1 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2 * new feature: a build-time option (controlled by the :SB-UNICODE
3 keyword feature, enabled by default) for building the system with
4 support for the entire 21-bit character space defined by the
6 * new feature: the system now has rudimentary external-format
7 support; the primary user-visible change at this time is that
8 characters with the high bit set (such as lower-case-e-acute) will
9 print correctly to a terminal in a UTF-8 environment.
10 * minor incompatible change: BASE-CHAR no longer names a class;
11 however, CHARACTER continues to do so, as required by ANSI.
12 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
13 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
14 has been moved to the SB-EXT package (temporarily re-exported from
16 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
17 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
19 * fixed bug #351: better error-handlind and reporting for malformed
21 * fixed bug #350: bignum-printing is now more memory-efficient,
22 allowing printing of very large bignums, eg. (expt 2 10000000).
23 (reported by Bruno Haible)
24 * fixed bug #302: better primitive-type selection for intersection
26 * fixed bug #308: non-graphic characters now all have names, as
27 required. (reported by Bruno Haible)
28 * bug fix: redefining a class with different superclasses now correctly
29 removes it from the direct-subclasses of its previous superclasses.
30 (reported by David Morse)
31 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
32 new value now works. (reported by Bruno Haible)
33 * bug fix: correct canonicalization of multiple non-standard slot
34 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
35 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
36 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
37 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
38 now exists, an signals an error.
39 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
40 during macroexpansion for non-iteration variables as well. (reported
41 by Bruno Haible for CMUCL)
42 * bug fix: Cyclic structures and unprintable objects in compiler
43 messages no longer cause errors. (reported by Bruno Haible)
44 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
45 and READ-FROM-STRING all now return a primary value of NIL if
46 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
47 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
48 (reported by Bruno Haible for CMUCL)
49 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
50 do not designate a condition. (reported by Bruno Haible for
52 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
53 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
54 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
55 permitting resolution in favour of any of the conflicting symbols.
56 (reported by Bruno Haible for CMUCL)
57 * bug fix: EQUAL compiler optimizations is less aggressive on
58 strings which can potentially compare true despite having distinct
59 specialized array element types.
60 * bug fix: unit enumerations can be defined without dividing by
61 zero. (thanks to Vincent Arkesteijn)
62 * FORMAT compile-time argument count checking has been enhanced.
63 (report from Bruno Haible for CMUCL)
64 * a partial workaround for the bug 262: the compiler does not try to
65 inline-expand a local function doing RETURN-FROM from a deleted
66 BLOCK. (thanks to Peter Denno for the bug report and to David
67 Wragg for the simple test case)
68 * fixed some bugs revealed by Paul Dietz' test suite:
69 ** RENAME-PACKAGE allows all package designators as new package
71 ** constraint propagation and lambda variable substitution are
72 more cautious in dealing with partially deleted code.
73 ** compiler sometimes lost reoptimization passes.
74 ** CERROR, given a condition as condition designator, passes its
75 remaining arguments to the continue format control without
77 ** Case-altering FORMAT directives work correctly on non-ASCII
79 ** The REAL type specifier handles bounds outside the floating
80 point ranges without signalling FLOATING-POINT-OVERFLOW.
81 ** Functions with IR1-transformations can create intercomponent
82 references to global functions.
83 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
85 ** FORMAT ~R treats a nil value for its first parameter correctly.
87 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
88 * enhancement: saving cores with foreign code loaded is now
89 supported on x86/NetBSD and sparc/Linux in addition to the previously
91 * bug fix: on some platforms repeated installations caused multiple
92 copies of HTML documentation to be installed -- should not happen
93 any more. (reported by Stefan Scholl)
94 * bug fix: parsing self-recursive alien record types multiple times
95 no longer causes infinite recursion. (reported by Thomas F. Burdick,
96 original patch by Helmut Eller for CMUCL)
97 * bug fix: stack-exhaustion detection works now on NetBSD as well.
98 (thanks to Richard Kreuter)
99 * bug fix: defining classes whose accessors are methods on existing
100 generic functions in other (locked) packages no longer signals
101 bogus package lock violations. (reported by François-René Rideau)
102 * bug fix: special variables as DEFMETHOD parameters no longer have
103 associated bogus type declarations. (reported by David Wragg and
105 * bug fix: read-write consistency on streams of element-type
106 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
107 * bug fix: redefiniton of the only method of a generic function with
108 no DEFGENERIC no longer emits a full WARNING. In addition,
109 redefinition of generic functions with no DEFGENERIC to an
110 incompatible lambda list now signals an error. (thanks to Zach
112 * bug fix: DEFGENERIC now works even when there's a function of the
113 same name in an enclosing lexical environment. (thanks to Zach
115 * fixed compiler failure, caused by instrumenting code during
116 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
117 * optimization: added loop analysis and improved register allocation
118 * fixed some bugs revealed by Paul Dietz' test suite:
119 ** POSITION on displaced vectors with non-zero displacement
120 returns the right answer.
121 ** (SIMPLE-STRING) is a valid type specifier for sequence
123 ** *PRINT-LEVEL* handling for slotless structures is pedantically
125 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
126 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
128 ** The justification version of the FORMAT ~< directive treats
129 non-zero minpad parameter correctly.
131 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
132 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
133 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
134 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
135 the supported interface.
136 * new feature: Single-stepping of code compiled with DEBUG 2 or
137 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
138 * new feature: saving cores with foreign code loaded is now
139 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
140 Timothy Moore's work for CMUCL)
141 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
142 parameters to * instead of NIL if no initform is supplied.
143 (reported by Johan Bockgaard)
144 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
145 correctly when FUNCALL appears as the car of the form. Note:
146 despite this FUNCALL forms are not currently subject to
147 compiler-macro expansion. (port of Raymond Toy's fix for the
148 same from CMUCL, reported by Johan Bockgaard)
149 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
150 (thanks for Teemu Kalvas)
151 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
152 case instead of terminating the iteration (thanks for Julian
153 Squires, Sean Champ and Raymond Toy)
154 * bug fix: incorrect expansion of defgeneric that caused a style
155 warning. (thanks for Zach Beane)
156 * bug fix: slot accessor effective method computation works properly
157 for classes with multiple non-standard applicable methods on
158 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
159 * on x86 compiler supports stack allocation of results of LIST and
160 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
161 CMUCL implementation by Gerd Moellmann)
162 * fixed some bugs revealed by Paul Dietz' test suite:
163 ** FORMAT strings with both the ~<~:;~> form of the justification
164 directive and pretty-printing directives cause an error.
166 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
167 * incompatible change: the internal functions
168 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
169 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
170 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
171 instead of the old functions.
172 * new feature: on platforms where "dladdr" is available foreign
173 function names now appear in backtraces. (based on Helmut Eller's
175 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
176 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
178 * documentation: ASDF manual is now distributed with SBCL.
179 * bug fix: SBCL can now load its contributed modules with REQUIRE
180 even if the system-provided entries have been removed from
181 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
183 * bug fix: backtraces involving undefined functions or assembly
184 routines are more informative. (thanks to Brian Downing)
185 * bug fix: mutually referent alien structures now work correctly.
186 (reported by Rick Taube)
187 * bug fix: structures defined by WITH-ALIEN can be referred to
188 within other definitions in the same WITH-ALIEN.
189 * bug fix: division operators (MOD, TRUNCATE and the like) with
190 constant zero divisors and integer dividends no longer generate
192 * bug fix: provide default methods for INPUT-STREAM-P and
193 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
194 * bug fix: improve the detection and resolution of MOP metacycles
195 (where effective method computation depends on the generic function
196 having its effective method computed). (reported by Bruno Haible)
197 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
198 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
199 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
200 * bug fix: pathnames with relative directory components can be
201 represented relative to default pathnames.
202 * optimization: in taking the GCD of bignums, reduce the two bignums
203 to approximately the same size (using Euclid's algorithm) before
204 applying the more sophisticated binary GCD. (thanks to Juho
206 * optimization: COUNT on bitvectors now operates word-at-a-time.
207 * optimization: ASH with a positive, but not necessarily constant,
208 (leftwards) shift, when in a modular context, is compiled to a
210 * fixed some bugs revealed by Paul Dietz' test suite:
211 ** FORMAT variable parameters ("~V<char>") are defaulted properly
212 if the corresponding argument is NIL.
213 ** FORMAT directives accepting a minpad parameter treat negative
215 ** literal commas as character parameters to FORMAT directives are
217 ** literal spaces directly after ~<Newline> directives within a
218 format-logical-block (~:< ~@:>) do not induce :FILL-style
219 conditional newlines.
220 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
221 as specified: it no longer includes conditional newlines).
222 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
224 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
225 :PER-LINE-PREFIX argument does not evaluate to a string.
227 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
228 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
229 the manual for details; for now, package locks can be disabled by
230 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
231 if no major problems are found then it is likely that they will be
232 compiled in unconditionally.
233 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
234 unsupported operators on all platforms. To load a shared library
235 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
236 object file, link it into a shared library outside of SBCL and
237 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
239 * bug workaround: ROOM T and the GC were not working together
240 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
241 reporting facility aren't preserved. That reporting has been
242 disabled, so now until and unless someone figures out how to make it
243 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
244 * minor incompatible change: as threatened around sbcl-0.8.0, the
245 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
246 The symbols are also exported from SB-PCL for backwards
247 compatibility, but more so than before SB-PCL should be treated as
248 an implementation-internal package.
249 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
251 * the SB-SPROF contrib now works on (most) non-x86 architectures.
252 It is known as of this release not to work on the Alpha, however.
253 * fixed bug #167: errors signalled due to illegal syntax in method
254 bodies are now more legible.
255 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
256 designators and can hence be used with TYPEP.
257 * fixed bug #333: CHECK-TYPE now ensures that the type error
258 signalled, if any, has the right object to be accessed by
259 TYPE-ERROR-DATUM. (reported by Tony Martinez)
260 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
261 5.1.2.3 for multiple-value place subforms. (reported by Kalle
263 * fixed bug #334: programmatic addition of slots using specialized
264 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
265 and :ALLOCATION :CLASS slots.
266 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
267 float exponents. (rereported by Peter Seibel)
268 * fixed a bug: #\Space (and other whitespace characters) are no
269 longer considered to be macro characters in standard syntax by
271 * fixed bug: initialization of condition class metaobjects no longer
272 causes an instance of the condition to be created. (reported by
274 * fixed bug: it is now possible to have slots such that
275 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
276 description is neither :INSTANCE nor :CLASS.
277 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
278 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
279 SLOT-VALUE-USING-CLASS), and no longer causes errors with
280 non-standard SLOT-DEFINITION-ALLOCATION values.
281 * fixed bugs: various race conditions handling exiting threads.
282 CL-PPCRE's thread torture test now passes.
283 * fixed arguably-a-bug: GC time proportional to number of threads in
284 system even when most of them are idle
285 * optimization: improved performance of BIT and SBIT on bit-vectors.
286 * .fasl file incompatibility: The fasl file version number has
287 been incremented because of changes associated with package locks.
289 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
290 * minor incompatible change: the system no longer provides
291 optimization hints (conditions which are TYPEP
292 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
293 mode; these hints are still emitted when higher SPEED optimization
295 * new contrib module: a sampling profiler (profiling by statistical
296 sampling, rather than by instrumenting functions) is available as
297 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
298 * the behaviour of the standard function ED is now customizeable by
299 third parties through a hook variable: see ED's documentation
300 string for information on the protocol.
301 * the compiler no longer emits efficiency notes for (FUNCALL X)
302 when the type of X is uncertain under default optimization
304 * fixed bug 276: mutating a binding of a specialized parameter to a
305 method to something that is not TYPEP the specializer is now
307 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
308 exhibit the required behaviour on the x86 platform. (thanks to
309 Peter van Eynde, Eric Marsden and Bruno Haible)
310 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
311 even for difficult arguments. (reported by Peter Graves)
312 * fixed bug 141a: the backquote printer now descends quoted
314 * fixed another bug in backquote printing: no more destructive
315 modification of the form's list structure. (reported by Brian
317 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
318 move between its address being taken and the call to
319 interrupt_thread, fixing a crashing race condition.
320 * the SB-POSIX contrib implementation has been adjusted so that it
321 no longer exhibits ridiculously poor performance when constructing
322 instances corresponding to C structs.
324 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
325 * minor incompatible change: the sb-grovel contrib now treats C
326 structures as alien (in the sense of SB-ALIEN) objects rather than
327 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
328 has implications for memory management of client code
329 (sb-grovel-returned objects must now be manually managed) and for
330 type safety (alien objects now have full types).
331 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
332 used to control emission of compiler diagnostics, rather than the
333 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
334 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
335 quality should be considered deprecated.
336 * install.sh now installs the user manual as well
337 * (not quite a new documentable feature, but worth considering in
338 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
339 beginnings of a semantically meaningful condition hierarchy is
340 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
342 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
343 designator as the defaults argument.
344 * fixed bug: Displaced arrays whose displaced-to array has become
345 too small now cause ARRAY-DIMENSION to signal an error, providing
346 for safer bounds-checking. (reported by Bruno Haible)
347 * fixed bug: DEFCLASS slot definitions with identical :READER and
348 :WRITER names now signal a reasonable error. (reported by Thomas
350 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
352 * fixed bug: Invalid slot specification errors now print correctly.
353 (thanks to Zach Beane)
354 * fixed bug 320: Shared to local slot value transfers in class
355 redefinitions now happen corrently from superclasses as
356 well. (reported by Bruno Haible)
357 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
359 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
360 improper lists correctly. (reported by Bruno Haible)
361 * fixed bug 313: source-transform for <fun-name> was erroneously
362 applied to a call of a value of a variable with name <fun-name>.
363 (reported by Antonio Menezes Leitao)
364 * fixed bug 307: The obsolete instance protocol ensures that
365 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
366 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
367 remove dead unknown-values globs from the middle of the stack.
