1 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2 * new feature: reloading changed shared object files with
3 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
4 * fixed bug #331: structure-class instances corresponding to
5 DEFSTRUCT forms are now created eagerly.
7 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
8 * new feature: a build-time option (controlled by the :SB-UNICODE
9 keyword feature, enabled by default) for building the system with
10 support for the entire 21-bit character space defined by the
12 * new feature: the system now has rudimentary external-format
13 support; the primary user-visible change at this time is that
14 characters with the high bit set (such as lower-case-e-acute) will
15 print correctly to a terminal in a UTF-8 environment.
16 * minor incompatible change: BASE-CHAR no longer names a class;
17 however, CHARACTER continues to do so, as required by ANSI.
18 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
19 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
20 has been moved to the SB-EXT package (temporarily re-exported from
22 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
23 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
25 * fixed bug #351: better error-handlind and reporting for malformed
27 * fixed bug #350: bignum-printing is now more memory-efficient,
28 allowing printing of very large bignums, eg. (expt 2 10000000).
29 (reported by Bruno Haible)
30 * fixed bug #302: better primitive-type selection for intersection
32 * fixed bug #308: non-graphic characters now all have names, as
33 required. (reported by Bruno Haible)
34 * bug fix: redefining a class with different superclasses now correctly
35 removes it from the direct-subclasses of its previous superclasses.
36 (reported by David Morse)
37 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
38 new value now works. (reported by Bruno Haible)
39 * bug fix: correct canonicalization of multiple non-standard slot
40 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
41 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
42 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
43 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
44 now exists, an signals an error.
45 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
46 during macroexpansion for non-iteration variables as well. (reported
47 by Bruno Haible for CMUCL)
48 * bug fix: Cyclic structures and unprintable objects in compiler
49 messages no longer cause errors. (reported by Bruno Haible)
50 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
51 and READ-FROM-STRING all now return a primary value of NIL if
52 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
53 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
54 (reported by Bruno Haible for CMUCL)
55 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
56 do not designate a condition. (reported by Bruno Haible for
58 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
59 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
60 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
61 permitting resolution in favour of any of the conflicting symbols.
62 (reported by Bruno Haible for CMUCL)
63 * bug fix: EQUAL compiler optimizations is less aggressive on
64 strings which can potentially compare true despite having distinct
65 specialized array element types.
66 * bug fix: unit enumerations can be defined without dividing by
67 zero. (thanks to Vincent Arkesteijn)
68 * FORMAT compile-time argument count checking has been enhanced.
69 (report from Bruno Haible for CMUCL)
70 * a partial workaround for the bug 262: the compiler does not try to
71 inline-expand a local function doing RETURN-FROM from a deleted
72 BLOCK. (thanks to Peter Denno for the bug report and to David
73 Wragg for the simple test case)
74 * fixed some bugs revealed by Paul Dietz' test suite:
75 ** RENAME-PACKAGE allows all package designators as new package
77 ** constraint propagation and lambda variable substitution are
78 more cautious in dealing with partially deleted code.
79 ** compiler sometimes lost reoptimization passes.
80 ** CERROR, given a condition as condition designator, passes its
81 remaining arguments to the continue format control without
83 ** Case-altering FORMAT directives work correctly on non-ASCII
85 ** The REAL type specifier handles bounds outside the floating
86 point ranges without signalling FLOATING-POINT-OVERFLOW.
87 ** Functions with IR1-transformations can create intercomponent
88 references to global functions.
89 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
91 ** FORMAT ~R treats a nil value for its first parameter correctly.
93 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
94 * enhancement: saving cores with foreign code loaded is now
95 supported on x86/NetBSD and sparc/Linux in addition to the previously
97 * bug fix: on some platforms repeated installations caused multiple
98 copies of HTML documentation to be installed -- should not happen
99 any more. (reported by Stefan Scholl)
100 * bug fix: parsing self-recursive alien record types multiple times
101 no longer causes infinite recursion. (reported by Thomas F. Burdick,
102 original patch by Helmut Eller for CMUCL)
103 * bug fix: stack-exhaustion detection works now on NetBSD as well.
104 (thanks to Richard Kreuter)
105 * bug fix: defining classes whose accessors are methods on existing
106 generic functions in other (locked) packages no longer signals
107 bogus package lock violations. (reported by François-René Rideau)
108 * bug fix: special variables as DEFMETHOD parameters no longer have
109 associated bogus type declarations. (reported by David Wragg and
111 * bug fix: read-write consistency on streams of element-type
112 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
113 * bug fix: redefiniton of the only method of a generic function with
114 no DEFGENERIC no longer emits a full WARNING. In addition,
115 redefinition of generic functions with no DEFGENERIC to an
116 incompatible lambda list now signals an error. (thanks to Zach
118 * bug fix: DEFGENERIC now works even when there's a function of the
119 same name in an enclosing lexical environment. (thanks to Zach
121 * fixed compiler failure, caused by instrumenting code during
122 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
123 * optimization: added loop analysis and improved register allocation
124 * fixed some bugs revealed by Paul Dietz' test suite:
125 ** POSITION on displaced vectors with non-zero displacement
126 returns the right answer.
127 ** (SIMPLE-STRING) is a valid type specifier for sequence
129 ** *PRINT-LEVEL* handling for slotless structures is pedantically
131 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
132 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
134 ** The justification version of the FORMAT ~< directive treats
135 non-zero minpad parameter correctly.
137 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
138 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
139 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
140 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
141 the supported interface.
142 * new feature: Single-stepping of code compiled with DEBUG 2 or
143 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
144 * new feature: saving cores with foreign code loaded is now
145 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
146 Timothy Moore's work for CMUCL)
147 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
148 parameters to * instead of NIL if no initform is supplied.
149 (reported by Johan Bockgaard)
150 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
151 correctly when FUNCALL appears as the car of the form. Note:
152 despite this FUNCALL forms are not currently subject to
153 compiler-macro expansion. (port of Raymond Toy's fix for the
154 same from CMUCL, reported by Johan Bockgaard)
155 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
156 (thanks for Teemu Kalvas)
157 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
158 case instead of terminating the iteration (thanks for Julian
159 Squires, Sean Champ and Raymond Toy)
160 * bug fix: incorrect expansion of defgeneric that caused a style
161 warning. (thanks for Zach Beane)
162 * bug fix: slot accessor effective method computation works properly
163 for classes with multiple non-standard applicable methods on
164 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
165 * on x86 compiler supports stack allocation of results of LIST and
166 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
167 CMUCL implementation by Gerd Moellmann)
168 * fixed some bugs revealed by Paul Dietz' test suite:
169 ** FORMAT strings with both the ~<~:;~> form of the justification
170 directive and pretty-printing directives cause an error.
172 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
173 * incompatible change: the internal functions
174 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
175 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
176 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
177 instead of the old functions.
178 * new feature: on platforms where "dladdr" is available foreign
179 function names now appear in backtraces. (based on Helmut Eller's
181 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
182 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
184 * documentation: ASDF manual is now distributed with SBCL.
185 * bug fix: SBCL can now load its contributed modules with REQUIRE
186 even if the system-provided entries have been removed from
187 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
189 * bug fix: backtraces involving undefined functions or assembly
190 routines are more informative. (thanks to Brian Downing)
191 * bug fix: mutually referent alien structures now work correctly.
192 (reported by Rick Taube)
193 * bug fix: structures defined by WITH-ALIEN can be referred to
194 within other definitions in the same WITH-ALIEN.
195 * bug fix: division operators (MOD, TRUNCATE and the like) with
196 constant zero divisors and integer dividends no longer generate
198 * bug fix: provide default methods for INPUT-STREAM-P and
199 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
200 * bug fix: improve the detection and resolution of MOP metacycles
201 (where effective method computation depends on the generic function
202 having its effective method computed). (reported by Bruno Haible)
203 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
204 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
205 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
206 * bug fix: pathnames with relative directory components can be
207 represented relative to default pathnames.
208 * optimization: in taking the GCD of bignums, reduce the two bignums
209 to approximately the same size (using Euclid's algorithm) before
210 applying the more sophisticated binary GCD. (thanks to Juho
212 * optimization: COUNT on bitvectors now operates word-at-a-time.
213 * optimization: ASH with a positive, but not necessarily constant,
214 (leftwards) shift, when in a modular context, is compiled to a
216 * fixed some bugs revealed by Paul Dietz' test suite:
217 ** FORMAT variable parameters ("~V<char>") are defaulted properly
218 if the corresponding argument is NIL.
219 ** FORMAT directives accepting a minpad parameter treat negative
221 ** literal commas as character parameters to FORMAT directives are
223 ** literal spaces directly after ~<Newline> directives within a
224 format-logical-block (~:< ~@:>) do not induce :FILL-style
225 conditional newlines.
226 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
227 as specified: it no longer includes conditional newlines).
228 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
230 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
231 :PER-LINE-PREFIX argument does not evaluate to a string.
233 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
234 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
235 the manual for details; for now, package locks can be disabled by
236 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
237 if no major problems are found then it is likely that they will be
238 compiled in unconditionally.
239 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
240 unsupported operators on all platforms. To load a shared library
241 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
242 object file, link it into a shared library outside of SBCL and
243 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
245 * bug workaround: ROOM T and the GC were not working together
246 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
247 reporting facility aren't preserved. That reporting has been
248 disabled, so now until and unless someone figures out how to make it
249 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
250 * minor incompatible change: as threatened around sbcl-0.8.0, the
251 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
252 The symbols are also exported from SB-PCL for backwards
253 compatibility, but more so than before SB-PCL should be treated as
254 an implementation-internal package.
255 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
257 * the SB-SPROF contrib now works on (most) non-x86 architectures.
258 It is known as of this release not to work on the Alpha, however.
259 * fixed bug #167: errors signalled due to illegal syntax in method
260 bodies are now more legible.
261 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
262 designators and can hence be used with TYPEP.
263 * fixed bug #333: CHECK-TYPE now ensures that the type error
264 signalled, if any, has the right object to be accessed by
265 TYPE-ERROR-DATUM. (reported by Tony Martinez)
266 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
267 5.1.2.3 for multiple-value place subforms. (reported by Kalle
269 * fixed bug #334: programmatic addition of slots using specialized
270 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
271 and :ALLOCATION :CLASS slots.
272 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
273 float exponents. (rereported by Peter Seibel)
274 * fixed a bug: #\Space (and other whitespace characters) are no
275 longer considered to be macro characters in standard syntax by
277 * fixed bug: initialization of condition class metaobjects no longer
278 causes an instance of the condition to be created. (reported by
280 * fixed bug: it is now possible to have slots such that
281 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
282 description is neither :INSTANCE nor :CLASS.
283 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
284 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
285 SLOT-VALUE-USING-CLASS), and no longer causes errors with
286 non-standard SLOT-DEFINITION-ALLOCATION values.
287 * fixed bugs: various race conditions handling exiting threads.
288 CL-PPCRE's thread torture test now passes.
289 * fixed arguably-a-bug: GC time proportional to number of threads in
290 system even when most of them are idle
291 * optimization: improved performance of BIT and SBIT on bit-vectors.
292 * .fasl file incompatibility: The fasl file version number has
293 been incremented because of changes associated with package locks.
295 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
296 * minor incompatible change: the system no longer provides
297 optimization hints (conditions which are TYPEP
298 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
299 mode; these hints are still emitted when higher SPEED optimization
301 * new contrib module: a sampling profiler (profiling by statistical
302 sampling, rather than by instrumenting functions) is available as
303 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
304 * the behaviour of the standard function ED is now customizeable by
305 third parties through a hook variable: see ED's documentation
306 string for information on the protocol.
307 * the compiler no longer emits efficiency notes for (FUNCALL X)
308 when the type of X is uncertain under default optimization
310 * fixed bug 276: mutating a binding of a specialized parameter to a
311 method to something that is not TYPEP the specializer is now
313 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
314 exhibit the required behaviour on the x86 platform. (thanks to
315 Peter van Eynde, Eric Marsden and Bruno Haible)
316 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
317 even for difficult arguments. (reported by Peter Graves)
318 * fixed bug 141a: the backquote printer now descends quoted
320 * fixed another bug in backquote printing: no more destructive
321 modification of the form's list structure. (reported by Brian
323 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
324 move between its address being taken and the call to
325 interrupt_thread, fixing a crashing race condition.
326 * the SB-POSIX contrib implementation has been adjusted so that it
327 no longer exhibits ridiculously poor performance when constructing
328 instances corresponding to C structs.
330 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
331 * minor incompatible change: the sb-grovel contrib now treats C
332 structures as alien (in the sense of SB-ALIEN) objects rather than
333 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
334 has implications for memory management of client code
335 (sb-grovel-returned objects must now be manually managed) and for
336 type safety (alien objects now have full types).
337 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
338 used to control emission of compiler diagnostics, rather than the
339 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
340 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
341 quality should be considered deprecated.
342 * install.sh now installs the user manual as well
343 * (not quite a new documentable feature, but worth considering in
344 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
345 beginnings of a semantically meaningful condition hierarchy is
346 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
348 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
349 designator as the defaults argument.
350 * fixed bug: Displaced arrays whose displaced-to array has become
351 too small now cause ARRAY-DIMENSION to signal an error, providing
352 for safer bounds-checking. (reported by Bruno Haible)
353 * fixed bug: DEFCLASS slot definitions with identical :READER and
354 :WRITER names now signal a reasonable error. (reported by Thomas
356 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
358 * fixed bug: Invalid slot specification errors now print correctly.
359 (thanks to Zach Beane)
360 * fixed bug 320: Shared to local slot value transfers in class
361 redefinitions now happen corrently from superclasses as
362 well. (reported by Bruno Haible)
363 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
365 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
366 improper lists correctly. (reported by Bruno Haible)
367 * fixed bug 313: source-transform for <fun-name> was erroneously
368 applied to a call of a value of a variable with name <fun-name>.
369 (reported by Antonio Menezes Leitao)
370 * fixed bug 307: The obsolete instance protocol ensures that
371 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
372 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
373 remove dead unknown-values globs from the middle of the stack.
