called when a slot is not present in an instance. (thanks
1258 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1259 longer signal an error on unreading a character following EOF on
1260 the previous constituent stream. (thanks to Tony Martinez)
1261 * fixed some bugs revealed by Paul Dietz' test suite:
1262 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1263 not just nonnegative fixnums;
1264 ** the logical bit-array operators such as BIT-AND now accept an
1265 explicit NIL for their "opt-arg" argument (to indicate a
1266 freshly-consed result bit-array);
1267 ** ELT now signals an error on an invalid sequence index in safe
1269 ** the type system is now cleverer about negations of numeric
1270 types, and consequently understands the BIGNUM and RATIO types
1272 ** the type system is now cleverer about the interaction between
1273 INTEGER and RATIO types: while bugs still remain, many more
1274 cases are accurately computed;
1275 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1276 if it is in the last clause;
1277 ** CONSTANTLY now correctly returns a side-effect-free function in
1279 ** DECLARE is no longer treated as a special-operator; in
1280 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1281 * incremented fasl file version number due to the change in the
1282 DEFSTRUCT-SLOT-DESCRIPTION structure.
1284 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1285 * minor incompatible change: code processed by the "interpreter" or
1286 EVAL now has a compilation optimization policy of (DEBUG 2)
1287 (changed from (DEBUG 1)) to improve debuggability of interactive
1288 development, and to allow the use of the debug RETURN command in
1290 * an experimental implementation of the RETURN command for the
1291 debugger has been included. (thanks to Frederik Kuivinen)
1292 * fixed bug 62: constraints were not propagated into a loop.
1293 * fixed bug in embedded calls of SORT (reported and investigated by
1295 * fixed some bugs revealed by Paul Dietz' test suite:
1296 ** printing and reading of arrays with some dimensions having
1297 length 0 (thanks to Gerd Moellmann);
1298 ** BOA constructor with &AUX argument without a default value does
1299 not cause a type error;
1300 ** CONSTANTP now returns true for all self-evaluating objects.
1302 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1303 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1304 accessors that are related by inheritance, as specified in the
1305 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1306 Valtteri Vuorikoski)
1307 * The compiler is now able to inline functions that were defined in
1308 a complex lexical environment (e.g. inside a MACROLET).
1309 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1311 * Support for the upcoming FreeBSD-5.0 release has been included.
1312 (thanks to Dag-Erling Smorgrav)
1313 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1314 effect when it is not in a toplevel context.
1315 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1316 stand a better chance of being correct. (thanks to Gerd
1318 * fixed bug in COERCE, which now signals an error on coercing a
1319 rational to a bounded real type which excludes the expected
1321 * The compiler is now able to derive types more accurately from the
1322 COERCE and COMPILE functions.
1323 * fixed bug 223: functional binding is considered to be constant
1324 only for symbols in the CL package.
1325 * fixed bug 231: SETQ did not check the type of a variable being set
1326 (reported by Robert E. Brown)
1327 * A new optimization for MAKE-INSTANCE has been included, fixing
1328 various bugs (including relating to :ALLOCATION :CLASS slots and
1329 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1331 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1332 ** As required by ANSI, LOOP now disallows anonymous collection
1333 clauses such as COLLECT I in conjunction with aggregate boolean
1334 clauses such as THEREIS (= I 1);
1335 ** LOOP now signals an error when any variable is reused in the
1336 same loop (including the potentially useful construct analogous
1337 to WITH A = 1 WITH A = (1+ A);
1338 ** IT is only a special loop symbol within the first clause of a
1339 conditional loop clause;
1340 ** LOOP with a typed iteration variable over a hashtable now
1341 signals a type error iff it should.
1342 * fixed some other bugs revealed by Paul Dietz' test suite:
1343 ** FILE-STREAM now names the class previously known as FD-STREAM;
1344 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1345 argument) no longer signals an error;
1346 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1347 of the slot symbol, rather than using the current package
1348 ((:CONC-NAME "") continues to intern the slot's name in the
1350 * incremented fasl file version number, because of the incompatible
1351 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1352 of the new implementation of DEFINE-COMPILER-MACRO.
1354 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1355 * Support for building SBCL for MIPS platforms running in
1356 little-endian mode has now been checked in, and basic
1357 functionality on said platforms verified.
1358 * minor incompatible change: PCL now records the pathname of a file
1359 in which methods and the like are defined, rather than its
1361 * minor incompatible change: TRUENAME now considers the truename of
1362 a file naming a directory to be the pathname with :DIRECTORY
1363 component indicating that directory.
1364 * minor incompatible change: a NAMED clause in the extended form of
1365 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1366 reason for the previous behaviour is unclear.
1367 * more systematization and improvement of CLOS and MOP conformance
1368 in PCL (thanks to Gerd Moellman and Pierre Mai):
1369 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1371 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1372 primary methods with no specializers;
1373 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1375 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1376 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1377 FORWARD-REFERENCED-CLASSes; error reporting on
1378 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1380 ** SXHASH on CLOS instances now uses a slot internal to the
1381 instance to return different numbers on distinct instances,
1382 while preserving the same return value through invocations of
1384 ** DEFMETHOD signals errors when methods with longer incongruent
1385 lambda lists are added to generic functions;
1386 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1387 CLASS, as specified in AMOP;
1388 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1389 on the order returned by the primary method for classes of
1390 class STANDARD-CLASS;
1391 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1392 * fixed some bugs shown by Paul Dietz' test suite:
1393 ** DOLIST puts its body in TAGBODY;
1394 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1396 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1397 value producing form;
1398 ** if more variables are given to PROGV than values, extra
1399 variables are bound and made to have no value;
1400 ** NSUBSTITUTE on list arguments gets the right answer with
1402 ** ELT signals an error of type TYPE-ERROR when the index argument
1403 is not a valid sequence index;
1404 ** LOOP signals (at macroexpansion time) an error of type
1405 PROGRAM-ERROR when duplicate variable names are found;
1406 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1407 ** FUNCALL of special-operators now cause an error of type
1409 ** PSETQ now works as required in the presence of side-effecting
1410 symbol-macro places;
1411 ** NCONC accepts any object as its last argument;
1412 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1414 ** loop-for-as-package does not require a package to be explicitely
1416 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1418 * fixed bug 166: compiler preserves "there is a way to go"
1419 invariant when deleting code.
1420 * fixed bug 172: macro lambda lists with required arguments after
1421 &REST arguments now cause an error to be signalled. (thanks to
1423 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1424 bug. (thanks to Matthew Danish)
1425 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1427 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1428 arguments are lexically rebound. (thanks to Gerd Moellmann and
1430 * fixed bug 194: error messages are now more informative when there
1431 is no primary method applicable in a call to a generic
1432 function. (thanks to Gerd Moellmann)
1433 * fixed bug in command line argument checking (thanks to Julian
1435 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1436 (thanks to Matthew Danish)
1437 * incremented fasl file version number, because of the
1438 SXHASH-related changes in the layout of CLOS data structures
1440 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1441 * minor incompatible change: The runtime (the Unix executable named
1442 "sbcl") is now much pickier about the .core files it will load.
1443 Essentially it now requires .core files to descend from the same
1444 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1445 as the runtime does. (The intent is to prevent the crashes which
1446 can occur, and which can even be reported as mysterious failures,
1447 when people patch the sources or change the build parameters
1448 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1449 sbcl and .core files.)
1450 * fixed bug: VALUES-LIST is no longer optimized away.
1451 * fixed bug 142: The FFI conversion of C string values to Lisp
1452 string values no longer conses excessively. (thanks to Nathan
1453 Froyd porting Raymond Toy's fix to CMU CL)
1454 * began to systematize and improve MOP conformance in PCL (thanks to
1455 Nathan Froyd, Gerd Moellman and Pierre Mai):
1456 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1458 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1459 ** FINALIZE-INHERITANCE is now called on class finalization;
1460 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1461 argument precedence order.