368 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
369 easier to resume long computations after using *BREAK-ON-SIGNALS*
370 to diagnose and fix failures. (thanks to Nikodemus Siivola)
371 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
372 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
373 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
374 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
375 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
377 * fixed a (fairly theoretical) bug in string printing: if
378 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
379 does not have array-element-type equal to the most general string
381 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
382 function name. (thanks to Nikodemus Siivola)
383 * fixed bug: compile-time format string checker failed on
384 non-closed ~{. (reported by Thomas F Burdick)
385 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
386 don't warn on legal loop code involving more than one aggregate
387 boolean. (thanks to Nikodemus Siivola)
388 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
389 in keyword position to MAKE-PACKAGE was not being reported as
391 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
392 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
394 * fixed bugs 280 and 312: the checking for multiple definitions in a
395 file is less likely to become confused by uses of inline
397 * fixed bug: the #S reader macro performs the keyword coercion
398 specified for slot names. (reported by Kalle Niemitalo)
399 * fixed bug: lambda lists may contain symbols whose names start with
400 & but are not lambda-list-keywords; their occurrence triggers a
402 * fixed bug 321: define-method-combination argument lambda lists do
403 not cause invalid code to be generated when &optional or &aux
404 variables have default values. (reported by Bruno Haible)
405 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
406 have CLOS classes; this appears also to have fixed reported
407 instability in stack exhaustion detection.
408 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
409 MUFFLE-WARNING when no associated restart is present is now
411 * optimization: rearranged the expansion of various defining macros
412 so that each expands into only one top-level form in a
413 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
415 * optimization: used a previously little-used slot in symbols to
416 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
418 * fixed some bugs revealed by Paul Dietz' test suite:
419 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
421 ** Fixed an optimization bug related to inheritance of initforms
422 from local to shared slots.
423 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
424 ** CAST optimizer forgot to flush argument derived type.
425 ** print/read consistency on floats is now orders of magnitude
426 more likely. (thanks also to Bruno Haible for a similar report
428 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
429 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
430 for objects of type REAL. Make it so.
431 ** SXHASH is specified (infelicitously) to respect similarity,
432 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
433 it so. (thanks to Markus Ziegler)
434 ** on the Alpha, the compiler succeeds in compiling functions
435 returning a known number of arguments greater than 63.
436 ** fixed handling of invalid NIL arguments in keyword position
438 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
440 ** provided a readably-printable representation for RANDOM-STATE
442 ** ensured that pathnames, where they have a namestring, always
443 print using #P"..." syntax.
445 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
446 * Support for the forthcoming 2.0 version of the NetBSD kernel
447 running on x86 hardware has been added. (thanks to Perry
448 E. Metzger most immediately, and others for their past work)
449 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
450 current release version; SBCL's OpenBSD support had been broken
451 since about the time of OpenBSD's switch to ELF binary format.)
452 * [placeholder for DX summary]
453 ** user code with &REST lists declared dynamic-extent, under high
454 speed or space and low safety and debug optimization policy.
455 * The manual has been converted to Texinfo format and the debugger
456 chapter from the cmucl manual has been added.
457 * A facility has been added to extract documentation strings from
458 sbcl and store them as Texinfo-formatted snippets for inclusion in
459 the manual (via Texinfo's @include directive)
460 * bug fix: compiler emitted division in optimized DEREF. (thanks for
461 the test case to Dave Roberts)
462 * bug fix: multidimensional simple arrays loaded from FASLs had fill
463 pointers. (reported by Sean Ross)
464 * bug fix: PROFILE output is printed nicely even for large numerical
465 values. (thanks to Zach Beane)
466 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
467 greater than 32 handle EOF correctly.
468 * bug fix: on X86 an immediate argument of the IMUL instruction is
469 correctly printed in disassembly. (thanks to Lutz Euler)
470 * bug fix: class slots in redefined classes preserve their old
471 values. (thanks to Bruno Haible and Nikodemus Siivola)
472 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
473 succeeds. (reported by Marco Baringer)
474 * fixed some bugs revealed by Paul Dietz' test suite:
475 ** READ-SEQUENCE now works on ECHO-STREAMs.
476 ** RATIONALIZE works more according to its specification. (thanks
479 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
480 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
481 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
482 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
483 should matter to you only if you rebind the printer control
484 variables and then find you want different bindings in the
485 debugger than in the ordinary execution of your program.)
486 * The runtime build system has been tweaked to support building
487 (on SPARC/SunOS) using a C compiler which invokes Sun's own
488 assembler and linker. (thanks to Nikodemus Siivola)
489 * Unbound, undefined, undeclared variables now trigger full
490 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
491 more often programmer error than deliberate exploitation of undefined
493 * optimization: the hash algorithm for strings has changed to one
494 that is less vulnerable to spurious collisions. (thanks to Juho
496 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
497 needless bounds checking. (thanks to Juho Snellman)
498 * optimization: implemented multiplication as a modular
499 (UNSIGNED-BYTE 32) operation on the PPC backend.
500 * fixed some bugs revealed by Paul Dietz' test suite:
501 ** ADJUST-ARRAY now copies the datum in a zero rank array if
503 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
504 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
505 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
507 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
509 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
511 ** ECHO-STREAMs no longer attempt to echo the end of file value to
512 their output stream on EOF from read.
513 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
514 have been read to end-of-file.
515 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
517 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
518 description of determination of which consecutive characters
520 ** Printing the "Space" character with escaping on now yields "#\\ ",
521 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
522 ** Reading floating-point numbers with *READ-BASE* set to a number
523 less than 10 works correctly.
524 ** Reading floating-point numbers with *READ-BASE* set to a number
525 more than 10 works correctly.
526 ** Printing with *PRINT-READABLY* targets the standard readtable, not
527 the readtable currently in effect.
529 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
530 * minor incompatible change: parsing of namestrings on a physical
531 (Unix) host has changed; numbers after the final #\. in a
532 namestring are no longer interpreted as a version field. This is
533 intented to be largely invisible to the user, except that the
534 meaning of the namestring "*.*.*" has changed: it now refers to a
535 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
536 should usually be replaced by
537 (make-pathname :name :wild :type :wild :version :wild)
538 with the added benefit that this is more likely to be portable.
539 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
540 option to OPEN now signals an error if the file being opened
541 exists; this may have an impact on existing code.
542 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
543 signals received, on the PowerPC platforms (both Linux and
544 Darwin). (thanks to Pierre Mai for pointing out the location of
546 * several fixes on OS X: The system now builds and runs cleanly on
547 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
548 (thanks to Brian Mastenbrook)
549 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
550 second-resolution: integer multiples of 1/3600 between -24 and 24.
551 (thanks to Vincent Arkesteijn)
552 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
553 type when called with 1 argument; PEEK-CHAR checked type of
554 PEEK-TYPE only after having read first character from a
555 stream. (reported by Peter Graves)
556 * bug fix: the garbage collector now has much better locality
557 behaviour, and in particular no longer treats all memory as being
558 exhausted when many small objects point to each other in a deeply
560 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
561 recognized as being TYPEP their class.
562 * bug fix: the PUSHNEW documentation string has been corrected.
563 (thanks to Vincent Arkesteijn)
564 * bug fix: defaulting of the value for the last of an atypically
565 large number of multiple values being bound was not being
566 performed correctly on the Alpha or PPC platforms
567 * optimization: implemented multiplication as a modular
568 (UNSIGNED-BYTE 32) operation on the x86 backend.
569 * optimization: SEARCH on simple-base-strings can now be open-coded.
570 (see also contrib/compiler-extras.lisp for inspiration for
571 teaching the compiler about the Boyer-Moore algorithm).
572 * value, returned by MAX (and MIN) called with several EQUALP, but
573 not EQL, arguments now does not depend on compiler settings.
574 * fixed some bugs revealed by Paul Dietz' test suite:
575 ** in stack analysis liveness information is propagated from
576 non-local entry points.
577 ** pathwise CAST removing failed when the CAST node did not start
579 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
580 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
582 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
583 host is already defined.
584 ** RENAME-FILE works on streams instead of signalling an internal
586 ** PEEK-CHAR uses the current readtable when determining whether
587 or not a character is whitespace.
588 ** MERGE-PATHNAMES handles the case when the pathname does not
589 specify a name while the default-pathname specifies a version.
590 ** Pathnames now stand a better chance of respecting print/read
592 ** Attempting to use standardized file system operators with a
593 pathname with invalid :DIRECTORY components signals a
595 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
597 ** DIRECTORY on logical pathnames is more correct.
598 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
599 signal a TYPE-ERROR if their argument is not a stream.
600 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
601 designator argument does not designate a stream.
602 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
603 examining the synonym.
604 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
606 ** OPEN and WITH-OPEN-STREAM allow opening streams with
607 element-type larger than ([UN]SIGNED-BYTE 32).
609 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
610 * When built with the :SB-FUTEX feature, threaded builds now take
611 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
612 for faster/more reliable mutex and condition variable support.
613 * Incompatible change (but one you probably shouldn't have been using
614 anyway): the interface and code for arbitrating between multiple
615 threads in the same user session has been redesigned.
616 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
617 error when called without an explicit environment argument.
618 (thanks to Nikodemus Siivola)
619 * bug fix: buffered :DIRECTION :IO streams are less likely to become
620 confused about their position. (thanks to Adam Warner and Gerd
622 * bug fix: Pretty printing backquoted forms with unquotations in the
623 argument list position of various code constructs such as LAMBDA
624 now works correctly. (reported by Paul Dietz)
625 * bug fix: Pretty printing unquotations no longer loses all
626 stream position information.
627 * optimization: performance of string output streams is now less
628 poor for multiple small sequence writes.
629 * optimization: performance of CSUBTYPEP in the presence of complex
630 expressions involving CONS and NOT many times has been improved.
631 (reported by Paul Dietz)
632 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
634 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
635 streams for the REPL, for the debugger, and so on. (thanks to
637 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
638 * fixed some bugs revealed by Paul Dietz' test suite:
639 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
640 is ignored if it is too small, rather than propagating through
641 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
642 ** extremely complex negations of CONS types were not being
643 sufficiently canonicalized, leading to inconsistencies in
645 ** VALUES tranformer lost derived type.
647 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
648 * fixed a bootstrapping bug: the build process no longer assumes
649 that the various BOOLE-related constants have the same value in
650 host and target lisps. (noted by Paul Dietz' test suite on an
651 SBCL binary built from CLISP)
652 * The system can now be dynamically linked on the MIPS platform,
653 which enables dynamic loading of foreign code from Lisp. (thanks
654 to Ralf Baechle for discussions on the MIPS ABI)
655 * The system now records debugging information for its own source
656 files in a filesystem-position-independent manner, relative to
657 the "SYS" logical host.
658 * fixed a compiler bug: MV-LET convertion did not check references
659 to the "max args" entry point. (reported by Brian Downing)
660 * tweaked disassembly notes to be less confident about proclaiming
661 some instruction as an LRA. (thanks to Brian Downing)
662 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
663 now each have their own history, command character, and other
664 characteristics. (thanks to David Lichteblau)
665 * fixed some bugs revealed by Paul Dietz' test suite:
666 ** compiler failure in compiling LOGAND expressions including a
668 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
669 shift greater than 32.
670 ** FLUSH-DEST did not mark blocks for type check regeneration.
671 ** HANDLER-CASE failed to accept declarations in handler clauses
672 in some circumstances.
674 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
675 * New code in contrib/sb-introspect (still probably not entirely
676 stable yet) provides some support for smart Lisp development
677 environments like SLIME.
678 * The conditions signalled for errors occurring when loading .fasl
679 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
680 in a way which should help ASDF recover gracefully.
681 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
682 stuff has been cleaned up. If you code contrib/ stuff, this might
683 affect you, and you can look at contrib/README, contrib/STANDARDS,
684 and/or the 0.8.4.27 diff to check.
685 * In full calls the compiler now does not generate checks for declared
686 argument types for all arguments.
687 * various threading fixes
688 ** and some experimental patches which didn't make it into
689 the main tree for this release, but which are shipped in
690 contrib/experimental-thread.patch as a possible fix for some
691 failures (deadlock, spinning...) in GC-intensive multithreaded
693 * fixed PPC build problem (source code incompatibility of different
694 library versions): added offsetof() hackery which attempts to divine
695 where glibc maintainers put uc_mcontext today
696 * fixed bug 282: compiler does not trust type assertions while passing
697 arguments to a full call.
698 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
699 &OPTIONAL VALUES type parameter.
700 * fix bug 214: algorithm for noting rejected templates is now more
701 similar to that of template seletion. (also reported by rydis on
703 * fixed bug 141b: printing backquoted information readably and prettily
704 inserts a space where necessary.
705 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
706 (CHARACTER 10) are now reported as errors, rather than propagated
707 as unknown types. (reported by piso on #lisp)
708 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
709 with values NIL and :ERROR. (thanks to Milan Zamazal)
710 * fixed bug 191c: CLOS now does proper keyword argument checking as
711 described in CLHS 7.6.5 and 7.6.5.1.
712 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
713 longer raise an error; further, using a list as a for-as-arithmetic
714 counter now raises a meaningful error.
715 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
716 now understood by sequence creation functions such as MAKE-SEQUENCE
718 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
719 a STRING-INPUT-STREAM.
720 * compiler enhancement: SIGNUM is now better able to derive the type
722 * type declarations inside WITH-SLOTS are checked. (reported by
724 * fixed some bugs revealed by Paul Dietz' test suite:
725 ** incorrect optimization of TRUNCATE for a positive first
726 argument and negative second.
727 ** compiler failure in let-convertion during flushing dead code.
728 ** compiler failure while deriving type of TRUNCATE on an
729 interval, containing 0.
730 ** ASH of a negative bignum by a negative bignum count now returns
732 ** intersection of CONS types now canonicalizes properly, fixing
733 inconsistencies in SUBTYPEP.
735 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
736 * incompatible change: The --disable-debugger command line
737 option now clobbers the debugger at a more fundamental
738 level, by redefining #'INVOKE-DEBUGGER instead of by
739 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
740 is specified by ANSI to ignore *DEBUGGER-HOOK* and
741 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
742 enter the debugger REPL and then suffer recursive errors
743 because *DEBUG-IO* is also messed up in --disable-debugger mode;
744 while under the new system, BREAK in --disable-debugger mode
745 terminates the system just as an unhandled error would.