374 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
375 easier to resume long computations after using *BREAK-ON-SIGNALS*
376 to diagnose and fix failures. (thanks to Nikodemus Siivola)
377 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
378 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
379 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
380 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
381 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
383 * fixed a (fairly theoretical) bug in string printing: if
384 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
385 does not have array-element-type equal to the most general string
387 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
388 function name. (thanks to Nikodemus Siivola)
389 * fixed bug: compile-time format string checker failed on
390 non-closed ~{. (reported by Thomas F Burdick)
391 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
392 don't warn on legal loop code involving more than one aggregate
393 boolean. (thanks to Nikodemus Siivola)
394 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
395 in keyword position to MAKE-PACKAGE was not being reported as
397 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
398 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
400 * fixed bugs 280 and 312: the checking for multiple definitions in a
401 file is less likely to become confused by uses of inline
403 * fixed bug: the #S reader macro performs the keyword coercion
404 specified for slot names. (reported by Kalle Niemitalo)
405 * fixed bug: lambda lists may contain symbols whose names start with
406 & but are not lambda-list-keywords; their occurrence triggers a
408 * fixed bug 321: define-method-combination argument lambda lists do
409 not cause invalid code to be generated when &optional or &aux
410 variables have default values. (reported by Bruno Haible)
411 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
412 have CLOS classes; this appears also to have fixed reported
413 instability in stack exhaustion detection.
414 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
415 MUFFLE-WARNING when no associated restart is present is now
417 * optimization: rearranged the expansion of various defining macros
418 so that each expands into only one top-level form in a
419 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
421 * optimization: used a previously little-used slot in symbols to
422 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
424 * fixed some bugs revealed by Paul Dietz' test suite:
425 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
427 ** Fixed an optimization bug related to inheritance of initforms
428 from local to shared slots.
429 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
430 ** CAST optimizer forgot to flush argument derived type.
431 ** print/read consistency on floats is now orders of magnitude
432 more likely. (thanks also to Bruno Haible for a similar report
434 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
435 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
436 for objects of type REAL. Make it so.
437 ** SXHASH is specified (infelicitously) to respect similarity,
438 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
439 it so. (thanks to Markus Ziegler)
440 ** on the Alpha, the compiler succeeds in compiling functions
441 returning a known number of arguments greater than 63.
442 ** fixed handling of invalid NIL arguments in keyword position
444 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
446 ** provided a readably-printable representation for RANDOM-STATE
448 ** ensured that pathnames, where they have a namestring, always
449 print using #P"..." syntax.
451 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
452 * Support for the forthcoming 2.0 version of the NetBSD kernel
453 running on x86 hardware has been added. (thanks to Perry
454 E. Metzger most immediately, and others for their past work)
455 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
456 current release version; SBCL's OpenBSD support had been broken
457 since about the time of OpenBSD's switch to ELF binary format.)
458 * [placeholder for DX summary]
459 ** user code with &REST lists declared dynamic-extent, under high
460 speed or space and low safety and debug optimization policy.
461 * The manual has been converted to Texinfo format and the debugger
462 chapter from the cmucl manual has been added.
463 * A facility has been added to extract documentation strings from
464 sbcl and store them as Texinfo-formatted snippets for inclusion in
465 the manual (via Texinfo's @include directive)
466 * bug fix: compiler emitted division in optimized DEREF. (thanks for
467 the test case to Dave Roberts)
468 * bug fix: multidimensional simple arrays loaded from FASLs had fill
469 pointers. (reported by Sean Ross)
470 * bug fix: PROFILE output is printed nicely even for large numerical
471 values. (thanks to Zach Beane)
472 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
473 greater than 32 handle EOF correctly.
474 * bug fix: on X86 an immediate argument of the IMUL instruction is
475 correctly printed in disassembly. (thanks to Lutz Euler)
476 * bug fix: class slots in redefined classes preserve their old
477 values. (thanks to Bruno Haible and Nikodemus Siivola)
478 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
479 succeeds. (reported by Marco Baringer)
480 * fixed some bugs revealed by Paul Dietz' test suite:
481 ** READ-SEQUENCE now works on ECHO-STREAMs.
482 ** RATIONALIZE works more according to its specification. (thanks
485 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
486 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
487 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
488 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
489 should matter to you only if you rebind the printer control
490 variables and then find you want different bindings in the
491 debugger than in the ordinary execution of your program.)
492 * The runtime build system has been tweaked to support building
493 (on SPARC/SunOS) using a C compiler which invokes Sun's own
494 assembler and linker. (thanks to Nikodemus Siivola)
495 * Unbound, undefined, undeclared variables now trigger full
496 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
497 more often programmer error than deliberate exploitation of undefined
499 * optimization: the hash algorithm for strings has changed to one
500 that is less vulnerable to spurious collisions. (thanks to Juho
502 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
503 needless bounds checking. (thanks to Juho Snellman)
504 * optimization: implemented multiplication as a modular
505 (UNSIGNED-BYTE 32) operation on the PPC backend.
506 * fixed some bugs revealed by Paul Dietz' test suite:
507 ** ADJUST-ARRAY now copies the datum in a zero rank array if
509 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
510 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
511 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
513 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
515 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
517 ** ECHO-STREAMs no longer attempt to echo the end of file value to
518 their output stream on EOF from read.
519 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
520 have been read to end-of-file.
521 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
523 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
524 description of determination of which consecutive characters
526 ** Printing the "Space" character with escaping on now yields "#\\ ",
527 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
528 ** Reading floating-point numbers with *READ-BASE* set to a number
529 less than 10 works correctly.
530 ** Reading floating-point numbers with *READ-BASE* set to a number
531 more than 10 works correctly.
532 ** Printing with *PRINT-READABLY* targets the standard readtable, not
533 the readtable currently in effect.
535 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
536 * minor incompatible change: parsing of namestrings on a physical
537 (Unix) host has changed; numbers after the final #\. in a
538 namestring are no longer interpreted as a version field. This is
539 intented to be largely invisible to the user, except that the
540 meaning of the namestring "*.*.*" has changed: it now refers to a
541 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
542 should usually be replaced by
543 (make-pathname :name :wild :type :wild :version :wild)
544 with the added benefit that this is more likely to be portable.
545 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
546 option to OPEN now signals an error if the file being opened
547 exists; this may have an impact on existing code.
548 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
549 signals received, on the PowerPC platforms (both Linux and
550 Darwin). (thanks to Pierre Mai for pointing out the location of
552 * several fixes on OS X: The system now builds and runs cleanly on
553 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
554 (thanks to Brian Mastenbrook)
555 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
556 second-resolution: integer multiples of 1/3600 between -24 and 24.
557 (thanks to Vincent Arkesteijn)
558 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
559 type when called with 1 argument; PEEK-CHAR checked type of
560 PEEK-TYPE only after having read first character from a
561 stream. (reported by Peter Graves)
562 * bug fix: the garbage collector now has much better locality
563 behaviour, and in particular no longer treats all memory as being
564 exhausted when many small objects point to each other in a deeply
566 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
567 recognized as being TYPEP their class.
568 * bug fix: the PUSHNEW documentation string has been corrected.
569 (thanks to Vincent Arkesteijn)
570 * bug fix: defaulting of the value for the last of an atypically
571 large number of multiple values being bound was not being
572 performed correctly on the Alpha or PPC platforms
573 * optimization: implemented multiplication as a modular
574 (UNSIGNED-BYTE 32) operation on the x86 backend.
575 * optimization: SEARCH on simple-base-strings can now be open-coded.
576 (see also contrib/compiler-extras.lisp for inspiration for
577 teaching the compiler about the Boyer-Moore algorithm).
578 * value, returned by MAX (and MIN) called with several EQUALP, but
579 not EQL, arguments now does not depend on compiler settings.
580 * fixed some bugs revealed by Paul Dietz' test suite:
581 ** in stack analysis liveness information is propagated from
582 non-local entry points.
583 ** pathwise CAST removing failed when the CAST node did not start
585 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
586 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
588 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
589 host is already defined.
590 ** RENAME-FILE works on streams instead of signalling an internal
592 ** PEEK-CHAR uses the current readtable when determining whether
593 or not a character is whitespace.
594 ** MERGE-PATHNAMES handles the case when the pathname does not
595 specify a name while the default-pathname specifies a version.
596 ** Pathnames now stand a better chance of respecting print/read
598 ** Attempting to use standardized file system operators with a
599 pathname with invalid :DIRECTORY components signals a
601 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
603 ** DIRECTORY on logical pathnames is more correct.
604 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
605 signal a TYPE-ERROR if their argument is not a stream.
606 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
607 designator argument does not designate a stream.
608 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
609 examining the synonym.
610 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
612 ** OPEN and WITH-OPEN-STREAM allow opening streams with
613 element-type larger than ([UN]SIGNED-BYTE 32).
615 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
616 * When built with the :SB-FUTEX feature, threaded builds now take
617 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
618 for faster/more reliable mutex and condition variable support.
619 * Incompatible change (but one you probably shouldn't have been using
620 anyway): the interface and code for arbitrating between multiple
621 threads in the same user session has been redesigned.
622 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
623 error when called without an explicit environment argument.
624 (thanks to Nikodemus Siivola)
625 * bug fix: buffered :DIRECTION :IO streams are less likely to become
626 confused about their position. (thanks to Adam Warner and Gerd
628 * bug fix: Pretty printing backquoted forms with unquotations in the
629 argument list position of various code constructs such as LAMBDA
630 now works correctly. (reported by Paul Dietz)
631 * bug fix: Pretty printing unquotations no longer loses all
632 stream position information.
633 * optimization: performance of string output streams is now less
634 poor for multiple small sequence writes.
635 * optimization: performance of CSUBTYPEP in the presence of complex
636 expressions involving CONS and NOT many times has been improved.
637 (reported by Paul Dietz)
638 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
640 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
641 streams for the REPL, for the debugger, and so on. (thanks to
643 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
644 * fixed some bugs revealed by Paul Dietz' test suite:
645 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
646 is ignored if it is too small, rather than propagating through
647 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
648 ** extremely complex negations of CONS types were not being
649 sufficiently canonicalized, leading to inconsistencies in
651 ** VALUES tranformer lost derived type.
653 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
654 * fixed a bootstrapping bug: the build process no longer assumes
655 that the various BOOLE-related constants have the same value in
656 host and target lisps. (noted by Paul Dietz' test suite on an
657 SBCL binary built from CLISP)
658 * The system can now be dynamically linked on the MIPS platform,
659 which enables dynamic loading of foreign code from Lisp. (thanks
660 to Ralf Baechle for discussions on the MIPS ABI)
661 * The system now records debugging information for its own source
662 files in a filesystem-position-independent manner, relative to
663 the "SYS" logical host.
664 * fixed a compiler bug: MV-LET convertion did not check references
665 to the "max args" entry point. (reported by Brian Downing)
666 * tweaked disassembly notes to be less confident about proclaiming
667 some instruction as an LRA. (thanks to Brian Downing)
668 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
669 now each have their own history, command character, and other
670 characteristics. (thanks to David Lichteblau)
671 * fixed some bugs revealed by Paul Dietz' test suite:
672 ** compiler failure in compiling LOGAND expressions including a
674 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
675 shift greater than 32.
676 ** FLUSH-DEST did not mark blocks for type check regeneration.
677 ** HANDLER-CASE failed to accept declarations in handler clauses
678 in some circumstances.
680 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
681 * New code in contrib/sb-introspect (still probably not entirely
682 stable yet) provides some support for smart Lisp development
683 environments like SLIME.
684 * The conditions signalled for errors occurring when loading .fasl
685 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
686 in a way which should help ASDF recover gracefully.
687 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
688 stuff has been cleaned up. If you code contrib/ stuff, this might
689 affect you, and you can look at contrib/README, contrib/STANDARDS,
690 and/or the 0.8.4.27 diff to check.
691 * In full calls the compiler now does not generate checks for declared
692 argument types for all arguments.
693 * various threading fixes
694 ** and some experimental patches which didn't make it into
695 the main tree for this release, but which are shipped in
696 contrib/experimental-thread.patch as a possible fix for some
697 failures (deadlock, spinning...) in GC-intensive multithreaded
699 * fixed PPC build problem (source code incompatibility of different
700 library versions): added offsetof() hackery which attempts to divine
701 where glibc maintainers put uc_mcontext today
702 * fixed bug 282: compiler does not trust type assertions while passing
703 arguments to a full call.
704 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
705 &OPTIONAL VALUES type parameter.
706 * fix bug 214: algorithm for noting rejected templates is now more
707 similar to that of template seletion. (also reported by rydis on
709 * fixed bug 141b: printing backquoted information readably and prettily
710 inserts a space where necessary.
711 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
712 (CHARACTER 10) are now reported as errors, rather than propagated
713 as unknown types. (reported by piso on #lisp)
714 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
715 with values NIL and :ERROR. (thanks to Milan Zamazal)
716 * fixed bug 191c: CLOS now does proper keyword argument checking as
717 described in CLHS 7.6.5 and 7.6.5.1.
718 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
719 longer raise an error; further, using a list as a for-as-arithmetic
720 counter now raises a meaningful error.
721 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
722 now understood by sequence creation functions such as MAKE-SEQUENCE
724 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
725 a STRING-INPUT-STREAM.
726 * compiler enhancement: SIGNUM is now better able to derive the type
728 * type declarations inside WITH-SLOTS are checked. (reported by
730 * fixed some bugs revealed by Paul Dietz' test suite:
731 ** incorrect optimization of TRUNCATE for a positive first
732 argument and negative second.
733 ** compiler failure in let-convertion during flushing dead code.
734 ** compiler failure while deriving type of TRUNCATE on an
735 interval, containing 0.
736 ** ASH of a negative bignum by a negative bignum count now returns
738 ** intersection of CONS types now canonicalizes properly, fixing
739 inconsistencies in SUBTYPEP.