1462 * fixed bug 202: The compiler no longer fails on functions whose
1463 derived types contradict their declared type.
1464 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1465 so it can be non-toplevel.
1466 * The fasl file version number has changed (because of the new
1467 implementation of DEFMACRO).
1468 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1469 safe code, that any length requirement by their type-specifier
1470 argument is valid. The exceptions to this are described in bug
1472 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1473 functions now check the types of their inputs as required by ANSI.
1474 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1475 introduced symbol is DECLAREd to be SPECIAL.
1476 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1477 * fixed bug: :COUNT argument to sequence functions may be negative
1478 * fixed bug: body of DO-SYMBOLS may contain declarations
1479 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1480 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1481 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1482 symbol macro only once
1483 * fixed printing of call frame when argument list is unavailable
1484 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1485 * compiler no longer signals WARNING on unknown keyword
1488 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1489 * A beta-quality port to the mips architecture running Linux,
1490 based on the old CMUCL backend, has been made. It has been tested
1491 on a big-endian kernel, and works sufficiently well to be able to
1492 rebuild itself; it has not been tested in little-endian mode.
1493 * fixed an inconsistency between gencgc.c and purify.c which made
1494 dumping/loading .core files unreliable
1495 * fixed bug 120a: The compiler now deals correctly with IFs where
1496 the consequent is the same as the alternative, instead of
1497 misderiving the return type. (thanks to Alexey Dejneka)
1498 * fixed bug 113: Logical pathnames are now dumpable (the logical
1499 host is resolved at load-time, throwing an error if it is not
1501 * fixed bug 174: FORMAT's error message is slightly clearer when a
1502 non-printing character is used in a format directive.
1503 * fixed several bugs in compiler checking of type declarations, i.e.
1504 violations of the Python "declarations are assertions" principle
1505 (thanks to Alexey Dejneka)
1506 * fixed several bugs in PCL's error checking (thanks to Gerd
1508 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1510 * fixed bug in compilation of functions as first class values
1511 (thanks to Antonio Martinez-Shotton)
1512 * The compiler's handling TYPE-ERRORs which it can prove will
1513 inevitably happen at runtime has been cleaned up and corrected
1514 in several ways. (thanks to Alexey Dejneka)
1515 * improved argument type checking for various basic arithmetic
1516 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1517 lavished on them in the past that they can be compiled in many
1518 ways in different special cases
1519 * fixed bug 181: compiler checks validity of user supplied type
1521 * cleaned up code flushing in optimization: Function calls which
1522 should signal errors for safety purposes (e.g. which ANSI says
1523 should signal errors when their arguments are of incorrect type)
1524 are no longer optimized away.
1525 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1526 * incremented fasl file version number, because changes in the
1527 implementation of sequence functions like COERCE caused
1528 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1529 in old inline expansions) to become undefined. (Actually these
1530 changes were later undone, so we might very well be binary
1531 compatible with 0.7.7 after all, but leaving the version number
1532 incremented seemed like the simplest and most conservative
1535 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1536 * An alpha-quality port to the parisc architecture running Linux,
1537 based on the old CMUCL backend, has been made. This, even more so
1538 than the other backends, should be considered still a work in
1539 progress; known problems include that the Linux kernel in 64-bit
1540 mode does not propagate the correct sigcontext structure to
1541 userspace, and consequently SBCL on a parisc64 kernel will not
1543 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1544 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1545 suppressed.) Also now that the compiler is looking at declarations
1546 in the environment, it checks optimization declarations as well,
1547 and suppresses inlining when (> DEBUG SPEED).
1548 * More fixes have been made to treatment of floating point exception
1549 treatment and other Unix signals. In particular, floating point
1550 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1551 * The detection and handling of control stack exhaustion (infinite
1552 or very deeply nested recursion) has changed. Stack exhaustion
1553 detection is now done by write-protecting pages at the OS level
1554 and applies at all optimization settings; when found, a
1555 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1556 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1557 be caught using IGNORE-ERRORS.
1558 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1559 correctly when transforming recursive LABELS functions to LETs.
1560 (thanks to Alexey Dejneka)
1561 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1562 that are names of constants or global variables.
1563 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1564 alien routines with docstrings.
1565 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1566 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1568 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1569 (thanks to Eric Marsden)
1570 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1571 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1572 * Bug fix: The SPARC backend can now compile functions involving
1573 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1574 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1575 to a functions expecting a PATHNAME-DESIGNATOR.
1576 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1577 lambda lists. (thanks to Alexey Dejneka)
1578 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1579 functions so that EXPORTing the name of the class doesn't cause
1580 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1581 to Antonio Martinez for reporting this)
1582 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1583 constant-folded without failing an assertion. (thanks to Einar
1584 Floystad Dorum for reporting this)
1585 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1586 types are now dealt with more correctly by the compiler.
1587 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1588 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1589 behaviour analogous to RENAME-FILE. This puts its behaviour more
1590 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1592 * The fasl file version number has changed again. (because of the
1593 bug fix involving the names of PCL MAKE-INSTANCE functions)
1595 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1596 * bug fix: Floating point exceptions are treated much more
1597 consistently on the x86/Linux and PPC/Linux platforms.
1598 * Array initialization with :INITIAL-ELEMENT is now much faster for
1599 cases when the compiler cannot open code the array creation, but
1600 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1601 array accesses have also seen a speed increase.
1602 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1603 specified. (This was at the root of some bad interactions between
1604 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1606 * bug fix: Internal error arguments for undefined functions are now
1607 computed correctly on the PPC/Linux platform.
1608 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1609 Raymond Toy's patch for CMU CL)
1610 * Support for the Solaris 9 operating environment has been included
1611 (thanks to Daniel Merritt)
1612 * A very ugly but hopefully complete draft of the missing FFI chapter
1613 of the manual has been created by reformatting the corresponding
1614 CMU CL manual chapter into (currently very ugly and incoherent)
1615 DocBook and bringing it up to date for SBCL behavior. Thus, the
1616 manual is now essentially complete, at least by my extreme
1617 once-and-only-once standards, whereby it's acceptable to refer to
1618 the doc strings of SB-EXT functions as the primary documentation.
1619 * The fasl file version number has changed again, due to cleanup of
1620 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1621 is no longer a static symbol.)
1623 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1624 * SBCL now builds with OpenMCL (version 0.12) as the
1625 cross-compilation host; also, more progress has been made toward
1626 bootstrapping under CLISP.
1627 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1629 * bug 158 fixed: The compiler can now deal with integer loop
1630 increments different from 1; fixing this turned out also to fix
1632 * bug 169 fixed: no more bogus warnings about using lexical bindings
1633 despite the presence of perfectly good SPECIAL declarations (thanks
1634 to David Lichteblau)
1635 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1636 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1637 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1639 * bug 180 fixed: Method combination specifications no longer ignore
1640 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1641 * bug fix: Structure type predicate functions now check their argument
1642 count as they should.
1643 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1644 correctly. (thanks to Pierre Mai)
1645 * minor incompatible change: The --noprogrammer option is deprecated
1646 in favor of the new --disable-debugger option, which is very similar.
1647 (The major difference is that it takes effect at a slightly different
1648 time at startup, causing handling of errors in --sysinit and
1649 --userinit files will be affected differently.) The
1650 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1651 been added to allow this functionality to be controlled from ordinary
1652 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1653 maintainers, who might want to run non-interactive scripts to
1654 build SBCL cores which will later be used interactively.)
1655 * minor incompatible change: The LOAD function no longer, when given
1656 a wild pathname to load, loads all files matching that pathname.
1657 Instead, an error of type FILE-ERROR is signalled.
1659 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1660 * bug 147 fixed: The compiler preserves its block link/count
1661 invariants more correctly now so that it doesn't crash. (thanks
1663 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1665 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1666 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1667 no-value-for-this-character as NIL (as specified by ANSI).