746 * fixed compiler performance when processing loops with a step >1;
747 * bug fix: DOCUMENTATION now retrieves generic function
748 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
749 support has been systematized, and now supports the methods
750 specified by ANSI, along with a default method and a method for
751 slot documentation. (reported by Nathan Froyd)
752 * bug fix: effective methods associated with a generic function are
753 no longer cached over a change of that generic function's method
754 combination. (reported by Andreas Fuchs)
755 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
756 names a non-existent directory. (thanks to Andreas Fuchs)
757 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
758 the PPC platform, lead to stack corruption; this has been fixed.
759 (reported by Rainer Joswig)
760 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
761 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
762 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
763 arithmetic in general, is now much more reliable on the PPC
765 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
766 platform now returns the right answer.
767 * optimization: restored some effective method precomputation in
768 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
769 precomputation is now tunable.
770 * optimization: compiler-internal data structure use has been
771 reviewed, and changes have been made that should improve the
772 performance of the compiler by about 20%.
773 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
774 simple-base-strings and simple-bit-vectors is improved.
775 * optimization: the optimization of 32-bit logical and arithmetic
776 functions introduced in version 0.8.3 on the x86 has been
777 implemented on the mips, ppc and sparc platforms; an
778 implementation of the same facility, but for 64-bit arithmetic,
779 has been added for the alpha.
780 * microoptimization: the compiler is better able to make use of the
781 x86 LEA instruction for multiplication by constants.
782 * bug fix: in some situations compiler did not report usage of
783 generic arithmetic in (SPEED 3) policy.
784 * bug 145b fix: compiler used wrong type specifier while converting
785 MEMBER-types to numeric.
786 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
788 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
789 index argument now works properly on the Alpha platform.
790 * bug fix: floating point exception treatment on the Alpha platform
792 * bug fix: FILE-POSITION works much better on string input and
793 output streams. (thanks to Nikodemus Siivola)
794 * bug fix: many threading/garbage collection symptoms sorted.
795 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
796 might be pseudo-atomic.
797 * internal change: Stopping for GC is now done with signals not
798 ptrace. GC is now done in whichever thread wanted it, instead of
800 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
802 * bug fix: result form in DO is not contained in the implicit
804 * incompatible change: ICR structure is changed; the value part of
805 CONTINUATION is now called LVAR; corresponding functions are
806 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
807 * added type deriver for ISQRT (thanks to Robert E. Brown).
808 * bug fix: better support for loading from the command line when an
809 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
811 * fixed some bugs revealed by Paul Dietz' test suite:
812 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
813 ** ROUND and FROUND now give the right answer when given very
814 small float arguments.
815 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
817 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
818 ** (GCD 0 <negative-integer>) returned <negative-integer>.
819 ** LCM should return a non-negative integer.
820 ** PARSE-INTEGER returned the index of a terminator instead of the
821 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
822 ** PARSE-INTEGER returned an incorrect index being applied to a
824 ** LCM with two arguments of 0 returns 0 rather than signalling
826 ** unsigned addition of a 32-bit constant with the high bit set no
827 longer causes an internal compiler error on the x86.
828 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
829 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
830 with negative last argument.
831 ** byte specifiers with zero size and position no longer cause
832 an error during type derivation.
833 ** bignum multiplication on the Alpha platform now returns the
835 * porting: The system now builds on SuSE AMD64, although it still
836 generates a 32-bit binary.
837 * .fasl file incompatibility: The fasl file version number has
838 been incremented (because of the changes to internal compiler
839 data structures referred to above).
841 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
842 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
843 more accurately, on the Darwin kernel running on PowerPC hardware.
844 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
845 * Compiler code deletion notes now signal a condition of type
846 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
847 an associated MUFFLE-WARNING restart.
848 * The compiler now performs limited argument count validation of
849 constant format strings in FORMAT, and where appropriate in ERROR,
850 CERROR and WARN. (thanks to Gerd Moellmann)
851 * New ASDF-INSTALL contrib can be used for automatic download and
852 installation of third-party Lisp code from CCLAN or other sites
854 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
855 which forces another thread to execute a function supplied by the
857 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
858 now accept and act upon their :ELEMENT-TYPE keyword argument.
859 (reported by Martin Atzmueller, Edi Weitz)
860 * bug fix: FILE-POSITION now accepts position designators up to
861 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
862 is the greater. (thanks to Patrik Nordebo)
863 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
864 Kalle Olavi Niemitalo)
865 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
866 printable. (reported by Eric Marsden)
867 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
868 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
869 * x86 bug fix in control stack exhaustion checking: now shows backtrace
870 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
871 (thanks to Stig Sandoe)
872 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
873 UNREAD-CHAR now correctly includes the unread character in the
874 target sequence. (thanks to Gerd Moellmann)
875 * bug fix in threaded builds: the system can now be suspended and
876 resumed by shell job control with minimal disruption.
877 * bug fixes in times and timezones >2038AD
878 * better handling of "where is GNU make?" problem in build scripts
879 (thanks to Nikodemus Siivola)
880 * new optimization: inside a named function any reference to a
881 function with the same name is considered to be a self-reference;
882 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
883 optimization quality.
884 * new optimization on x86: logical functions and + now have
885 optimized (UNSIGNED-BYTE 32) versions, which are automatically
886 used when the result is truncated to 32 bits.
887 * VALUES declaration is partially enabled.
888 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
889 * fixed some bugs revealed by Paul Dietz' test suite:
890 ** The system now obeys the constraint imposed by
891 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
892 types form a lattice under type intersection.
893 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
894 ** ASSOC now ignores NIL elements in an alist.
895 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
896 and (1+ MOST-POSITIVE-FIXNUM) answers.
897 ** The addition of a method with invalid qualifiers to a generic
898 function does not cause an error to be signalled immediately;
899 a warning is signalled, and the error is generated only on
900 calling the generic function.
901 * changed .fasl file version number, in part to add type codes for
902 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
903 obscure ANSI requirements
905 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
906 * fixed bug 148: failure to inline-expand a local function left
907 garbage, confusing the compiler.
908 * fixed bugs 3cd: structure slot readers perform type check if the
909 slot can have an invalid value (i.e. it is either not initialized
910 or can be written with a less specific slot writer).
911 * bug fix: the compiler now traps array references to elements off
912 the end of an array; previously, the bounds checking in some
913 circumstances could go off-by-one.
914 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
916 * type declarations for array element types now obey the description
917 on the CLHS page "Declaration TYPE", as per discussions on
918 sbcl-help around 2003-05-08. This means that a declaration
919 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
920 declaration, all references to BAR will be asserted or assumed
921 (with THE, so dependent on compiler policy) to involve objects of
922 type FOO. Note that no such declaration is implied in
923 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
924 * declared types of functions from the "Conditions"
925 chapter. (reported by Paul Dietz)
926 * bug fix: CERROR accepts a function as its first argument.
927 * bug fix: NTH an NTHCDR accept a bignum as index
928 arguments. (reported by Adam Warner)
929 * optimization: character compare routines now optimize comparing
930 against a constant character. (reported by Gilbert Baumann)
931 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
932 argument now works properly on the MIPS platform.
933 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
934 * repeated evaluation of the same DEFSTRUCT, a slot of which is
935 declared to have a functional type, does not cause an error
937 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
938 many :DEFAULT-INITARGS. (reported by Istvan Marko)
939 * fixed bug: if last continuation of a deleted block has a
940 destination, this destination should be deleted too. (reported by
942 * fixed a bug in the bootstrap process: the host compiler's values
943 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
944 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
945 test case from Patrik Nordebo)
946 * improved the ability of the disassembler on the PPC platform to
947 provide helpful disassembly notes.
948 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
949 the class in more cases than previously.
950 * bug fix: FILE-POSITION now understands :START and :END for
951 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
952 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
953 8) data. (thanks to David Lichteblau)
954 * bug fix: it is possible to add a method to a generic function
956 * bug fix: reader failed to signal END-OF-FILE inside an
957 object representation. (reported by Nikodemus Siivola)
958 * fixed some bugs revealed by Paul Dietz' test suite:
959 ** LAST and [N]BUTLAST should accept a bignum.
960 ** condition slot accessors are methods.
961 ** (VECTOR NIL) is a subtype of STRING.
963 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
964 * minor incompatible change: some nonsensical specialized lambda
965 lists (used in DEFMETHOD) which were previously ignored now signal
967 * minor incompatible change: the system is now aware of the types of
968 variables in the COMMON-LISP package, and will signal errors for
969 most violations of these type constraints (where previously they
970 were silently accepted).
971 * minor incompatible change: COMPILE-FILE now uses the freedom
972 afforded (ANSI 3.2.2.3) to use derived function types for
973 functions defined in the same file. This also permits the system
974 to warn on static type mismatches and function
975 redefinition. (Currently it does not work with high DEBUG level.)
976 * minor incompatible change: VALUES declaration is disabled.
977 * When issuing notes, the compiler now signals a condition of type
978 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
979 restart for use in user handlers. It is expected that the
980 COMPILER-NOTE condition will eventually become a condition
981 supertype to a hierarchy of note types, which will then be
982 handleable in a similar fashion. However, at the moment, no such
983 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
984 but it's an implementation detail, not a classification for the
986 * Changes in type checking closed the following bugs:
987 ** type checking of unused values (192b, 194d, 203);
988 ** template selection based on unsafe type assertions (192c, 236);
989 ** type checking in branches (194bc).
990 * A short form of VALUES type specifier has ANSI meaning (it has
991 increased the number of situations when SBCL cannot perform type
993 * fixed bug in DEFSTRUCT: once again, naming structure slots with
994 keywords or constants is permissible.
995 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
996 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
997 classes. (thanks to Antonio Martinez)
998 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
999 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1000 argument to a pretty-print stream. Instead, it leaves any such
1001 operations to DESCRIBE-OBJECT methods.
1002 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1004 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1005 lambda lists. (thanks to David Lichteblau)
1006 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1007 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1008 (thanks to Gerd Moellmann)
1009 * fixed bug 246: increased compilation speed of long
1010 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1012 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1014 * DEFCONSTANT now throws a condition of type
1015 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1016 constant to a non-EQL value; CONTINUE and ABORT restarts
1017 respectively change and preserve the value.
1018 * fixed bug 63: The code walker, part of the implementation of CLOS,
1019 is now better at handling symbol macros.
1020 * bug fix: There is no longer an internal implementation type named
1021 CL:LENGTH. (reported by Raymond Toy)
1022 * bug fix: In macro-like defining macros/special operators the
1023 implicit block does not enclose the lambda list.
1024 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1025 enclosed in parentheses) are not suitable as type specifiers, and
1026 their use properly signals an error now.
1027 * bug fix: An argument count mismatch for a type specifier in code
1028 being compiled no longer causes an unhandled error at compile
1029 time, but signals a compile-time warning.
1030 * fixed simple vector readable printing
1031 * bug fix: DESCRIBE takes more care over whether the class
1032 precedence list slot of a class is bound before accessing it.
1033 (reported by Markus Krummenacker)
1034 * bug fix: FORMATTER can successfully compile pretty-printer format
1035 strings which use variants of the ~* directive inside.
1036 * bug fix: SEARCH now applies its TEST predicate to the elements of
1037 the arguments in the correct order. (thanks to Wolfhard Buss)
1038 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1039 in type inference. (thanks to Robert E. Brown)
1040 * bug fix: Reading in symbols with an explicit package name of ""
1041 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1042 * fixed some bugs revealed by Paul Dietz' test suite:
1043 ** NIL is now allowed as a structure slot name.
1044 ** Arbitrary numbers, not just REALs, are allowed in certain
1045 circumstances in LOOP for-as-arithmetic clauses.
1046 ** Multiple class redefinitions before slot access no longer
1047 causes a type error.
1048 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1049 association between the name and a class.
1050 ** Generic functions with non-standard method-combination and over
1051 five methods all of which return constants no longer return NIL
1052 after the first few invocations. (thanks to Gerd Moellmann)
1053 ** CALL-NEXT-METHOD with no arguments now passes the original
1054 values of the arguments, even in the presence of assignment.
1055 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1057 ** DISASSEMBLE works with closures and funcallable instances.
1058 ** ADD-METHOD now returns the generic function, not the new method.
1059 ** FIND-METHOD signals an error if the lengths of the specializers
1060 is incompatible with the generic function, even if the ERRORP
1062 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1063 which its argument is a member.
1064 ** DEFCLASS only redefines the class named by its class-name
1065 argument if that name is the proper name of the class;
1066 otherwise, it creates a new class.
1067 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1068 of the UNBOUND-SLOT condition to the name of the slot.
1069 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1070 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1071 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1072 SLOT-MAKUNBOUND in the specified fashion.
1074 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1075 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1076 cross-compilation host. As a consequence, we can now bootstrap our
1077 way up to SBCL starting with a bare gcc toolchain and human-readable
1078 source code (first the source to CLISP, then the source to SBCL).
1079 * A contributed module containing a partial implementation of the
1080 simple-streams interface has been included. (thanks to Rudi
1082 * A contributed module implementing the RFC1321 Message Digest
1083 Algorithm, known as MD5, has been included.
1084 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1085 no longer has any effect, as the code controlled by this feature
1086 has been deleted. (As far as we know, no-one has ever built using
1087 this feature, and its semantics were confused in any case).
1088 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1089 work on conditions (as required by the ANSI specification),
1090 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1091 expected behaviour on conditions. Users should note, however,
1092 that such behaviour is not required by the ANSI specification,
1093 and so use of this behaviour may render their code unportable.
1094 * fixed some bugs revealed by Paul Dietz' test suite:
1095 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1097 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1098 ** SXHASH on condition objects no longer returns NIL.
1099 ** :ALLOCATION :CLASS slots are better treated; their values are
1100 updated on class redefinition, and initforms inherited from
1101 superclasses are applied.