741 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
742 * incompatible change: The --disable-debugger command line
743 option now clobbers the debugger at a more fundamental
744 level, by redefining #'INVOKE-DEBUGGER instead of by
745 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
746 is specified by ANSI to ignore *DEBUGGER-HOOK* and
747 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
748 enter the debugger REPL and then suffer recursive errors
749 because *DEBUG-IO* is also messed up in --disable-debugger mode;
750 while under the new system, BREAK in --disable-debugger mode
751 terminates the system just as an unhandled error would.
752 * fixed compiler performance when processing loops with a step >1;
753 * bug fix: DOCUMENTATION now retrieves generic function
754 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
755 support has been systematized, and now supports the methods
756 specified by ANSI, along with a default method and a method for
757 slot documentation. (reported by Nathan Froyd)
758 * bug fix: effective methods associated with a generic function are
759 no longer cached over a change of that generic function's method
760 combination. (reported by Andreas Fuchs)
761 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
762 names a non-existent directory. (thanks to Andreas Fuchs)
763 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
764 the PPC platform, lead to stack corruption; this has been fixed.
765 (reported by Rainer Joswig)
766 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
767 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
768 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
769 arithmetic in general, is now much more reliable on the PPC
771 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
772 platform now returns the right answer.
773 * optimization: restored some effective method precomputation in
774 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
775 precomputation is now tunable.
776 * optimization: compiler-internal data structure use has been
777 reviewed, and changes have been made that should improve the
778 performance of the compiler by about 20%.
779 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
780 simple-base-strings and simple-bit-vectors is improved.
781 * optimization: the optimization of 32-bit logical and arithmetic
782 functions introduced in version 0.8.3 on the x86 has been
783 implemented on the mips, ppc and sparc platforms; an
784 implementation of the same facility, but for 64-bit arithmetic,
785 has been added for the alpha.
786 * microoptimization: the compiler is better able to make use of the
787 x86 LEA instruction for multiplication by constants.
788 * bug fix: in some situations compiler did not report usage of
789 generic arithmetic in (SPEED 3) policy.
790 * bug 145b fix: compiler used wrong type specifier while converting
791 MEMBER-types to numeric.
792 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
794 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
795 index argument now works properly on the Alpha platform.
796 * bug fix: floating point exception treatment on the Alpha platform
798 * bug fix: FILE-POSITION works much better on string input and
799 output streams. (thanks to Nikodemus Siivola)
800 * bug fix: many threading/garbage collection symptoms sorted.
801 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
802 might be pseudo-atomic.
803 * internal change: Stopping for GC is now done with signals not
804 ptrace. GC is now done in whichever thread wanted it, instead of
806 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
808 * bug fix: result form in DO is not contained in the implicit
810 * incompatible change: ICR structure is changed; the value part of
811 CONTINUATION is now called LVAR; corresponding functions are
812 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
813 * added type deriver for ISQRT (thanks to Robert E. Brown).
814 * bug fix: better support for loading from the command line when an
815 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
817 * fixed some bugs revealed by Paul Dietz' test suite:
818 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
819 ** ROUND and FROUND now give the right answer when given very
820 small float arguments.
821 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
823 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
824 ** (GCD 0 <negative-integer>) returned <negative-integer>.
825 ** LCM should return a non-negative integer.
826 ** PARSE-INTEGER returned the index of a terminator instead of the
827 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
828 ** PARSE-INTEGER returned an incorrect index being applied to a
830 ** LCM with two arguments of 0 returns 0 rather than signalling
832 ** unsigned addition of a 32-bit constant with the high bit set no
833 longer causes an internal compiler error on the x86.
834 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
835 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
836 with negative last argument.
837 ** byte specifiers with zero size and position no longer cause
838 an error during type derivation.
839 ** bignum multiplication on the Alpha platform now returns the
841 * porting: The system now builds on SuSE AMD64, although it still
842 generates a 32-bit binary.
843 * .fasl file incompatibility: The fasl file version number has
844 been incremented (because of the changes to internal compiler
845 data structures referred to above).
847 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
848 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
849 more accurately, on the Darwin kernel running on PowerPC hardware.
850 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
851 * Compiler code deletion notes now signal a condition of type
852 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
853 an associated MUFFLE-WARNING restart.
854 * The compiler now performs limited argument count validation of
855 constant format strings in FORMAT, and where appropriate in ERROR,
856 CERROR and WARN. (thanks to Gerd Moellmann)
857 * New ASDF-INSTALL contrib can be used for automatic download and
858 installation of third-party Lisp code from CCLAN or other sites
860 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
861 which forces another thread to execute a function supplied by the
863 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
864 now accept and act upon their :ELEMENT-TYPE keyword argument.
865 (reported by Martin Atzmueller, Edi Weitz)
866 * bug fix: FILE-POSITION now accepts position designators up to
867 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
868 is the greater. (thanks to Patrik Nordebo)
869 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
870 Kalle Olavi Niemitalo)
871 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
872 printable. (reported by Eric Marsden)
873 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
874 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
875 * x86 bug fix in control stack exhaustion checking: now shows backtrace
876 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
877 (thanks to Stig Sandoe)
878 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
879 UNREAD-CHAR now correctly includes the unread character in the
880 target sequence. (thanks to Gerd Moellmann)
881 * bug fix in threaded builds: the system can now be suspended and
882 resumed by shell job control with minimal disruption.
883 * bug fixes in times and timezones >2038AD
884 * better handling of "where is GNU make?" problem in build scripts
885 (thanks to Nikodemus Siivola)
886 * new optimization: inside a named function any reference to a
887 function with the same name is considered to be a self-reference;
888 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
889 optimization quality.
890 * new optimization on x86: logical functions and + now have
891 optimized (UNSIGNED-BYTE 32) versions, which are automatically
892 used when the result is truncated to 32 bits.
893 * VALUES declaration is partially enabled.
894 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
895 * fixed some bugs revealed by Paul Dietz' test suite:
896 ** The system now obeys the constraint imposed by
897 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
898 types form a lattice under type intersection.
899 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
900 ** ASSOC now ignores NIL elements in an alist.
901 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
902 and (1+ MOST-POSITIVE-FIXNUM) answers.
903 ** The addition of a method with invalid qualifiers to a generic
904 function does not cause an error to be signalled immediately;
905 a warning is signalled, and the error is generated only on
906 calling the generic function.
907 * changed .fasl file version number, in part to add type codes for
908 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
909 obscure ANSI requirements
911 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
912 * fixed bug 148: failure to inline-expand a local function left
913 garbage, confusing the compiler.
914 * fixed bugs 3cd: structure slot readers perform type check if the
915 slot can have an invalid value (i.e. it is either not initialized
916 or can be written with a less specific slot writer).
917 * bug fix: the compiler now traps array references to elements off
918 the end of an array; previously, the bounds checking in some
919 circumstances could go off-by-one.
920 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
922 * type declarations for array element types now obey the description
923 on the CLHS page "Declaration TYPE", as per discussions on
924 sbcl-help around 2003-05-08. This means that a declaration
925 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
926 declaration, all references to BAR will be asserted or assumed
927 (with THE, so dependent on compiler policy) to involve objects of
928 type FOO. Note that no such declaration is implied in
929 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
930 * declared types of functions from the "Conditions"
931 chapter. (reported by Paul Dietz)
932 * bug fix: CERROR accepts a function as its first argument.
933 * bug fix: NTH an NTHCDR accept a bignum as index
934 arguments. (reported by Adam Warner)
935 * optimization: character compare routines now optimize comparing
936 against a constant character. (reported by Gilbert Baumann)
937 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
938 argument now works properly on the MIPS platform.
939 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
940 * repeated evaluation of the same DEFSTRUCT, a slot of which is
941 declared to have a functional type, does not cause an error
943 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
944 many :DEFAULT-INITARGS. (reported by Istvan Marko)
945 * fixed bug: if last continuation of a deleted block has a
946 destination, this destination should be deleted too. (reported by
948 * fixed a bug in the bootstrap process: the host compiler's values
949 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
950 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
951 test case from Patrik Nordebo)
952 * improved the ability of the disassembler on the PPC platform to
953 provide helpful disassembly notes.
954 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
955 the class in more cases than previously.
956 * bug fix: FILE-POSITION now understands :START and :END for
957 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
958 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
959 8) data. (thanks to David Lichteblau)
960 * bug fix: it is possible to add a method to a generic function
962 * bug fix: reader failed to signal END-OF-FILE inside an
963 object representation. (reported by Nikodemus Siivola)
964 * fixed some bugs revealed by Paul Dietz' test suite:
965 ** LAST and [N]BUTLAST should accept a bignum.
966 ** condition slot accessors are methods.
967 ** (VECTOR NIL) is a subtype of STRING.
969 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
970 * minor incompatible change: some nonsensical specialized lambda
971 lists (used in DEFMETHOD) which were previously ignored now signal
973 * minor incompatible change: the system is now aware of the types of
974 variables in the COMMON-LISP package, and will signal errors for
975 most violations of these type constraints (where previously they
976 were silently accepted).
977 * minor incompatible change: COMPILE-FILE now uses the freedom
978 afforded (ANSI 3.2.2.3) to use derived function types for
979 functions defined in the same file. This also permits the system
980 to warn on static type mismatches and function
981 redefinition. (Currently it does not work with high DEBUG level.)
982 * minor incompatible change: VALUES declaration is disabled.
983 * When issuing notes, the compiler now signals a condition of type
984 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
985 restart for use in user handlers. It is expected that the
986 COMPILER-NOTE condition will eventually become a condition
987 supertype to a hierarchy of note types, which will then be
988 handleable in a similar fashion. However, at the moment, no such
989 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
990 but it's an implementation detail, not a classification for the
992 * Changes in type checking closed the following bugs:
993 ** type checking of unused values (192b, 194d, 203);
994 ** template selection based on unsafe type assertions (192c, 236);
995 ** type checking in branches (194bc).
996 * A short form of VALUES type specifier has ANSI meaning (it has
997 increased the number of situations when SBCL cannot perform type
999 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1000 keywords or constants is permissible.
1001 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1002 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1003 classes. (thanks to Antonio Martinez)
1004 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1005 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1006 argument to a pretty-print stream. Instead, it leaves any such
1007 operations to DESCRIBE-OBJECT methods.
1008 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1010 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1011 lambda lists. (thanks to David Lichteblau)
1012 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1013 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1014 (thanks to Gerd Moellmann)
1015 * fixed bug 246: increased compilation speed of long
1016 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1018 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1020 * DEFCONSTANT now throws a condition of type
1021 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1022 constant to a non-EQL value; CONTINUE and ABORT restarts
1023 respectively change and preserve the value.
1024 * fixed bug 63: The code walker, part of the implementation of CLOS,
1025 is now better at handling symbol macros.
1026 * bug fix: There is no longer an internal implementation type named
1027 CL:LENGTH. (reported by Raymond Toy)
1028 * bug fix: In macro-like defining macros/special operators the
1029 implicit block does not enclose the lambda list.
1030 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1031 enclosed in parentheses) are not suitable as type specifiers, and
1032 their use properly signals an error now.
1033 * bug fix: An argument count mismatch for a type specifier in code
1034 being compiled no longer causes an unhandled error at compile
1035 time, but signals a compile-time warning.
1036 * fixed simple vector readable printing
1037 * bug fix: DESCRIBE takes more care over whether the class
1038 precedence list slot of a class is bound before accessing it.
1039 (reported by Markus Krummenacker)
1040 * bug fix: FORMATTER can successfully compile pretty-printer format
1041 strings which use variants of the ~* directive inside.
1042 * bug fix: SEARCH now applies its TEST predicate to the elements of
1043 the arguments in the correct order. (thanks to Wolfhard Buss)
1044 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1045 in type inference. (thanks to Robert E. Brown)
1046 * bug fix: Reading in symbols with an explicit package name of ""
1047 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1048 * fixed some bugs revealed by Paul Dietz' test suite:
1049 ** NIL is now allowed as a structure slot name.
1050 ** Arbitrary numbers, not just REALs, are allowed in certain
1051 circumstances in LOOP for-as-arithmetic clauses.
1052 ** Multiple class redefinitions before slot access no longer
1053 causes a type error.
1054 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1055 association between the name and a class.
1056 ** Generic functions with non-standard method-combination and over
1057 five methods all of which return constants no longer return NIL
1058 after the first few invocations. (thanks to Gerd Moellmann)
1059 ** CALL-NEXT-METHOD with no arguments now passes the original
1060 values of the arguments, even in the presence of assignment.
1061 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1063 ** DISASSEMBLE works with closures and funcallable instances.
1064 ** ADD-METHOD now returns the generic function, not the new method.
1065 ** FIND-METHOD signals an error if the lengths of the specializers
1066 is incompatible with the generic function, even if the ERRORP
1068 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1069 which its argument is a member.
1070 ** DEFCLASS only redefines the class named by its class-name
1071 argument if that name is the proper name of the class;
1072 otherwise, it creates a new class.
1073 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1074 of the UNBOUND-SLOT condition to the name of the slot.
1075 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1076 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1077 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1078 SLOT-MAKUNBOUND in the specified fashion.
1080 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1081 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1082 cross-compilation host. As a consequence, we can now bootstrap our
1083 way up to SBCL starting with a bare gcc toolchain and human-readable
1084 source code (first the source to CLISP, then the source to SBCL).
1085 * A contributed module containing a partial implementation of the
1086 simple-streams interface has been included. (thanks to Rudi
1088 * A contributed module implementing the RFC1321 Message Digest
1089 Algorithm, known as MD5, has been included.
1090 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1091 no longer has any effect, as the code controlled by this feature
1092 has been deleted. (As far as we know, no-one has ever built using
1093 this feature, and its semantics were confused in any case).
1094 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1095 work on conditions (as required by the ANSI specification),
1096 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1097 expected behaviour on conditions. Users should note, however,
1098 that such behaviour is not required by the ANSI specification,
1099 and so use of this behaviour may render their code unportable.
1100 * fixed some bugs revealed by Paul Dietz' test suite:
1101 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1103 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1104 ** SXHASH on condition objects no longer returns NIL.
1105 ** :ALLOCATION :CLASS slots are better treated; their values are
1106 updated on class redefinition, and initforms inherited from
1107 superclasses are applied.