1668 * HOST-NAMESTRING on physical pathnames now returns a string that is
1669 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1670 (thanks to Christophe Rhodes)
1671 * The Alpha port handles icache flushing more correctly. (thanks to
1673 * More progress has been made toward bootstrapping under CLISP. (thanks
1674 to Christophe Rhodes)
1675 * The fasl file format has changed again, because dynamic loading
1676 on OpenBSD (which has non-ELF object files) motivated some cleanups
1677 in the way that foreign symbols are transformed and passed around.
1678 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1679 is no longer treated as whitespace by the reader, but instead as
1680 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1681 |A
\7fB|, instead of A as it used to.
1683 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1684 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1685 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1686 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1687 this as of 0.7.1.45, but was left out of the previous news section
1688 (thanks to Dan Barlow)
1689 * SBCL now runs on the Solaris operating system on SPARC architectures
1690 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1691 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1692 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1693 * SPARC backend cleanups, allowing builds of cores optimized for V8
1694 and V9 SPARCS, and also emission of code targeted to a particular
1695 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1697 * SBCL is closer to bootstrapping under CLISP, thanks to various
1698 fixes by Christophe Rhodes.
1699 * The fasl file format has changed again, to allow the compiler's
1700 INFO database to support symbol macros.
1701 * The user manual (in doc/) is formatted into HTML more nicely.
1702 (thanks to coreythomas)
1703 * The system is smarter about SUBTYPEP relationships, especially
1704 those involving NOT types (including types such as ATOM which are
1705 represented internally using NOT types). Thus SUBTYPEP is less
1706 likely to return (VALUES NIL NIL) in general, and in particular
1707 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1709 * The fasl file format has changed again, because the internal
1710 representation of types now includes a new slot to support the new
1711 SUBTYPEP-of-NOT-types logic.
1712 * (not a change in the main branch of SBCL, but a related prototype
1713 which can hopefully be merged into the main branch of SBCL in the
1714 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1715 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1716 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1717 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1718 dispatch macro character combinations. (thanks to Alexey Dejneka)
1719 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1720 that superficially look like logical namestrings correctly.
1721 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1723 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1724 default supplied-p) for &optional and &key arguments. (thanks to
1727 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1728 * incompatible change: The compiler is now less aggressive about
1729 tail call optimization, doing it only when (> SPACE DEBUG) or
1730 (> SPEED DEBUG). (This is an incompatible change because there are
1731 programs which relied on the old CMU-CL-style behavior to optimize
1732 away their unbounded recursion which will now die of stack overflow.)
1733 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1734 for non-GENCGC systems has been increased to 20M (since that
1735 seems much closer to the likely performance optimum for modern
1736 systems than the old 4M value was)
1737 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1738 and *DEBUG-PRINT-LEVEL*
1739 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1740 of CMU CL's support for SPARC, and various endianness and other
1741 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1742 * new syntactic sugar for the Unix command line: --load foo.bar is now
1743 an alternate notation for --eval '(load "foo.bar")'.
1745 ** The system now detects stack overflow and handles it gracefully,
1746 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1747 optimization settings. (This is a good thing in general, and
1748 its introduction in this version should be particularly timely
1749 for anyone whose code fails because of suppression of tail
1751 ** The system now hunts for the C variable "environ" in a more
1752 devious way, to avoid segfaults when the C library version differs
1753 between compile time and run time. (thanks to Christophe Rhodes)
1754 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1755 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1756 ** The compiler no longer issues bogus style warnings for undefined
1757 classes in the same source file as the DEFCLASSes which defined
1758 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1760 ** fixes in CONDITION class precedence list for undefined function
1761 errors (thanks to Alexei Dejneka)
1762 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1763 correctly. (thanks to Dan Barlow)
1764 ** portability fixes aiming at bootstrapping under CLISP (thanks
1765 to Dave McDonald and Christophe Rhodes)
1766 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1767 ** fixes in type translation and and type inference (thanks to
1769 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1770 ** various fixes in the new ports (thanks to Dan Barlow)
1771 * several changes related to debugging:
1772 ** suppression of tail recursion, as noted above
1773 ** stack overflow detection, as noted above
1774 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1775 is now the default. (For some time encapsulation has been more
1776 reliable than the breakpoint-based :ENCAPSULATE NIL
1777 implementation, at least on X86 systems; and I just noticed that
1778 encapsulation also seems closer to the spirit of the ANSI
1781 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1783 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1784 up properly again. (There was a packaging bug in 0.7.0 which
1785 left their definitions in SB-SYS::LOAD-FOREIGN and
1786 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1787 vital for most things which interface to C-level interfaces,
1788 like extensions working with sockets or databases or
1789 Perl-compatible regexes or whatever, and the need to fix
1790 this bug is the main reason that 0.7.1 was released so
1792 ** DEFGENERIC is now choosier about the methods it redefines, so that
1793 reLOADing a previously-LOADed file containing DEFGENERICs does
1794 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1795 cycle now works as it should. (thanks to Alexey Dejneka)
1796 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1797 fixed by Christophe Rhodes. (He actually submitted this patch
1798 months ago, and I delayed until after 0.7.0.)
1799 ** Bug 111 (internal compiler confusion about runtime checks on
1800 FUNCTION types) was fixed by Alexey Dejneka.
1801 * Some internal cleanups (getting rid of variables which aren't
1802 needed now that the byte interpreter is gone) caused the fasl
1803 file format number to change again.
1805 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1806 * major incompatible change: The default fasl file extension, i.e. the
1807 default extension for files produced by COMPILE-FILE, has changed
1808 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1810 ** There are many changes in the implementation of the compiler.
1811 SBCL is now essentially a compiler-only implementation of ANSI
1812 Common Lisp. EVAL still "interprets" a few special cases, but
1813 almost all the interesting cases are handled by creating
1814 a LAMBDA expression, calling COMPILE on it, then calling
1815 FUNCALL on the result.
1816 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1817 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1818 Since the code is newer, there might still be some new bugs
1819 (though not as many as before Martin Atzmueller's fixes:-). But
1820 the new code is substantially simpler and clearer, and hopefully
1821 any remaining bugs will be simpler, less fundamental, and more
1822 fixable then the bugs in the old code.
1823 ** The revised compiler is still a little unsteady on its feet.
1825 *** The debugging information it produces (particularly the names
1826 of FUNCTION objects) is sometimes much less useful than what
1827 the old compiler produced.
1828 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1829 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1830 within a MACROLET) has been temporarily weakened.
1831 ** There are new compiler optimizations for various functions:
1832 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1833 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1834 *** the math functions TRUNCATE, FLOOR, and CEILING
1835 *** the function-of-all-trades COERCE
1836 Mostly these should be transparent, but there's one
1837 potentially-annoying problem (bug 117): when the compiler
1838 inline-expands a function and does type analysis on the result,
1839 it can create control paths which have type mismatches, and
1840 when it can't prove that those control paths aren't taken,
1841 it will issue WARNINGs about the type mismatches. This is
1842 a particular problem in practice for the new sequence functions.
1843 It's not clear how this should be fixed, and for now, a
1844 workaround is given in the entry for 117 in the BUGS file.
1845 ** (Because of the interaction between the two previous items --
1846 occasional inlining problems and new inline expansions -- some
1847 of the new sequence function optimizations won't really kick in
1848 completely until debugging information, and then inlining, are
1849 straightened out in some future version.)