1102 ** REMOVE-METHOD returns its generic function argument even when
1103 no method was removed.
1104 ** SHARED-INITIALIZE now initializes the values of the requested
1105 slots, including those with :ALLOCATION :CLASS.
1106 ** ALLOCATE-INSTANCE now works on structure classes defined via
1107 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1109 ** SLOT-EXISTS-P now works on conditions, as well as structures
1111 ** MAKE-LOAD-FORM now has the required methods on
1112 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1113 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1114 keyword, and now implements the SLOT-NAMES argument.
1115 ** methods with &OPTIONAL arguments no longer allow too many
1116 arguments to be passed in the call without error.
1117 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1118 option is consistent with the required arguments of the generic
1119 function lambda list.
1120 * bug fix: REQUIRE accepts a string designator. (Thanks to
1122 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1123 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1124 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1125 * bug fix: The compiler now checks for duplicated variables in macro
1127 * bug fix: SETQ on globals returns the correct value.
1128 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1129 not a COMPILER-ERROR (followed by some other strange error on
1130 choosing the CONTINUE restart).
1131 * bug fix: make.sh and friends are now more consistent in the way that
1132 they look for GNU "make".
1134 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1135 * experimental native threads support (on x86 Linux >=2.4 only).
1136 This is not compiled in by default: you need to add :SB-THREAD to
1137 the target features. See the "Beyond ANSI" chapter of the manual
1139 * fix for longstanding nonANSIism: The old distinction between
1140 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1141 The return value from CL:FIND-CLASS is now a CLOS class, and
1142 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1143 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1144 * An interface to the MetaObject Protocol, as described in Kiczales,
1145 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1146 MIT Press, 1991, is available from the SB-MOP package.
1147 * incompatible change: the SB-PCL package should now be considered
1148 a private implementation detail, and no longer a semi-private MOP
1150 * minor incompatible change: due to rearrangement for threads, the
1151 control stack and binding stack are now allocated at arbitrary
1152 addresses instead of being hardcoded per-port. Users affected by
1153 this probably have to be doing advanced things with shared
1154 libraries, and will know who they are.
1155 * minor incompatible change: Previously, all --eval forms used were
1156 processed with READ before any of them were processed with EVAL.
1157 Now each --eval form is processed with both READ and EVAL before
1158 the next --eval form is processed. (Thus package operations like
1159 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1160 work as the user might reasonably expect.)
1161 * minor incompatible change: *STANDARD-INPUT* is now only an
1162 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1164 * minor incompatible change: Y-OR-N-P is now character-oriented,
1165 not line oriented. Also, YES-OR-NO-P now works without errors.
1166 (thanks to Antonio Martinez)
1167 * sb-aclrepl module improvements: an integrated inspector, added
1168 repl features, and a bug fix to :trace command.
1169 * Known functions, which cannot be open coded by the backend, are
1170 considered to be able to check types of their arguments. (fixing
1171 a bug report by Nathan J. Froyd)
1172 * fixed a bug in computing method discriminating functions: It is
1173 now possible to define methods specialized on classes which have
1174 forward-referenced superclasses. (thanks to Gerd Moellmann)
1175 * fixed evaluation order in optional entries (reported by Gilbert
1177 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1178 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1179 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1180 required-by-AMOP NAME argument, as well as CLASS and
1181 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1182 * fixed bug 20: DEFMETHOD can define methods using names that are
1183 not the proper names of classes to designate class specializers.
1184 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1185 Unix file descriptors, instead of blowing up. (thanks to Antonio
1187 * Garbage collection refactoring: user-visible change is that a
1188 call to the GC function during WITHOUT-GCING will not do garbage
1189 collection until the end of the WITHOUT-GCING. If you were doing
1190 this you were probably losing anyway.
1191 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1192 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1193 TYPEP the latter but not the former.
1194 * The compiler issues a full WARNING for calls to undefined functions
1195 with names from the CL package.
1196 * MAP-INTO for a vector destination is open coded. (reported by
1197 Brian Downing on c.l.l)
1198 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1199 documentation string.
1200 * fixed some bugs revealed by Paul Dietz' test suite:
1201 ** COPY-ALIST now signals an error if its argument is a dotted
1203 ** Condition slots are now accessed more correctly in the presence
1204 of multiple initargs for a given slot.
1205 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1206 exclude restarts of the same name associated with a different
1208 ** DEFCLASS of forward-referenced classes with another
1209 forward-referenced class in the superclasses list no longer
1211 ** Condition slots are now initialized once each, not multiple
1212 times. (thanks to Gerd Moellmann)
1213 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1215 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1216 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1217 arguments contain duplicated elements.
1218 ** RESTART-CASE understands local macros.
1219 ** RESTART-CASE associates exactly its own restarts with a condition.
1220 ** ENDP in safe mode checks its argument to be of type LIST.
1221 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1222 in question is unbound.
1223 ** Optimized MAKE-INSTANCE functions no longer cause internal
1224 assertion failures in the presence of duplicate initargs.
1225 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1226 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1227 propagated even in the ctor optimized implementation of
1229 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1231 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1232 * a better implementation of SXHASH on (simple) bit vectors,
1233 measured both in execution speed and in distribution of results
1234 over the positive fixnums, has been installed. Likewise, a better
1235 implementation of EQUAL for simple bit vectors is now available.
1236 * fixed CEILING optimization for a divisor of form 2^k.
1237 * fixed bug 240 (emitting extra style warnings "using the lexical
1238 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1239 by Antonio Martinez)
1240 * fixed SXHASH, giving different results for NIL depending on type
1241 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1242 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1243 variable. (found by Rolf Wester)
1244 * SBCL does not ignore type declarations for special
1245 variables. (reported by rif on c.l.l 2003-03-05)
1246 * some bug fixes in contrib/sb-aclrepl/
1247 * fixed some bugs revealed by Paul Dietz' test suite:
1248 ** a bug in the CONS type specifier, whereby the CAR and CDR
1249 types got intertwined, has been fixed;
1250 ** the type system is now able to reason about the interaction
1251 between INTEGER and RATIO types more completely;
1252 ** APPEND, [N]REVERSE and NRECONC check that those their
1253 arguments, which must be proper lists, are really so;
1254 ** An array specialized to be unable to hold elements has been
1255 implemented, as required -- yes, really -- by ANSI;
1256 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1257 on malformed property lists;
1259 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1260 * incompatible packaging change: in line with Unix convention,
1261 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1262 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1263 * REQUIRE and PROVIDE are now optionally capable of doing something
1264 useful. See the documentation string for REQUIRE.
1265 * infrastructure for a managed SBCL contrib system: contributed
1266 modules in this release include:
1267 ** the ASDF system definition facility;
1268 ** an interface to the BSD Sockets API;
1269 ** an ACL-like convenience interface to the repl;
1270 (thanks to Kevin Rosenberg)
1271 ** an implementation of ROTATE-BYTE, with efficient implementation
1273 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1274 gives a closer approximation to the right answer than previously.
1275 (thanks to Raymond Toy)
1276 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1277 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1278 optional environment arguments, as required by ANSI.
1279 * fixed bugs in other functions taking environment objects, allowing
1280 calls with an explicit NIL environment argument to be compiled
1282 * fixed bug 228: primary return values from
1283 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1284 COMPILE or FUNCTION.
1285 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1286 :TYPE will no longer signal errors on innocuous objects.
1287 * fixed bug 231b: SETQ is better at respecting type declarations in
1288 the lexical environment.
1289 * fixed a bug in DEFCLASS: classes named by symbols with no or
1290 unprintable packages can now be defined.
1291 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1292 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1293 * fixed compiler failure related to checking types of functions.
1294 (reported by Robert E. Brown)
1295 * the compiler is now much more consistent in its error-checking
1296 treatment of bounding index arguments to sequence functions: in
1297 (SAFETY 3) code, errors will be signalled in almost all cases if
1298 invalid sequence bounding indices are passed to functions defined
1299 by ANSI to operate on sequences.
1300 * fixed a bug in the build procedure: documentation of SBCL-specific
1301 packages is now preserved and available in the final Lisp image.
1302 * lifted FDEFINITION lookup out of loops in the implementation of
1303 many list operations. (thanks to Robert E. Brown)
1304 * fixed a bug in the reader: the #n# reader macro now works for
1305 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1306 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1307 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1308 E. Brown and Tony Martinez respectively)
1309 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1310 SLOT-BOUNDP: in particular, it is now able to optimize them much
1311 better, and is now not vulnerable to having packages renamed.
1312 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1313 always called when a slot is not present in an instance. (thanks
1315 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1316 longer signal an error on unreading a character following EOF on
1317 the previous constituent stream. (thanks to Tony Martinez)
1318 * fixed some bugs revealed by Paul Dietz' test suite:
1319 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1320 not just nonnegative fixnums;
1321 ** the logical bit-array operators such as BIT-AND now accept an
1322 explicit NIL for their "opt-arg" argument (to indicate a
1323 freshly-consed result bit-array);
1324 ** ELT now signals an error on an invalid sequence index in safe
1326 ** the type system is now cleverer about negations of numeric
1327 types, and consequently understands the BIGNUM and RATIO types
1329 ** the type system is now cleverer about the interaction between
1330 INTEGER and RATIO types: while bugs still remain, many more
1331 cases are accurately computed;
1332 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1333 if it is in the last clause;
1334 ** CONSTANTLY now correctly returns a side-effect-free function in
1336 ** DECLARE is no longer treated as a special-operator; in
1337 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1338 * incremented fasl file version number due to the change in the
1339 DEFSTRUCT-SLOT-DESCRIPTION structure.
1341 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1342 * minor incompatible change: code processed by the "interpreter" or
1343 EVAL now has a compilation optimization policy of (DEBUG 2)
1344 (changed from (DEBUG 1)) to improve debuggability of interactive
1345 development, and to allow the use of the debug RETURN command in
1347 * an experimental implementation of the RETURN command for the
1348 debugger has been included. (thanks to Frederik Kuivinen)
1349 * fixed bug 62: constraints were not propagated into a loop.
1350 * fixed bug in embedded calls of SORT (reported and investigated by
1352 * fixed some bugs revealed by Paul Dietz' test suite:
1353 ** printing and reading of arrays with some dimensions having
1354 length 0 (thanks to Gerd Moellmann);
1355 ** BOA constructor with &AUX argument without a default value does
1356 not cause a type error;
1357 ** CONSTANTP now returns true for all self-evaluating objects.
1359 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1360 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1361 accessors that are related by inheritance, as specified in the
1362 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1363 Valtteri Vuorikoski)
1364 * The compiler is now able to inline functions that were defined in
1365 a complex lexical environment (e.g. inside a MACROLET).
1366 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1368 * Support for the upcoming FreeBSD-5.0 release has been included.
1369 (thanks to Dag-Erling Smorgrav)
1370 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1371 effect when it is not in a toplevel context.
1372 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1373 stand a better chance of being correct. (thanks to Gerd
1375 * fixed bug in COERCE, which now signals an error on coercing a
1376 rational to a bounded real type which excludes the expected
1378 * The compiler is now able to derive types more accurately from the
1379 COERCE and COMPILE functions.
1380 * fixed bug 223: functional binding is considered to be constant
1381 only for symbols in the CL package.
1382 * fixed bug 231: SETQ did not check the type of a variable being set
1383 (reported by Robert E. Brown)
1384 * A new optimization for MAKE-INSTANCE has been included, fixing
1385 various bugs (including relating to :ALLOCATION :CLASS slots and
1386 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1388 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1389 ** As required by ANSI, LOOP now disallows anonymous collection
1390 clauses such as COLLECT I in conjunction with aggregate boolean
1391 clauses such as THEREIS (= I 1);
1392 ** LOOP now signals an error when any variable is reused in the
1393 same loop (including the potentially useful construct analogous
1394 to WITH A = 1 WITH A = (1+ A);
1395 ** IT is only a special loop symbol within the first clause of a
1396 conditional loop clause;
1397 ** LOOP with a typed iteration variable over a hashtable now
1398 signals a type error iff it should.
1399 * fixed some other bugs revealed by Paul Dietz' test suite:
1400 ** FILE-STREAM now names the class previously known as FD-STREAM;
1401 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1402 argument) no longer signals an error;
1403 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1404 of the slot symbol, rather than using the current package
1405 ((:CONC-NAME "") continues to intern the slot's name in the
1407 * incremented fasl file version number, because of the incompatible
1408 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1409 of the new implementation of DEFINE-COMPILER-MACRO.
1411 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1412 * Support for building SBCL for MIPS platforms running in
1413 little-endian mode has now been checked in, and basic
1414 functionality on said platforms verified.
1415 * minor incompatible change: PCL now records the pathname of a file
1416 in which methods and the like are defined, rather than its
1418 * minor incompatible change: TRUENAME now considers the truename of
1419 a file naming a directory to be the pathname with :DIRECTORY
1420 component indicating that directory.
1421 * minor incompatible change: a NAMED clause in the extended form of
1422 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1423 reason for the previous behaviour is unclear.
1424 * more systematization and improvement of CLOS and MOP conformance
1425 in PCL (thanks to Gerd Moellman and Pierre Mai):
1426 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1428 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1429 primary methods with no specializers;
1430 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1432 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1433 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1434 FORWARD-REFERENCED-CLASSes; error reporting on
1435 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1437 ** SXHASH on CLOS instances now uses a slot internal to the
1438 instance to return different numbers on distinct instances,
1439 while preserving the same return value through invocations of
1441 ** DEFMETHOD signals errors when methods with longer incongruent
1442 lambda lists are added to generic functions;
1443 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1444 CLASS, as specified in AMOP;
1445 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1446 on the order returned by the primary method for classes of
1447 class STANDARD-CLASS;
1448 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1449 * fixed some bugs shown by Paul Dietz' test suite:
1450 ** DOLIST puts its body in TAGBODY;
1451 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1453 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1454 value producing form;
1455 ** if more variables are given to PROGV than values, extra
1456 variables are bound and made to have no value;
1457 ** NSUBSTITUTE on list arguments gets the right answer with
1459 ** ELT signals an error of type TYPE-ERROR when the index argument
1460 is not a valid sequence index;
1461 ** LOOP signals (at macroexpansion time) an error of type
1462 PROGRAM-ERROR when duplicate variable names are found;
1463 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1464 ** FUNCALL of special-operators now cause an error of type
1466 ** PSETQ now works as required in the presence of side-effecting
1467 symbol-macro places;
1468 ** NCONC accepts any object as its last argument;
1469 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1471 ** loop-for-as-package does not require a package to be explicitely
1473 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1475 * fixed bug 166: compiler preserves "there is a way to go"
1476 invariant when deleting code.