1108 ** REMOVE-METHOD returns its generic function argument even when
1109 no method was removed.
1110 ** SHARED-INITIALIZE now initializes the values of the requested
1111 slots, including those with :ALLOCATION :CLASS.
1112 ** ALLOCATE-INSTANCE now works on structure classes defined via
1113 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1115 ** SLOT-EXISTS-P now works on conditions, as well as structures
1117 ** MAKE-LOAD-FORM now has the required methods on
1118 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1119 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1120 keyword, and now implements the SLOT-NAMES argument.
1121 ** methods with &OPTIONAL arguments no longer allow too many
1122 arguments to be passed in the call without error.
1123 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1124 option is consistent with the required arguments of the generic
1125 function lambda list.
1126 * bug fix: REQUIRE accepts a string designator. (Thanks to
1128 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1129 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1130 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1131 * bug fix: The compiler now checks for duplicated variables in macro
1133 * bug fix: SETQ on globals returns the correct value.
1134 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1135 not a COMPILER-ERROR (followed by some other strange error on
1136 choosing the CONTINUE restart).
1137 * bug fix: make.sh and friends are now more consistent in the way that
1138 they look for GNU "make".
1140 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1141 * experimental native threads support (on x86 Linux >=2.4 only).
1142 This is not compiled in by default: you need to add :SB-THREAD to
1143 the target features. See the "Beyond ANSI" chapter of the manual
1145 * fix for longstanding nonANSIism: The old distinction between
1146 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1147 The return value from CL:FIND-CLASS is now a CLOS class, and
1148 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1149 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1150 * An interface to the MetaObject Protocol, as described in Kiczales,
1151 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1152 MIT Press, 1991, is available from the SB-MOP package.
1153 * incompatible change: the SB-PCL package should now be considered
1154 a private implementation detail, and no longer a semi-private MOP
1156 * minor incompatible change: due to rearrangement for threads, the
1157 control stack and binding stack are now allocated at arbitrary
1158 addresses instead of being hardcoded per-port. Users affected by
1159 this probably have to be doing advanced things with shared
1160 libraries, and will know who they are.
1161 * minor incompatible change: Previously, all --eval forms used were
1162 processed with READ before any of them were processed with EVAL.
1163 Now each --eval form is processed with both READ and EVAL before
1164 the next --eval form is processed. (Thus package operations like
1165 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1166 work as the user might reasonably expect.)
1167 * minor incompatible change: *STANDARD-INPUT* is now only an
1168 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1170 * minor incompatible change: Y-OR-N-P is now character-oriented,
1171 not line oriented. Also, YES-OR-NO-P now works without errors.
1172 (thanks to Antonio Martinez)
1173 * sb-aclrepl module improvements: an integrated inspector, added
1174 repl features, and a bug fix to :trace command.
1175 * Known functions, which cannot be open coded by the backend, are
1176 considered to be able to check types of their arguments. (fixing
1177 a bug report by Nathan J. Froyd)
1178 * fixed a bug in computing method discriminating functions: It is
1179 now possible to define methods specialized on classes which have
1180 forward-referenced superclasses. (thanks to Gerd Moellmann)
1181 * fixed evaluation order in optional entries (reported by Gilbert
1183 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1184 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1185 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1186 required-by-AMOP NAME argument, as well as CLASS and
1187 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1188 * fixed bug 20: DEFMETHOD can define methods using names that are
1189 not the proper names of classes to designate class specializers.
1190 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1191 Unix file descriptors, instead of blowing up. (thanks to Antonio
1193 * Garbage collection refactoring: user-visible change is that a
1194 call to the GC function during WITHOUT-GCING will not do garbage
1195 collection until the end of the WITHOUT-GCING. If you were doing
1196 this you were probably losing anyway.
1197 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1198 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1199 TYPEP the latter but not the former.
1200 * The compiler issues a full WARNING for calls to undefined functions
1201 with names from the CL package.
1202 * MAP-INTO for a vector destination is open coded. (reported by
1203 Brian Downing on c.l.l)
1204 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1205 documentation string.
1206 * fixed some bugs revealed by Paul Dietz' test suite:
1207 ** COPY-ALIST now signals an error if its argument is a dotted
1209 ** Condition slots are now accessed more correctly in the presence
1210 of multiple initargs for a given slot.
1211 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1212 exclude restarts of the same name associated with a different
1214 ** DEFCLASS of forward-referenced classes with another
1215 forward-referenced class in the superclasses list no longer
1217 ** Condition slots are now initialized once each, not multiple
1218 times. (thanks to Gerd Moellmann)
1219 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1221 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1222 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1223 arguments contain duplicated elements.
1224 ** RESTART-CASE understands local macros.
1225 ** RESTART-CASE associates exactly its own restarts with a condition.
1226 ** ENDP in safe mode checks its argument to be of type LIST.
1227 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1228 in question is unbound.
1229 ** Optimized MAKE-INSTANCE functions no longer cause internal
1230 assertion failures in the presence of duplicate initargs.
1231 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1232 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1233 propagated even in the ctor optimized implementation of
1235 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1237 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1238 * a better implementation of SXHASH on (simple) bit vectors,
1239 measured both in execution speed and in distribution of results
1240 over the positive fixnums, has been installed. Likewise, a better
1241 implementation of EQUAL for simple bit vectors is now available.
1242 * fixed CEILING optimization for a divisor of form 2^k.
1243 * fixed bug 240 (emitting extra style warnings "using the lexical
1244 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1245 by Antonio Martinez)
1246 * fixed SXHASH, giving different results for NIL depending on type
1247 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1248 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1249 variable. (found by Rolf Wester)
1250 * SBCL does not ignore type declarations for special
1251 variables. (reported by rif on c.l.l 2003-03-05)
1252 * some bug fixes in contrib/sb-aclrepl/
1253 * fixed some bugs revealed by Paul Dietz' test suite:
1254 ** a bug in the CONS type specifier, whereby the CAR and CDR
1255 types got intertwined, has been fixed;
1256 ** the type system is now able to reason about the interaction
1257 between INTEGER and RATIO types more completely;
1258 ** APPEND, [N]REVERSE and NRECONC check that those their
1259 arguments, which must be proper lists, are really so;
1260 ** An array specialized to be unable to hold elements has been
1261 implemented, as required -- yes, really -- by ANSI;
1262 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1263 on malformed property lists;
1265 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1266 * incompatible packaging change: in line with Unix convention,
1267 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1268 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1269 * REQUIRE and PROVIDE are now optionally capable of doing something
1270 useful. See the documentation string for REQUIRE.
1271 * infrastructure for a managed SBCL contrib system: contributed
1272 modules in this release include:
1273 ** the ASDF system definition facility;
1274 ** an interface to the BSD Sockets API;
1275 ** an ACL-like convenience interface to the repl;
1276 (thanks to Kevin Rosenberg)
1277 ** an implementation of ROTATE-BYTE, with efficient implementation
1279 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1280 gives a closer approximation to the right answer than previously.
1281 (thanks to Raymond Toy)
1282 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1283 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1284 optional environment arguments, as required by ANSI.
1285 * fixed bugs in other functions taking environment objects, allowing
1286 calls with an explicit NIL environment argument to be compiled
1288 * fixed bug 228: primary return values from
1289 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1290 COMPILE or FUNCTION.
1291 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1292 :TYPE will no longer signal errors on innocuous objects.
1293 * fixed bug 231b: SETQ is better at respecting type declarations in
1294 the lexical environment.
1295 * fixed a bug in DEFCLASS: classes named by symbols with no or
1296 unprintable packages can now be defined.
1297 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1298 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1299 * fixed compiler failure related to checking types of functions.
1300 (reported by Robert E. Brown)
1301 * the compiler is now much more consistent in its error-checking
1302 treatment of bounding index arguments to sequence functions: in
1303 (SAFETY 3) code, errors will be signalled in almost all cases if
1304 invalid sequence bounding indices are passed to functions defined
1305 by ANSI to operate on sequences.
1306 * fixed a bug in the build procedure: documentation of SBCL-specific
1307 packages is now preserved and available in the final Lisp image.
1308 * lifted FDEFINITION lookup out of loops in the implementation of
1309 many list operations. (thanks to Robert E. Brown)
1310 * fixed a bug in the reader: the #n# reader macro now works for
1311 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1312 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1313 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1314 E. Brown and Tony Martinez respectively)
1315 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1316 SLOT-BOUNDP: in particular, it is now able to optimize them much
1317 better, and is now not vulnerable to having packages renamed.
1318 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1319 always called when a slot is not present in an instance. (thanks
1321 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1322 longer signal an error on unreading a character following EOF on
1323 the previous constituent stream. (thanks to Tony Martinez)
1324 * fixed some bugs revealed by Paul Dietz' test suite:
1325 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1326 not just nonnegative fixnums;
1327 ** the logical bit-array operators such as BIT-AND now accept an
1328 explicit NIL for their "opt-arg" argument (to indicate a
1329 freshly-consed result bit-array);
1330 ** ELT now signals an error on an invalid sequence index in safe
1332 ** the type system is now cleverer about negations of numeric
1333 types, and consequently understands the BIGNUM and RATIO types
1335 ** the type system is now cleverer about the interaction between
1336 INTEGER and RATIO types: while bugs still remain, many more
1337 cases are accurately computed;
1338 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1339 if it is in the last clause;
1340 ** CONSTANTLY now correctly returns a side-effect-free function in
1342 ** DECLARE is no longer treated as a special-operator; in
1343 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1344 * incremented fasl file version number due to the change in the
1345 DEFSTRUCT-SLOT-DESCRIPTION structure.
1347 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1348 * minor incompatible change: code processed by the "interpreter" or
1349 EVAL now has a compilation optimization policy of (DEBUG 2)
1350 (changed from (DEBUG 1)) to improve debuggability of interactive
1351 development, and to allow the use of the debug RETURN command in
1353 * an experimental implementation of the RETURN command for the
1354 debugger has been included. (thanks to Frederik Kuivinen)
1355 * fixed bug 62: constraints were not propagated into a loop.
1356 * fixed bug in embedded calls of SORT (reported and investigated by
1358 * fixed some bugs revealed by Paul Dietz' test suite:
1359 ** printing and reading of arrays with some dimensions having
1360 length 0 (thanks to Gerd Moellmann);
1361 ** BOA constructor with &AUX argument without a default value does
1362 not cause a type error;
1363 ** CONSTANTP now returns true for all self-evaluating objects.
1365 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1366 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1367 accessors that are related by inheritance, as specified in the
1368 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1369 Valtteri Vuorikoski)
1370 * The compiler is now able to inline functions that were defined in
1371 a complex lexical environment (e.g. inside a MACROLET).
1372 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1374 * Support for the upcoming FreeBSD-5.0 release has been included.
1375 (thanks to Dag-Erling Smorgrav)
1376 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1377 effect when it is not in a toplevel context.
1378 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1379 stand a better chance of being correct. (thanks to Gerd
1381 * fixed bug in COERCE, which now signals an error on coercing a
1382 rational to a bounded real type which excludes the expected
1384 * The compiler is now able to derive types more accurately from the
1385 COERCE and COMPILE functions.
1386 * fixed bug 223: functional binding is considered to be constant
1387 only for symbols in the CL package.
1388 * fixed bug 231: SETQ did not check the type of a variable being set
1389 (reported by Robert E. Brown)
1390 * A new optimization for MAKE-INSTANCE has been included, fixing
1391 various bugs (including relating to :ALLOCATION :CLASS slots and
1392 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1394 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1395 ** As required by ANSI, LOOP now disallows anonymous collection
1396 clauses such as COLLECT I in conjunction with aggregate boolean
1397 clauses such as THEREIS (= I 1);
1398 ** LOOP now signals an error when any variable is reused in the
1399 same loop (including the potentially useful construct analogous
1400 to WITH A = 1 WITH A = (1+ A);
1401 ** IT is only a special loop symbol within the first clause of a
1402 conditional loop clause;
1403 ** LOOP with a typed iteration variable over a hashtable now
1404 signals a type error iff it should.
1405 * fixed some other bugs revealed by Paul Dietz' test suite:
1406 ** FILE-STREAM now names the class previously known as FD-STREAM;
1407 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1408 argument) no longer signals an error;
1409 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1410 of the slot symbol, rather than using the current package
1411 ((:CONC-NAME "") continues to intern the slot's name in the
1413 * incremented fasl file version number, because of the incompatible
1414 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1415 of the new implementation of DEFINE-COMPILER-MACRO.
1417 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1418 * Support for building SBCL for MIPS platforms running in
1419 little-endian mode has now been checked in, and basic
1420 functionality on said platforms verified.
1421 * minor incompatible change: PCL now records the pathname of a file
1422 in which methods and the like are defined, rather than its
1424 * minor incompatible change: TRUENAME now considers the truename of
1425 a file naming a directory to be the pathname with :DIRECTORY
1426 component indicating that directory.
1427 * minor incompatible change: a NAMED clause in the extended form of
1428 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1429 reason for the previous behaviour is unclear.
1430 * more systematization and improvement of CLOS and MOP conformance
1431 in PCL (thanks to Gerd Moellman and Pierre Mai):
1432 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1434 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1435 primary methods with no specializers;
1436 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1438 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1439 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1440 FORWARD-REFERENCED-CLASSes; error reporting on
1441 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1443 ** SXHASH on CLOS instances now uses a slot internal to the
1444 instance to return different numbers on distinct instances,
1445 while preserving the same return value through invocations of
1447 ** DEFMETHOD signals errors when methods with longer incongruent
1448 lambda lists are added to generic functions;
1449 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1450 CLASS, as specified in AMOP;
1451 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1452 on the order returned by the primary method for classes of
1453 class STANDARD-CLASS;
1454 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1455 * fixed some bugs shown by Paul Dietz' test suite:
1456 ** DOLIST puts its body in TAGBODY;
1457 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1459 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1460 value producing form;
1461 ** if more variables are given to PROGV than values, extra
1462 variables are bound and made to have no value;
1463 ** NSUBSTITUTE on list arguments gets the right answer with
1465 ** ELT signals an error of type TYPE-ERROR when the index argument
1466 is not a valid sequence index;
1467 ** LOOP signals (at macroexpansion time) an error of type
1468 PROGRAM-ERROR when duplicate variable names are found;
1469 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1470 ** FUNCALL of special-operators now cause an error of type
1472 ** PSETQ now works as required in the presence of side-effecting
1473 symbol-macro places;
1474 ** NCONC accepts any object as its last argument;
1475 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1477 ** loop-for-as-package does not require a package to be explicitely
1479 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1481 * fixed bug 166: compiler preserves "there is a way to go"
1482 invariant when deleting code.