1850 * minor incompatible changes:
1851 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1852 DIRECTORY no longer implicitly promotes NIL slots of its
1853 pathname argument to :WILD. In particular, when you ask for the
1854 contents of a directory (which you used to be able to do without
1855 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1856 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1857 ** changes in behavior that ANSI explicitly defines to be
1858 implementation dependent:
1859 *** The new compiler-only implementation still conforms with ANSI,
1860 but acts a little different than before. Besides the obvious
1861 changes in performance tradeoffs (that the cost per form passed
1862 to EVAL has gone up, and the cost per form executed by EVAL
1863 has gone down), the behavior of the system changes a little
1864 because there are no longer any interpreted function objects.
1865 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1866 e.g. doing COMPILE on the output of interactive DEFUN is
1868 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1870 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1871 has changed from (:CL) to NIL.
1872 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1873 e.g. #\Newline and #\Tab, don't have names specified in the
1874 ANSI Common Lisp standard, is now based on their ASCII symbolic
1875 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1876 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1877 are no longer used for output.
1878 ** changes in internal implementation constants:
1879 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1880 4 million. (If your application spends a lot of time GCing and
1881 you have a lot of RAM, you might want to experiment with
1882 increasing it even more.)
1883 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1884 However, almost all old code should still continue to work without
1885 immediate update, as SB-C-CALL is now a (deprecated) nickname
1887 ** Old operator names in the style DEF-FOO are now deprecated in
1888 favor of new corresponding names DEFINE-FOO, for consistency with
1889 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1890 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1891 internal symbols, but a few supported extensions like
1892 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1893 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1894 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1895 etc. as you get deeper into recursive calls to the debugger
1896 command loop, instead of the old "5]", "5]]", "5]]]"
1897 sequence. (I was motivated to do this when squabbles between
1898 ILISP and SBCL left me very deeply nested in the debugger. In the
1899 short term, this change will probably provoke more ILISP/SBCL
1900 squabbles, but hopefully it will be an improvement in the long run.)
1901 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1902 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1903 function, since declaiming return types involving aliens is
1904 (1) annoyingly messy to do by hand and (2) vital to efficient
1905 compilation of code which calls such functions.
1906 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1907 longer reexported by the SB-EXT package. They're solely useful
1908 for alien code, so it seems more logical that you should get
1909 them from the SB-ALIEN package, not in SB-EXT.
1910 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1911 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1912 features. Instead, the code that they used to control is always
1913 built into the system.
1914 * many other bug fixes
1915 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1916 advantage of the new EVAL-WHEN stuff and to clean them up in
1917 general, and they are now more ANSI-compliant in a number of
1918 ways. Martin Atzmueller is responsible for a lot of this.
1919 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1921 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1922 *** correct ERROR type for various file operations
1923 *** some fixes for Lisp streams
1924 *** DEFMETHOD syntax checking
1925 *** changing old weird representation of debug information as
1926 strings (which, among their other deficiencies, don't transform
1927 correctly when you rename packages, and don't change their
1928 print representation when you change things like *PACKAGE*
1929 and *PRINT-LENGTH*) to symbols and lists of symbols
1930 He also made several improvements and fixed several bugs in DESCRIBE.
1931 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1933 *** misbehavior of WRITE-STRING/WRITE-LINE
1934 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1935 and several other LOOP problems as well
1936 *** DIRECTORY when similar filenames are present
1937 *** DEFGENERIC with :METHOD options
1938 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1939 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1940 *** argument ordering in FIND with :TEST option
1941 *** mishandled package designator argument in APROPOS-LIST
1942 *** various problems in the backquote readmacro
1943 *** a bug in APROPOS
1944 *** probably some others that I'm not describing very well here,
1945 since the CVS log documents them by reference to sbcl-devel
1946 messages, and the SourceForge archives aren't working well.:-(
1947 ** Dan Barlow improved the Alpha port (and is making progress on the
1948 PPC port, for those of you who think different).
1949 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
1950 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
1951 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
1952 continues to work on the SPARC port (for those of you in a position
1953 to look down upon our little PC-compatible boxes from a great height).
1954 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
1955 and uses that copy, rather than the current dynamic value, when
1956 it's trying to decide whether to truncate output. Thus e.g.
1957 (let ((*print-lines* 50))
1958 (pprint-logical-block (stream nil)
1960 (let ((*print-lines* 8))
1961 (print (aref possiblybigthings i) stream)))))
1962 should now truncate the logical block only at 50 lines, instead of
1963 often truncating it at 8 lines, as it did before.
1964 * The doc/cmucl/ directory, containing old CMU CL documentation
1965 from the time of the fork, is no longer part of the base system.
1966 SourceForge has shut down its anonymous FTP service, and with it
1967 my original plan for distributing the old CMU CL documentation
1968 there. For now, if you need these files you can download an old
1969 SBCL source release and extract them from it.
1970 * The fasl file version number changed again, for dozens of reasons,
1971 some of which are apparent above.
1973 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
1974 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
1975 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
1976 patch, so that DISASSEMBLE output is much nicer.
1977 * The code in the SB-PROFILE package now seems reasonably stable.
1978 I still haven't decided what the final interface should look like
1979 (I'd like PROFILE to interact cleanly with TRACE, since both
1980 facilities use function encapsulation) but if you have a need
1981 for profiling now, you can probably use it successfully with
1982 the current CMU-CL-style interface.
1983 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
1984 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
1985 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
1986 link now returns the dangling link itself, and for similar
1987 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
1988 link itself. (In these cases the old code signalled an error and
1989 looped endlessly, respectively.) Thus, DIRECTORY now works even
1990 in the presence of dangling and cyclic symbolic links.
1991 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
1992 is now a supported extension again, since the consensus on
1993 sbcl-devel was that it can be useful for ordinary development
1994 work, not just for debugging SBCL itself.
1995 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
1996 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
1997 that currently-defined functions might be redefined later with
1998 different return types.
1999 * Hash tables can be printed readably, as inspired by CMU CL code
2000 of Eric Marsden and SBCL code of Martin Atzmueller.
2001 * better error handling in CLOS method combination, thanks to
2002 Martin Atzmueller porting Pierre Mai's CMU CL patches
2003 * more overflow fixes for >16Mbyte I/O buffers
2004 * A bug in READ has been fixed, so that now a single Ctrl-D
2005 character suffices to cause end-of-file on character streams.
2006 In particular, now you only need one Ctrl-D at the command
2007 line (not two) to exit SBCL.
2008 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2010 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2011 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2012 does the right thing.
2013 * The implementation of some type tests, especially for CONDITION
2014 types, is now tidier and maybe faster, due to CMU CL code
2015 originally by Douglas Crosher, ported by Martin Atzmueller.
2016 * Some math functions have been fixed, and there are new
2017 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2018 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2019 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2020 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2021 likely to be built into sbcl-0.7.0.)
2022 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2023 CL, ported to SBCL by Martin Atzmueller)
2024 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2025 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2026 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2027 * There's a new slam.sh hack to shorten the edit/compile/debug
2028 cycle for low-level changes to SBCL itself, and a new
2029 :SB-AFTER-XC-CORE target feature to control the generation of
2030 the after-xc.core file needed by slam.sh.
2031 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2032 COMPILE-FILE is no longer supported, so that now every function
2033 gets an entry point, so that block compilation looks a little
2034 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2035 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2036 returns the number of bytes consed since the system started,
2037 rather than the number consed since the first time the function
2038 was called. (The new definition parallels ANSI functions like
2039 CL:GET-INTERNAL-RUN-TIME.)
2040 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2041 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2042 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2043 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2044 * Fasl file version numbers are now independent of the target CPU,
2045 since historically most system changes which required version
2046 number changes have affected all CPUs equally. Similarly,
2047 the byte fasl file version is now equal to the ordinary
2050 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2051 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2052 is no longer recognized. I apologize for this, because it was
2053 listed in SB-EXT as a supported extension, but I found that
2054 its existing behavior was poorly specified, as well as incorrectly
2055 specified, and it looked like too much of a mess to straighten it
2056 out. I have enough on my hands trying to get ANSI stuff to work..
2057 * many patches ported from CMU CL by Martin Atzmueller, with
2058 half a dozen bug fixes in pretty-printing and the debugger, and
2059 half a dozen others elsewhere
2060 * fixed bug 13: Floating point infinities are now supported again.
2061 They might still be a little bit flaky, but thanks to bug reports
2062 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2063 as flaky as they were.
2064 * The --noprogrammer command line option is now supported. (Its
2065 behavior is slightly different in detail from what the old man
2066 page claimed it would do, but it's still appropriate under the
2067 same circumstances that the man page talks about.)
2068 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2069 are now supported, and enabled by default. Thus, the compiler can
2070 handle many floating point and complex operations much less
2071 inefficiently. (Thus e.g. you can implement a complex FFT
2073 * The compiler now detects type mismatches between DECLAIM FTYPE
2074 and DEFUN better, and implements CHECK-TYPE more correctly, and
2075 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2076 thanks to patches from Martin Atzmueller.
2077 * various fixes to make the cross-compiler more portable to
2078 ANSI-conforming-but-different cross-compilation hosts (notably
2079 Lispworks for Windows, following bug reports from Arthur Lemmens)
2080 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2081 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2082 fixed thanks to Pierre Mai's CMU CL patches.
2083 * fixes to keep the system from overflowing internal counters when
2084 it tries to use i/o buffers larger than 16M bytes
2085 * fixed bug 45a: Various internal functions required to support
2086 complex special functions have been merged from CMU CL sources.
2087 (When I was first setting up SBCL, I misunderstood a compile-time
2088 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2089 * improved support for type intersection and union, fixing bug 12
2090 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2091 more obscure bugs as well
2092 * some steps toward byte-compiling non-performance-critical
2093 parts of the system, courtesy of patches from Martin Atzmueller
2094 * Christophe Rhodes has made some debian packages of sbcl at
2095 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2096 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2097 stable, but are nonetheless usable. When he's ready, I'd be happy
2098 to add them to the SourceForge "File Releases" section. (And if
2099 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2100 * new fasl file format version number (because of changes in
2101 internal representation of (OR ..) types to accommodate the new
2102 support for (AND ..) types, among other things)
2104 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2105 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2107 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2109 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2110 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2111 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2112 * DESCRIBE now works on structure objects again.
2113 * Most function call argument type mismatches are now handled as
2114 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2115 know whether the function will be redefined before the call is
2116 executed. (The compiler could flag local calls with full WARNINGs,
2117 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2118 it doesn't keep track of enough information to know whether calls
2119 are local in this sense.)
2120 * Compiler output is now more verbose, with messages truncated
2121 later than before. (There should be some supported way for users
2122 to override the default verbosity, but I haven't decided how to
2123 provide it yet, so this behavior is still controlled by the internal
2124 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2125 src/compiler/ir1util.lisp.)
2126 * Fasl file format version numbers have increased again, because
2127 support for the Gray streams extension changes the layout of the
2128 system's STREAM objects.
2129 * The Gray subclassable streams extension now works, thanks to a
2130 patch from Martin Atzmueller.
2131 * The full LOAD-FOREIGN extension (not just the primitive
2132 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2133 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2134 but like most other programs, it defaults to copying the Unix
2135 environment from the original process instead of starting the
2136 new process in an empty environment.
2137 * Extensions which manipulate the Unix environment now support
2138 an :ENVIRONMENT keyword option which doesn't smash case or
2139 do other bad things. The CMU-CL-style :ENV option is retained
2140 for porting convenience.
2141 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2142 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2144 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2146 * A patch from Martin Atzmueller seems to have solved the SIGINT
2147 problem, and as far as we know, signal-handling now works cleanly.
2148 (If you find any new bugs, please report them!)
2149 * The system no longer defaults Lisp source file names to types
2150 ".l", ".cl", or ".lsp", but only to ".lisp".
2151 * The compiler no longer uses special default file extensions for
2152 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2153 seems to expect a single default extension for all compiled code,
2154 and there's no compelling reason to try to stretch the standard
2155 to allow two different extensions.) Instead, byte-compiled files
2156 default to the same extension as native-compiled files.
2157 * Fasl file format version numbers have increased again, because
2158 a rearrangement of internal implementation packages made some
2159 dumped symbols in old fasl files unreadable in new cores.
2160 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2161 many fewer weird special cases.
2162 * Bug #17 (differing COMPILE-FILE behavior between logical and
2163 physical pathnames) has been fixed, and some related misbehavior too,
2164 thanks to a patch from Martin Atzmueller.
2165 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2166 by Tim Moore, ported to SBCL by Martin Atzmueller.
2167 * Martin Atzmueller fixed several filesystem-related problems,
2168 including bug #36, in part by porting CMU CL patches, which were
2169 written in part by Paul Werkowski.
2170 * More compiler warnings in src/runtime/ are gone, thanks to
2171 more patches from Martin Atzmueller.
2172 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2175 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2177 * DESCRIBE now works on CONDITION objects.
2178 * The debugger now handles errors which arise when trying to print
2179 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2181 * The build system now uses an additional file, customize-target-features.lisp,
2182 to allow local modifications to the target *FEATURES* list. (The point of
2183 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2184 debugging features enabled, without having to worry about propagating it
2185 into everyone's system when I do a "cvs update".) When no
2186 customize-target-features.lisp file exists, the target *FEATURES* list
2187 should be constructed the same way as before.
2188 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2189 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2190 warn about behavior which is undefined under the ANSI spec. Note: This
2191 is specified by ANSI, but it's not very popular with programmers.
2192 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2193 macro in the SBCL sources for an example of a workaround which you
2194 might use to make portable ANSI-standard code which does what you want.
2195 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2196 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2197 when it's used as a non-toplevel form. (This is required in order
2198 to implement the DEFCONSTANT-EQX macro.)
2199 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2200 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2201 funny things, due to bugs in the implementation of EVAL-WHEN
2202 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2203 non-toplevel DEFCONSTANTs are uncommon.)
2204 * The core file version number and fasl file version number have been
2205 incremented, because the old noncompliant DEFCONSTANT behavior involved
2206 calling functions which no longer exist, and because I also took the
2207 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2208 * fixed bug 1 (error handling before read-eval-print loop starts), and
2209 redid debugger restarts and related debugger commands somewhat while
2211 ** The QUIT debugger command is gone, since it did something
2212 rather different than the SB-EXT:QUIT command, and since it never
2213 worked properly outside the main toplevel read/eval/print loop.
2214 Invoking the new TOPLEVEL restart provides the same functionality.
2215 ** The GO debugger command is also gone, since you can just invoke
2216 the CONTINUE restart directly instead.
2217 ** The TOP debugger command is also gone, since it's redundant with the
2218 FRAME 0 command, and since it interfered with abbreviations for the
2220 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2222 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2223 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2224 some porting work by Martin Atzmueller)
2225 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2226 with a boatload of patches which he ported to SBCL. Now that those
2228 ** The system tries to make sure that its low-priority messages
2229 are prefixed by semicolons, to help people who like to use
2230 syntax highlighting in their ILISP buffer. (This patch
2231 was originally due to Raymond Toy.)
2232 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2233 patches originally written by Raymond Toy.
2234 ** The compiler understands coercion between single-value and
2235 multiple-VALUES type expressions better, getting rid of some very
2236 weird behavior, thanks to patches originally by Robert MacLachlan
2237 and Douglas Crosher.
2238 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2239 lambda lists, thanks to a patch originally by Pierre Mai.
2240 ** The system no longer bogusly warns about "abbreviated type
2242 ** The compiler gets less confused by inlining and RETURN-FROM,
2243 thanks to some patches originally by Tim Moore.
2244 ** The system no longer hangs when dumping circular lists to fasl
2245 files, thanks to a patch originally from Douglas Crosher.