1477 * fixed bug 172: macro lambda lists with required arguments after
1478 &REST arguments now cause an error to be signalled. (thanks to
1480 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1481 bug. (thanks to Matthew Danish)
1482 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1484 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1485 arguments are lexically rebound. (thanks to Gerd Moellmann and
1487 * fixed bug 194: error messages are now more informative when there
1488 is no primary method applicable in a call to a generic
1489 function. (thanks to Gerd Moellmann)
1490 * fixed bug in command line argument checking (thanks to Julian
1492 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1493 (thanks to Matthew Danish)
1494 * incremented fasl file version number, because of the
1495 SXHASH-related changes in the layout of CLOS data structures
1497 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1498 * minor incompatible change: The runtime (the Unix executable named
1499 "sbcl") is now much pickier about the .core files it will load.
1500 Essentially it now requires .core files to descend from the same
1501 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1502 as the runtime does. (The intent is to prevent the crashes which
1503 can occur, and which can even be reported as mysterious failures,
1504 when people patch the sources or change the build parameters
1505 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1506 sbcl and .core files.)
1507 * fixed bug: VALUES-LIST is no longer optimized away.
1508 * fixed bug 142: The FFI conversion of C string values to Lisp
1509 string values no longer conses excessively. (thanks to Nathan
1510 Froyd porting Raymond Toy's fix to CMU CL)
1511 * began to systematize and improve MOP conformance in PCL (thanks to
1512 Nathan Froyd, Gerd Moellman and Pierre Mai):
1513 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1515 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1516 ** FINALIZE-INHERITANCE is now called on class finalization;
1517 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1518 argument precedence order.
1519 * fixed bug 202: The compiler no longer fails on functions whose
1520 derived types contradict their declared type.
1521 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1522 so it can be non-toplevel.
1523 * The fasl file version number has changed (because of the new
1524 implementation of DEFMACRO).
1525 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1526 safe code, that any length requirement by their type-specifier
1527 argument is valid. The exceptions to this are described in bug
1529 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1530 functions now check the types of their inputs as required by ANSI.
1531 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1532 introduced symbol is DECLAREd to be SPECIAL.
1533 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1534 * fixed bug: :COUNT argument to sequence functions may be negative
1535 * fixed bug: body of DO-SYMBOLS may contain declarations
1536 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1537 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1538 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1539 symbol macro only once
1540 * fixed printing of call frame when argument list is unavailable
1541 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1542 * compiler no longer signals WARNING on unknown keyword
1545 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1546 * A beta-quality port to the mips architecture running Linux,
1547 based on the old CMUCL backend, has been made. It has been tested
1548 on a big-endian kernel, and works sufficiently well to be able to
1549 rebuild itself; it has not been tested in little-endian mode.
1550 * fixed an inconsistency between gencgc.c and purify.c which made
1551 dumping/loading .core files unreliable
1552 * fixed bug 120a: The compiler now deals correctly with IFs where
1553 the consequent is the same as the alternative, instead of
1554 misderiving the return type. (thanks to Alexey Dejneka)
1555 * fixed bug 113: Logical pathnames are now dumpable (the logical
1556 host is resolved at load-time, throwing an error if it is not
1558 * fixed bug 174: FORMAT's error message is slightly clearer when a
1559 non-printing character is used in a format directive.
1560 * fixed several bugs in compiler checking of type declarations, i.e.
1561 violations of the Python "declarations are assertions" principle
1562 (thanks to Alexey Dejneka)
1563 * fixed several bugs in PCL's error checking (thanks to Gerd
1565 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1567 * fixed bug in compilation of functions as first class values
1568 (thanks to Antonio Martinez-Shotton)
1569 * The compiler's handling TYPE-ERRORs which it can prove will
1570 inevitably happen at runtime has been cleaned up and corrected
1571 in several ways. (thanks to Alexey Dejneka)
1572 * improved argument type checking for various basic arithmetic
1573 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1574 lavished on them in the past that they can be compiled in many
1575 ways in different special cases
1576 * fixed bug 181: compiler checks validity of user supplied type
1578 * cleaned up code flushing in optimization: Function calls which
1579 should signal errors for safety purposes (e.g. which ANSI says
1580 should signal errors when their arguments are of incorrect type)
1581 are no longer optimized away.
1582 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1583 * incremented fasl file version number, because changes in the
1584 implementation of sequence functions like COERCE caused
1585 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1586 in old inline expansions) to become undefined. (Actually these
1587 changes were later undone, so we might very well be binary
1588 compatible with 0.7.7 after all, but leaving the version number
1589 incremented seemed like the simplest and most conservative
1592 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1593 * An alpha-quality port to the parisc architecture running Linux,
1594 based on the old CMUCL backend, has been made. This, even more so
1595 than the other backends, should be considered still a work in
1596 progress; known problems include that the Linux kernel in 64-bit
1597 mode does not propagate the correct sigcontext structure to
1598 userspace, and consequently SBCL on a parisc64 kernel will not
1600 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1601 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1602 suppressed.) Also now that the compiler is looking at declarations
1603 in the environment, it checks optimization declarations as well,
1604 and suppresses inlining when (> DEBUG SPEED).
1605 * More fixes have been made to treatment of floating point exception
1606 treatment and other Unix signals. In particular, floating point
1607 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1608 * The detection and handling of control stack exhaustion (infinite
1609 or very deeply nested recursion) has changed. Stack exhaustion
1610 detection is now done by write-protecting pages at the OS level
1611 and applies at all optimization settings; when found, a
1612 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1613 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1614 be caught using IGNORE-ERRORS.
1615 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1616 correctly when transforming recursive LABELS functions to LETs.
1617 (thanks to Alexey Dejneka)
1618 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1619 that are names of constants or global variables.
1620 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1621 alien routines with docstrings.
1622 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1623 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1625 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1626 (thanks to Eric Marsden)
1627 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1628 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1629 * Bug fix: The SPARC backend can now compile functions involving
1630 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1631 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1632 to a functions expecting a PATHNAME-DESIGNATOR.
1633 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1634 lambda lists. (thanks to Alexey Dejneka)
1635 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1636 functions so that EXPORTing the name of the class doesn't cause
1637 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1638 to Antonio Martinez for reporting this)
1639 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1640 constant-folded without failing an assertion. (thanks to Einar
1641 Floystad Dorum for reporting this)
1642 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1643 types are now dealt with more correctly by the compiler.
1644 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1645 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1646 behaviour analogous to RENAME-FILE. This puts its behaviour more
1647 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1649 * The fasl file version number has changed again. (because of the
1650 bug fix involving the names of PCL MAKE-INSTANCE functions)
1652 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1653 * bug fix: Floating point exceptions are treated much more
1654 consistently on the x86/Linux and PPC/Linux platforms.
1655 * Array initialization with :INITIAL-ELEMENT is now much faster for
1656 cases when the compiler cannot open code the array creation, but
1657 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1658 array accesses have also seen a speed increase.
1659 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1660 specified. (This was at the root of some bad interactions between
1661 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1663 * bug fix: Internal error arguments for undefined functions are now
1664 computed correctly on the PPC/Linux platform.
1665 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1666 Raymond Toy's patch for CMU CL)
1667 * Support for the Solaris 9 operating environment has been included
1668 (thanks to Daniel Merritt)
1669 * A very ugly but hopefully complete draft of the missing FFI chapter
1670 of the manual has been created by reformatting the corresponding
1671 CMU CL manual chapter into (currently very ugly and incoherent)
1672 DocBook and bringing it up to date for SBCL behavior. Thus, the
1673 manual is now essentially complete, at least by my extreme
1674 once-and-only-once standards, whereby it's acceptable to refer to
1675 the doc strings of SB-EXT functions as the primary documentation.
1676 * The fasl file version number has changed again, due to cleanup of
1677 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1678 is no longer a static symbol.)
1680 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1681 * SBCL now builds with OpenMCL (version 0.12) as the
1682 cross-compilation host; also, more progress has been made toward
1683 bootstrapping under CLISP.
1684 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1686 * bug 158 fixed: The compiler can now deal with integer loop
1687 increments different from 1; fixing this turned out also to fix
1689 * bug 169 fixed: no more bogus warnings about using lexical bindings
1690 despite the presence of perfectly good SPECIAL declarations (thanks
1691 to David Lichteblau)
1692 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1693 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1694 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1696 * bug 180 fixed: Method combination specifications no longer ignore
1697 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1698 * bug fix: Structure type predicate functions now check their argument
1699 count as they should.
1700 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1701 correctly. (thanks to Pierre Mai)
1702 * minor incompatible change: The --noprogrammer option is deprecated
1703 in favor of the new --disable-debugger option, which is very similar.
1704 (The major difference is that it takes effect at a slightly different
1705 time at startup, causing handling of errors in --sysinit and
1706 --userinit files will be affected differently.) The
1707 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1708 been added to allow this functionality to be controlled from ordinary
1709 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1710 maintainers, who might want to run non-interactive scripts to
1711 build SBCL cores which will later be used interactively.)
1712 * minor incompatible change: The LOAD function no longer, when given
1713 a wild pathname to load, loads all files matching that pathname.
1714 Instead, an error of type FILE-ERROR is signalled.
1716 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1717 * bug 147 fixed: The compiler preserves its block link/count
1718 invariants more correctly now so that it doesn't crash. (thanks
1720 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1722 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1723 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1724 no-value-for-this-character as NIL (as specified by ANSI).
1725 * HOST-NAMESTRING on physical pathnames now returns a string that is
1726 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1727 (thanks to Christophe Rhodes)
1728 * The Alpha port handles icache flushing more correctly. (thanks to
1730 * More progress has been made toward bootstrapping under CLISP. (thanks
1731 to Christophe Rhodes)
1732 * The fasl file format has changed again, because dynamic loading
1733 on OpenBSD (which has non-ELF object files) motivated some cleanups
1734 in the way that foreign symbols are transformed and passed around.
1735 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1736 is no longer treated as whitespace by the reader, but instead as
1737 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1738 |A
\7fB|, instead of A as it used to.
1740 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1741 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1742 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1743 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1744 this as of 0.7.1.45, but was left out of the previous news section
1745 (thanks to Dan Barlow)
1746 * SBCL now runs on the Solaris operating system on SPARC architectures
1747 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1748 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1749 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1750 * SPARC backend cleanups, allowing builds of cores optimized for V8
1751 and V9 SPARCS, and also emission of code targeted to a particular
1752 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1754 * SBCL is closer to bootstrapping under CLISP, thanks to various
1755 fixes by Christophe Rhodes.
1756 * The fasl file format has changed again, to allow the compiler's
1757 INFO database to support symbol macros.
1758 * The user manual (in doc/) is formatted into HTML more nicely.
1759 (thanks to coreythomas)
1760 * The system is smarter about SUBTYPEP relationships, especially
1761 those involving NOT types (including types such as ATOM which are
1762 represented internally using NOT types). Thus SUBTYPEP is less
1763 likely to return (VALUES NIL NIL) in general, and in particular
1764 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1766 * The fasl file format has changed again, because the internal
1767 representation of types now includes a new slot to support the new
1768 SUBTYPEP-of-NOT-types logic.
1769 * (not a change in the main branch of SBCL, but a related prototype
1770 which can hopefully be merged into the main branch of SBCL in the
1771 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1772 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1773 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1774 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1775 dispatch macro character combinations. (thanks to Alexey Dejneka)
1776 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1777 that superficially look like logical namestrings correctly.
1778 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1780 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1781 default supplied-p) for &optional and &key arguments. (thanks to
1784 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1785 * incompatible change: The compiler is now less aggressive about
1786 tail call optimization, doing it only when (> SPACE DEBUG) or
1787 (> SPEED DEBUG). (This is an incompatible change because there are
1788 programs which relied on the old CMU-CL-style behavior to optimize
1789 away their unbounded recursion which will now die of stack overflow.)
1790 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1791 for non-GENCGC systems has been increased to 20M (since that
1792 seems much closer to the likely performance optimum for modern
1793 systems than the old 4M value was)
1794 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1795 and *DEBUG-PRINT-LEVEL*
1796 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1797 of CMU CL's support for SPARC, and various endianness and other
1798 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1799 * new syntactic sugar for the Unix command line: --load foo.bar is now
1800 an alternate notation for --eval '(load "foo.bar")'.
1802 ** The system now detects stack overflow and handles it gracefully,
1803 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1804 optimization settings. (This is a good thing in general, and
1805 its introduction in this version should be particularly timely
1806 for anyone whose code fails because of suppression of tail
1808 ** The system now hunts for the C variable "environ" in a more
1809 devious way, to avoid segfaults when the C library version differs
1810 between compile time and run time. (thanks to Christophe Rhodes)
1811 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1812 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1813 ** The compiler no longer issues bogus style warnings for undefined
1814 classes in the same source file as the DEFCLASSes which defined
1815 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1817 ** fixes in CONDITION class precedence list for undefined function
1818 errors (thanks to Alexei Dejneka)
1819 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1820 correctly. (thanks to Dan Barlow)
1821 ** portability fixes aiming at bootstrapping under CLISP (thanks
1822 to Dave McDonald and Christophe Rhodes)
1823 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1824 ** fixes in type translation and and type inference (thanks to
1826 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1827 ** various fixes in the new ports (thanks to Dan Barlow)
1828 * several changes related to debugging:
1829 ** suppression of tail recursion, as noted above
1830 ** stack overflow detection, as noted above
1831 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1832 is now the default. (For some time encapsulation has been more
1833 reliable than the breakpoint-based :ENCAPSULATE NIL
1834 implementation, at least on X86 systems; and I just noticed that
1835 encapsulation also seems closer to the spirit of the ANSI
1838 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1840 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1841 up properly again. (There was a packaging bug in 0.7.0 which
1842 left their definitions in SB-SYS::LOAD-FOREIGN and
1843 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1844 vital for most things which interface to C-level interfaces,
1845 like extensions working with sockets or databases or
1846 Perl-compatible regexes or whatever, and the need to fix
1847 this bug is the main reason that 0.7.1 was released so
1849 ** DEFGENERIC is now choosier about the methods it redefines, so that
1850 reLOADing a previously-LOADed file containing DEFGENERICs does
1851 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1852 cycle now works as it should. (thanks to Alexey Dejneka)
1853 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1854 fixed by Christophe Rhodes. (He actually submitted this patch
1855 months ago, and I delayed until after 0.7.0.)