1483 * fixed bug 172: macro lambda lists with required arguments after
1484 &REST arguments now cause an error to be signalled. (thanks to
1486 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1487 bug. (thanks to Matthew Danish)
1488 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1490 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1491 arguments are lexically rebound. (thanks to Gerd Moellmann and
1493 * fixed bug 194: error messages are now more informative when there
1494 is no primary method applicable in a call to a generic
1495 function. (thanks to Gerd Moellmann)
1496 * fixed bug in command line argument checking (thanks to Julian
1498 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1499 (thanks to Matthew Danish)
1500 * incremented fasl file version number, because of the
1501 SXHASH-related changes in the layout of CLOS data structures
1503 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1504 * minor incompatible change: The runtime (the Unix executable named
1505 "sbcl") is now much pickier about the .core files it will load.
1506 Essentially it now requires .core files to descend from the same
1507 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1508 as the runtime does. (The intent is to prevent the crashes which
1509 can occur, and which can even be reported as mysterious failures,
1510 when people patch the sources or change the build parameters
1511 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1512 sbcl and .core files.)
1513 * fixed bug: VALUES-LIST is no longer optimized away.
1514 * fixed bug 142: The FFI conversion of C string values to Lisp
1515 string values no longer conses excessively. (thanks to Nathan
1516 Froyd porting Raymond Toy's fix to CMU CL)
1517 * began to systematize and improve MOP conformance in PCL (thanks to
1518 Nathan Froyd, Gerd Moellman and Pierre Mai):
1519 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1521 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1522 ** FINALIZE-INHERITANCE is now called on class finalization;
1523 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1524 argument precedence order.
1525 * fixed bug 202: The compiler no longer fails on functions whose
1526 derived types contradict their declared type.
1527 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1528 so it can be non-toplevel.
1529 * The fasl file version number has changed (because of the new
1530 implementation of DEFMACRO).
1531 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1532 safe code, that any length requirement by their type-specifier
1533 argument is valid. The exceptions to this are described in bug
1535 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1536 functions now check the types of their inputs as required by ANSI.
1537 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1538 introduced symbol is DECLAREd to be SPECIAL.
1539 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1540 * fixed bug: :COUNT argument to sequence functions may be negative
1541 * fixed bug: body of DO-SYMBOLS may contain declarations
1542 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1543 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1544 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1545 symbol macro only once
1546 * fixed printing of call frame when argument list is unavailable
1547 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1548 * compiler no longer signals WARNING on unknown keyword
1551 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1552 * A beta-quality port to the mips architecture running Linux,
1553 based on the old CMUCL backend, has been made. It has been tested
1554 on a big-endian kernel, and works sufficiently well to be able to
1555 rebuild itself; it has not been tested in little-endian mode.
1556 * fixed an inconsistency between gencgc.c and purify.c which made
1557 dumping/loading .core files unreliable
1558 * fixed bug 120a: The compiler now deals correctly with IFs where
1559 the consequent is the same as the alternative, instead of
1560 misderiving the return type. (thanks to Alexey Dejneka)
1561 * fixed bug 113: Logical pathnames are now dumpable (the logical
1562 host is resolved at load-time, throwing an error if it is not
1564 * fixed bug 174: FORMAT's error message is slightly clearer when a
1565 non-printing character is used in a format directive.
1566 * fixed several bugs in compiler checking of type declarations, i.e.
1567 violations of the Python "declarations are assertions" principle
1568 (thanks to Alexey Dejneka)
1569 * fixed several bugs in PCL's error checking (thanks to Gerd
1571 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1573 * fixed bug in compilation of functions as first class values
1574 (thanks to Antonio Martinez-Shotton)
1575 * The compiler's handling TYPE-ERRORs which it can prove will
1576 inevitably happen at runtime has been cleaned up and corrected
1577 in several ways. (thanks to Alexey Dejneka)
1578 * improved argument type checking for various basic arithmetic
1579 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1580 lavished on them in the past that they can be compiled in many
1581 ways in different special cases
1582 * fixed bug 181: compiler checks validity of user supplied type
1584 * cleaned up code flushing in optimization: Function calls which
1585 should signal errors for safety purposes (e.g. which ANSI says
1586 should signal errors when their arguments are of incorrect type)
1587 are no longer optimized away.
1588 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1589 * incremented fasl file version number, because changes in the
1590 implementation of sequence functions like COERCE caused
1591 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1592 in old inline expansions) to become undefined. (Actually these
1593 changes were later undone, so we might very well be binary
1594 compatible with 0.7.7 after all, but leaving the version number
1595 incremented seemed like the simplest and most conservative
1598 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1599 * An alpha-quality port to the parisc architecture running Linux,
1600 based on the old CMUCL backend, has been made. This, even more so
1601 than the other backends, should be considered still a work in
1602 progress; known problems include that the Linux kernel in 64-bit
1603 mode does not propagate the correct sigcontext structure to
1604 userspace, and consequently SBCL on a parisc64 kernel will not
1606 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1607 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1608 suppressed.) Also now that the compiler is looking at declarations
1609 in the environment, it checks optimization declarations as well,
1610 and suppresses inlining when (> DEBUG SPEED).
1611 * More fixes have been made to treatment of floating point exception
1612 treatment and other Unix signals. In particular, floating point
1613 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1614 * The detection and handling of control stack exhaustion (infinite
1615 or very deeply nested recursion) has changed. Stack exhaustion
1616 detection is now done by write-protecting pages at the OS level
1617 and applies at all optimization settings; when found, a
1618 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1619 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1620 be caught using IGNORE-ERRORS.
1621 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1622 correctly when transforming recursive LABELS functions to LETs.
1623 (thanks to Alexey Dejneka)
1624 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1625 that are names of constants or global variables.
1626 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1627 alien routines with docstrings.
1628 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1629 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1631 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1632 (thanks to Eric Marsden)
1633 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1634 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1635 * Bug fix: The SPARC backend can now compile functions involving
1636 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1637 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1638 to a functions expecting a PATHNAME-DESIGNATOR.
1639 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1640 lambda lists. (thanks to Alexey Dejneka)
1641 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1642 functions so that EXPORTing the name of the class doesn't cause
1643 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1644 to Antonio Martinez for reporting this)
1645 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1646 constant-folded without failing an assertion. (thanks to Einar
1647 Floystad Dorum for reporting this)
1648 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1649 types are now dealt with more correctly by the compiler.
1650 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1651 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1652 behaviour analogous to RENAME-FILE. This puts its behaviour more
1653 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1655 * The fasl file version number has changed again. (because of the
1656 bug fix involving the names of PCL MAKE-INSTANCE functions)
1658 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1659 * bug fix: Floating point exceptions are treated much more
1660 consistently on the x86/Linux and PPC/Linux platforms.
1661 * Array initialization with :INITIAL-ELEMENT is now much faster for
1662 cases when the compiler cannot open code the array creation, but
1663 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1664 array accesses have also seen a speed increase.
1665 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1666 specified. (This was at the root of some bad interactions between
1667 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1669 * bug fix: Internal error arguments for undefined functions are now
1670 computed correctly on the PPC/Linux platform.
1671 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1672 Raymond Toy's patch for CMU CL)
1673 * Support for the Solaris 9 operating environment has been included
1674 (thanks to Daniel Merritt)
1675 * A very ugly but hopefully complete draft of the missing FFI chapter
1676 of the manual has been created by reformatting the corresponding
1677 CMU CL manual chapter into (currently very ugly and incoherent)
1678 DocBook and bringing it up to date for SBCL behavior. Thus, the
1679 manual is now essentially complete, at least by my extreme
1680 once-and-only-once standards, whereby it's acceptable to refer to
1681 the doc strings of SB-EXT functions as the primary documentation.
1682 * The fasl file version number has changed again, due to cleanup of
1683 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1684 is no longer a static symbol.)
1686 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1687 * SBCL now builds with OpenMCL (version 0.12) as the
1688 cross-compilation host; also, more progress has been made toward
1689 bootstrapping under CLISP.
1690 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1692 * bug 158 fixed: The compiler can now deal with integer loop
1693 increments different from 1; fixing this turned out also to fix
1695 * bug 169 fixed: no more bogus warnings about using lexical bindings
1696 despite the presence of perfectly good SPECIAL declarations (thanks
1697 to David Lichteblau)
1698 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1699 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1700 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1702 * bug 180 fixed: Method combination specifications no longer ignore
1703 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1704 * bug fix: Structure type predicate functions now check their argument
1705 count as they should.
1706 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1707 correctly. (thanks to Pierre Mai)
1708 * minor incompatible change: The --noprogrammer option is deprecated
1709 in favor of the new --disable-debugger option, which is very similar.
1710 (The major difference is that it takes effect at a slightly different
1711 time at startup, causing handling of errors in --sysinit and
1712 --userinit files will be affected differently.) The
1713 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1714 been added to allow this functionality to be controlled from ordinary
1715 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1716 maintainers, who might want to run non-interactive scripts to
1717 build SBCL cores which will later be used interactively.)
1718 * minor incompatible change: The LOAD function no longer, when given
1719 a wild pathname to load, loads all files matching that pathname.
1720 Instead, an error of type FILE-ERROR is signalled.
1722 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1723 * bug 147 fixed: The compiler preserves its block link/count
1724 invariants more correctly now so that it doesn't crash. (thanks
1726 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1728 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1729 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1730 no-value-for-this-character as NIL (as specified by ANSI).
1731 * HOST-NAMESTRING on physical pathnames now returns a string that is
1732 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1733 (thanks to Christophe Rhodes)
1734 * The Alpha port handles icache flushing more correctly. (thanks to
1736 * More progress has been made toward bootstrapping under CLISP. (thanks
1737 to Christophe Rhodes)
1738 * The fasl file format has changed again, because dynamic loading
1739 on OpenBSD (which has non-ELF object files) motivated some cleanups
1740 in the way that foreign symbols are transformed and passed around.
1741 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1742 is no longer treated as whitespace by the reader, but instead as
1743 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1744 |A
\7fB|, instead of A as it used to.
1746 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1747 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1748 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1749 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1750 this as of 0.7.1.45, but was left out of the previous news section
1751 (thanks to Dan Barlow)
1752 * SBCL now runs on the Solaris operating system on SPARC architectures
1753 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1754 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1755 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1756 * SPARC backend cleanups, allowing builds of cores optimized for V8
1757 and V9 SPARCS, and also emission of code targeted to a particular
1758 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1760 * SBCL is closer to bootstrapping under CLISP, thanks to various
1761 fixes by Christophe Rhodes.
1762 * The fasl file format has changed again, to allow the compiler's
1763 INFO database to support symbol macros.
1764 * The user manual (in doc/) is formatted into HTML more nicely.
1765 (thanks to coreythomas)
1766 * The system is smarter about SUBTYPEP relationships, especially
1767 those involving NOT types (including types such as ATOM which are
1768 represented internally using NOT types). Thus SUBTYPEP is less
1769 likely to return (VALUES NIL NIL) in general, and in particular
1770 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1772 * The fasl file format has changed again, because the internal
1773 representation of types now includes a new slot to support the new
1774 SUBTYPEP-of-NOT-types logic.
1775 * (not a change in the main branch of SBCL, but a related prototype
1776 which can hopefully be merged into the main branch of SBCL in the
1777 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1778 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1779 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1780 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1781 dispatch macro character combinations. (thanks to Alexey Dejneka)
1782 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1783 that superficially look like logical namestrings correctly.
1784 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1786 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1787 default supplied-p) for &optional and &key arguments. (thanks to
1790 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1791 * incompatible change: The compiler is now less aggressive about
1792 tail call optimization, doing it only when (> SPACE DEBUG) or
1793 (> SPEED DEBUG). (This is an incompatible change because there are
1794 programs which relied on the old CMU-CL-style behavior to optimize
1795 away their unbounded recursion which will now die of stack overflow.)
1796 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1797 for non-GENCGC systems has been increased to 20M (since that
1798 seems much closer to the likely performance optimum for modern
1799 systems than the old 4M value was)
1800 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1801 and *DEBUG-PRINT-LEVEL*
1802 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1803 of CMU CL's support for SPARC, and various endianness and other
1804 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1805 * new syntactic sugar for the Unix command line: --load foo.bar is now
1806 an alternate notation for --eval '(load "foo.bar")'.
1808 ** The system now detects stack overflow and handles it gracefully,
1809 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1810 optimization settings. (This is a good thing in general, and
1811 its introduction in this version should be particularly timely
1812 for anyone whose code fails because of suppression of tail
1814 ** The system now hunts for the C variable "environ" in a more
1815 devious way, to avoid segfaults when the C library version differs
1816 between compile time and run time. (thanks to Christophe Rhodes)
1817 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1818 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1819 ** The compiler no longer issues bogus style warnings for undefined
1820 classes in the same source file as the DEFCLASSes which defined
1821 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1823 ** fixes in CONDITION class precedence list for undefined function
1824 errors (thanks to Alexei Dejneka)
1825 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1826 correctly. (thanks to Dan Barlow)
1827 ** portability fixes aiming at bootstrapping under CLISP (thanks
1828 to Dave McDonald and Christophe Rhodes)
1829 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1830 ** fixes in type translation and and type inference (thanks to
1832 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1833 ** various fixes in the new ports (thanks to Dan Barlow)
1834 * several changes related to debugging:
1835 ** suppression of tail recursion, as noted above
1836 ** stack overflow detection, as noted above
1837 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1838 is now the default. (For some time encapsulation has been more
1839 reliable than the breakpoint-based :ENCAPSULATE NIL
1840 implementation, at least on X86 systems; and I just noticed that
1841 encapsulation also seems closer to the spirit of the ANSI
1844 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1846 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1847 up properly again. (There was a packaging bug in 0.7.0 which
1848 left their definitions in SB-SYS::LOAD-FOREIGN and
1849 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1850 vital for most things which interface to C-level interfaces,
1851 like extensions working with sockets or databases or
1852 Perl-compatible regexes or whatever, and the need to fix
1853 this bug is the main reason that 0.7.1 was released so
1855 ** DEFGENERIC is now choosier about the methods it redefines, so that
1856 reLOADing a previously-LOADed file containing DEFGENERICs does
1857 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1858 cycle now works as it should. (thanks to Alexey Dejneka)
1859 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1860 fixed by Christophe Rhodes. (He actually submitted this patch
1861 months ago, and I delayed until after 0.7.0.)