2246 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2247 undefined function error.
2248 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2249 in structure slot initforms) for now, documented workaround instead:-|
2250 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2251 * fixed bug 5: added stubs for various Gray stream functions called
2252 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2253 installed, at least appropriate type errors are generated
2254 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2255 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2256 consistently in DEFMETHOD forms.
2257 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2258 it doesn't seem to affect SBCL after all
2259 * The C runtime system now builds with better optimization and many
2260 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2262 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2264 * The system is now under CVS at SourceForge (instead of the
2265 CVS repository on my home machine).
2266 * The new signal handling code has been tweaked to treat register
2267 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2268 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2269 downstream, e.g. in debug-int.lisp, has implicit dependencies
2270 on the unsignedness of integer representation of machine words,
2271 and that caused the system to bomb out with infinite regress
2272 when trying to recover from type errors involving signed values,
2273 e.g. (BUTLAST '(1 2 3) -1).)
2274 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2275 (This was one of the bugs Peter Van Eynde reported back in July.)
2276 * The system now uses code inspired by Colin Walters' O(N)
2277 implementation of MAP (from the cmucl-imp@cons.org mailing
2278 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2279 inline the MAP operation, and there is more than one
2280 sequence argument to the MAP call (so that it can't just
2281 do ETYPECASE once and for all based on the type of the
2282 single sequence argument). (The old non-inline implementation
2283 of the general M-argument sequence-of-length-N case required
2284 O(M*N*N) time when any of the sequence arguments were LISTs.)
2285 * The QUIT :UNIX-CODE keyword argument has been renamed to
2286 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2287 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2289 * Raymond Wiker's patches to port dynamic loading from Linux to
2290 FreeBSD have been added.
2291 * The BUGS file is now more nearly up to date, thanks in large part
2292 to Martin Atzmueller's review of it.
2293 * The debugger now flushes standard output streams before it begins
2294 its output ("debugger invoked" and so forth).
2295 * The core version number and fasl file version number have both
2296 been incremented, because of incompatible changes in the layout
2298 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2299 used to not be called for a saved Lisp image.)
2300 * Martin Atzmueller's version of a patch to fix a compiler crash,
2301 as posted on sbcl-devel 13 September 2000, has been installed.
2302 * Instead of installing Martin Atzmueller's patch for the
2303 compiler transform for SUBSEQ, I deleted the compiler transform,
2304 and transforms for some similar consing operations.
2305 * A bug in signal handling which kept TRACE from working on OpenBSD
2307 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2308 away by constant folding
2309 * The system now defines its address space constants in one place
2310 (in the Lisp sources), and propagates them automatically elsewhere
2311 (through GENESIS and the sbcl.h file). Therefore, patching the
2312 address map is less unnecessarily tedious and error-prone. The
2313 Lisp names of address space constants have also been systematized.
2314 * CVS tags like dollar-Header-dollar have been removed from
2315 the sources, because they have never saved me trouble and
2316 they've been source of trouble working with patches and other
2317 diff-related operations.
2318 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2319 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2321 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2323 * The system has been ported to OpenBSD.
2324 * The system now compiles with a simple "sh make.sh" on the systems
2325 that it's supported on. I.e., now you no longer need to tweak
2326 text in the target-features.lisp-expr and symlinks in src/runtime/
2327 by hand, the make.sh takes care of it for you.
2328 * The system is no longer so grossly inefficient when compiling code
2329 involving vectors implemented as general (not simple) vectors (VECTOR T),
2330 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2331 which dares to use the various sequence functions on non-simple
2332 vectors, takes less of a performance hit.
2333 * There is now a primitive type predicate VECTOR-T-P
2334 to test for the (VECTOR T) type, so that e.g.
2335 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2336 can now be compiled with some semblance of efficiency. (The old code
2337 turned the type declaration into a full call to %TYPEP at runtime!)
2338 * AREF on (VECTOR T) is still not fast, since it's still compiled
2339 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2340 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2341 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2342 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2343 without making full calls to SUBTYPEP at runtime.
2344 (Some analogous efficiency issues for non-simple vectors specialized to
2345 element types other than T, or for non-simple multidimensional arrays,
2346 have not been addressed. They could almost certainly be handled the
2347 same way if anyone is motivated to do so.)
2348 * The changes in array handling break binary compatibility, so
2349 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2350 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2351 instead of (VALUES T T).
2352 * By following the instructions that Dan Barlow posted to sbcl-devel
2353 on 2 July 2000, I was able to enable primitive dynamic object
2354 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2355 functionality is not implemented (since it calls ld to resolve
2356 library references automatically, requiring RUN-PROGRAM for its
2357 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2358 try to resolve library references) is now supported.
2359 * The system now flushes the standard output streams when it terminates,
2360 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2361 them at several other probably-convenient times, e.g. in each pass of
2362 the toplevel read-eval-print loop, and after evaluating a form given
2363 as an "--eval" command-line option. (These changes were motivated by a
2364 discussion of stream flushing issues on cmucl-imp in August 2000.)
2365 * The source transform for TYPEP of array types no longer assumes
2366 that an array whose element type is a not-yet-defined type
2367 is implemented as an array of T, but instead punts, so that the
2368 type will be interpreted at runtime.
2369 * There is now some support for cross-compiling in make.sh: each of
2370 the phases of make.sh has its own script. (This should be transparent
2371 to people doing ordinary, non-cross-compile builds.)
2372 * Since my laptop doesn't have hundreds of megabytes of memory like
2373 my desktop machine, I became more motivated to do some items on
2374 my to-do list in order to reduce the size of the system a little:
2375 ** Arrange for various needed-only-at-cold-init things to be
2376 uninterned after cold init. To support this, those things have
2377 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2378 symbols with such names are now uninterned after cold init).
2379 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2380 fasl files for cold load.
2381 ** Remove the old compiler structure pooling code (which used to
2382 be conditional on the target feature :SB-ALLOC) completely.
2383 ** Redo the representation of some data in cold init to be more compact.
2384 (I also looked into supporting byte compiled code at bootstrap time,
2385 which would probably reduce the size of the system a lot, but that
2386 looked too complicated, so I punted for now.)
2387 * The maximum signal nesting depth in the src/runtime/ support code has
2388 been reduced from 4096 to 256. (I don't know any reason for the very
2389 large old value. If the new smaller value turns out to break something,
2390 I'll probably just bump it back up.)
2391 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2393 * Many, many bugs reported by Peter Van Eynde have been added to
2394 the BUGS list; some have even been fixed.
2395 * While enabling dynamic object file loading, I tried to make the
2396 code easier to understand, renaming various functions and variables
2397 with less ambiguous names, and changing some function calling
2398 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2399 * While trying to figure out how to do the OpenBSD port, I tried to
2400 clean up some of the code in src/runtime/. In particular, I dropped
2401 support for non-POSIX signal handling, added various comments,
2402 tweaked the code to reduce the number of compilation warnings, and
2403 renamed some files to increase consistency.
2404 * To support the new automatic configuration functionality in make.sh,
2405 the source file target-features.lisp-expr has been replaced with the
2406 source file base-target-features.lisp-expr and the machine-generated
2407 file local-target-features.lisp-expr.
2408 * fixed a stupid quoting error in make.sh so that using CMU CL
2409 "lisp -batch" as cross-compilation host works again
2411 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2413 * DESCRIBE no longer tries to call itself recursively to describe
2414 bound/fbound values, so that it no longer fails on symbols which are
2415 bound to themselves (like keywords, T, and NIL).
2416 * DESCRIBE now works on generic functions.
2417 * The printer now prints less-screwed-up representations of closures
2418 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2419 * A private symbol is used instead of the :EMPTY keyword previously
2420 used to mark empty slots in hash tables. Thus
2421 (DEFVAR *HT* (MAKE-HASH-TABLE))
2422 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2423 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2424 now does what ANSI says that it should. (You can still get
2425 similar noncompliant behavior if bang on the hash table
2426 implementation with all the symbols you get back from
2427 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2428 This breaks binary compatibility, since tests for equality to
2429 :EMPTY are wired into things like the macroexpansion of
2430 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2432 * There's now a minimal placeholder implementation for CL:STEP,
2433 as required by ANSI.
2434 * An obscure bug in the interaction of the normal compiler, the byte
2435 compiler, inlining, and structure predicates has been patched
2436 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2437 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2439 * Missing ordinary arguments in a macro call are now detected even
2440 when the macro lambda list contains &KEY or &REST.
2441 * The debugger no longer complains about encountering the top of the
2442 stack when you type "FRAME 0" to explicitly instruct it to go to
2443 the top of the stack. And it now prints the frame you request even
2444 if it's the current frame (instead of saying "You are here.").
2445 * As specified by ANSI, the system now always prints keywords
2446 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2448 * The default initial SIZE of HASH-TABLEs is now smaller.
2449 * Type information from CLOS class dispatch is now propagated
2450 into DEFMETHOD bodies, so that e.g.
2451 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2453 is now basically equivalent to
2454 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2455 (DECLARE (TYPE SINGLE-FLOAT X))
2457 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2458 operation, without having to do run-time type dispatch.
2459 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2460 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2461 * Since I don't seem to be making big file reorganizations very often
2462 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2463 is overflowing my ability to conveniently back them up), I've finally
2464 checked the system into CVS. (The CVS repository is on my home system,
2465 not at SourceForge -- putting it on SourceForge might come later.)
2466 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2467 high-level GC-NOTIFY-FOO functions send their output. (There's
2468 still very little control of where low-level verbose GC functions
2469 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2470 less than it used to -- the GC-NOTIFY-FOO functions are now under
2471 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2472 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2473 in only one place in the source code, and propagates it automatically
2474 everywhere that it's needed. Thus e.g. when I bump the version from
2475 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2476 * The C source files now include boilerplate legalese and documentation
2477 at the head of each file (just as the Lisp source files already did).
2478 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2479 to his page will be replaced with a link to his new CLiki service.
2481 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2483 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2484 * The build process now looks for GNU make under the default name "gmake",
2485 instead of "make" as it used to. If GNU make is not available as "gmake"
2486 on your system, you can change this default behavior by setting the
2487 GNUMAKE environment variable.
2488 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2489 can build without error under CMU CL.
2491 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2493 * There is now a partial SBCL user manual (with some new text and some
2494 text cribbed from the CMU CL manual).
2495 * The beginnings of a profiler have been added (starting with the
2496 CMU CL profiler and simplifying and cleaning up). Eventually the
2497 main interface should be through the TRACE macro, but for now,
2498 it's still accessed through vaguely CMU-CL-style functions and macros
2499 exported from the package SB-PROFILE.
2500 * Some problems left over from porting CMU CL to the new
2501 cross-compilation bootstrap process have been cleaned up:
2502 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2503 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2504 part when getting rid of the compiler *BACKEND* stuff.)
2505 ** The value of *NULL-TYPE* was screwed up, because it was
2506 being initialized before the type system knew the final
2507 definition of the 'NULL type. This screwed up several key
2508 optimizations in the compiler, causing inefficiency in all sorts
2509 of places. (I found it because I wanted to understand why
2510 GET-INTERNAL-RUN-TIME was consing.)
2511 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2512 PROCLAIM FTYPE information. Unfortunately this broke binary
2513 compatibility again, since now the forms output by DEFGENERIC
2514 to refer to functions which didn't exist in 0.6.3.
2515 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2516 can use the new (as of 0.6.3) transform for SOME into MAP into
2518 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2519 values to (INTEGER 0 1000000), so that the time code will no longer
2520 occasionally get blown up by Linux returning 1000000 microseconds
2521 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2522 its output conform to the ANSI spec. (Alas, this makes its output
2523 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2524 * A full call to MAP NIL with a single sequence argument no longer conses.
2525 * fixes to problems pointed out by Martin Atzmueller:
2526 * The manual page no longer talks about multiprocessing as though
2527 it were currently supported.
2528 * The ILISP support patches have been removed from the distribution,
2529 because as of version 5.10.1, ILISP now supports SBCL without us
2530 having to maintain patches.
2531 * added a modified version of Raymond Toy's recent CMU CL patch for
2532 EQUALP comparison of HASH-TABLE
2534 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2536 * The system still can't cross-compile itself with
2537 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2538 entails), but at least it can compile more of itself that way
2539 than it used to be able to, and various buglets which were uncovered
2540 by trying to cross-compile itself that way have now been fixed.
2541 * This release breaks binary compatibility again. This time
2542 at least I've incremented the FASL file format version to 2, so that the
2543 problem can be detected reliably instead of just causing weird errors.
2544 * various new style warnings:
2545 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2546 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2547 ** using the lexical binding of a variable named in the *FOO* style
2548 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2549 as specified by ANSI.
2550 * *RANDOM-STATE* is no longer automatically initialized from
2551 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2552 default behavior of the system is to repeat its behavior every time
2553 it's run. If you'd like to change this behavior, you can always
2554 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2555 old convention there was no comparably easy way to get the system to
2556 repeat its behavior every time it was run.
2557 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2558 FTYPE declarations has been removed, in favor of their ANSI
2559 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2560 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2561 the types of their sequence arguments can be determined at compile time.
2562 This is done through a new open code expansion for MAP which eliminates
2563 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2564 argument types can be determined at compile time.
2565 * The optimizer now transforms COERCE into an identity operation when it
2566 can prove that the coerced object is already of the correct type. (This
2567 can be a win for machine generated code, including the output of other
2568 optimization transforms, such as the MAP transform above.)
2569 * Credit information has been moved from source file headers into CREDITS.
2570 * Source file headers have been made more standard.
2571 * The CASE macro now compiles without complaining even when it has
2574 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2576 * (Note that the way that the PCL macroexpansions were rewritten
2577 to accommodate the change in DEFGENERIC below breaks binary
2578 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2579 not run under sbcl-0.6.2. Once we get out of alpha releases,
2580 i.e. hit release 1.0.0, we'll probably try to maintain binary
2581 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2582 and sbcl-1.4.4. Until then, however, it might be fairly common
2583 for maintenance releases to break binary compatibility.)
2584 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2586 * The reporting of unhandled signals has been changed to print some
2587 explanatory text as well as the report form. (Previously only
2588 the report form was printed.)
2589 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2590 it defines, so that the compiler no longer issues undefined function
2591 warnings for compiled-but-not-yet-loaded generic functions.
2592 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2593 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2594 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2595 section of the ANSI spec are supported.
2596 * The "" nickname for the "KEYWORD" package has been removed.
2597 The reader still handles symbol tokens which begin with a package marker
2598 as keywords, but it doesn't expose its mechanism for doing so in the
2599 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2600 * The system now issues STYLE-WARNINGs for contradictory TYPE
2601 proclamations. (Warnings for contradictory FTYPE proclamations would
2602 be nice too, but those can't be done usefully unless the type system
2603 is made smarter about FUNCTION types.)
2604 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2605 systematized, so that "*target-*.lisp is supposed to exist only on the
2606 target and imply that there's a related file which exists on the
2607 host, and *host-*.lisp is supposed to exist only on the host and imply
2608 that there's a related file which exists on the target. This involves a
2609 lot of renaming. Hopefully the acute confusion caused by the renaming
2610 will be justified by the reduction in chronic confusion..
2611 ** runtime-type.lisp -> early-target-type.lisp
2612 ** target-type.lisp -> late-target-type.lisp
2613 ** early-host-format.lisp -> early-format.lisp
2614 ** late-host-format.lisp -> late-format.lisp
2615 ** host-error.lisp -> misc-error.lisp
2616 ** early-error.lisp -> early-target-error.lisp
2617 ** late-error.lisp -> late-target-error.lisp
2618 ** host-defboot.lisp -> early-defboot.lisp
2619 ** code/misc.lisp -> code/target-misc.lisp
2620 ** code/host-misc.lisp -> code/misc.lisp
2621 ** code/numbers.lisp -> code/target-numbers.lisp
2622 ** code/early-numbers.lisp -> numbers.lisp
2623 ** early-host-type.lisp -> early-type.lisp
2624 ** late-host-type.lisp -> late-type.lisp
2625 ** host-typep.lisp -> typep.lisp
2626 ** load.lisp -> target-load.lisp
2627 ** host-load.lisp -> load.lisp
2628 ** host-disassem.lisp -> disassem.lisp
2629 ** host-insts.lisp -> insts.lisp
2630 ** byte-comp.lisp -> target-byte-comp.lisp
2631 ** host-byte-comp.lisp -> byte-comp.lisp
2632 ** host-signal.lisp -> signal.lisp
2633 ** host-defstruct.lisp -> defstruct.lisp
2634 ** late-target-type.lisp -> deftypes-for-target.lisp
2635 Furthermore, several other previously target-only files foo.lisp (e.g.
2636 hash-table.lisp and random.lisp) have been split into a target-and-host
2637 foo.lisp file and a target-only target-foo.lisp file, with their key type
2638 definitions in the target-and-host part, so that the cross-compiler will
2639 know more about target types.
2640 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2641 gone away. In their place are various *BACKEND-FOO* variables
2642 corresponding to the slots of the old structure.
2643 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2644 into the target SBCL has been fixed.
2645 * The chill.lisp system for loading cold code into a running SBCL
2647 * Support for the CMU CL "scavenger hook" extension has been removed.
2648 (It was undocumented and unused in the CMU CL sources that SBCL was
2649 derived from, and stale in sbcl-0.6.1.)
2650 * Various errors in the cross-compiler type system were detected
2651 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2652 (enabling various consistency checks). Many of them were fixed,
2653 but some hard problems remain, so the compiler is back to
2654 running without *TYPE-SYSTEM-INITIALIZED* for now.
2655 * As part of the cross-compiler type system cleanup, I implemented
2656 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2657 * I have started adding UNCROSS calls throughout the type system
2658 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2659 on cross-compiler input files will be able to go away, and various
2661 * CONSTANTP now returns true for quoted forms (as explicitly required
2664 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2666 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2667 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2668 (MAKE-LIST -1). In the longer term, it ought to become true
2669 that declarations are assertions even at SAFETY 1. For now, it's not
2670 quite true even at SAFETY 3, but it's at least more nearly true..
2671 (Note that this change seems to increases the size of the system by
2672 O(5%) and to decrease the speed of the compiler by 20% or more.)
2673 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2674 for the problem of printing dozens of lines of distracting information
2675 about low-level system machinery as part of the top stack frame
2676 on entry to the debugger when an undefined function was called.
2677 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2678 is not reset to COMMON-LISP-USER.
2679 * Compilation of stuff related to dyncount.lisp has been made conditional
2680 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2681 smaller. The various dyncount-related symbols have been moved into
2682 a new "SB-DYNCOUNT" package.
2683 * tty-inspect.lisp has been renamed to inspect.lisp.
2684 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2685 feature has gone away. (When we eventually port to other flavors of
2686 libc and/or Unix, we'll try to make the differences between flavors
2687 invisible at the user level.)
2688 * Various other *FEATURES* tags, and/or their associated conditionals,
2689 have been removed if obsolescent, or given better documentation, or
2690 sometimes given more-mnemonic names.
2692 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2694 * tidied up "make.sh" script
2695 * tidied up system directory structure
2696 * better "clean.sh" behavior
2697 * added doc/FOR-CMUCL-DEVELOPERS
2698 * many many small tweaks to output format, e.g. removing possibly-confusing
2699 trailing #\. character in DESCRIBE-INSTANCE
2700 * (EQUALP #\A 'A) no longer signals an error.
2701 * new hashing code, including EQUALP hashing
2702 * tidied up Lisp initialization and toplevel
2703 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2704 * command line argument processing
2705 * added POSIX-GETENV function to deal with Unix-ish environment variables
2706 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2707 terminating SBCL on EOF
2708 * non-verbose GC by default
2709 * There is no more "sbcl" shell script; the sbcl file is now the C
2710 runtime executable (just like CMU CL).
2711 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2713 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2714 symbol and package information as Lisp native symbol and package objects
2715 instead of strings naming symbols and strings naming packages. This way,
2716 whenever packages are renamed (as in warm init), debug information is
2717 transformed along with everything else.
2718 * tweaked the optimization policy declarations which control the building
2719 of SBCL itself. Now, among other things, the system no longer saves
2720 source location debugging information. (This helps two problems at once
2721 by reducing SBCL size and by keeping SBCL from trying to look for its
2722 sources -- which may not exist -- when reporting errors.)
2723 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2724 debugging and testing purposes
2725 * cleaned up printing, making the printer call PRINT-OBJECT for
2726 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2727 methods, giving nearly-ANSI behavior
2728 * converted almost all special variables to use *FOO* naming convention
2729 * deleted PARSE-TIME functionality, since it can be done portably
2730 * moved some files out of cold init into warm init
2731 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2733 * regularized formatting of source files
2734 * added an install.sh script
2735 * fixed ridiculous memory usage of cross-compiler by making
2736 compiler/alloc.lisp not try to do pooling unless it can hook
2737 itself into the GC of the cross-compilation host. Now the system
2738 builds nicely on my old laptop.
2739 * added :SB-ALLOC in target-features.lisp-expr
2740 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2741 was not implemented)
2742 * re-did condition handling and note reporting in the compiler. Notes
2743 are no longer handled by signalling conditions. Style warnings
2744 and warnings are handled more correctly and reported in such a way
2745 that it's easy to find one or the other in your output (so that you
2746 can e.g. figure out which of many problems caused COMPILE-FILE to
2748 * changed the severity of several compiler warnings from full WARNING
2749 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2750 compiler note reporting so that it doesn't use the condition system
2751 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2752 COMPILE-FILE command)
2753 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2754 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2755 no longer supported, since their implementation was deeply intertwingled
2756 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2757 * removed bogus "support" for compiler macros named (SETF FOO), and
2758 removed the compiler macro for SETF INFO (but only after making a fool
2759 of myself on the cmucl-imp mailing list by posting a bogus patch for
2760 DEFINE-COMPILER-MACRO..)
2761 * Compiled files containing forms which have side effects on the Lisp
2762 reader (such as DEFPACKAGE forms) are now handled more correctly.
2763 (Compiler queuing of top level lambdas has been suppressed by setting
2764 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2765 * deleted various currently-unused source files, e.g. gengc.lisp. They
2766 may be added back at some point e.g. when porting to other architectures,
2767 but until they are it's distracting to distribute them and to try to
2769 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2770 there were so many of them they're just distractions, and UNCROSS is
2771 known to be able to handle the current sources
2772 * moved PROFILE functionality into TRACE, so that it will be clear
2773 how the wrapping and unwrapping of functions when you profile them
2774 interacts with the wrapping and unwrapping of functions when you
2775 trace them. (Actually, the functionality isn't there yet, but at least
2776 the interface specification is there. Hopefully, the functionality will
2777 arrive with some future maintenance release.)
2778 * removed host-oops.lisp
2779 * changed signature of QUIT function to allow UNIX-CODE argument
2780 * fixed READ-SEQUENCE bug
2781 * tweaked verbose GC output so that it looks more like the progress
2782 output that ANSI specifies for functions like LOAD
2783 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2784 * added <http://sbcl.sourceforge.com> to the banner information printed by