1856 ** Bug 111 (internal compiler confusion about runtime checks on
1857 FUNCTION types) was fixed by Alexey Dejneka.
1858 * Some internal cleanups (getting rid of variables which aren't
1859 needed now that the byte interpreter is gone) caused the fasl
1860 file format number to change again.
1862 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1863 * major incompatible change: The default fasl file extension, i.e. the
1864 default extension for files produced by COMPILE-FILE, has changed
1865 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1867 ** There are many changes in the implementation of the compiler.
1868 SBCL is now essentially a compiler-only implementation of ANSI
1869 Common Lisp. EVAL still "interprets" a few special cases, but
1870 almost all the interesting cases are handled by creating
1871 a LAMBDA expression, calling COMPILE on it, then calling
1872 FUNCALL on the result.
1873 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1874 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1875 Since the code is newer, there might still be some new bugs
1876 (though not as many as before Martin Atzmueller's fixes:-). But
1877 the new code is substantially simpler and clearer, and hopefully
1878 any remaining bugs will be simpler, less fundamental, and more
1879 fixable then the bugs in the old code.
1880 ** The revised compiler is still a little unsteady on its feet.
1882 *** The debugging information it produces (particularly the names
1883 of FUNCTION objects) is sometimes much less useful than what
1884 the old compiler produced.
1885 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1886 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1887 within a MACROLET) has been temporarily weakened.
1888 ** There are new compiler optimizations for various functions:
1889 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1890 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1891 *** the math functions TRUNCATE, FLOOR, and CEILING
1892 *** the function-of-all-trades COERCE
1893 Mostly these should be transparent, but there's one
1894 potentially-annoying problem (bug 117): when the compiler
1895 inline-expands a function and does type analysis on the result,
1896 it can create control paths which have type mismatches, and
1897 when it can't prove that those control paths aren't taken,
1898 it will issue WARNINGs about the type mismatches. This is
1899 a particular problem in practice for the new sequence functions.
1900 It's not clear how this should be fixed, and for now, a
1901 workaround is given in the entry for 117 in the BUGS file.
1902 ** (Because of the interaction between the two previous items --
1903 occasional inlining problems and new inline expansions -- some
1904 of the new sequence function optimizations won't really kick in
1905 completely until debugging information, and then inlining, are
1906 straightened out in some future version.)
1907 * minor incompatible changes:
1908 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1909 DIRECTORY no longer implicitly promotes NIL slots of its
1910 pathname argument to :WILD. In particular, when you ask for the
1911 contents of a directory (which you used to be able to do without
1912 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1913 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1914 ** changes in behavior that ANSI explicitly defines to be
1915 implementation dependent:
1916 *** The new compiler-only implementation still conforms with ANSI,
1917 but acts a little different than before. Besides the obvious
1918 changes in performance tradeoffs (that the cost per form passed
1919 to EVAL has gone up, and the cost per form executed by EVAL
1920 has gone down), the behavior of the system changes a little
1921 because there are no longer any interpreted function objects.
1922 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1923 e.g. doing COMPILE on the output of interactive DEFUN is
1925 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1927 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1928 has changed from (:CL) to NIL.
1929 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1930 e.g. #\Newline and #\Tab, don't have names specified in the
1931 ANSI Common Lisp standard, is now based on their ASCII symbolic
1932 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1933 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1934 are no longer used for output.
1935 ** changes in internal implementation constants:
1936 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1937 4 million. (If your application spends a lot of time GCing and
1938 you have a lot of RAM, you might want to experiment with
1939 increasing it even more.)
1940 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1941 However, almost all old code should still continue to work without
1942 immediate update, as SB-C-CALL is now a (deprecated) nickname
1944 ** Old operator names in the style DEF-FOO are now deprecated in
1945 favor of new corresponding names DEFINE-FOO, for consistency with
1946 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1947 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1948 internal symbols, but a few supported extensions like
1949 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1950 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1951 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1952 etc. as you get deeper into recursive calls to the debugger
1953 command loop, instead of the old "5]", "5]]", "5]]]"
1954 sequence. (I was motivated to do this when squabbles between
1955 ILISP and SBCL left me very deeply nested in the debugger. In the
1956 short term, this change will probably provoke more ILISP/SBCL
1957 squabbles, but hopefully it will be an improvement in the long run.)
1958 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1959 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1960 function, since declaiming return types involving aliens is
1961 (1) annoyingly messy to do by hand and (2) vital to efficient
1962 compilation of code which calls such functions.
1963 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1964 longer reexported by the SB-EXT package. They're solely useful
1965 for alien code, so it seems more logical that you should get
1966 them from the SB-ALIEN package, not in SB-EXT.
1967 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1968 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1969 features. Instead, the code that they used to control is always
1970 built into the system.
1971 * many other bug fixes
1972 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1973 advantage of the new EVAL-WHEN stuff and to clean them up in
1974 general, and they are now more ANSI-compliant in a number of
1975 ways. Martin Atzmueller is responsible for a lot of this.
1976 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1978 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1979 *** correct ERROR type for various file operations
1980 *** some fixes for Lisp streams
1981 *** DEFMETHOD syntax checking
1982 *** changing old weird representation of debug information as
1983 strings (which, among their other deficiencies, don't transform
1984 correctly when you rename packages, and don't change their
1985 print representation when you change things like *PACKAGE*
1986 and *PRINT-LENGTH*) to symbols and lists of symbols
1987 He also made several improvements and fixed several bugs in DESCRIBE.
1988 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1990 *** misbehavior of WRITE-STRING/WRITE-LINE
1991 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1992 and several other LOOP problems as well
1993 *** DIRECTORY when similar filenames are present
1994 *** DEFGENERIC with :METHOD options
1995 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1996 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1997 *** argument ordering in FIND with :TEST option
1998 *** mishandled package designator argument in APROPOS-LIST
1999 *** various problems in the backquote readmacro
2000 *** a bug in APROPOS
2001 *** probably some others that I'm not describing very well here,
2002 since the CVS log documents them by reference to sbcl-devel
2003 messages, and the SourceForge archives aren't working well.:-(
2004 ** Dan Barlow improved the Alpha port (and is making progress on the
2005 PPC port, for those of you who think different).
2006 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2007 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2008 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2009 continues to work on the SPARC port (for those of you in a position
2010 to look down upon our little PC-compatible boxes from a great height).
2011 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2012 and uses that copy, rather than the current dynamic value, when
2013 it's trying to decide whether to truncate output. Thus e.g.
2014 (let ((*print-lines* 50))
2015 (pprint-logical-block (stream nil)
2017 (let ((*print-lines* 8))
2018 (print (aref possiblybigthings i) stream)))))
2019 should now truncate the logical block only at 50 lines, instead of
2020 often truncating it at 8 lines, as it did before.
2021 * The doc/cmucl/ directory, containing old CMU CL documentation
2022 from the time of the fork, is no longer part of the base system.
2023 SourceForge has shut down its anonymous FTP service, and with it
2024 my original plan for distributing the old CMU CL documentation
2025 there. For now, if you need these files you can download an old
2026 SBCL source release and extract them from it.
2027 * The fasl file version number changed again, for dozens of reasons,
2028 some of which are apparent above.
2030 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2031 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2032 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2033 patch, so that DISASSEMBLE output is much nicer.
2034 * The code in the SB-PROFILE package now seems reasonably stable.
2035 I still haven't decided what the final interface should look like
2036 (I'd like PROFILE to interact cleanly with TRACE, since both
2037 facilities use function encapsulation) but if you have a need
2038 for profiling now, you can probably use it successfully with
2039 the current CMU-CL-style interface.
2040 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2041 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2042 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2043 link now returns the dangling link itself, and for similar
2044 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2045 link itself. (In these cases the old code signalled an error and
2046 looped endlessly, respectively.) Thus, DIRECTORY now works even
2047 in the presence of dangling and cyclic symbolic links.
2048 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2049 is now a supported extension again, since the consensus on
2050 sbcl-devel was that it can be useful for ordinary development
2051 work, not just for debugging SBCL itself.
2052 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2053 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2054 that currently-defined functions might be redefined later with
2055 different return types.
2056 * Hash tables can be printed readably, as inspired by CMU CL code
2057 of Eric Marsden and SBCL code of Martin Atzmueller.
2058 * better error handling in CLOS method combination, thanks to
2059 Martin Atzmueller porting Pierre Mai's CMU CL patches
2060 * more overflow fixes for >16Mbyte I/O buffers
2061 * A bug in READ has been fixed, so that now a single Ctrl-D
2062 character suffices to cause end-of-file on character streams.
2063 In particular, now you only need one Ctrl-D at the command
2064 line (not two) to exit SBCL.
2065 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2067 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2068 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2069 does the right thing.
2070 * The implementation of some type tests, especially for CONDITION
2071 types, is now tidier and maybe faster, due to CMU CL code
2072 originally by Douglas Crosher, ported by Martin Atzmueller.
2073 * Some math functions have been fixed, and there are new
2074 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2075 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2076 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2077 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2078 likely to be built into sbcl-0.7.0.)
2079 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2080 CL, ported to SBCL by Martin Atzmueller)
2081 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2082 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2083 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2084 * There's a new slam.sh hack to shorten the edit/compile/debug
2085 cycle for low-level changes to SBCL itself, and a new
2086 :SB-AFTER-XC-CORE target feature to control the generation of
2087 the after-xc.core file needed by slam.sh.
2088 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2089 COMPILE-FILE is no longer supported, so that now every function
2090 gets an entry point, so that block compilation looks a little
2091 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2092 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2093 returns the number of bytes consed since the system started,
2094 rather than the number consed since the first time the function
2095 was called. (The new definition parallels ANSI functions like
2096 CL:GET-INTERNAL-RUN-TIME.)
2097 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2098 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2099 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2100 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2101 * Fasl file version numbers are now independent of the target CPU,
2102 since historically most system changes which required version
2103 number changes have affected all CPUs equally. Similarly,
2104 the byte fasl file version is now equal to the ordinary
2107 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2108 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2109 is no longer recognized. I apologize for this, because it was
2110 listed in SB-EXT as a supported extension, but I found that
2111 its existing behavior was poorly specified, as well as incorrectly
2112 specified, and it looked like too much of a mess to straighten it
2113 out. I have enough on my hands trying to get ANSI stuff to work..
2114 * many patches ported from CMU CL by Martin Atzmueller, with
2115 half a dozen bug fixes in pretty-printing and the debugger, and
2116 half a dozen others elsewhere
2117 * fixed bug 13: Floating point infinities are now supported again.
2118 They might still be a little bit flaky, but thanks to bug reports
2119 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2120 as flaky as they were.
2121 * The --noprogrammer command line option is now supported. (Its
2122 behavior is slightly different in detail from what the old man
2123 page claimed it would do, but it's still appropriate under the
2124 same circumstances that the man page talks about.)
2125 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2126 are now supported, and enabled by default. Thus, the compiler can
2127 handle many floating point and complex operations much less
2128 inefficiently. (Thus e.g. you can implement a complex FFT
2130 * The compiler now detects type mismatches between DECLAIM FTYPE
2131 and DEFUN better, and implements CHECK-TYPE more correctly, and
2132 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2133 thanks to patches from Martin Atzmueller.
2134 * various fixes to make the cross-compiler more portable to
2135 ANSI-conforming-but-different cross-compilation hosts (notably
2136 Lispworks for Windows, following bug reports from Arthur Lemmens)
2137 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2138 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2139 fixed thanks to Pierre Mai's CMU CL patches.
2140 * fixes to keep the system from overflowing internal counters when
2141 it tries to use i/o buffers larger than 16M bytes
2142 * fixed bug 45a: Various internal functions required to support
2143 complex special functions have been merged from CMU CL sources.
2144 (When I was first setting up SBCL, I misunderstood a compile-time
2145 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2146 * improved support for type intersection and union, fixing bug 12
2147 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2148 more obscure bugs as well
2149 * some steps toward byte-compiling non-performance-critical
2150 parts of the system, courtesy of patches from Martin Atzmueller
2151 * Christophe Rhodes has made some debian packages of sbcl at
2152 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2153 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2154 stable, but are nonetheless usable. When he's ready, I'd be happy
2155 to add them to the SourceForge "File Releases" section. (And if
2156 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2157 * new fasl file format version number (because of changes in
2158 internal representation of (OR ..) types to accommodate the new
2159 support for (AND ..) types, among other things)
2161 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2162 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2164 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2166 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2167 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2168 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2169 * DESCRIBE now works on structure objects again.
2170 * Most function call argument type mismatches are now handled as
2171 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2172 know whether the function will be redefined before the call is
2173 executed. (The compiler could flag local calls with full WARNINGs,
2174 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2175 it doesn't keep track of enough information to know whether calls
2176 are local in this sense.)
2177 * Compiler output is now more verbose, with messages truncated
2178 later than before. (There should be some supported way for users
2179 to override the default verbosity, but I haven't decided how to
2180 provide it yet, so this behavior is still controlled by the internal
2181 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2182 src/compiler/ir1util.lisp.)