1862 ** Bug 111 (internal compiler confusion about runtime checks on
1863 FUNCTION types) was fixed by Alexey Dejneka.
1864 * Some internal cleanups (getting rid of variables which aren't
1865 needed now that the byte interpreter is gone) caused the fasl
1866 file format number to change again.
1868 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1869 * major incompatible change: The default fasl file extension, i.e. the
1870 default extension for files produced by COMPILE-FILE, has changed
1871 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1873 ** There are many changes in the implementation of the compiler.
1874 SBCL is now essentially a compiler-only implementation of ANSI
1875 Common Lisp. EVAL still "interprets" a few special cases, but
1876 almost all the interesting cases are handled by creating
1877 a LAMBDA expression, calling COMPILE on it, then calling
1878 FUNCALL on the result.
1879 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1880 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1881 Since the code is newer, there might still be some new bugs
1882 (though not as many as before Martin Atzmueller's fixes:-). But
1883 the new code is substantially simpler and clearer, and hopefully
1884 any remaining bugs will be simpler, less fundamental, and more
1885 fixable then the bugs in the old code.
1886 ** The revised compiler is still a little unsteady on its feet.
1888 *** The debugging information it produces (particularly the names
1889 of FUNCTION objects) is sometimes much less useful than what
1890 the old compiler produced.
1891 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1892 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1893 within a MACROLET) has been temporarily weakened.
1894 ** There are new compiler optimizations for various functions:
1895 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1896 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1897 *** the math functions TRUNCATE, FLOOR, and CEILING
1898 *** the function-of-all-trades COERCE
1899 Mostly these should be transparent, but there's one
1900 potentially-annoying problem (bug 117): when the compiler
1901 inline-expands a function and does type analysis on the result,
1902 it can create control paths which have type mismatches, and
1903 when it can't prove that those control paths aren't taken,
1904 it will issue WARNINGs about the type mismatches. This is
1905 a particular problem in practice for the new sequence functions.
1906 It's not clear how this should be fixed, and for now, a
1907 workaround is given in the entry for 117 in the BUGS file.
1908 ** (Because of the interaction between the two previous items --
1909 occasional inlining problems and new inline expansions -- some
1910 of the new sequence function optimizations won't really kick in
1911 completely until debugging information, and then inlining, are
1912 straightened out in some future version.)
1913 * minor incompatible changes:
1914 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1915 DIRECTORY no longer implicitly promotes NIL slots of its
1916 pathname argument to :WILD. In particular, when you ask for the
1917 contents of a directory (which you used to be able to do without
1918 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1919 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1920 ** changes in behavior that ANSI explicitly defines to be
1921 implementation dependent:
1922 *** The new compiler-only implementation still conforms with ANSI,
1923 but acts a little different than before. Besides the obvious
1924 changes in performance tradeoffs (that the cost per form passed
1925 to EVAL has gone up, and the cost per form executed by EVAL
1926 has gone down), the behavior of the system changes a little
1927 because there are no longer any interpreted function objects.
1928 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1929 e.g. doing COMPILE on the output of interactive DEFUN is
1931 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1933 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1934 has changed from (:CL) to NIL.
1935 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1936 e.g. #\Newline and #\Tab, don't have names specified in the
1937 ANSI Common Lisp standard, is now based on their ASCII symbolic
1938 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1939 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1940 are no longer used for output.
1941 ** changes in internal implementation constants:
1942 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1943 4 million. (If your application spends a lot of time GCing and
1944 you have a lot of RAM, you might want to experiment with
1945 increasing it even more.)
1946 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1947 However, almost all old code should still continue to work without
1948 immediate update, as SB-C-CALL is now a (deprecated) nickname
1950 ** Old operator names in the style DEF-FOO are now deprecated in
1951 favor of new corresponding names DEFINE-FOO, for consistency with
1952 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1953 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1954 internal symbols, but a few supported extensions like
1955 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1956 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1957 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1958 etc. as you get deeper into recursive calls to the debugger
1959 command loop, instead of the old "5]", "5]]", "5]]]"
1960 sequence. (I was motivated to do this when squabbles between
1961 ILISP and SBCL left me very deeply nested in the debugger. In the
1962 short term, this change will probably provoke more ILISP/SBCL
1963 squabbles, but hopefully it will be an improvement in the long run.)
1964 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1965 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1966 function, since declaiming return types involving aliens is
1967 (1) annoyingly messy to do by hand and (2) vital to efficient
1968 compilation of code which calls such functions.
1969 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1970 longer reexported by the SB-EXT package. They're solely useful
1971 for alien code, so it seems more logical that you should get
1972 them from the SB-ALIEN package, not in SB-EXT.
1973 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1974 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1975 features. Instead, the code that they used to control is always
1976 built into the system.
1977 * many other bug fixes
1978 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1979 advantage of the new EVAL-WHEN stuff and to clean them up in
1980 general, and they are now more ANSI-compliant in a number of
1981 ways. Martin Atzmueller is responsible for a lot of this.
1982 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1984 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1985 *** correct ERROR type for various file operations
1986 *** some fixes for Lisp streams
1987 *** DEFMETHOD syntax checking
1988 *** changing old weird representation of debug information as
1989 strings (which, among their other deficiencies, don't transform
1990 correctly when you rename packages, and don't change their
1991 print representation when you change things like *PACKAGE*
1992 and *PRINT-LENGTH*) to symbols and lists of symbols
1993 He also made several improvements and fixed several bugs in DESCRIBE.
1994 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1996 *** misbehavior of WRITE-STRING/WRITE-LINE
1997 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1998 and several other LOOP problems as well
1999 *** DIRECTORY when similar filenames are present
2000 *** DEFGENERIC with :METHOD options
2001 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2002 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2003 *** argument ordering in FIND with :TEST option
2004 *** mishandled package designator argument in APROPOS-LIST
2005 *** various problems in the backquote readmacro
2006 *** a bug in APROPOS
2007 *** probably some others that I'm not describing very well here,
2008 since the CVS log documents them by reference to sbcl-devel
2009 messages, and the SourceForge archives aren't working well.:-(
2010 ** Dan Barlow improved the Alpha port (and is making progress on the
2011 PPC port, for those of you who think different).
2012 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2013 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2014 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2015 continues to work on the SPARC port (for those of you in a position
2016 to look down upon our little PC-compatible boxes from a great height).
2017 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2018 and uses that copy, rather than the current dynamic value, when
2019 it's trying to decide whether to truncate output. Thus e.g.
2020 (let ((*print-lines* 50))
2021 (pprint-logical-block (stream nil)
2023 (let ((*print-lines* 8))
2024 (print (aref possiblybigthings i) stream)))))
2025 should now truncate the logical block only at 50 lines, instead of
2026 often truncating it at 8 lines, as it did before.
2027 * The doc/cmucl/ directory, containing old CMU CL documentation
2028 from the time of the fork, is no longer part of the base system.
2029 SourceForge has shut down its anonymous FTP service, and with it
2030 my original plan for distributing the old CMU CL documentation
2031 there. For now, if you need these files you can download an old
2032 SBCL source release and extract them from it.
2033 * The fasl file version number changed again, for dozens of reasons,
2034 some of which are apparent above.
2036 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2037 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2038 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2039 patch, so that DISASSEMBLE output is much nicer.
2040 * The code in the SB-PROFILE package now seems reasonably stable.
2041 I still haven't decided what the final interface should look like
2042 (I'd like PROFILE to interact cleanly with TRACE, since both
2043 facilities use function encapsulation) but if you have a need
2044 for profiling now, you can probably use it successfully with
2045 the current CMU-CL-style interface.
2046 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2047 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2048 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2049 link now returns the dangling link itself, and for similar
2050 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2051 link itself. (In these cases the old code signalled an error and
2052 looped endlessly, respectively.) Thus, DIRECTORY now works even
2053 in the presence of dangling and cyclic symbolic links.
2054 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2055 is now a supported extension again, since the consensus on
2056 sbcl-devel was that it can be useful for ordinary development
2057 work, not just for debugging SBCL itself.
2058 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2059 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2060 that currently-defined functions might be redefined later with
2061 different return types.
2062 * Hash tables can be printed readably, as inspired by CMU CL code
2063 of Eric Marsden and SBCL code of Martin Atzmueller.
2064 * better error handling in CLOS method combination, thanks to
2065 Martin Atzmueller porting Pierre Mai's CMU CL patches
2066 * more overflow fixes for >16Mbyte I/O buffers
2067 * A bug in READ has been fixed, so that now a single Ctrl-D
2068 character suffices to cause end-of-file on character streams.
2069 In particular, now you only need one Ctrl-D at the command
2070 line (not two) to exit SBCL.
2071 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2073 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2074 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2075 does the right thing.
2076 * The implementation of some type tests, especially for CONDITION
2077 types, is now tidier and maybe faster, due to CMU CL code
2078 originally by Douglas Crosher, ported by Martin Atzmueller.
2079 * Some math functions have been fixed, and there are new
2080 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2081 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2082 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2083 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2084 likely to be built into sbcl-0.7.0.)
2085 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2086 CL, ported to SBCL by Martin Atzmueller)
2087 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2088 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2089 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2090 * There's a new slam.sh hack to shorten the edit/compile/debug
2091 cycle for low-level changes to SBCL itself, and a new
2092 :SB-AFTER-XC-CORE target feature to control the generation of
2093 the after-xc.core file needed by slam.sh.
2094 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2095 COMPILE-FILE is no longer supported, so that now every function
2096 gets an entry point, so that block compilation looks a little
2097 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2098 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2099 returns the number of bytes consed since the system started,
2100 rather than the number consed since the first time the function
2101 was called. (The new definition parallels ANSI functions like
2102 CL:GET-INTERNAL-RUN-TIME.)
2103 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2104 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2105 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2106 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2107 * Fasl file version numbers are now independent of the target CPU,
2108 since historically most system changes which required version
2109 number changes have affected all CPUs equally. Similarly,
2110 the byte fasl file version is now equal to the ordinary
2113 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2114 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2115 is no longer recognized. I apologize for this, because it was
2116 listed in SB-EXT as a supported extension, but I found that
2117 its existing behavior was poorly specified, as well as incorrectly
2118 specified, and it looked like too much of a mess to straighten it
2119 out. I have enough on my hands trying to get ANSI stuff to work..
2120 * many patches ported from CMU CL by Martin Atzmueller, with
2121 half a dozen bug fixes in pretty-printing and the debugger, and
2122 half a dozen others elsewhere
2123 * fixed bug 13: Floating point infinities are now supported again.
2124 They might still be a little bit flaky, but thanks to bug reports
2125 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2126 as flaky as they were.
2127 * The --noprogrammer command line option is now supported. (Its
2128 behavior is slightly different in detail from what the old man
2129 page claimed it would do, but it's still appropriate under the
2130 same circumstances that the man page talks about.)
2131 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2132 are now supported, and enabled by default. Thus, the compiler can
2133 handle many floating point and complex operations much less
2134 inefficiently. (Thus e.g. you can implement a complex FFT
2136 * The compiler now detects type mismatches between DECLAIM FTYPE
2137 and DEFUN better, and implements CHECK-TYPE more correctly, and
2138 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2139 thanks to patches from Martin Atzmueller.
2140 * various fixes to make the cross-compiler more portable to
2141 ANSI-conforming-but-different cross-compilation hosts (notably
2142 Lispworks for Windows, following bug reports from Arthur Lemmens)
2143 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2144 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2145 fixed thanks to Pierre Mai's CMU CL patches.
2146 * fixes to keep the system from overflowing internal counters when
2147 it tries to use i/o buffers larger than 16M bytes
2148 * fixed bug 45a: Various internal functions required to support
2149 complex special functions have been merged from CMU CL sources.
2150 (When I was first setting up SBCL, I misunderstood a compile-time
2151 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2152 * improved support for type intersection and union, fixing bug 12
2153 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2154 more obscure bugs as well
2155 * some steps toward byte-compiling non-performance-critical
2156 parts of the system, courtesy of patches from Martin Atzmueller
2157 * Christophe Rhodes has made some debian packages of sbcl at
2158 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2159 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2160 stable, but are nonetheless usable. When he's ready, I'd be happy
2161 to add them to the SourceForge "File Releases" section. (And if
2162 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2163 * new fasl file format version number (because of changes in
2164 internal representation of (OR ..) types to accommodate the new
2165 support for (AND ..) types, among other things)
2167 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2168 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2170 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2172 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2173 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2174 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2175 * DESCRIBE now works on structure objects again.
2176 * Most function call argument type mismatches are now handled as
2177 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2178 know whether the function will be redefined before the call is
2179 executed. (The compiler could flag local calls with full WARNINGs,
2180 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2181 it doesn't keep track of enough information to know whether calls
2182 are local in this sense.)
2183 * Compiler output is now more verbose, with messages truncated
2184 later than before. (There should be some supported way for users
2185 to override the default verbosity, but I haven't decided how to
2186 provide it yet, so this behavior is still controlled by the internal
2187 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2188 src/compiler/ir1util.lisp.)
2189 * Fasl file format version numbers have increased again, because
2190 support for the Gray streams extension changes the layout of the
2191 system's STREAM objects.
2192 * The Gray subclassable streams extension now works, thanks to a
2193 patch from Martin Atzmueller.
2194 * The full LOAD-FOREIGN extension (not just the primitive
2195 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2196 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2197 but like most other programs, it defaults to copying the Unix
2198 environment from the original process instead of starting the
2199 new process in an empty environment.
2200 * Extensions which manipulate the Unix environment now support
2201 an :ENVIRONMENT keyword option which doesn't smash case or
2202 do other bad things. The CMU-CL-style :ENV option is retained
2203 for porting convenience.
2204 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2205 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2207 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2209 * A patch from Martin Atzmueller seems to have solved the SIGINT
2210 problem, and as far as we know, signal-handling now works cleanly.
2211 (If you find any new bugs, please report them!)
2212 * The system no longer defaults Lisp source file names to types
2213 ".l", ".cl", or ".lsp", but only to ".lisp".
2214 * The compiler no longer uses special default file extensions for
2215 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2216 seems to expect a single default extension for all compiled code,
2217 and there's no compelling reason to try to stretch the standard
2218 to allow two different extensions.) Instead, byte-compiled files
2219 default to the same extension as native-compiled files.
2220 * Fasl file format version numbers have increased again, because
2221 a rearrangement of internal implementation packages made some
2222 dumped symbols in old fasl files unreadable in new cores.
2223 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2224 many fewer weird special cases.
2225 * Bug #17 (differing COMPILE-FILE behavior between logical and
2226 physical pathnames) has been fixed, and some related misbehavior too,
2227 thanks to a patch from Martin Atzmueller.
2228 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2229 by Tim Moore, ported to SBCL by Martin Atzmueller.
2230 * Martin Atzmueller fixed several filesystem-related problems,
2231 including bug #36, in part by porting CMU CL patches, which were
2232 written in part by Paul Werkowski.
2233 * More compiler warnings in src/runtime/ are gone, thanks to
2234 more patches from Martin Atzmueller.
2235 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2238 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2240 * DESCRIBE now works on CONDITION objects.
2241 * The debugger now handles errors which arise when trying to print
2242 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2244 * The build system now uses an additional file, customize-target-features.lisp,
2245 to allow local modifications to the target *FEATURES* list. (The point of
2246 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2247 debugging features enabled, without having to worry about propagating it
2248 into everyone's system when I do a "cvs update".) When no
2249 customize-target-features.lisp file exists, the target *FEATURES* list
2250 should be constructed the same way as before.
2251 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2252 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2253 warn about behavior which is undefined under the ANSI spec. Note: This
2254 is specified by ANSI, but it's not very popular with programmers.
2255 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2256 macro in the SBCL sources for an example of a workaround which you
2257 might use to make portable ANSI-standard code which does what you want.
2258 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2259 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2260 when it's used as a non-toplevel form. (This is required in order
2261 to implement the DEFCONSTANT-EQX macro.)
2262 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2263 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2264 funny things, due to bugs in the implementation of EVAL-WHEN
2265 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2266 non-toplevel DEFCONSTANTs are uncommon.)
2267 * The core file version number and fasl file version number have been
2268 incremented, because the old noncompliant DEFCONSTANT behavior involved
2269 calling functions which no longer exist, and because I also took the
2270 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2271 * fixed bug 1 (error handling before read-eval-print loop starts), and
2272 redid debugger restarts and related debugger commands somewhat while
2274 ** The QUIT debugger command is gone, since it did something
2275 rather different than the SB-EXT:QUIT command, and since it never
2276 worked properly outside the main toplevel read/eval/print loop.
2277 Invoking the new TOPLEVEL restart provides the same functionality.
2278 ** The GO debugger command is also gone, since you can just invoke
2279 the CONTINUE restart directly instead.
2280 ** The TOP debugger command is also gone, since it's redundant with the
2281 FRAME 0 command, and since it interfered with abbreviations for the
2283 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2285 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2286 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2287 some porting work by Martin Atzmueller)
2288 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2289 with a boatload of patches which he ported to SBCL. Now that those
2291 ** The system tries to make sure that its low-priority messages
2292 are prefixed by semicolons, to help people who like to use
2293 syntax highlighting in their ILISP buffer. (This patch
2294 was originally due to Raymond Toy.)
2295 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2296 patches originally written by Raymond Toy.
2297 ** The compiler understands coercion between single-value and
2298 multiple-VALUES type expressions better, getting rid of some very
2299 weird behavior, thanks to patches originally by Robert MacLachlan
2300 and Douglas Crosher.
2301 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2302 lambda lists, thanks to a patch originally by Pierre Mai.
2303 ** The system no longer bogusly warns about "abbreviated type
2305 ** The compiler gets less confused by inlining and RETURN-FROM,
2306 thanks to some patches originally by Tim Moore.
2307 ** The system no longer hangs when dumping circular lists to fasl
2308 files, thanks to a patch originally from Douglas Crosher.
2309 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2310 undefined function error.
2311 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2312 in structure slot initforms) for now, documented workaround instead:-|
2313 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2314 * fixed bug 5: added stubs for various Gray stream functions called
2315 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2316 installed, at least appropriate type errors are generated
2317 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2318 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2319 consistently in DEFMETHOD forms.
2320 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2321 it doesn't seem to affect SBCL after all
2322 * The C runtime system now builds with better optimization and many
2323 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2325 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2327 * The system is now under CVS at SourceForge (instead of the
2328 CVS repository on my home machine).
2329 * The new signal handling code has been tweaked to treat register
2330 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2331 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2332 downstream, e.g. in debug-int.lisp, has implicit dependencies
2333 on the unsignedness of integer representation of machine words,
2334 and that caused the system to bomb out with infinite regress
2335 when trying to recover from type errors involving signed values,
2336 e.g. (BUTLAST '(1 2 3) -1).)
2337 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2338 (This was one of the bugs Peter Van Eynde reported back in July.)
2339 * The system now uses code inspired by Colin Walters' O(N)
2340 implementation of MAP (from the cmucl-imp@cons.org mailing
2341 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2342 inline the MAP operation, and there is more than one
2343 sequence argument to the MAP call (so that it can't just
2344 do ETYPECASE once and for all based on the type of the
2345 single sequence argument). (The old non-inline implementation
2346 of the general M-argument sequence-of-length-N case required
2347 O(M*N*N) time when any of the sequence arguments were LISTs.)
2348 * The QUIT :UNIX-CODE keyword argument has been renamed to
2349 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2350 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2352 * Raymond Wiker's patches to port dynamic loading from Linux to
2353 FreeBSD have been added.
2354 * The BUGS file is now more nearly up to date, thanks in large part
2355 to Martin Atzmueller's review of it.
2356 * The debugger now flushes standard output streams before it begins
2357 its output ("debugger invoked" and so forth).
2358 * The core version number and fasl file version number have both
2359 been incremented, because of incompatible changes in the layout
2361 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2362 used to not be called for a saved Lisp image.)
2363 * Martin Atzmueller's version of a patch to fix a compiler crash,
2364 as posted on sbcl-devel 13 September 2000, has been installed.
2365 * Instead of installing Martin Atzmueller's patch for the
2366 compiler transform for SUBSEQ, I deleted the compiler transform,
2367 and transforms for some similar consing operations.
2368 * A bug in signal handling which kept TRACE from working on OpenBSD
2370 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2371 away by constant folding
2372 * The system now defines its address space constants in one place
2373 (in the Lisp sources), and propagates them automatically elsewhere
2374 (through GENESIS and the sbcl.h file). Therefore, patching the
2375 address map is less unnecessarily tedious and error-prone. The
2376 Lisp names of address space constants have also been systematized.
2377 * CVS tags like dollar-Header-dollar have been removed from
2378 the sources, because they have never saved me trouble and
2379 they've been source of trouble working with patches and other
2380 diff-related operations.
2381 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2382 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2384 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2386 * The system has been ported to OpenBSD.
2387 * The system now compiles with a simple "sh make.sh" on the systems
2388 that it's supported on. I.e., now you no longer need to tweak
2389 text in the target-features.lisp-expr and symlinks in src/runtime/
2390 by hand, the make.sh takes care of it for you.
2391 * The system is no longer so grossly inefficient when compiling code
2392 involving vectors implemented as general (not simple) vectors (VECTOR T),
2393 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2394 which dares to use the various sequence functions on non-simple
2395 vectors, takes less of a performance hit.
2396 * There is now a primitive type predicate VECTOR-T-P
2397 to test for the (VECTOR T) type, so that e.g.
2398 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2399 can now be compiled with some semblance of efficiency. (The old code
2400 turned the type declaration into a full call to %TYPEP at runtime!)
2401 * AREF on (VECTOR T) is still not fast, since it's still compiled
2402 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2403 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2404 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2405 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2406 without making full calls to SUBTYPEP at runtime.
2407 (Some analogous efficiency issues for non-simple vectors specialized to
2408 element types other than T, or for non-simple multidimensional arrays,
2409 have not been addressed. They could almost certainly be handled the
2410 same way if anyone is motivated to do so.)
2411 * The changes in array handling break binary compatibility, so
2412 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2413 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2414 instead of (VALUES T T).
2415 * By following the instructions that Dan Barlow posted to sbcl-devel
2416 on 2 July 2000, I was able to enable primitive dynamic object
2417 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2418 functionality is not implemented (since it calls ld to resolve
2419 library references automatically, requiring RUN-PROGRAM for its
2420 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2421 try to resolve library references) is now supported.
2422 * The system now flushes the standard output streams when it terminates,
2423 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2424 them at several other probably-convenient times, e.g. in each pass of
2425 the toplevel read-eval-print loop, and after evaluating a form given
2426 as an "--eval" command-line option. (These changes were motivated by a
2427 discussion of stream flushing issues on cmucl-imp in August 2000.)
2428 * The source transform for TYPEP of array types no longer assumes
2429 that an array whose element type is a not-yet-defined type
2430 is implemented as an array of T, but instead punts, so that the
2431 type will be interpreted at runtime.
2432 * There is now some support for cross-compiling in make.sh: each of
2433 the phases of make.sh has its own script. (This should be transparent
2434 to people doing ordinary, non-cross-compile builds.)
2435 * Since my laptop doesn't have hundreds of megabytes of memory like
2436 my desktop machine, I became more motivated to do some items on
2437 my to-do list in order to reduce the size of the system a little:
2438 ** Arrange for various needed-only-at-cold-init things to be
2439 uninterned after cold init. To support this, those things have
2440 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2441 symbols with such names are now uninterned after cold init).
2442 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2443 fasl files for cold load.
2444 ** Remove the old compiler structure pooling code (which used to
2445 be conditional on the target feature :SB-ALLOC) completely.
2446 ** Redo the representation of some data in cold init to be more compact.
2447 (I also looked into supporting byte compiled code at bootstrap time,
2448 which would probably reduce the size of the system a lot, but that
2449 looked too complicated, so I punted for now.)
2450 * The maximum signal nesting depth in the src/runtime/ support code has
2451 been reduced from 4096 to 256. (I don't know any reason for the very
2452 large old value. If the new smaller value turns out to break something,
2453 I'll probably just bump it back up.)
2454 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2456 * Many, many bugs reported by Peter Van Eynde have been added to
2457 the BUGS list; some have even been fixed.
2458 * While enabling dynamic object file loading, I tried to make the
2459 code easier to understand, renaming various functions and variables
2460 with less ambiguous names, and changing some function calling
2461 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2462 * While trying to figure out how to do the OpenBSD port, I tried to
2463 clean up some of the code in src/runtime/. In particular, I dropped
2464 support for non-POSIX signal handling, added various comments,
2465 tweaked the code to reduce the number of compilation warnings, and
2466 renamed some files to increase consistency.
2467 * To support the new automatic configuration functionality in make.sh,
2468 the source file target-features.lisp-expr has been replaced with the
2469 source file base-target-features.lisp-expr and the machine-generated
2470 file local-target-features.lisp-expr.
2471 * fixed a stupid quoting error in make.sh so that using CMU CL
2472 "lisp -batch" as cross-compilation host works again
2474 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2476 * DESCRIBE no longer tries to call itself recursively to describe
2477 bound/fbound values, so that it no longer fails on symbols which are
2478 bound to themselves (like keywords, T, and NIL).
2479 * DESCRIBE now works on generic functions.
2480 * The printer now prints less-screwed-up representations of closures
2481 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2482 * A private symbol is used instead of the :EMPTY keyword previously
2483 used to mark empty slots in hash tables. Thus
2484 (DEFVAR *HT* (MAKE-HASH-TABLE))
2485 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2486 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2487 now does what ANSI says that it should. (You can still get
2488 similar noncompliant behavior if bang on the hash table
2489 implementation with all the symbols you get back from
2490 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2491 This breaks binary compatibility, since tests for equality to
2492 :EMPTY are wired into things like the macroexpansion of
2493 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2495 * There's now a minimal placeholder implementation for CL:STEP,
2496 as required by ANSI.
2497 * An obscure bug in the interaction of the normal compiler, the byte
2498 compiler, inlining, and structure predicates has been patched
2499 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2500 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2502 * Missing ordinary arguments in a macro call are now detected even
2503 when the macro lambda list contains &KEY or &REST.
2504 * The debugger no longer complains about encountering the top of the
2505 stack when you type "FRAME 0" to explicitly instruct it to go to
2506 the top of the stack. And it now prints the frame you request even
2507 if it's the current frame (instead of saying "You are here.").
2508 * As specified by ANSI, the system now always prints keywords
2509 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2511 * The default initial SIZE of HASH-TABLEs is now smaller.
2512 * Type information from CLOS class dispatch is now propagated
2513 into DEFMETHOD bodies, so that e.g.
2514 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2516 is now basically equivalent to
2517 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2518 (DECLARE (TYPE SINGLE-FLOAT X))
2520 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2521 operation, without having to do run-time type dispatch.