2183 * Fasl file format version numbers have increased again, because
2184 support for the Gray streams extension changes the layout of the
2185 system's STREAM objects.
2186 * The Gray subclassable streams extension now works, thanks to a
2187 patch from Martin Atzmueller.
2188 * The full LOAD-FOREIGN extension (not just the primitive
2189 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2190 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2191 but like most other programs, it defaults to copying the Unix
2192 environment from the original process instead of starting the
2193 new process in an empty environment.
2194 * Extensions which manipulate the Unix environment now support
2195 an :ENVIRONMENT keyword option which doesn't smash case or
2196 do other bad things. The CMU-CL-style :ENV option is retained
2197 for porting convenience.
2198 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2199 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2201 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2203 * A patch from Martin Atzmueller seems to have solved the SIGINT
2204 problem, and as far as we know, signal-handling now works cleanly.
2205 (If you find any new bugs, please report them!)
2206 * The system no longer defaults Lisp source file names to types
2207 ".l", ".cl", or ".lsp", but only to ".lisp".
2208 * The compiler no longer uses special default file extensions for
2209 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2210 seems to expect a single default extension for all compiled code,
2211 and there's no compelling reason to try to stretch the standard
2212 to allow two different extensions.) Instead, byte-compiled files
2213 default to the same extension as native-compiled files.
2214 * Fasl file format version numbers have increased again, because
2215 a rearrangement of internal implementation packages made some
2216 dumped symbols in old fasl files unreadable in new cores.
2217 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2218 many fewer weird special cases.
2219 * Bug #17 (differing COMPILE-FILE behavior between logical and
2220 physical pathnames) has been fixed, and some related misbehavior too,
2221 thanks to a patch from Martin Atzmueller.
2222 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2223 by Tim Moore, ported to SBCL by Martin Atzmueller.
2224 * Martin Atzmueller fixed several filesystem-related problems,
2225 including bug #36, in part by porting CMU CL patches, which were
2226 written in part by Paul Werkowski.
2227 * More compiler warnings in src/runtime/ are gone, thanks to
2228 more patches from Martin Atzmueller.
2229 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2232 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2234 * DESCRIBE now works on CONDITION objects.
2235 * The debugger now handles errors which arise when trying to print
2236 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2238 * The build system now uses an additional file, customize-target-features.lisp,
2239 to allow local modifications to the target *FEATURES* list. (The point of
2240 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2241 debugging features enabled, without having to worry about propagating it
2242 into everyone's system when I do a "cvs update".) When no
2243 customize-target-features.lisp file exists, the target *FEATURES* list
2244 should be constructed the same way as before.
2245 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2246 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2247 warn about behavior which is undefined under the ANSI spec. Note: This
2248 is specified by ANSI, but it's not very popular with programmers.
2249 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2250 macro in the SBCL sources for an example of a workaround which you
2251 might use to make portable ANSI-standard code which does what you want.
2252 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2253 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2254 when it's used as a non-toplevel form. (This is required in order
2255 to implement the DEFCONSTANT-EQX macro.)
2256 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2257 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2258 funny things, due to bugs in the implementation of EVAL-WHEN
2259 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2260 non-toplevel DEFCONSTANTs are uncommon.)
2261 * The core file version number and fasl file version number have been
2262 incremented, because the old noncompliant DEFCONSTANT behavior involved
2263 calling functions which no longer exist, and because I also took the
2264 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2265 * fixed bug 1 (error handling before read-eval-print loop starts), and
2266 redid debugger restarts and related debugger commands somewhat while
2268 ** The QUIT debugger command is gone, since it did something
2269 rather different than the SB-EXT:QUIT command, and since it never
2270 worked properly outside the main toplevel read/eval/print loop.
2271 Invoking the new TOPLEVEL restart provides the same functionality.
2272 ** The GO debugger command is also gone, since you can just invoke
2273 the CONTINUE restart directly instead.
2274 ** The TOP debugger command is also gone, since it's redundant with the
2275 FRAME 0 command, and since it interfered with abbreviations for the
2277 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2279 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2280 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2281 some porting work by Martin Atzmueller)
2282 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2283 with a boatload of patches which he ported to SBCL. Now that those
2285 ** The system tries to make sure that its low-priority messages
2286 are prefixed by semicolons, to help people who like to use
2287 syntax highlighting in their ILISP buffer. (This patch
2288 was originally due to Raymond Toy.)
2289 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2290 patches originally written by Raymond Toy.
2291 ** The compiler understands coercion between single-value and
2292 multiple-VALUES type expressions better, getting rid of some very
2293 weird behavior, thanks to patches originally by Robert MacLachlan
2294 and Douglas Crosher.
2295 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2296 lambda lists, thanks to a patch originally by Pierre Mai.
2297 ** The system no longer bogusly warns about "abbreviated type
2299 ** The compiler gets less confused by inlining and RETURN-FROM,
2300 thanks to some patches originally by Tim Moore.
2301 ** The system no longer hangs when dumping circular lists to fasl
2302 files, thanks to a patch originally from Douglas Crosher.
2303 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2304 undefined function error.
2305 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2306 in structure slot initforms) for now, documented workaround instead:-|
2307 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2308 * fixed bug 5: added stubs for various Gray stream functions called
2309 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2310 installed, at least appropriate type errors are generated
2311 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2312 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2313 consistently in DEFMETHOD forms.
2314 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2315 it doesn't seem to affect SBCL after all
2316 * The C runtime system now builds with better optimization and many
2317 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2319 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2321 * The system is now under CVS at SourceForge (instead of the
2322 CVS repository on my home machine).
2323 * The new signal handling code has been tweaked to treat register
2324 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2325 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2326 downstream, e.g. in debug-int.lisp, has implicit dependencies
2327 on the unsignedness of integer representation of machine words,
2328 and that caused the system to bomb out with infinite regress
2329 when trying to recover from type errors involving signed values,
2330 e.g. (BUTLAST '(1 2 3) -1).)
2331 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2332 (This was one of the bugs Peter Van Eynde reported back in July.)
2333 * The system now uses code inspired by Colin Walters' O(N)
2334 implementation of MAP (from the cmucl-imp@cons.org mailing
2335 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2336 inline the MAP operation, and there is more than one
2337 sequence argument to the MAP call (so that it can't just
2338 do ETYPECASE once and for all based on the type of the
2339 single sequence argument). (The old non-inline implementation
2340 of the general M-argument sequence-of-length-N case required
2341 O(M*N*N) time when any of the sequence arguments were LISTs.)
2342 * The QUIT :UNIX-CODE keyword argument has been renamed to
2343 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2344 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2346 * Raymond Wiker's patches to port dynamic loading from Linux to
2347 FreeBSD have been added.
2348 * The BUGS file is now more nearly up to date, thanks in large part
2349 to Martin Atzmueller's review of it.
2350 * The debugger now flushes standard output streams before it begins
2351 its output ("debugger invoked" and so forth).
2352 * The core version number and fasl file version number have both
2353 been incremented, because of incompatible changes in the layout
2355 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2356 used to not be called for a saved Lisp image.)
2357 * Martin Atzmueller's version of a patch to fix a compiler crash,
2358 as posted on sbcl-devel 13 September 2000, has been installed.
2359 * Instead of installing Martin Atzmueller's patch for the
2360 compiler transform for SUBSEQ, I deleted the compiler transform,
2361 and transforms for some similar consing operations.
2362 * A bug in signal handling which kept TRACE from working on OpenBSD
2364 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2365 away by constant folding
2366 * The system now defines its address space constants in one place
2367 (in the Lisp sources), and propagates them automatically elsewhere
2368 (through GENESIS and the sbcl.h file). Therefore, patching the
2369 address map is less unnecessarily tedious and error-prone. The
2370 Lisp names of address space constants have also been systematized.
2371 * CVS tags like dollar-Header-dollar have been removed from
2372 the sources, because they have never saved me trouble and
2373 they've been source of trouble working with patches and other
2374 diff-related operations.
2375 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2376 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2378 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2380 * The system has been ported to OpenBSD.
2381 * The system now compiles with a simple "sh make.sh" on the systems
2382 that it's supported on. I.e., now you no longer need to tweak
2383 text in the target-features.lisp-expr and symlinks in src/runtime/
2384 by hand, the make.sh takes care of it for you.
2385 * The system is no longer so grossly inefficient when compiling code
2386 involving vectors implemented as general (not simple) vectors (VECTOR T),
2387 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2388 which dares to use the various sequence functions on non-simple
2389 vectors, takes less of a performance hit.
2390 * There is now a primitive type predicate VECTOR-T-P
2391 to test for the (VECTOR T) type, so that e.g.
2392 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2393 can now be compiled with some semblance of efficiency. (The old code
2394 turned the type declaration into a full call to %TYPEP at runtime!)
2395 * AREF on (VECTOR T) is still not fast, since it's still compiled
2396 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2397 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2398 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2399 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2400 without making full calls to SUBTYPEP at runtime.
2401 (Some analogous efficiency issues for non-simple vectors specialized to
2402 element types other than T, or for non-simple multidimensional arrays,
2403 have not been addressed. They could almost certainly be handled the
2404 same way if anyone is motivated to do so.)
2405 * The changes in array handling break binary compatibility, so
2406 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2407 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2408 instead of (VALUES T T).
2409 * By following the instructions that Dan Barlow posted to sbcl-devel
2410 on 2 July 2000, I was able to enable primitive dynamic object
2411 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2412 functionality is not implemented (since it calls ld to resolve
2413 library references automatically, requiring RUN-PROGRAM for its
2414 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2415 try to resolve library references) is now supported.
2416 * The system now flushes the standard output streams when it terminates,
2417 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2418 them at several other probably-convenient times, e.g. in each pass of
2419 the toplevel read-eval-print loop, and after evaluating a form given
2420 as an "--eval" command-line option. (These changes were motivated by a
2421 discussion of stream flushing issues on cmucl-imp in August 2000.)
2422 * The source transform for TYPEP of array types no longer assumes
2423 that an array whose element type is a not-yet-defined type
2424 is implemented as an array of T, but instead punts, so that the
2425 type will be interpreted at runtime.
2426 * There is now some support for cross-compiling in make.sh: each of
2427 the phases of make.sh has its own script. (This should be transparent
2428 to people doing ordinary, non-cross-compile builds.)
2429 * Since my laptop doesn't have hundreds of megabytes of memory like
2430 my desktop machine, I became more motivated to do some items on
2431 my to-do list in order to reduce the size of the system a little:
2432 ** Arrange for various needed-only-at-cold-init things to be
2433 uninterned after cold init. To support this, those things have
2434 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2435 symbols with such names are now uninterned after cold init).
2436 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2437 fasl files for cold load.
2438 ** Remove the old compiler structure pooling code (which used to
2439 be conditional on the target feature :SB-ALLOC) completely.
2440 ** Redo the representation of some data in cold init to be more compact.
2441 (I also looked into supporting byte compiled code at bootstrap time,
2442 which would probably reduce the size of the system a lot, but that
2443 looked too complicated, so I punted for now.)
2444 * The maximum signal nesting depth in the src/runtime/ support code has
2445 been reduced from 4096 to 256. (I don't know any reason for the very
2446 large old value. If the new smaller value turns out to break something,
2447 I'll probably just bump it back up.)
2448 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2450 * Many, many bugs reported by Peter Van Eynde have been added to
2451 the BUGS list; some have even been fixed.
2452 * While enabling dynamic object file loading, I tried to make the
2453 code easier to understand, renaming various functions and variables
2454 with less ambiguous names, and changing some function calling
2455 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2456 * While trying to figure out how to do the OpenBSD port, I tried to
2457 clean up some of the code in src/runtime/. In particular, I dropped
2458 support for non-POSIX signal handling, added various comments,
2459 tweaked the code to reduce the number of compilation warnings, and
2460 renamed some files to increase consistency.
2461 * To support the new automatic configuration functionality in make.sh,
2462 the source file target-features.lisp-expr has been replaced with the
2463 source file base-target-features.lisp-expr and the machine-generated
2464 file local-target-features.lisp-expr.
2465 * fixed a stupid quoting error in make.sh so that using CMU CL
2466 "lisp -batch" as cross-compilation host works again
2468 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2470 * DESCRIBE no longer tries to call itself recursively to describe
2471 bound/fbound values, so that it no longer fails on symbols which are
2472 bound to themselves (like keywords, T, and NIL).
2473 * DESCRIBE now works on generic functions.
2474 * The printer now prints less-screwed-up representations of closures
2475 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2476 * A private symbol is used instead of the :EMPTY keyword previously
2477 used to mark empty slots in hash tables. Thus
2478 (DEFVAR *HT* (MAKE-HASH-TABLE))
2479 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2480 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2481 now does what ANSI says that it should. (You can still get
2482 similar noncompliant behavior if bang on the hash table
2483 implementation with all the symbols you get back from
2484 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2485 This breaks binary compatibility, since tests for equality to
2486 :EMPTY are wired into things like the macroexpansion of
2487 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2489 * There's now a minimal placeholder implementation for CL:STEP,
2490 as required by ANSI.
2491 * An obscure bug in the interaction of the normal compiler, the byte
2492 compiler, inlining, and structure predicates has been patched
2493 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2494 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2496 * Missing ordinary arguments in a macro call are now detected even
2497 when the macro lambda list contains &KEY or &REST.
2498 * The debugger no longer complains about encountering the top of the
2499 stack when you type "FRAME 0" to explicitly instruct it to go to
2500 the top of the stack. And it now prints the frame you request even
2501 if it's the current frame (instead of saying "You are here.").
2502 * As specified by ANSI, the system now always prints keywords
2503 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2505 * The default initial SIZE of HASH-TABLEs is now smaller.
2506 * Type information from CLOS class dispatch is now propagated
2507 into DEFMETHOD bodies, so that e.g.
2508 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2510 is now basically equivalent to
2511 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2512 (DECLARE (TYPE SINGLE-FLOAT X))
2514 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2515 operation, without having to do run-time type dispatch.