2522 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2523 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2524 * Since I don't seem to be making big file reorganizations very often
2525 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2526 is overflowing my ability to conveniently back them up), I've finally
2527 checked the system into CVS. (The CVS repository is on my home system,
2528 not at SourceForge -- putting it on SourceForge might come later.)
2529 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2530 high-level GC-NOTIFY-FOO functions send their output. (There's
2531 still very little control of where low-level verbose GC functions
2532 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2533 less than it used to -- the GC-NOTIFY-FOO functions are now under
2534 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2535 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2536 in only one place in the source code, and propagates it automatically
2537 everywhere that it's needed. Thus e.g. when I bump the version from
2538 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2539 * The C source files now include boilerplate legalese and documentation
2540 at the head of each file (just as the Lisp source files already did).
2541 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2542 to his page will be replaced with a link to his new CLiki service.
2544 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2546 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2547 * The build process now looks for GNU make under the default name "gmake",
2548 instead of "make" as it used to. If GNU make is not available as "gmake"
2549 on your system, you can change this default behavior by setting the
2550 GNUMAKE environment variable.
2551 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2552 can build without error under CMU CL.
2554 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2556 * There is now a partial SBCL user manual (with some new text and some
2557 text cribbed from the CMU CL manual).
2558 * The beginnings of a profiler have been added (starting with the
2559 CMU CL profiler and simplifying and cleaning up). Eventually the
2560 main interface should be through the TRACE macro, but for now,
2561 it's still accessed through vaguely CMU-CL-style functions and macros
2562 exported from the package SB-PROFILE.
2563 * Some problems left over from porting CMU CL to the new
2564 cross-compilation bootstrap process have been cleaned up:
2565 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2566 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2567 part when getting rid of the compiler *BACKEND* stuff.)
2568 ** The value of *NULL-TYPE* was screwed up, because it was
2569 being initialized before the type system knew the final
2570 definition of the 'NULL type. This screwed up several key
2571 optimizations in the compiler, causing inefficiency in all sorts
2572 of places. (I found it because I wanted to understand why
2573 GET-INTERNAL-RUN-TIME was consing.)
2574 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2575 PROCLAIM FTYPE information. Unfortunately this broke binary
2576 compatibility again, since now the forms output by DEFGENERIC
2577 to refer to functions which didn't exist in 0.6.3.
2578 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2579 can use the new (as of 0.6.3) transform for SOME into MAP into
2581 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2582 values to (INTEGER 0 1000000), so that the time code will no longer
2583 occasionally get blown up by Linux returning 1000000 microseconds
2584 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2585 its output conform to the ANSI spec. (Alas, this makes its output
2586 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2587 * A full call to MAP NIL with a single sequence argument no longer conses.
2588 * fixes to problems pointed out by Martin Atzmueller:
2589 * The manual page no longer talks about multiprocessing as though
2590 it were currently supported.
2591 * The ILISP support patches have been removed from the distribution,
2592 because as of version 5.10.1, ILISP now supports SBCL without us
2593 having to maintain patches.
2594 * added a modified version of Raymond Toy's recent CMU CL patch for
2595 EQUALP comparison of HASH-TABLE
2597 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2599 * The system still can't cross-compile itself with
2600 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2601 entails), but at least it can compile more of itself that way
2602 than it used to be able to, and various buglets which were uncovered
2603 by trying to cross-compile itself that way have now been fixed.
2604 * This release breaks binary compatibility again. This time
2605 at least I've incremented the FASL file format version to 2, so that the
2606 problem can be detected reliably instead of just causing weird errors.
2607 * various new style warnings:
2608 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2609 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2610 ** using the lexical binding of a variable named in the *FOO* style
2611 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2612 as specified by ANSI.
2613 * *RANDOM-STATE* is no longer automatically initialized from
2614 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2615 default behavior of the system is to repeat its behavior every time
2616 it's run. If you'd like to change this behavior, you can always
2617 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2618 old convention there was no comparably easy way to get the system to
2619 repeat its behavior every time it was run.
2620 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2621 FTYPE declarations has been removed, in favor of their ANSI
2622 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2623 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2624 the types of their sequence arguments can be determined at compile time.
2625 This is done through a new open code expansion for MAP which eliminates
2626 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2627 argument types can be determined at compile time.
2628 * The optimizer now transforms COERCE into an identity operation when it
2629 can prove that the coerced object is already of the correct type. (This
2630 can be a win for machine generated code, including the output of other
2631 optimization transforms, such as the MAP transform above.)
2632 * Credit information has been moved from source file headers into CREDITS.
2633 * Source file headers have been made more standard.
2634 * The CASE macro now compiles without complaining even when it has
2637 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2639 * (Note that the way that the PCL macroexpansions were rewritten
2640 to accommodate the change in DEFGENERIC below breaks binary
2641 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2642 not run under sbcl-0.6.2. Once we get out of alpha releases,
2643 i.e. hit release 1.0.0, we'll probably try to maintain binary
2644 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2645 and sbcl-1.4.4. Until then, however, it might be fairly common
2646 for maintenance releases to break binary compatibility.)
2647 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2649 * The reporting of unhandled signals has been changed to print some
2650 explanatory text as well as the report form. (Previously only
2651 the report form was printed.)
2652 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2653 it defines, so that the compiler no longer issues undefined function
2654 warnings for compiled-but-not-yet-loaded generic functions.
2655 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2656 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2657 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2658 section of the ANSI spec are supported.
2659 * The "" nickname for the "KEYWORD" package has been removed.
2660 The reader still handles symbol tokens which begin with a package marker
2661 as keywords, but it doesn't expose its mechanism for doing so in the
2662 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2663 * The system now issues STYLE-WARNINGs for contradictory TYPE
2664 proclamations. (Warnings for contradictory FTYPE proclamations would
2665 be nice too, but those can't be done usefully unless the type system
2666 is made smarter about FUNCTION types.)
2667 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2668 systematized, so that "*target-*.lisp is supposed to exist only on the
2669 target and imply that there's a related file which exists on the
2670 host, and *host-*.lisp is supposed to exist only on the host and imply
2671 that there's a related file which exists on the target. This involves a
2672 lot of renaming. Hopefully the acute confusion caused by the renaming
2673 will be justified by the reduction in chronic confusion..
2674 ** runtime-type.lisp -> early-target-type.lisp
2675 ** target-type.lisp -> late-target-type.lisp
2676 ** early-host-format.lisp -> early-format.lisp
2677 ** late-host-format.lisp -> late-format.lisp
2678 ** host-error.lisp -> misc-error.lisp
2679 ** early-error.lisp -> early-target-error.lisp
2680 ** late-error.lisp -> late-target-error.lisp
2681 ** host-defboot.lisp -> early-defboot.lisp
2682 ** code/misc.lisp -> code/target-misc.lisp
2683 ** code/host-misc.lisp -> code/misc.lisp
2684 ** code/numbers.lisp -> code/target-numbers.lisp
2685 ** code/early-numbers.lisp -> numbers.lisp
2686 ** early-host-type.lisp -> early-type.lisp
2687 ** late-host-type.lisp -> late-type.lisp
2688 ** host-typep.lisp -> typep.lisp
2689 ** load.lisp -> target-load.lisp
2690 ** host-load.lisp -> load.lisp
2691 ** host-disassem.lisp -> disassem.lisp
2692 ** host-insts.lisp -> insts.lisp
2693 ** byte-comp.lisp -> target-byte-comp.lisp
2694 ** host-byte-comp.lisp -> byte-comp.lisp
2695 ** host-signal.lisp -> signal.lisp
2696 ** host-defstruct.lisp -> defstruct.lisp
2697 ** late-target-type.lisp -> deftypes-for-target.lisp
2698 Furthermore, several other previously target-only files foo.lisp (e.g.
2699 hash-table.lisp and random.lisp) have been split into a target-and-host
2700 foo.lisp file and a target-only target-foo.lisp file, with their key type
2701 definitions in the target-and-host part, so that the cross-compiler will
2702 know more about target types.
2703 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2704 gone away. In their place are various *BACKEND-FOO* variables
2705 corresponding to the slots of the old structure.
2706 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2707 into the target SBCL has been fixed.
2708 * The chill.lisp system for loading cold code into a running SBCL
2710 * Support for the CMU CL "scavenger hook" extension has been removed.
2711 (It was undocumented and unused in the CMU CL sources that SBCL was
2712 derived from, and stale in sbcl-0.6.1.)
2713 * Various errors in the cross-compiler type system were detected
2714 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2715 (enabling various consistency checks). Many of them were fixed,
2716 but some hard problems remain, so the compiler is back to
2717 running without *TYPE-SYSTEM-INITIALIZED* for now.
2718 * As part of the cross-compiler type system cleanup, I implemented
2719 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2720 * I have started adding UNCROSS calls throughout the type system
2721 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2722 on cross-compiler input files will be able to go away, and various
2724 * CONSTANTP now returns true for quoted forms (as explicitly required
2727 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2729 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2730 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2731 (MAKE-LIST -1). In the longer term, it ought to become true
2732 that declarations are assertions even at SAFETY 1. For now, it's not
2733 quite true even at SAFETY 3, but it's at least more nearly true..
2734 (Note that this change seems to increases the size of the system by
2735 O(5%) and to decrease the speed of the compiler by 20% or more.)
2736 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2737 for the problem of printing dozens of lines of distracting information
2738 about low-level system machinery as part of the top stack frame
2739 on entry to the debugger when an undefined function was called.
2740 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2741 is not reset to COMMON-LISP-USER.
2742 * Compilation of stuff related to dyncount.lisp has been made conditional
2743 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2744 smaller. The various dyncount-related symbols have been moved into
2745 a new "SB-DYNCOUNT" package.
2746 * tty-inspect.lisp has been renamed to inspect.lisp.
2747 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2748 feature has gone away. (When we eventually port to other flavors of
2749 libc and/or Unix, we'll try to make the differences between flavors
2750 invisible at the user level.)
2751 * Various other *FEATURES* tags, and/or their associated conditionals,
2752 have been removed if obsolescent, or given better documentation, or
2753 sometimes given more-mnemonic names.
2755 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2757 * tidied up "make.sh" script
2758 * tidied up system directory structure
2759 * better "clean.sh" behavior
2760 * added doc/FOR-CMUCL-DEVELOPERS
2761 * many many small tweaks to output format, e.g. removing possibly-confusing
2762 trailing #\. character in DESCRIBE-INSTANCE
2763 * (EQUALP #\A 'A) no longer signals an error.
2764 * new hashing code, including EQUALP hashing
2765 * tidied up Lisp initialization and toplevel
2766 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2767 * command line argument processing
2768 * added POSIX-GETENV function to deal with Unix-ish environment variables
2769 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2770 terminating SBCL on EOF
2771 * non-verbose GC by default
2772 * There is no more "sbcl" shell script; the sbcl file is now the C
2773 runtime executable (just like CMU CL).
2774 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2776 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2777 symbol and package information as Lisp native symbol and package objects
2778 instead of strings naming symbols and strings naming packages. This way,
2779 whenever packages are renamed (as in warm init), debug information is
2780 transformed along with everything else.
2781 * tweaked the optimization policy declarations which control the building
2782 of SBCL itself. Now, among other things, the system no longer saves
2783 source location debugging information. (This helps two problems at once
2784 by reducing SBCL size and by keeping SBCL from trying to look for its
2785 sources -- which may not exist -- when reporting errors.)
2786 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2787 debugging and testing purposes
2788 * cleaned up printing, making the printer call PRINT-OBJECT for
2789 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2790 methods, giving nearly-ANSI behavior
2791 * converted almost all special variables to use *FOO* naming convention
2792 * deleted PARSE-TIME functionality, since it can be done portably
2793 * moved some files out of cold init into warm init
2794 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2796 * regularized formatting of source files
2797 * added an install.sh script
2798 * fixed ridiculous memory usage of cross-compiler by making
2799 compiler/alloc.lisp not try to do pooling unless it can hook
2800 itself into the GC of the cross-compilation host. Now the system
2801 builds nicely on my old laptop.
2802 * added :SB-ALLOC in target-features.lisp-expr
2803 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2804 was not implemented)
2805 * re-did condition handling and note reporting in the compiler. Notes
2806 are no longer handled by signalling conditions. Style warnings
2807 and warnings are handled more correctly and reported in such a way
2808 that it's easy to find one or the other in your output (so that you
2809 can e.g. figure out which of many problems caused COMPILE-FILE to
2811 * changed the severity of several compiler warnings from full WARNING
2812 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2813 compiler note reporting so that it doesn't use the condition system
2814 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2815 COMPILE-FILE command)
2816 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2817 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2818 no longer supported, since their implementation was deeply intertwingled
2819 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2820 * removed bogus "support" for compiler macros named (SETF FOO), and
2821 removed the compiler macro for SETF INFO (but only after making a fool
2822 of myself on the cmucl-imp mailing list by posting a bogus patch for
2823 DEFINE-COMPILER-MACRO..)
2824 * Compiled files containing forms which have side effects on the Lisp
2825 reader (such as DEFPACKAGE forms) are now handled more correctly.
2826 (Compiler queuing of top level lambdas has been suppressed by setting
2827 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2828 * deleted various currently-unused source files, e.g. gengc.lisp. They
2829 may be added back at some point e.g. when porting to other architectures,
2830 but until they are it's distracting to distribute them and to try to
2832 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2833 there were so many of them they're just distractions, and UNCROSS is
2834 known to be able to handle the current sources
2835 * moved PROFILE functionality into TRACE, so that it will be clear
2836 how the wrapping and unwrapping of functions when you profile them
2837 interacts with the wrapping and unwrapping of functions when you
2838 trace them. (Actually, the functionality isn't there yet, but at least
2839 the interface specification is there. Hopefully, the functionality will
2840 arrive with some future maintenance release.)
2841 * removed host-oops.lisp
2842 * changed signature of QUIT function to allow UNIX-CODE argument
2843 * fixed READ-SEQUENCE bug
2844 * tweaked verbose GC output so that it looks more like the progress
2845 output that ANSI specifies for functions like LOAD
2846 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2847 * added <http://sbcl.sourceforge.com> to the banner information printed by