2516 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2517 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2518 * Since I don't seem to be making big file reorganizations very often
2519 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2520 is overflowing my ability to conveniently back them up), I've finally
2521 checked the system into CVS. (The CVS repository is on my home system,
2522 not at SourceForge -- putting it on SourceForge might come later.)
2523 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2524 high-level GC-NOTIFY-FOO functions send their output. (There's
2525 still very little control of where low-level verbose GC functions
2526 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2527 less than it used to -- the GC-NOTIFY-FOO functions are now under
2528 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2529 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2530 in only one place in the source code, and propagates it automatically
2531 everywhere that it's needed. Thus e.g. when I bump the version from
2532 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2533 * The C source files now include boilerplate legalese and documentation
2534 at the head of each file (just as the Lisp source files already did).
2535 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2536 to his page will be replaced with a link to his new CLiki service.
2538 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2540 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2541 * The build process now looks for GNU make under the default name "gmake",
2542 instead of "make" as it used to. If GNU make is not available as "gmake"
2543 on your system, you can change this default behavior by setting the
2544 GNUMAKE environment variable.
2545 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2546 can build without error under CMU CL.
2548 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2550 * There is now a partial SBCL user manual (with some new text and some
2551 text cribbed from the CMU CL manual).
2552 * The beginnings of a profiler have been added (starting with the
2553 CMU CL profiler and simplifying and cleaning up). Eventually the
2554 main interface should be through the TRACE macro, but for now,
2555 it's still accessed through vaguely CMU-CL-style functions and macros
2556 exported from the package SB-PROFILE.
2557 * Some problems left over from porting CMU CL to the new
2558 cross-compilation bootstrap process have been cleaned up:
2559 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2560 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2561 part when getting rid of the compiler *BACKEND* stuff.)
2562 ** The value of *NULL-TYPE* was screwed up, because it was
2563 being initialized before the type system knew the final
2564 definition of the 'NULL type. This screwed up several key
2565 optimizations in the compiler, causing inefficiency in all sorts
2566 of places. (I found it because I wanted to understand why
2567 GET-INTERNAL-RUN-TIME was consing.)
2568 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2569 PROCLAIM FTYPE information. Unfortunately this broke binary
2570 compatibility again, since now the forms output by DEFGENERIC
2571 to refer to functions which didn't exist in 0.6.3.
2572 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2573 can use the new (as of 0.6.3) transform for SOME into MAP into
2575 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2576 values to (INTEGER 0 1000000), so that the time code will no longer
2577 occasionally get blown up by Linux returning 1000000 microseconds
2578 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2579 its output conform to the ANSI spec. (Alas, this makes its output
2580 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2581 * A full call to MAP NIL with a single sequence argument no longer conses.
2582 * fixes to problems pointed out by Martin Atzmueller:
2583 * The manual page no longer talks about multiprocessing as though
2584 it were currently supported.
2585 * The ILISP support patches have been removed from the distribution,
2586 because as of version 5.10.1, ILISP now supports SBCL without us
2587 having to maintain patches.
2588 * added a modified version of Raymond Toy's recent CMU CL patch for
2589 EQUALP comparison of HASH-TABLE
2591 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2593 * The system still can't cross-compile itself with
2594 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2595 entails), but at least it can compile more of itself that way
2596 than it used to be able to, and various buglets which were uncovered
2597 by trying to cross-compile itself that way have now been fixed.
2598 * This release breaks binary compatibility again. This time
2599 at least I've incremented the FASL file format version to 2, so that the
2600 problem can be detected reliably instead of just causing weird errors.
2601 * various new style warnings:
2602 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2603 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2604 ** using the lexical binding of a variable named in the *FOO* style
2605 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2606 as specified by ANSI.
2607 * *RANDOM-STATE* is no longer automatically initialized from
2608 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2609 default behavior of the system is to repeat its behavior every time
2610 it's run. If you'd like to change this behavior, you can always
2611 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2612 old convention there was no comparably easy way to get the system to
2613 repeat its behavior every time it was run.
2614 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2615 FTYPE declarations has been removed, in favor of their ANSI
2616 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2617 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2618 the types of their sequence arguments can be determined at compile time.
2619 This is done through a new open code expansion for MAP which eliminates
2620 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2621 argument types can be determined at compile time.
2622 * The optimizer now transforms COERCE into an identity operation when it
2623 can prove that the coerced object is already of the correct type. (This
2624 can be a win for machine generated code, including the output of other
2625 optimization transforms, such as the MAP transform above.)
2626 * Credit information has been moved from source file headers into CREDITS.
2627 * Source file headers have been made more standard.
2628 * The CASE macro now compiles without complaining even when it has
2631 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2633 * (Note that the way that the PCL macroexpansions were rewritten
2634 to accommodate the change in DEFGENERIC below breaks binary
2635 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2636 not run under sbcl-0.6.2. Once we get out of alpha releases,
2637 i.e. hit release 1.0.0, we'll probably try to maintain binary
2638 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2639 and sbcl-1.4.4. Until then, however, it might be fairly common
2640 for maintenance releases to break binary compatibility.)
2641 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2643 * The reporting of unhandled signals has been changed to print some
2644 explanatory text as well as the report form. (Previously only
2645 the report form was printed.)
2646 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2647 it defines, so that the compiler no longer issues undefined function
2648 warnings for compiled-but-not-yet-loaded generic functions.
2649 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2650 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2651 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2652 section of the ANSI spec are supported.
2653 * The "" nickname for the "KEYWORD" package has been removed.
2654 The reader still handles symbol tokens which begin with a package marker
2655 as keywords, but it doesn't expose its mechanism for doing so in the
2656 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2657 * The system now issues STYLE-WARNINGs for contradictory TYPE
2658 proclamations. (Warnings for contradictory FTYPE proclamations would
2659 be nice too, but those can't be done usefully unless the type system
2660 is made smarter about FUNCTION types.)
2661 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2662 systematized, so that "*target-*.lisp is supposed to exist only on the
2663 target and imply that there's a related file which exists on the
2664 host, and *host-*.lisp is supposed to exist only on the host and imply
2665 that there's a related file which exists on the target. This involves a
2666 lot of renaming. Hopefully the acute confusion caused by the renaming
2667 will be justified by the reduction in chronic confusion..
2668 ** runtime-type.lisp -> early-target-type.lisp
2669 ** target-type.lisp -> late-target-type.lisp
2670 ** early-host-format.lisp -> early-format.lisp
2671 ** late-host-format.lisp -> late-format.lisp
2672 ** host-error.lisp -> misc-error.lisp
2673 ** early-error.lisp -> early-target-error.lisp
2674 ** late-error.lisp -> late-target-error.lisp
2675 ** host-defboot.lisp -> early-defboot.lisp
2676 ** code/misc.lisp -> code/target-misc.lisp
2677 ** code/host-misc.lisp -> code/misc.lisp
2678 ** code/numbers.lisp -> code/target-numbers.lisp
2679 ** code/early-numbers.lisp -> numbers.lisp
2680 ** early-host-type.lisp -> early-type.lisp
2681 ** late-host-type.lisp -> late-type.lisp
2682 ** host-typep.lisp -> typep.lisp
2683 ** load.lisp -> target-load.lisp
2684 ** host-load.lisp -> load.lisp
2685 ** host-disassem.lisp -> disassem.lisp
2686 ** host-insts.lisp -> insts.lisp
2687 ** byte-comp.lisp -> target-byte-comp.lisp
2688 ** host-byte-comp.lisp -> byte-comp.lisp
2689 ** host-signal.lisp -> signal.lisp
2690 ** host-defstruct.lisp -> defstruct.lisp
2691 ** late-target-type.lisp -> deftypes-for-target.lisp
2692 Furthermore, several other previously target-only files foo.lisp (e.g.
2693 hash-table.lisp and random.lisp) have been split into a target-and-host
2694 foo.lisp file and a target-only target-foo.lisp file, with their key type
2695 definitions in the target-and-host part, so that the cross-compiler will
2696 know more about target types.
2697 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2698 gone away. In their place are various *BACKEND-FOO* variables
2699 corresponding to the slots of the old structure.
2700 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2701 into the target SBCL has been fixed.
2702 * The chill.lisp system for loading cold code into a running SBCL
2704 * Support for the CMU CL "scavenger hook" extension has been removed.
2705 (It was undocumented and unused in the CMU CL sources that SBCL was
2706 derived from, and stale in sbcl-0.6.1.)
2707 * Various errors in the cross-compiler type system were detected
2708 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2709 (enabling various consistency checks). Many of them were fixed,
2710 but some hard problems remain, so the compiler is back to
2711 running without *TYPE-SYSTEM-INITIALIZED* for now.
2712 * As part of the cross-compiler type system cleanup, I implemented
2713 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2714 * I have started adding UNCROSS calls throughout the type system
2715 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2716 on cross-compiler input files will be able to go away, and various
2718 * CONSTANTP now returns true for quoted forms (as explicitly required
2721 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2723 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2724 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2725 (MAKE-LIST -1). In the longer term, it ought to become true
2726 that declarations are assertions even at SAFETY 1. For now, it's not
2727 quite true even at SAFETY 3, but it's at least more nearly true..
2728 (Note that this change seems to increases the size of the system by
2729 O(5%) and to decrease the speed of the compiler by 20% or more.)
2730 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2731 for the problem of printing dozens of lines of distracting information
2732 about low-level system machinery as part of the top stack frame
2733 on entry to the debugger when an undefined function was called.
2734 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2735 is not reset to COMMON-LISP-USER.
2736 * Compilation of stuff related to dyncount.lisp has been made conditional
2737 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2738 smaller. The various dyncount-related symbols have been moved into
2739 a new "SB-DYNCOUNT" package.
2740 * tty-inspect.lisp has been renamed to inspect.lisp.
2741 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2742 feature has gone away. (When we eventually port to other flavors of
2743 libc and/or Unix, we'll try to make the differences between flavors
2744 invisible at the user level.)
2745 * Various other *FEATURES* tags, and/or their associated conditionals,
2746 have been removed if obsolescent, or given better documentation, or
2747 sometimes given more-mnemonic names.
2749 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2751 * tidied up "make.sh" script
2752 * tidied up system directory structure
2753 * better "clean.sh" behavior
2754 * added doc/FOR-CMUCL-DEVELOPERS
2755 * many many small tweaks to output format, e.g. removing possibly-confusing
2756 trailing #\. character in DESCRIBE-INSTANCE
2757 * (EQUALP #\A 'A) no longer signals an error.
2758 * new hashing code, including EQUALP hashing
2759 * tidied up Lisp initialization and toplevel
2760 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2761 * command line argument processing
2762 * added POSIX-GETENV function to deal with Unix-ish environment variables
2763 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2764 terminating SBCL on EOF
2765 * non-verbose GC by default
2766 * There is no more "sbcl" shell script; the sbcl file is now the C
2767 runtime executable (just like CMU CL).
2768 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2770 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2771 symbol and package information as Lisp native symbol and package objects
2772 instead of strings naming symbols and strings naming packages. This way,
2773 whenever packages are renamed (as in warm init), debug information is
2774 transformed along with everything else.
2775 * tweaked the optimization policy declarations which control the building
2776 of SBCL itself. Now, among other things, the system no longer saves
2777 source location debugging information. (This helps two problems at once
2778 by reducing SBCL size and by keeping SBCL from trying to look for its
2779 sources -- which may not exist -- when reporting errors.)
2780 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2781 debugging and testing purposes
2782 * cleaned up printing, making the printer call PRINT-OBJECT for
2783 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2784 methods, giving nearly-ANSI behavior
2785 * converted almost all special variables to use *FOO* naming convention
2786 * deleted PARSE-TIME functionality, since it can be done portably
2787 * moved some files out of cold init into warm init
2788 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2790 * regularized formatting of source files
2791 * added an install.sh script
2792 * fixed ridiculous memory usage of cross-compiler by making
2793 compiler/alloc.lisp not try to do pooling unless it can hook
2794 itself into the GC of the cross-compilation host. Now the system
2795 builds nicely on my old laptop.
2796 * added :SB-ALLOC in target-features.lisp-expr
2797 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2798 was not implemented)
2799 * re-did condition handling and note reporting in the compiler. Notes
2800 are no longer handled by signalling conditions. Style warnings
2801 and warnings are handled more correctly and reported in such a way
2802 that it's easy to find one or the other in your output (so that you
2803 can e.g. figure out which of many problems caused COMPILE-FILE to
2805 * changed the severity of several compiler warnings from full WARNING
2806 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2807 compiler note reporting so that it doesn't use the condition system
2808 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2809 COMPILE-FILE command)
2810 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2811 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2812 no longer supported, since their implementation was deeply intertwingled
2813 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2814 * removed bogus "support" for compiler macros named (SETF FOO), and
2815 removed the compiler macro for SETF INFO (but only after making a fool
2816 of myself on the cmucl-imp mailing list by posting a bogus patch for
2817 DEFINE-COMPILER-MACRO..)
2818 * Compiled files containing forms which have side effects on the Lisp
2819 reader (such as DEFPACKAGE forms) are now handled more correctly.
2820 (Compiler queuing of top level lambdas has been suppressed by setting
2821 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2822 * deleted various currently-unused source files, e.g. gengc.lisp. They
2823 may be added back at some point e.g. when porting to other architectures,
2824 but until they are it's distracting to distribute them and to try to
2826 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2827 there were so many of them they're just distractions, and UNCROSS is
2828 known to be able to handle the current sources
2829 * moved PROFILE functionality into TRACE, so that it will be clear
2830 how the wrapping and unwrapping of functions when you profile them
2831 interacts with the wrapping and unwrapping of functions when you
2832 trace them. (Actually, the functionality isn't there yet, but at least
2833 the interface specification is there. Hopefully, the functionality will
2834 arrive with some future maintenance release.)
2835 * removed host-oops.lisp
2836 * changed signature of QUIT function to allow UNIX-CODE argument
2837 * fixed READ-SEQUENCE bug
2838 * tweaked verbose GC output so that it looks more like the progress
2839 output that ANSI specifies for functions like LOAD
2840 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2841 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob