1 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2 * enhancement: saving cores with foreign code loaded is now
3 supported on ppc/Darwin in addition to the previously supported
6 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
7 * new feature: reloading changed shared object files with
8 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
9 * new feature: references to foreign variables and functions
10 can now be compiled and loaded before the corresponding shared
11 object file is loaded, as long as the foreign definitions are
13 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
14 supported. (thanks to Dan Debertin)
15 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
16 just keywords. (thanks to Vincent Arkesteijn)
17 * fixed bug #331: structure-class instances corresponding to
18 DEFSTRUCT forms are now created eagerly.
19 * fixed bug #345: backtraces from calls to undefined functions work
20 on x86 as well. Related bug #61 is now also partially fixed on x86
21 (backtraces from throws to unknown catch tags.)
22 * bug fix: lambda-list parsing is now stricter vrt. order and number
23 of lambda-list keywords.
24 * bug fix: as specified by AMOP, an error is signalled if a
25 class-option appears multiple times in a DEFCLASS form. (reported
27 * bug fix: RANDOM can be compiled when the compiler derives the type
28 of its numeric argument as a disjoint set of small integers.
29 (reported by Paul Dietz)
30 * bug fix: starting a core saved with shared objects loaded when
31 those objects are not available no longer causes threaded SBCL to
32 hang. (reported by Sean Ross)
33 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
35 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
36 argument, which must always be NIL. (reported by Kalle Niemitalo)
37 * bug fix: printing 1.0d+23 no longer results in an error.
38 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
39 * bug fix: structure slot setters preserve evaluation order. (thanks
41 * fixed some bugs related to Unicode integration:
42 ** RUN-PROGRAM can allow its child to take input from a Lisp
43 stream. (reported by Stefan Scholl)
44 ** ASDF-INSTALL successfully downloads .tar.gz files in all
45 locales. (reported by Ken Causey)
46 * fixed some bugs revealed by Paul Dietz' test suite:
47 ** INCF, DECF and REMF evaluate their place form as specified in
49 ** FORMATTER expands ~{ ~} iteration directives with V or #
51 ** FORMATTER deals with the ~@[ ~] conditional directive where the
52 consequent uses no arguments correctly.
53 ** the system has a partial understanding of the (COMPLEX RATIO)
55 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
58 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
59 * new feature: a build-time option (controlled by the :SB-UNICODE
60 keyword feature, enabled by default) for building the system with
61 support for the entire 21-bit character space defined by the
63 * new feature: the system now has rudimentary external-format
64 support; the primary user-visible change at this time is that
65 characters with the high bit set (such as lower-case-e-acute) will
66 print correctly to a terminal in a UTF-8 environment.
67 * minor incompatible change: BASE-CHAR no longer names a class;
68 however, CHARACTER continues to do so, as required by ANSI.
69 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
70 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
71 has been moved to the SB-EXT package (temporarily re-exported from
73 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
74 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
76 * fixed bug #351: better error-handlind and reporting for malformed
78 * fixed bug #350: bignum-printing is now more memory-efficient,
79 allowing printing of very large bignums, eg. (expt 2 10000000).
80 (reported by Bruno Haible)
81 * fixed bug #302: better primitive-type selection for intersection
83 * fixed bug #308: non-graphic characters now all have names, as
84 required. (reported by Bruno Haible)
85 * bug fix: redefining a class with different superclasses now correctly
86 removes it from the direct-subclasses of its previous superclasses.
87 (reported by David Morse)
88 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
89 new value now works. (reported by Bruno Haible)
90 * bug fix: correct canonicalization of multiple non-standard slot
91 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
92 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
93 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
94 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
95 now exists, an signals an error.
96 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
97 during macroexpansion for non-iteration variables as well. (reported
98 by Bruno Haible for CMUCL)
99 * bug fix: Cyclic structures and unprintable objects in compiler
100 messages no longer cause errors. (reported by Bruno Haible)
101 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
102 and READ-FROM-STRING all now return a primary value of NIL if
103 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
104 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
105 (reported by Bruno Haible for CMUCL)
106 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
107 do not designate a condition. (reported by Bruno Haible for
109 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
110 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
111 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
112 permitting resolution in favour of any of the conflicting symbols.
113 (reported by Bruno Haible for CMUCL)
114 * bug fix: EQUAL compiler optimizations is less aggressive on
115 strings which can potentially compare true despite having distinct
116 specialized array element types.
117 * bug fix: unit enumerations can be defined without dividing by
118 zero. (thanks to Vincent Arkesteijn)
119 * FORMAT compile-time argument count checking has been enhanced.
120 (report from Bruno Haible for CMUCL)
121 * a partial workaround for the bug 262: the compiler does not try to
122 inline-expand a local function doing RETURN-FROM from a deleted
123 BLOCK. (thanks to Peter Denno for the bug report and to David
124 Wragg for the simple test case)
125 * fixed some bugs revealed by Paul Dietz' test suite:
126 ** RENAME-PACKAGE allows all package designators as new package
128 ** constraint propagation and lambda variable substitution are
129 more cautious in dealing with partially deleted code.
130 ** compiler sometimes lost reoptimization passes.
131 ** CERROR, given a condition as condition designator, passes its
132 remaining arguments to the continue format control without
134 ** Case-altering FORMAT directives work correctly on non-ASCII
136 ** The REAL type specifier handles bounds outside the floating
137 point ranges without signalling FLOATING-POINT-OVERFLOW.
138 ** Functions with IR1-transformations can create intercomponent
139 references to global functions.
140 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
142 ** FORMAT ~R treats a nil value for its first parameter correctly.
144 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
145 * enhancement: saving cores with foreign code loaded is now
146 supported on x86/NetBSD and sparc/Linux in addition to the previously
148 * bug fix: on some platforms repeated installations caused multiple
149 copies of HTML documentation to be installed -- should not happen
150 any more. (reported by Stefan Scholl)
151 * bug fix: parsing self-recursive alien record types multiple times
152 no longer causes infinite recursion. (reported by Thomas F. Burdick,
153 original patch by Helmut Eller for CMUCL)
154 * bug fix: stack-exhaustion detection works now on NetBSD as well.
155 (thanks to Richard Kreuter)
156 * bug fix: defining classes whose accessors are methods on existing
157 generic functions in other (locked) packages no longer signals
158 bogus package lock violations. (reported by François-René Rideau)
159 * bug fix: special variables as DEFMETHOD parameters no longer have
160 associated bogus type declarations. (reported by David Wragg and
162 * bug fix: read-write consistency on streams of element-type
163 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
164 * bug fix: redefiniton of the only method of a generic function with
165 no DEFGENERIC no longer emits a full WARNING. In addition,
166 redefinition of generic functions with no DEFGENERIC to an
167 incompatible lambda list now signals an error. (thanks to Zach
169 * bug fix: DEFGENERIC now works even when there's a function of the
170 same name in an enclosing lexical environment. (thanks to Zach
172 * fixed compiler failure, caused by instrumenting code during
173 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
174 * optimization: added loop analysis and improved register allocation
175 * fixed some bugs revealed by Paul Dietz' test suite:
176 ** POSITION on displaced vectors with non-zero displacement
177 returns the right answer.
178 ** (SIMPLE-STRING) is a valid type specifier for sequence
180 ** *PRINT-LEVEL* handling for slotless structures is pedantically
182 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
183 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
185 ** The justification version of the FORMAT ~< directive treats
186 non-zero minpad parameter correctly.
188 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
189 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
190 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
191 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
192 the supported interface.
193 * new feature: Single-stepping of code compiled with DEBUG 2 or
194 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
195 * new feature: saving cores with foreign code loaded is now
196 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
197 Timothy Moore's work for CMUCL)
198 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
199 parameters to * instead of NIL if no initform is supplied.
200 (reported by Johan Bockgaard)
201 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
202 correctly when FUNCALL appears as the car of the form. Note:
203 despite this FUNCALL forms are not currently subject to
204 compiler-macro expansion. (port of Raymond Toy's fix for the
205 same from CMUCL, reported by Johan Bockgaard)
206 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
207 (thanks for Teemu Kalvas)
208 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
209 case instead of terminating the iteration (thanks for Julian
210 Squires, Sean Champ and Raymond Toy)
211 * bug fix: incorrect expansion of defgeneric that caused a style
212 warning. (thanks for Zach Beane)
213 * bug fix: slot accessor effective method computation works properly
214 for classes with multiple non-standard applicable methods on
215 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
216 * on x86 compiler supports stack allocation of results of LIST and
217 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
218 CMUCL implementation by Gerd Moellmann)
219 * fixed some bugs revealed by Paul Dietz' test suite:
220 ** FORMAT strings with both the ~<~:;~> form of the justification
221 directive and pretty-printing directives cause an error.
223 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
224 * incompatible change: the internal functions
225 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
226 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
227 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
228 instead of the old functions.
229 * new feature: on platforms where "dladdr" is available foreign
230 function names now appear in backtraces. (based on Helmut Eller's
232 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
233 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
235 * documentation: ASDF manual is now distributed with SBCL.
236 * bug fix: SBCL can now load its contributed modules with REQUIRE
237 even if the system-provided entries have been removed from
238 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
240 * bug fix: backtraces involving undefined functions or assembly
241 routines are more informative. (thanks to Brian Downing)
242 * bug fix: mutually referent alien structures now work correctly.
243 (reported by Rick Taube)
244 * bug fix: structures defined by WITH-ALIEN can be referred to
245 within other definitions in the same WITH-ALIEN.
246 * bug fix: division operators (MOD, TRUNCATE and the like) with
247 constant zero divisors and integer dividends no longer generate
249 * bug fix: provide default methods for INPUT-STREAM-P and
250 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
251 * bug fix: improve the detection and resolution of MOP metacycles
252 (where effective method computation depends on the generic function
253 having its effective method computed). (reported by Bruno Haible)
254 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
255 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
256 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
257 * bug fix: pathnames with relative directory components can be
258 represented relative to default pathnames.
259 * optimization: in taking the GCD of bignums, reduce the two bignums
260 to approximately the same size (using Euclid's algorithm) before
261 applying the more sophisticated binary GCD. (thanks to Juho
263 * optimization: COUNT on bitvectors now operates word-at-a-time.
264 * optimization: ASH with a positive, but not necessarily constant,
265 (leftwards) shift, when in a modular context, is compiled to a
267 * fixed some bugs revealed by Paul Dietz' test suite:
268 ** FORMAT variable parameters ("~V<char>") are defaulted properly
269 if the corresponding argument is NIL.
270 ** FORMAT directives accepting a minpad parameter treat negative
272 ** literal commas as character parameters to FORMAT directives are
274 ** literal spaces directly after ~<Newline> directives within a
275 format-logical-block (~:< ~@:>) do not induce :FILL-style
276 conditional newlines.
277 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
278 as specified: it no longer includes conditional newlines).
279 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
281 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
282 :PER-LINE-PREFIX argument does not evaluate to a string.
284 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
285 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
286 the manual for details; for now, package locks can be disabled by
287 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
288 if no major problems are found then it is likely that they will be
289 compiled in unconditionally.
290 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
291 unsupported operators on all platforms. To load a shared library
292 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
293 object file, link it into a shared library outside of SBCL and
294 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
296 * bug workaround: ROOM T and the GC were not working together
297 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
298 reporting facility aren't preserved. That reporting has been
299 disabled, so now until and unless someone figures out how to make it
300 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
301 * minor incompatible change: as threatened around sbcl-0.8.0, the
302 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
303 The symbols are also exported from SB-PCL for backwards
304 compatibility, but more so than before SB-PCL should be treated as
305 an implementation-internal package.
306 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
308 * the SB-SPROF contrib now works on (most) non-x86 architectures.
309 It is known as of this release not to work on the Alpha, however.
310 * fixed bug #167: errors signalled due to illegal syntax in method
311 bodies are now more legible.
312 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
313 designators and can hence be used with TYPEP.
314 * fixed bug #333: CHECK-TYPE now ensures that the type error
315 signalled, if any, has the right object to be accessed by
316 TYPE-ERROR-DATUM. (reported by Tony Martinez)
317 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
318 5.1.2.3 for multiple-value place subforms. (reported by Kalle
320 * fixed bug #334: programmatic addition of slots using specialized
321 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
322 and :ALLOCATION :CLASS slots.
323 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
324 float exponents. (rereported by Peter Seibel)
325 * fixed a bug: #\Space (and other whitespace characters) are no
326 longer considered to be macro characters in standard syntax by
328 * fixed bug: initialization of condition class metaobjects no longer
329 causes an instance of the condition to be created. (reported by
331 * fixed bug: it is now possible to have slots such that
332 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
333 description is neither :INSTANCE nor :CLASS.
334 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
335 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
336 SLOT-VALUE-USING-CLASS), and no longer causes errors with
337 non-standard SLOT-DEFINITION-ALLOCATION values.
338 * fixed bugs: various race conditions handling exiting threads.
339 CL-PPCRE's thread torture test now passes.
340 * fixed arguably-a-bug: GC time proportional to number of threads in
341 system even when most of them are idle
342 * optimization: improved performance of BIT and SBIT on bit-vectors.
343 * .fasl file incompatibility: The fasl file version number has
344 been incremented because of changes associated with package locks.
346 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
347 * minor incompatible change: the system no longer provides
348 optimization hints (conditions which are TYPEP
349 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
350 mode; these hints are still emitted when higher SPEED optimization
352 * new contrib module: a sampling profiler (profiling by statistical
353 sampling, rather than by instrumenting functions) is available as
354 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
355 * the behaviour of the standard function ED is now customizeable by
356 third parties through a hook variable: see ED's documentation
357 string for information on the protocol.
358 * the compiler no longer emits efficiency notes for (FUNCALL X)
359 when the type of X is uncertain under default optimization
361 * fixed bug 276: mutating a binding of a specialized parameter to a
362 method to something that is not TYPEP the specializer is now
364 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
365 exhibit the required behaviour on the x86 platform. (thanks to
366 Peter van Eynde, Eric Marsden and Bruno Haible)
367 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
368 even for difficult arguments. (reported by Peter Graves)
369 * fixed bug 141a: the backquote printer now descends quoted
371 * fixed another bug in backquote printing: no more destructive
372 modification of the form's list structure. (reported by Brian
374 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
375 move between its address being taken and the call to
376 interrupt_thread, fixing a crashing race condition.
377 * the SB-POSIX contrib implementation has been adjusted so that it
378 no longer exhibits ridiculously poor performance when constructing
379 instances corresponding to C structs.
381 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
382 * minor incompatible change: the sb-grovel contrib now treats C
383 structures as alien (in the sense of SB-ALIEN) objects rather than
384 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
385 has implications for memory management of client code
386 (sb-grovel-returned objects must now be manually managed) and for
387 type safety (alien objects now have full types).
388 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
389 used to control emission of compiler diagnostics, rather than the
390 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
391 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
392 quality should be considered deprecated.
393 * install.sh now installs the user manual as well
394 * (not quite a new documentable feature, but worth considering in
395 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
396 beginnings of a semantically meaningful condition hierarchy is
397 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
399 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
400 designator as the defaults argument.
401 * fixed bug: Displaced arrays whose displaced-to array has become
402 too small now cause ARRAY-DIMENSION to signal an error, providing
403 for safer bounds-checking. (reported by Bruno Haible)
404 * fixed bug: DEFCLASS slot definitions with identical :READER and
405 :WRITER names now signal a reasonable error. (reported by Thomas
407 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
409 * fixed bug: Invalid slot specification errors now print correctly.
410 (thanks to Zach Beane)
411 * fixed bug 320: Shared to local slot value transfers in class
412 redefinitions now happen corrently from superclasses as
413 well. (reported by Bruno Haible)
414 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
416 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
417 improper lists correctly. (reported by Bruno Haible)
418 * fixed bug 313: source-transform for <fun-name> was erroneously
419 applied to a call of a value of a variable with name <fun-name>.
420 (reported by Antonio Menezes Leitao)
421 * fixed bug 307: The obsolete instance protocol ensures that
422 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
423 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
424 remove dead unknown-values globs from the middle of the stack.
425 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
426 easier to resume long computations after using *BREAK-ON-SIGNALS*
427 to diagnose and fix failures. (thanks to Nikodemus Siivola)
428 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
429 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
430 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
431 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
432 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
434 * fixed a (fairly theoretical) bug in string printing: if
435 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
436 does not have array-element-type equal to the most general string
438 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
439 function name. (thanks to Nikodemus Siivola)
440 * fixed bug: compile-time format string checker failed on
441 non-closed ~{. (reported by Thomas F Burdick)
442 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
443 don't warn on legal loop code involving more than one aggregate
444 boolean. (thanks to Nikodemus Siivola)
445 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
446 in keyword position to MAKE-PACKAGE was not being reported as
448 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
449 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
451 * fixed bugs 280 and 312: the checking for multiple definitions in a
452 file is less likely to become confused by uses of inline
454 * fixed bug: the #S reader macro performs the keyword coercion
455 specified for slot names. (reported by Kalle Niemitalo)
456 * fixed bug: lambda lists may contain symbols whose names start with
457 & but are not lambda-list-keywords; their occurrence triggers a
459 * fixed bug 321: define-method-combination argument lambda lists do
460 not cause invalid code to be generated when &optional or &aux
461 variables have default values. (reported by Bruno Haible)
462 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
463 have CLOS classes; this appears also to have fixed reported
464 instability in stack exhaustion detection.
465 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
466 MUFFLE-WARNING when no associated restart is present is now
468 * optimization: rearranged the expansion of various defining macros
469 so that each expands into only one top-level form in a
470 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
472 * optimization: used a previously little-used slot in symbols to
473 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
475 * fixed some bugs revealed by Paul Dietz' test suite:
476 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
478 ** Fixed an optimization bug related to inheritance of initforms
479 from local to shared slots.
480 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
481 ** CAST optimizer forgot to flush argument derived type.
482 ** print/read consistency on floats is now orders of magnitude
483 more likely. (thanks also to Bruno Haible for a similar report
485 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
486 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
487 for objects of type REAL. Make it so.
488 ** SXHASH is specified (infelicitously) to respect similarity,
489 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
490 it so. (thanks to Markus Ziegler)
491 ** on the Alpha, the compiler succeeds in compiling functions
492 returning a known number of arguments greater than 63.
493 ** fixed handling of invalid NIL arguments in keyword position
495 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
497 ** provided a readably-printable representation for RANDOM-STATE
499 ** ensured that pathnames, where they have a namestring, always
500 print using #P"..." syntax.
502 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
503 * Support for the forthcoming 2.0 version of the NetBSD kernel
504 running on x86 hardware has been added. (thanks to Perry
505 E. Metzger most immediately, and others for their past work)
506 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
507 current release version; SBCL's OpenBSD support had been broken
508 since about the time of OpenBSD's switch to ELF binary format.)
509 * [placeholder for DX summary]
510 ** user code with &REST lists declared dynamic-extent, under high
511 speed or space and low safety and debug optimization policy.
512 * The manual has been converted to Texinfo format and the debugger
513 chapter from the cmucl manual has been added.
514 * A facility has been added to extract documentation strings from
515 sbcl and store them as Texinfo-formatted snippets for inclusion in
516 the manual (via Texinfo's @include directive)
517 * bug fix: compiler emitted division in optimized DEREF. (thanks for
518 the test case to Dave Roberts)
519 * bug fix: multidimensional simple arrays loaded from FASLs had fill
520 pointers. (reported by Sean Ross)
521 * bug fix: PROFILE output is printed nicely even for large numerical
522 values. (thanks to Zach Beane)
523 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
524 greater than 32 handle EOF correctly.
525 * bug fix: on X86 an immediate argument of the IMUL instruction is
526 correctly printed in disassembly. (thanks to Lutz Euler)
527 * bug fix: class slots in redefined classes preserve their old
528 values. (thanks to Bruno Haible and Nikodemus Siivola)
529 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
530 succeeds. (reported by Marco Baringer)
531 * fixed some bugs revealed by Paul Dietz' test suite:
532 ** READ-SEQUENCE now works on ECHO-STREAMs.
533 ** RATIONALIZE works more according to its specification. (thanks
536 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
537 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
538 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
539 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
540 should matter to you only if you rebind the printer control
541 variables and then find you want different bindings in the
542 debugger than in the ordinary execution of your program.)
543 * The runtime build system has been tweaked to support building
544 (on SPARC/SunOS) using a C compiler which invokes Sun's own
545 assembler and linker. (thanks to Nikodemus Siivola)
546 * Unbound, undefined, undeclared variables now trigger full
547 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
548 more often programmer error than deliberate exploitation of undefined
550 * optimization: the hash algorithm for strings has changed to one
551 that is less vulnerable to spurious collisions. (thanks to Juho
553 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
554 needless bounds checking. (thanks to Juho Snellman)
555 * optimization: implemented multiplication as a modular
556 (UNSIGNED-BYTE 32) operation on the PPC backend.
557 * fixed some bugs revealed by Paul Dietz' test suite:
558 ** ADJUST-ARRAY now copies the datum in a zero rank array if
560 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
561 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
562 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
564 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
566 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
568 ** ECHO-STREAMs no longer attempt to echo the end of file value to
569 their output stream on EOF from read.
570 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
571 have been read to end-of-file.
572 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
574 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
575 description of determination of which consecutive characters
577 ** Printing the "Space" character with escaping on now yields "#\\ ",
578 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
579 ** Reading floating-point numbers with *READ-BASE* set to a number
580 less than 10 works correctly.
581 ** Reading floating-point numbers with *READ-BASE* set to a number
582 more than 10 works correctly.
583 ** Printing with *PRINT-READABLY* targets the standard readtable, not
584 the readtable currently in effect.
586 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
587 * minor incompatible change: parsing of namestrings on a physical
588 (Unix) host has changed; numbers after the final #\. in a
589 namestring are no longer interpreted as a version field. This is
590 intented to be largely invisible to the user, except that the
591 meaning of the namestring "*.*.*" has changed: it now refers to a
592 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
593 should usually be replaced by
594 (make-pathname :name :wild :type :wild :version :wild)
595 with the added benefit that this is more likely to be portable.
596 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
597 option to OPEN now signals an error if the file being opened
598 exists; this may have an impact on existing code.
599 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
600 signals received, on the PowerPC platforms (both Linux and
601 Darwin). (thanks to Pierre Mai for pointing out the location of
603 * several fixes on OS X: The system now builds and runs cleanly on
604 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
605 (thanks to Brian Mastenbrook)
606 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
607 second-resolution: integer multiples of 1/3600 between -24 and 24.
608 (thanks to Vincent Arkesteijn)
609 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
610 type when called with 1 argument; PEEK-CHAR checked type of
611 PEEK-TYPE only after having read first character from a
612 stream. (reported by Peter Graves)
613 * bug fix: the garbage collector now has much better locality
614 behaviour, and in particular no longer treats all memory as being
615 exhausted when many small objects point to each other in a deeply
617 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
618 recognized as being TYPEP their class.
619 * bug fix: the PUSHNEW documentation string has been corrected.
620 (thanks to Vincent Arkesteijn)
621 * bug fix: defaulting of the value for the last of an atypically
622 large number of multiple values being bound was not being
623 performed correctly on the Alpha or PPC platforms
624 * optimization: implemented multiplication as a modular
625 (UNSIGNED-BYTE 32) operation on the x86 backend.
626 * optimization: SEARCH on simple-base-strings can now be open-coded.
627 (see also contrib/compiler-extras.lisp for inspiration for
628 teaching the compiler about the Boyer-Moore algorithm).
629 * value, returned by MAX (and MIN) called with several EQUALP, but
630 not EQL, arguments now does not depend on compiler settings.
631 * fixed some bugs revealed by Paul Dietz' test suite:
632 ** in stack analysis liveness information is propagated from
633 non-local entry points.
634 ** pathwise CAST removing failed when the CAST node did not start
636 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
637 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
639 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
640 host is already defined.
641 ** RENAME-FILE works on streams instead of signalling an internal
643 ** PEEK-CHAR uses the current readtable when determining whether
644 or not a character is whitespace.
645 ** MERGE-PATHNAMES handles the case when the pathname does not
646 specify a name while the default-pathname specifies a version.
647 ** Pathnames now stand a better chance of respecting print/read
649 ** Attempting to use standardized file system operators with a
650 pathname with invalid :DIRECTORY components signals a
652 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
654 ** DIRECTORY on logical pathnames is more correct.
655 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
656 signal a TYPE-ERROR if their argument is not a stream.
657 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
658 designator argument does not designate a stream.
659 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
660 examining the synonym.
661 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
663 ** OPEN and WITH-OPEN-STREAM allow opening streams with
664 element-type larger than ([UN]SIGNED-BYTE 32).
666 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
667 * When built with the :SB-FUTEX feature, threaded builds now take
668 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
669 for faster/more reliable mutex and condition variable support.
670 * Incompatible change (but one you probably shouldn't have been using
671 anyway): the interface and code for arbitrating between multiple
672 threads in the same user session has been redesigned.
673 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
674 error when called without an explicit environment argument.
675 (thanks to Nikodemus Siivola)
676 * bug fix: buffered :DIRECTION :IO streams are less likely to become
677 confused about their position. (thanks to Adam Warner and Gerd
679 * bug fix: Pretty printing backquoted forms with unquotations in the
680 argument list position of various code constructs such as LAMBDA
681 now works correctly. (reported by Paul Dietz)
682 * bug fix: Pretty printing unquotations no longer loses all
683 stream position information.
684 * optimization: performance of string output streams is now less
685 poor for multiple small sequence writes.
686 * optimization: performance of CSUBTYPEP in the presence of complex
687 expressions involving CONS and NOT many times has been improved.
688 (reported by Paul Dietz)
689 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
691 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
692 streams for the REPL, for the debugger, and so on. (thanks to
694 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
695 * fixed some bugs revealed by Paul Dietz' test suite:
696 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
697 is ignored if it is too small, rather than propagating through
698 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
699 ** extremely complex negations of CONS types were not being
700 sufficiently canonicalized, leading to inconsistencies in
702 ** VALUES tranformer lost derived type.
704 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
705 * fixed a bootstrapping bug: the build process no longer assumes
706 that the various BOOLE-related constants have the same value in
707 host and target lisps. (noted by Paul Dietz' test suite on an
708 SBCL binary built from CLISP)
709 * The system can now be dynamically linked on the MIPS platform,
710 which enables dynamic loading of foreign code from Lisp. (thanks
711 to Ralf Baechle for discussions on the MIPS ABI)
712 * The system now records debugging information for its own source
713 files in a filesystem-position-independent manner, relative to
714 the "SYS" logical host.
715 * fixed a compiler bug: MV-LET convertion did not check references
716 to the "max args" entry point. (reported by Brian Downing)
717 * tweaked disassembly notes to be less confident about proclaiming
718 some instruction as an LRA. (thanks to Brian Downing)
719 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
720 now each have their own history, command character, and other
721 characteristics. (thanks to David Lichteblau)
722 * fixed some bugs revealed by Paul Dietz' test suite:
723 ** compiler failure in compiling LOGAND expressions including a
725 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
726 shift greater than 32.
727 ** FLUSH-DEST did not mark blocks for type check regeneration.
728 ** HANDLER-CASE failed to accept declarations in handler clauses
729 in some circumstances.
731 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
732 * New code in contrib/sb-introspect (still probably not entirely
733 stable yet) provides some support for smart Lisp development
734 environments like SLIME.
735 * The conditions signalled for errors occurring when loading .fasl
736 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
737 in a way which should help ASDF recover gracefully.
738 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
739 stuff has been cleaned up. If you code contrib/ stuff, this might
740 affect you, and you can look at contrib/README, contrib/STANDARDS,
741 and/or the 0.8.4.27 diff to check.
742 * In full calls the compiler now does not generate checks for declared
743 argument types for all arguments.
744 * various threading fixes
745 ** and some experimental patches which didn't make it into
746 the main tree for this release, but which are shipped in
747 contrib/experimental-thread.patch as a possible fix for some
748 failures (deadlock, spinning...) in GC-intensive multithreaded
750 * fixed PPC build problem (source code incompatibility of different
751 library versions): added offsetof() hackery which attempts to divine
752 where glibc maintainers put uc_mcontext today
753 * fixed bug 282: compiler does not trust type assertions while passing
754 arguments to a full call.
755 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
756 &OPTIONAL VALUES type parameter.
757 * fix bug 214: algorithm for noting rejected templates is now more
758 similar to that of template seletion. (also reported by rydis on
760 * fixed bug 141b: printing backquoted information readably and prettily
761 inserts a space where necessary.
762 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
763 (CHARACTER 10) are now reported as errors, rather than propagated
764 as unknown types. (reported by piso on #lisp)
765 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
766 with values NIL and :ERROR. (thanks to Milan Zamazal)
767 * fixed bug 191c: CLOS now does proper keyword argument checking as
768 described in CLHS 7.6.5 and 7.6.5.1.
769 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
770 longer raise an error; further, using a list as a for-as-arithmetic
771 counter now raises a meaningful error.
772 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
773 now understood by sequence creation functions such as MAKE-SEQUENCE
775 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
776 a STRING-INPUT-STREAM.
777 * compiler enhancement: SIGNUM is now better able to derive the type
779 * type declarations inside WITH-SLOTS are checked. (reported by
781 * fixed some bugs revealed by Paul Dietz' test suite:
782 ** incorrect optimization of TRUNCATE for a positive first
783 argument and negative second.
784 ** compiler failure in let-convertion during flushing dead code.
785 ** compiler failure while deriving type of TRUNCATE on an
786 interval, containing 0.
787 ** ASH of a negative bignum by a negative bignum count now returns
789 ** intersection of CONS types now canonicalizes properly, fixing
790 inconsistencies in SUBTYPEP.
792 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
793 * incompatible change: The --disable-debugger command line
794 option now clobbers the debugger at a more fundamental
795 level, by redefining #'INVOKE-DEBUGGER instead of by
796 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
797 is specified by ANSI to ignore *DEBUGGER-HOOK* and
798 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
799 enter the debugger REPL and then suffer recursive errors
800 because *DEBUG-IO* is also messed up in --disable-debugger mode;
801 while under the new system, BREAK in --disable-debugger mode
802 terminates the system just as an unhandled error would.
803 * fixed compiler performance when processing loops with a step >1;
804 * bug fix: DOCUMENTATION now retrieves generic function
805 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
806 support has been systematized, and now supports the methods
807 specified by ANSI, along with a default method and a method for
808 slot documentation. (reported by Nathan Froyd)
809 * bug fix: effective methods associated with a generic function are
810 no longer cached over a change of that generic function's method
811 combination. (reported by Andreas Fuchs)
812 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
813 names a non-existent directory. (thanks to Andreas Fuchs)
814 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
815 the PPC platform, lead to stack corruption; this has been fixed.
816 (reported by Rainer Joswig)
817 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
818 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
819 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
820 arithmetic in general, is now much more reliable on the PPC
822 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
823 platform now returns the right answer.
824 * optimization: restored some effective method precomputation in
825 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
826 precomputation is now tunable.
827 * optimization: compiler-internal data structure use has been
828 reviewed, and changes have been made that should improve the
829 performance of the compiler by about 20%.
830 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
831 simple-base-strings and simple-bit-vectors is improved.
832 * optimization: the optimization of 32-bit logical and arithmetic
833 functions introduced in version 0.8.3 on the x86 has been
834 implemented on the mips, ppc and sparc platforms; an
835 implementation of the same facility, but for 64-bit arithmetic,
836 has been added for the alpha.
837 * microoptimization: the compiler is better able to make use of the
838 x86 LEA instruction for multiplication by constants.
839 * bug fix: in some situations compiler did not report usage of
840 generic arithmetic in (SPEED 3) policy.
841 * bug 145b fix: compiler used wrong type specifier while converting
842 MEMBER-types to numeric.
843 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
845 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
846 index argument now works properly on the Alpha platform.
847 * bug fix: floating point exception treatment on the Alpha platform
849 * bug fix: FILE-POSITION works much better on string input and
850 output streams. (thanks to Nikodemus Siivola)
851 * bug fix: many threading/garbage collection symptoms sorted.
852 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
853 might be pseudo-atomic.
854 * internal change: Stopping for GC is now done with signals not
855 ptrace. GC is now done in whichever thread wanted it, instead of
857 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
859 * bug fix: result form in DO is not contained in the implicit
861 * incompatible change: ICR structure is changed; the value part of
862 CONTINUATION is now called LVAR; corresponding functions are
863 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
864 * added type deriver for ISQRT (thanks to Robert E. Brown).
865 * bug fix: better support for loading from the command line when an
866 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
868 * fixed some bugs revealed by Paul Dietz' test suite:
869 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
870 ** ROUND and FROUND now give the right answer when given very
871 small float arguments.
872 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
874 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
875 ** (GCD 0 <negative-integer>) returned <negative-integer>.
876 ** LCM should return a non-negative integer.
877 ** PARSE-INTEGER returned the index of a terminator instead of the
878 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
879 ** PARSE-INTEGER returned an incorrect index being applied to a
881 ** LCM with two arguments of 0 returns 0 rather than signalling
883 ** unsigned addition of a 32-bit constant with the high bit set no
884 longer causes an internal compiler error on the x86.
885 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
886 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
887 with negative last argument.
888 ** byte specifiers with zero size and position no longer cause
889 an error during type derivation.
890 ** bignum multiplication on the Alpha platform now returns the
892 * porting: The system now builds on SuSE AMD64, although it still
893 generates a 32-bit binary.
894 * .fasl file incompatibility: The fasl file version number has
895 been incremented (because of the changes to internal compiler
896 data structures referred to above).
898 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
899 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
900 more accurately, on the Darwin kernel running on PowerPC hardware.
901 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
902 * Compiler code deletion notes now signal a condition of type
903 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
904 an associated MUFFLE-WARNING restart.
905 * The compiler now performs limited argument count validation of
906 constant format strings in FORMAT, and where appropriate in ERROR,
907 CERROR and WARN. (thanks to Gerd Moellmann)
908 * New ASDF-INSTALL contrib can be used for automatic download and
909 installation of third-party Lisp code from CCLAN or other sites
911 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
912 which forces another thread to execute a function supplied by the
914 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
915 now accept and act upon their :ELEMENT-TYPE keyword argument.
916 (reported by Martin Atzmueller, Edi Weitz)
917 * bug fix: FILE-POSITION now accepts position designators up to
918 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
919 is the greater. (thanks to Patrik Nordebo)
920 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
921 Kalle Olavi Niemitalo)
922 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
923 printable. (reported by Eric Marsden)
924 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
925 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
926 * x86 bug fix in control stack exhaustion checking: now shows backtrace
927 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
928 (thanks to Stig Sandoe)
929 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
930 UNREAD-CHAR now correctly includes the unread character in the
931 target sequence. (thanks to Gerd Moellmann)
932 * bug fix in threaded builds: the system can now be suspended and
933 resumed by shell job control with minimal disruption.
934 * bug fixes in times and timezones >2038AD
935 * better handling of "where is GNU make?" problem in build scripts
936 (thanks to Nikodemus Siivola)
937 * new optimization: inside a named function any reference to a
938 function with the same name is considered to be a self-reference;
939 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
940 optimization quality.
941 * new optimization on x86: logical functions and + now have
942 optimized (UNSIGNED-BYTE 32) versions, which are automatically
943 used when the result is truncated to 32 bits.
944 * VALUES declaration is partially enabled.
945 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
946 * fixed some bugs revealed by Paul Dietz' test suite:
947 ** The system now obeys the constraint imposed by
948 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
949 types form a lattice under type intersection.
950 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
951 ** ASSOC now ignores NIL elements in an alist.
952 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
953 and (1+ MOST-POSITIVE-FIXNUM) answers.
954 ** The addition of a method with invalid qualifiers to a generic
955 function does not cause an error to be signalled immediately;
956 a warning is signalled, and the error is generated only on
957 calling the generic function.
958 * changed .fasl file version number, in part to add type codes for
959 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
960 obscure ANSI requirements
962 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
963 * fixed bug 148: failure to inline-expand a local function left
964 garbage, confusing the compiler.
965 * fixed bugs 3cd: structure slot readers perform type check if the
966 slot can have an invalid value (i.e. it is either not initialized
967 or can be written with a less specific slot writer).
968 * bug fix: the compiler now traps array references to elements off
969 the end of an array; previously, the bounds checking in some
970 circumstances could go off-by-one.
971 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
973 * type declarations for array element types now obey the description
974 on the CLHS page "Declaration TYPE", as per discussions on
975 sbcl-help around 2003-05-08. This means that a declaration
976 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
977 declaration, all references to BAR will be asserted or assumed
978 (with THE, so dependent on compiler policy) to involve objects of
979 type FOO. Note that no such declaration is implied in
980 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
981 * declared types of functions from the "Conditions"
982 chapter. (reported by Paul Dietz)
983 * bug fix: CERROR accepts a function as its first argument.
984 * bug fix: NTH an NTHCDR accept a bignum as index
985 arguments. (reported by Adam Warner)
986 * optimization: character compare routines now optimize comparing
987 against a constant character. (reported by Gilbert Baumann)
988 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
989 argument now works properly on the MIPS platform.
990 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
991 * repeated evaluation of the same DEFSTRUCT, a slot of which is
992 declared to have a functional type, does not cause an error
994 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
995 many :DEFAULT-INITARGS. (reported by Istvan Marko)
996 * fixed bug: if last continuation of a deleted block has a
997 destination, this destination should be deleted too. (reported by
999 * fixed a bug in the bootstrap process: the host compiler's values
1000 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1001 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1002 test case from Patrik Nordebo)
1003 * improved the ability of the disassembler on the PPC platform to
1004 provide helpful disassembly notes.
1005 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1006 the class in more cases than previously.
1007 * bug fix: FILE-POSITION now understands :START and :END for
1008 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1009 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1010 8) data. (thanks to David Lichteblau)
1011 * bug fix: it is possible to add a method to a generic function
1012 without lambda list.
1013 * bug fix: reader failed to signal END-OF-FILE inside an
1014 object representation. (reported by Nikodemus Siivola)
1015 * fixed some bugs revealed by Paul Dietz' test suite:
1016 ** LAST and [N]BUTLAST should accept a bignum.
1017 ** condition slot accessors are methods.
1018 ** (VECTOR NIL) is a subtype of STRING.
1020 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1021 * minor incompatible change: some nonsensical specialized lambda
1022 lists (used in DEFMETHOD) which were previously ignored now signal
1024 * minor incompatible change: the system is now aware of the types of
1025 variables in the COMMON-LISP package, and will signal errors for
1026 most violations of these type constraints (where previously they
1027 were silently accepted).
1028 * minor incompatible change: COMPILE-FILE now uses the freedom
1029 afforded (ANSI 3.2.2.3) to use derived function types for
1030 functions defined in the same file. This also permits the system
1031 to warn on static type mismatches and function
1032 redefinition. (Currently it does not work with high DEBUG level.)
1033 * minor incompatible change: VALUES declaration is disabled.
1034 * When issuing notes, the compiler now signals a condition of type
1035 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1036 restart for use in user handlers. It is expected that the
1037 COMPILER-NOTE condition will eventually become a condition
1038 supertype to a hierarchy of note types, which will then be
1039 handleable in a similar fashion. However, at the moment, no such
1040 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1041 but it's an implementation detail, not a classification for the
1043 * Changes in type checking closed the following bugs:
1044 ** type checking of unused values (192b, 194d, 203);
1045 ** template selection based on unsafe type assertions (192c, 236);
1046 ** type checking in branches (194bc).
1047 * A short form of VALUES type specifier has ANSI meaning (it has
1048 increased the number of situations when SBCL cannot perform type
1050 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1051 keywords or constants is permissible.
1052 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1053 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1054 classes. (thanks to Antonio Martinez)
1055 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1056 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1057 argument to a pretty-print stream. Instead, it leaves any such
1058 operations to DESCRIBE-OBJECT methods.
1059 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1061 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1062 lambda lists. (thanks to David Lichteblau)
1063 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1064 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1065 (thanks to Gerd Moellmann)
1066 * fixed bug 246: increased compilation speed of long
1067 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1069 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1071 * DEFCONSTANT now throws a condition of type
1072 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1073 constant to a non-EQL value; CONTINUE and ABORT restarts
1074 respectively change and preserve the value.
1075 * fixed bug 63: The code walker, part of the implementation of CLOS,
1076 is now better at handling symbol macros.
1077 * bug fix: There is no longer an internal implementation type named
1078 CL:LENGTH. (reported by Raymond Toy)
1079 * bug fix: In macro-like defining macros/special operators the
1080 implicit block does not enclose the lambda list.
1081 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1082 enclosed in parentheses) are not suitable as type specifiers, and
1083 their use properly signals an error now.
1084 * bug fix: An argument count mismatch for a type specifier in code
1085 being compiled no longer causes an unhandled error at compile
1086 time, but signals a compile-time warning.
1087 * fixed simple vector readable printing
1088 * bug fix: DESCRIBE takes more care over whether the class
1089 precedence list slot of a class is bound before accessing it.
1090 (reported by Markus Krummenacker)
1091 * bug fix: FORMATTER can successfully compile pretty-printer format
1092 strings which use variants of the ~* directive inside.
1093 * bug fix: SEARCH now applies its TEST predicate to the elements of
1094 the arguments in the correct order. (thanks to Wolfhard Buss)
1095 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1096 in type inference. (thanks to Robert E. Brown)
1097 * bug fix: Reading in symbols with an explicit package name of ""
1098 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1099 * fixed some bugs revealed by Paul Dietz' test suite:
1100 ** NIL is now allowed as a structure slot name.
1101 ** Arbitrary numbers, not just REALs, are allowed in certain
1102 circumstances in LOOP for-as-arithmetic clauses.
1103 ** Multiple class redefinitions before slot access no longer
1104 causes a type error.
1105 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1106 association between the name and a class.
1107 ** Generic functions with non-standard method-combination and over
1108 five methods all of which return constants no longer return NIL
1109 after the first few invocations. (thanks to Gerd Moellmann)
1110 ** CALL-NEXT-METHOD with no arguments now passes the original
1111 values of the arguments, even in the presence of assignment.
1112 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1114 ** DISASSEMBLE works with closures and funcallable instances.
1115 ** ADD-METHOD now returns the generic function, not the new method.
1116 ** FIND-METHOD signals an error if the lengths of the specializers
1117 is incompatible with the generic function, even if the ERRORP
1119 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1120 which its argument is a member.
1121 ** DEFCLASS only redefines the class named by its class-name
1122 argument if that name is the proper name of the class;
1123 otherwise, it creates a new class.
1124 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1125 of the UNBOUND-SLOT condition to the name of the slot.
1126 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1127 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1128 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1129 SLOT-MAKUNBOUND in the specified fashion.
1131 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1132 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1133 cross-compilation host. As a consequence, we can now bootstrap our
1134 way up to SBCL starting with a bare gcc toolchain and human-readable
1135 source code (first the source to CLISP, then the source to SBCL).
1136 * A contributed module containing a partial implementation of the
1137 simple-streams interface has been included. (thanks to Rudi
1139 * A contributed module implementing the RFC1321 Message Digest
1140 Algorithm, known as MD5, has been included.
1141 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1142 no longer has any effect, as the code controlled by this feature
1143 has been deleted. (As far as we know, no-one has ever built using
1144 this feature, and its semantics were confused in any case).
1145 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1146 work on conditions (as required by the ANSI specification),
1147 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1148 expected behaviour on conditions. Users should note, however,
1149 that such behaviour is not required by the ANSI specification,
1150 and so use of this behaviour may render their code unportable.
1151 * fixed some bugs revealed by Paul Dietz' test suite:
1152 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1154 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1155 ** SXHASH on condition objects no longer returns NIL.
1156 ** :ALLOCATION :CLASS slots are better treated; their values are
1157 updated on class redefinition, and initforms inherited from
1158 superclasses are applied.
1159 ** REMOVE-METHOD returns its generic function argument even when
1160 no method was removed.
1161 ** SHARED-INITIALIZE now initializes the values of the requested
1162 slots, including those with :ALLOCATION :CLASS.
1163 ** ALLOCATE-INSTANCE now works on structure classes defined via
1164 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1166 ** SLOT-EXISTS-P now works on conditions, as well as structures
1168 ** MAKE-LOAD-FORM now has the required methods on
1169 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1170 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1171 keyword, and now implements the SLOT-NAMES argument.
1172 ** methods with &OPTIONAL arguments no longer allow too many
1173 arguments to be passed in the call without error.
1174 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1175 option is consistent with the required arguments of the generic
1176 function lambda list.
1177 * bug fix: REQUIRE accepts a string designator. (Thanks to
1179 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1180 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1181 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1182 * bug fix: The compiler now checks for duplicated variables in macro
1184 * bug fix: SETQ on globals returns the correct value.
1185 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1186 not a COMPILER-ERROR (followed by some other strange error on
1187 choosing the CONTINUE restart).
1188 * bug fix: make.sh and friends are now more consistent in the way that
1189 they look for GNU "make".
1191 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1192 * experimental native threads support (on x86 Linux >=2.4 only).
1193 This is not compiled in by default: you need to add :SB-THREAD to
1194 the target features. See the "Beyond ANSI" chapter of the manual
1196 * fix for longstanding nonANSIism: The old distinction between
1197 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1198 The return value from CL:FIND-CLASS is now a CLOS class, and
1199 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1200 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1201 * An interface to the MetaObject Protocol, as described in Kiczales,
1202 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1203 MIT Press, 1991, is available from the SB-MOP package.
1204 * incompatible change: the SB-PCL package should now be considered
1205 a private implementation detail, and no longer a semi-private MOP
1207 * minor incompatible change: due to rearrangement for threads, the
1208 control stack and binding stack are now allocated at arbitrary
1209 addresses instead of being hardcoded per-port. Users affected by
1210 this probably have to be doing advanced things with shared
1211 libraries, and will know who they are.
1212 * minor incompatible change: Previously, all --eval forms used were
1213 processed with READ before any of them were processed with EVAL.
1214 Now each --eval form is processed with both READ and EVAL before
1215 the next --eval form is processed. (Thus package operations like
1216 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1217 work as the user might reasonably expect.)
1218 * minor incompatible change: *STANDARD-INPUT* is now only an
1219 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1221 * minor incompatible change: Y-OR-N-P is now character-oriented,
1222 not line oriented. Also, YES-OR-NO-P now works without errors.
1223 (thanks to Antonio Martinez)
1224 * sb-aclrepl module improvements: an integrated inspector, added
1225 repl features, and a bug fix to :trace command.
1226 * Known functions, which cannot be open coded by the backend, are
1227 considered to be able to check types of their arguments. (fixing
1228 a bug report by Nathan J. Froyd)
1229 * fixed a bug in computing method discriminating functions: It is
1230 now possible to define methods specialized on classes which have
1231 forward-referenced superclasses. (thanks to Gerd Moellmann)
1232 * fixed evaluation order in optional entries (reported by Gilbert
1234 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1235 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1236 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1237 required-by-AMOP NAME argument, as well as CLASS and
1238 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1239 * fixed bug 20: DEFMETHOD can define methods using names that are
1240 not the proper names of classes to designate class specializers.
1241 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1242 Unix file descriptors, instead of blowing up. (thanks to Antonio
1244 * Garbage collection refactoring: user-visible change is that a
1245 call to the GC function during WITHOUT-GCING will not do garbage
1246 collection until the end of the WITHOUT-GCING. If you were doing
1247 this you were probably losing anyway.
1248 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1249 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1250 TYPEP the latter but not the former.
1251 * The compiler issues a full WARNING for calls to undefined functions
1252 with names from the CL package.
1253 * MAP-INTO for a vector destination is open coded. (reported by
1254 Brian Downing on c.l.l)
1255 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1256 documentation string.
1257 * fixed some bugs revealed by Paul Dietz' test suite:
1258 ** COPY-ALIST now signals an error if its argument is a dotted
1260 ** Condition slots are now accessed more correctly in the presence
1261 of multiple initargs for a given slot.
1262 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1263 exclude restarts of the same name associated with a different
1265 ** DEFCLASS of forward-referenced classes with another
1266 forward-referenced class in the superclasses list no longer
1268 ** Condition slots are now initialized once each, not multiple
1269 times. (thanks to Gerd Moellmann)
1270 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1272 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1273 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1274 arguments contain duplicated elements.
1275 ** RESTART-CASE understands local macros.
1276 ** RESTART-CASE associates exactly its own restarts with a condition.
1277 ** ENDP in safe mode checks its argument to be of type LIST.
1278 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1279 in question is unbound.
1280 ** Optimized MAKE-INSTANCE functions no longer cause internal
1281 assertion failures in the presence of duplicate initargs.
1282 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1283 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1284 propagated even in the ctor optimized implementation of
1286 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1288 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1289 * a better implementation of SXHASH on (simple) bit vectors,
1290 measured both in execution speed and in distribution of results
1291 over the positive fixnums, has been installed. Likewise, a better
1292 implementation of EQUAL for simple bit vectors is now available.
1293 * fixed CEILING optimization for a divisor of form 2^k.
1294 * fixed bug 240 (emitting extra style warnings "using the lexical
1295 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1296 by Antonio Martinez)
1297 * fixed SXHASH, giving different results for NIL depending on type
1298 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1299 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1300 variable. (found by Rolf Wester)
1301 * SBCL does not ignore type declarations for special
1302 variables. (reported by rif on c.l.l 2003-03-05)
1303 * some bug fixes in contrib/sb-aclrepl/
1304 * fixed some bugs revealed by Paul Dietz' test suite:
1305 ** a bug in the CONS type specifier, whereby the CAR and CDR
1306 types got intertwined, has been fixed;
1307 ** the type system is now able to reason about the interaction
1308 between INTEGER and RATIO types more completely;
1309 ** APPEND, [N]REVERSE and NRECONC check that those their
1310 arguments, which must be proper lists, are really so;
1311 ** An array specialized to be unable to hold elements has been
1312 implemented, as required -- yes, really -- by ANSI;
1313 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1314 on malformed property lists;
1316 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1317 * incompatible packaging change: in line with Unix convention,
1318 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1319 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1320 * REQUIRE and PROVIDE are now optionally capable of doing something
1321 useful. See the documentation string for REQUIRE.
1322 * infrastructure for a managed SBCL contrib system: contributed
1323 modules in this release include:
1324 ** the ASDF system definition facility;
1325 ** an interface to the BSD Sockets API;
1326 ** an ACL-like convenience interface to the repl;
1327 (thanks to Kevin Rosenberg)
1328 ** an implementation of ROTATE-BYTE, with efficient implementation
1330 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1331 gives a closer approximation to the right answer than previously.
1332 (thanks to Raymond Toy)
1333 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1334 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1335 optional environment arguments, as required by ANSI.
1336 * fixed bugs in other functions taking environment objects, allowing
1337 calls with an explicit NIL environment argument to be compiled
1339 * fixed bug 228: primary return values from
1340 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1341 COMPILE or FUNCTION.
1342 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1343 :TYPE will no longer signal errors on innocuous objects.
1344 * fixed bug 231b: SETQ is better at respecting type declarations in
1345 the lexical environment.
1346 * fixed a bug in DEFCLASS: classes named by symbols with no or
1347 unprintable packages can now be defined.
1348 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1349 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1350 * fixed compiler failure related to checking types of functions.
1351 (reported by Robert E. Brown)
1352 * the compiler is now much more consistent in its error-checking
1353 treatment of bounding index arguments to sequence functions: in
1354 (SAFETY 3) code, errors will be signalled in almost all cases if
1355 invalid sequence bounding indices are passed to functions defined
1356 by ANSI to operate on sequences.
1357 * fixed a bug in the build procedure: documentation of SBCL-specific
1358 packages is now preserved and available in the final Lisp image.
1359 * lifted FDEFINITION lookup out of loops in the implementation of
1360 many list operations. (thanks to Robert E. Brown)
1361 * fixed a bug in the reader: the #n# reader macro now works for
1362 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1363 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1364 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1365 E. Brown and Tony Martinez respectively)
1366 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1367 SLOT-BOUNDP: in particular, it is now able to optimize them much
1368 better, and is now not vulnerable to having packages renamed.
1369 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1370 always called when a slot is not present in an instance. (thanks
1372 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1373 longer signal an error on unreading a character following EOF on
1374 the previous constituent stream. (thanks to Tony Martinez)
1375 * fixed some bugs revealed by Paul Dietz' test suite:
1376 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1377 not just nonnegative fixnums;
1378 ** the logical bit-array operators such as BIT-AND now accept an
1379 explicit NIL for their "opt-arg" argument (to indicate a
1380 freshly-consed result bit-array);
1381 ** ELT now signals an error on an invalid sequence index in safe
1383 ** the type system is now cleverer about negations of numeric
1384 types, and consequently understands the BIGNUM and RATIO types
1386 ** the type system is now cleverer about the interaction between
1387 INTEGER and RATIO types: while bugs still remain, many more
1388 cases are accurately computed;
1389 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1390 if it is in the last clause;
1391 ** CONSTANTLY now correctly returns a side-effect-free function in
1393 ** DECLARE is no longer treated as a special-operator; in
1394 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1395 * incremented fasl file version number due to the change in the
1396 DEFSTRUCT-SLOT-DESCRIPTION structure.
1398 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1399 * minor incompatible change: code processed by the "interpreter" or
1400 EVAL now has a compilation optimization policy of (DEBUG 2)
1401 (changed from (DEBUG 1)) to improve debuggability of interactive
1402 development, and to allow the use of the debug RETURN command in
1404 * an experimental implementation of the RETURN command for the
1405 debugger has been included. (thanks to Frederik Kuivinen)
1406 * fixed bug 62: constraints were not propagated into a loop.
1407 * fixed bug in embedded calls of SORT (reported and investigated by
1409 * fixed some bugs revealed by Paul Dietz' test suite:
1410 ** printing and reading of arrays with some dimensions having
1411 length 0 (thanks to Gerd Moellmann);
1412 ** BOA constructor with &AUX argument without a default value does
1413 not cause a type error;
1414 ** CONSTANTP now returns true for all self-evaluating objects.
1416 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1417 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1418 accessors that are related by inheritance, as specified in the
1419 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1420 Valtteri Vuorikoski)
1421 * The compiler is now able to inline functions that were defined in
1422 a complex lexical environment (e.g. inside a MACROLET).
1423 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1425 * Support for the upcoming FreeBSD-5.0 release has been included.
1426 (thanks to Dag-Erling Smorgrav)
1427 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1428 effect when it is not in a toplevel context.
1429 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1430 stand a better chance of being correct. (thanks to Gerd
1432 * fixed bug in COERCE, which now signals an error on coercing a
1433 rational to a bounded real type which excludes the expected
1435 * The compiler is now able to derive types more accurately from the
1436 COERCE and COMPILE functions.
1437 * fixed bug 223: functional binding is considered to be constant
1438 only for symbols in the CL package.
1439 * fixed bug 231: SETQ did not check the type of a variable being set
1440 (reported by Robert E. Brown)
1441 * A new optimization for MAKE-INSTANCE has been included, fixing
1442 various bugs (including relating to :ALLOCATION :CLASS slots and
1443 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1445 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1446 ** As required by ANSI, LOOP now disallows anonymous collection
1447 clauses such as COLLECT I in conjunction with aggregate boolean
1448 clauses such as THEREIS (= I 1);
1449 ** LOOP now signals an error when any variable is reused in the
1450 same loop (including the potentially useful construct analogous
1451 to WITH A = 1 WITH A = (1+ A);
1452 ** IT is only a special loop symbol within the first clause of a
1453 conditional loop clause;
1454 ** LOOP with a typed iteration variable over a hashtable now
1455 signals a type error iff it should.
1456 * fixed some other bugs revealed by Paul Dietz' test suite:
1457 ** FILE-STREAM now names the class previously known as FD-STREAM;
1458 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1459 argument) no longer signals an error;
1460 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1461 of the slot symbol, rather than using the current package
1462 ((:CONC-NAME "") continues to intern the slot's name in the
1464 * incremented fasl file version number, because of the incompatible
1465 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1466 of the new implementation of DEFINE-COMPILER-MACRO.
1468 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1469 * Support for building SBCL for MIPS platforms running in
1470 little-endian mode has now been checked in, and basic
1471 functionality on said platforms verified.
1472 * minor incompatible change: PCL now records the pathname of a file
1473 in which methods and the like are defined, rather than its
1475 * minor incompatible change: TRUENAME now considers the truename of
1476 a file naming a directory to be the pathname with :DIRECTORY
1477 component indicating that directory.
1478 * minor incompatible change: a NAMED clause in the extended form of
1479 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1480 reason for the previous behaviour is unclear.
1481 * more systematization and improvement of CLOS and MOP conformance
1482 in PCL (thanks to Gerd Moellman and Pierre Mai):
1483 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1485 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1486 primary methods with no specializers;
1487 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1489 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1490 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1491 FORWARD-REFERENCED-CLASSes; error reporting on
1492 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1494 ** SXHASH on CLOS instances now uses a slot internal to the
1495 instance to return different numbers on distinct instances,
1496 while preserving the same return value through invocations of
1498 ** DEFMETHOD signals errors when methods with longer incongruent
1499 lambda lists are added to generic functions;
1500 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1501 CLASS, as specified in AMOP;
1502 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1503 on the order returned by the primary method for classes of
1504 class STANDARD-CLASS;
1505 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1506 * fixed some bugs shown by Paul Dietz' test suite:
1507 ** DOLIST puts its body in TAGBODY;
1508 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1510 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1511 value producing form;
1512 ** if more variables are given to PROGV than values, extra
1513 variables are bound and made to have no value;
1514 ** NSUBSTITUTE on list arguments gets the right answer with
1516 ** ELT signals an error of type TYPE-ERROR when the index argument
1517 is not a valid sequence index;
1518 ** LOOP signals (at macroexpansion time) an error of type
1519 PROGRAM-ERROR when duplicate variable names are found;
1520 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1521 ** FUNCALL of special-operators now cause an error of type
1523 ** PSETQ now works as required in the presence of side-effecting
1524 symbol-macro places;
1525 ** NCONC accepts any object as its last argument;
1526 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1528 ** loop-for-as-package does not require a package to be explicitely
1530 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1532 * fixed bug 166: compiler preserves "there is a way to go"
1533 invariant when deleting code.
1534 * fixed bug 172: macro lambda lists with required arguments after
1535 &REST arguments now cause an error to be signalled. (thanks to
1537 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1538 bug. (thanks to Matthew Danish)
1539 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1541 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1542 arguments are lexically rebound. (thanks to Gerd Moellmann and
1544 * fixed bug 194: error messages are now more informative when there
1545 is no primary method applicable in a call to a generic
1546 function. (thanks to Gerd Moellmann)
1547 * fixed bug in command line argument checking (thanks to Julian
1549 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1550 (thanks to Matthew Danish)
1551 * incremented fasl file version number, because of the
1552 SXHASH-related changes in the layout of CLOS data structures
1554 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1555 * minor incompatible change: The runtime (the Unix executable named
1556 "sbcl") is now much pickier about the .core files it will load.
1557 Essentially it now requires .core files to descend from the same
1558 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1559 as the runtime does. (The intent is to prevent the crashes which
1560 can occur, and which can even be reported as mysterious failures,
1561 when people patch the sources or change the build parameters
1562 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1563 sbcl and .core files.)
1564 * fixed bug: VALUES-LIST is no longer optimized away.
1565 * fixed bug 142: The FFI conversion of C string values to Lisp
1566 string values no longer conses excessively. (thanks to Nathan
1567 Froyd porting Raymond Toy's fix to CMU CL)
1568 * began to systematize and improve MOP conformance in PCL (thanks to
1569 Nathan Froyd, Gerd Moellman and Pierre Mai):
1570 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1572 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1573 ** FINALIZE-INHERITANCE is now called on class finalization;
1574 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1575 argument precedence order.
1576 * fixed bug 202: The compiler no longer fails on functions whose
1577 derived types contradict their declared type.
1578 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1579 so it can be non-toplevel.
1580 * The fasl file version number has changed (because of the new
1581 implementation of DEFMACRO).
1582 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1583 safe code, that any length requirement by their type-specifier
1584 argument is valid. The exceptions to this are described in bug
1586 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1587 functions now check the types of their inputs as required by ANSI.
1588 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1589 introduced symbol is DECLAREd to be SPECIAL.
1590 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1591 * fixed bug: :COUNT argument to sequence functions may be negative
1592 * fixed bug: body of DO-SYMBOLS may contain declarations
1593 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1594 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1595 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1596 symbol macro only once
1597 * fixed printing of call frame when argument list is unavailable
1598 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1599 * compiler no longer signals WARNING on unknown keyword
1602 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1603 * A beta-quality port to the mips architecture running Linux,
1604 based on the old CMUCL backend, has been made. It has been tested
1605 on a big-endian kernel, and works sufficiently well to be able to
1606 rebuild itself; it has not been tested in little-endian mode.
1607 * fixed an inconsistency between gencgc.c and purify.c which made
1608 dumping/loading .core files unreliable
1609 * fixed bug 120a: The compiler now deals correctly with IFs where
1610 the consequent is the same as the alternative, instead of
1611 misderiving the return type. (thanks to Alexey Dejneka)
1612 * fixed bug 113: Logical pathnames are now dumpable (the logical
1613 host is resolved at load-time, throwing an error if it is not
1615 * fixed bug 174: FORMAT's error message is slightly clearer when a
1616 non-printing character is used in a format directive.
1617 * fixed several bugs in compiler checking of type declarations, i.e.
1618 violations of the Python "declarations are assertions" principle
1619 (thanks to Alexey Dejneka)
1620 * fixed several bugs in PCL's error checking (thanks to Gerd
1622 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1624 * fixed bug in compilation of functions as first class values
1625 (thanks to Antonio Martinez-Shotton)
1626 * The compiler's handling TYPE-ERRORs which it can prove will
1627 inevitably happen at runtime has been cleaned up and corrected
1628 in several ways. (thanks to Alexey Dejneka)
1629 * improved argument type checking for various basic arithmetic
1630 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1631 lavished on them in the past that they can be compiled in many
1632 ways in different special cases
1633 * fixed bug 181: compiler checks validity of user supplied type
1635 * cleaned up code flushing in optimization: Function calls which
1636 should signal errors for safety purposes (e.g. which ANSI says
1637 should signal errors when their arguments are of incorrect type)
1638 are no longer optimized away.
1639 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1640 * incremented fasl file version number, because changes in the
1641 implementation of sequence functions like COERCE caused
1642 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1643 in old inline expansions) to become undefined. (Actually these
1644 changes were later undone, so we might very well be binary
1645 compatible with 0.7.7 after all, but leaving the version number
1646 incremented seemed like the simplest and most conservative
1649 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1650 * An alpha-quality port to the parisc architecture running Linux,
1651 based on the old CMUCL backend, has been made. This, even more so
1652 than the other backends, should be considered still a work in
1653 progress; known problems include that the Linux kernel in 64-bit
1654 mode does not propagate the correct sigcontext structure to
1655 userspace, and consequently SBCL on a parisc64 kernel will not
1657 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1658 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1659 suppressed.) Also now that the compiler is looking at declarations
1660 in the environment, it checks optimization declarations as well,
1661 and suppresses inlining when (> DEBUG SPEED).
1662 * More fixes have been made to treatment of floating point exception
1663 treatment and other Unix signals. In particular, floating point
1664 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1665 * The detection and handling of control stack exhaustion (infinite
1666 or very deeply nested recursion) has changed. Stack exhaustion
1667 detection is now done by write-protecting pages at the OS level
1668 and applies at all optimization settings; when found, a
1669 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1670 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1671 be caught using IGNORE-ERRORS.
1672 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1673 correctly when transforming recursive LABELS functions to LETs.
1674 (thanks to Alexey Dejneka)
1675 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1676 that are names of constants or global variables.
1677 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1678 alien routines with docstrings.
1679 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1680 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1682 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1683 (thanks to Eric Marsden)
1684 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1685 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1686 * Bug fix: The SPARC backend can now compile functions involving
1687 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1688 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1689 to a functions expecting a PATHNAME-DESIGNATOR.
1690 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1691 lambda lists. (thanks to Alexey Dejneka)
1692 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1693 functions so that EXPORTing the name of the class doesn't cause
1694 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1695 to Antonio Martinez for reporting this)
1696 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1697 constant-folded without failing an assertion. (thanks to Einar
1698 Floystad Dorum for reporting this)
1699 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1700 types are now dealt with more correctly by the compiler.
1701 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1702 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1703 behaviour analogous to RENAME-FILE. This puts its behaviour more
1704 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1706 * The fasl file version number has changed again. (because of the
1707 bug fix involving the names of PCL MAKE-INSTANCE functions)
1709 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1710 * bug fix: Floating point exceptions are treated much more
1711 consistently on the x86/Linux and PPC/Linux platforms.
1712 * Array initialization with :INITIAL-ELEMENT is now much faster for
1713 cases when the compiler cannot open code the array creation, but
1714 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1715 array accesses have also seen a speed increase.
1716 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1717 specified. (This was at the root of some bad interactions between
1718 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1720 * bug fix: Internal error arguments for undefined functions are now
1721 computed correctly on the PPC/Linux platform.
1722 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1723 Raymond Toy's patch for CMU CL)
1724 * Support for the Solaris 9 operating environment has been included
1725 (thanks to Daniel Merritt)
1726 * A very ugly but hopefully complete draft of the missing FFI chapter
1727 of the manual has been created by reformatting the corresponding
1728 CMU CL manual chapter into (currently very ugly and incoherent)
1729 DocBook and bringing it up to date for SBCL behavior. Thus, the
1730 manual is now essentially complete, at least by my extreme
1731 once-and-only-once standards, whereby it's acceptable to refer to
1732 the doc strings of SB-EXT functions as the primary documentation.
1733 * The fasl file version number has changed again, due to cleanup of
1734 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1735 is no longer a static symbol.)
1737 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1738 * SBCL now builds with OpenMCL (version 0.12) as the
1739 cross-compilation host; also, more progress has been made toward
1740 bootstrapping under CLISP.
1741 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1743 * bug 158 fixed: The compiler can now deal with integer loop
1744 increments different from 1; fixing this turned out also to fix
1746 * bug 169 fixed: no more bogus warnings about using lexical bindings
1747 despite the presence of perfectly good SPECIAL declarations (thanks
1748 to David Lichteblau)
1749 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1750 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1751 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1753 * bug 180 fixed: Method combination specifications no longer ignore
1754 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1755 * bug fix: Structure type predicate functions now check their argument
1756 count as they should.
1757 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1758 correctly. (thanks to Pierre Mai)
1759 * minor incompatible change: The --noprogrammer option is deprecated
1760 in favor of the new --disable-debugger option, which is very similar.
1761 (The major difference is that it takes effect at a slightly different
1762 time at startup, causing handling of errors in --sysinit and
1763 --userinit files will be affected differently.) The
1764 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1765 been added to allow this functionality to be controlled from ordinary
1766 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1767 maintainers, who might want to run non-interactive scripts to
1768 build SBCL cores which will later be used interactively.)
1769 * minor incompatible change: The LOAD function no longer, when given
1770 a wild pathname to load, loads all files matching that pathname.
1771 Instead, an error of type FILE-ERROR is signalled.
1773 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1774 * bug 147 fixed: The compiler preserves its block link/count
1775 invariants more correctly now so that it doesn't crash. (thanks
1777 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1779 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1780 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1781 no-value-for-this-character as NIL (as specified by ANSI).
1782 * HOST-NAMESTRING on physical pathnames now returns a string that is
1783 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1784 (thanks to Christophe Rhodes)
1785 * The Alpha port handles icache flushing more correctly. (thanks to
1787 * More progress has been made toward bootstrapping under CLISP. (thanks
1788 to Christophe Rhodes)
1789 * The fasl file format has changed again, because dynamic loading
1790 on OpenBSD (which has non-ELF object files) motivated some cleanups
1791 in the way that foreign symbols are transformed and passed around.
1792 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1793 is no longer treated as whitespace by the reader, but instead as
1794 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1795 |A
\7fB|, instead of A as it used to.
1797 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1798 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1799 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1800 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1801 this as of 0.7.1.45, but was left out of the previous news section
1802 (thanks to Dan Barlow)
1803 * SBCL now runs on the Solaris operating system on SPARC architectures
1804 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1805 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1806 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1807 * SPARC backend cleanups, allowing builds of cores optimized for V8
1808 and V9 SPARCS, and also emission of code targeted to a particular
1809 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1811 * SBCL is closer to bootstrapping under CLISP, thanks to various
1812 fixes by Christophe Rhodes.
1813 * The fasl file format has changed again, to allow the compiler's
1814 INFO database to support symbol macros.
1815 * The user manual (in doc/) is formatted into HTML more nicely.
1816 (thanks to coreythomas)
1817 * The system is smarter about SUBTYPEP relationships, especially
1818 those involving NOT types (including types such as ATOM which are
1819 represented internally using NOT types). Thus SUBTYPEP is less
1820 likely to return (VALUES NIL NIL) in general, and in particular
1821 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1823 * The fasl file format has changed again, because the internal
1824 representation of types now includes a new slot to support the new
1825 SUBTYPEP-of-NOT-types logic.
1826 * (not a change in the main branch of SBCL, but a related prototype
1827 which can hopefully be merged into the main branch of SBCL in the
1828 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1829 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1830 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1831 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1832 dispatch macro character combinations. (thanks to Alexey Dejneka)
1833 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1834 that superficially look like logical namestrings correctly.
1835 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1837 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1838 default supplied-p) for &optional and &key arguments. (thanks to
1841 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1842 * incompatible change: The compiler is now less aggressive about
1843 tail call optimization, doing it only when (> SPACE DEBUG) or
1844 (> SPEED DEBUG). (This is an incompatible change because there are
1845 programs which relied on the old CMU-CL-style behavior to optimize
1846 away their unbounded recursion which will now die of stack overflow.)
1847 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1848 for non-GENCGC systems has been increased to 20M (since that
1849 seems much closer to the likely performance optimum for modern
1850 systems than the old 4M value was)
1851 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1852 and *DEBUG-PRINT-LEVEL*
1853 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1854 of CMU CL's support for SPARC, and various endianness and other
1855 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1856 * new syntactic sugar for the Unix command line: --load foo.bar is now
1857 an alternate notation for --eval '(load "foo.bar")'.
1859 ** The system now detects stack overflow and handles it gracefully,
1860 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1861 optimization settings. (This is a good thing in general, and
1862 its introduction in this version should be particularly timely
1863 for anyone whose code fails because of suppression of tail
1865 ** The system now hunts for the C variable "environ" in a more
1866 devious way, to avoid segfaults when the C library version differs
1867 between compile time and run time. (thanks to Christophe Rhodes)
1868 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1869 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1870 ** The compiler no longer issues bogus style warnings for undefined
1871 classes in the same source file as the DEFCLASSes which defined
1872 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1874 ** fixes in CONDITION class precedence list for undefined function
1875 errors (thanks to Alexei Dejneka)
1876 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1877 correctly. (thanks to Dan Barlow)
1878 ** portability fixes aiming at bootstrapping under CLISP (thanks
1879 to Dave McDonald and Christophe Rhodes)
1880 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1881 ** fixes in type translation and and type inference (thanks to
1883 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1884 ** various fixes in the new ports (thanks to Dan Barlow)
1885 * several changes related to debugging:
1886 ** suppression of tail recursion, as noted above
1887 ** stack overflow detection, as noted above
1888 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1889 is now the default. (For some time encapsulation has been more
1890 reliable than the breakpoint-based :ENCAPSULATE NIL
1891 implementation, at least on X86 systems; and I just noticed that
1892 encapsulation also seems closer to the spirit of the ANSI
1895 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1897 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1898 up properly again. (There was a packaging bug in 0.7.0 which
1899 left their definitions in SB-SYS::LOAD-FOREIGN and
1900 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1901 vital for most things which interface to C-level interfaces,
1902 like extensions working with sockets or databases or
1903 Perl-compatible regexes or whatever, and the need to fix
1904 this bug is the main reason that 0.7.1 was released so
1906 ** DEFGENERIC is now choosier about the methods it redefines, so that
1907 reLOADing a previously-LOADed file containing DEFGENERICs does
1908 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1909 cycle now works as it should. (thanks to Alexey Dejneka)
1910 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1911 fixed by Christophe Rhodes. (He actually submitted this patch
1912 months ago, and I delayed until after 0.7.0.)
1913 ** Bug 111 (internal compiler confusion about runtime checks on
1914 FUNCTION types) was fixed by Alexey Dejneka.
1915 * Some internal cleanups (getting rid of variables which aren't
1916 needed now that the byte interpreter is gone) caused the fasl
1917 file format number to change again.
1919 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1920 * major incompatible change: The default fasl file extension, i.e. the
1921 default extension for files produced by COMPILE-FILE, has changed
1922 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1924 ** There are many changes in the implementation of the compiler.
1925 SBCL is now essentially a compiler-only implementation of ANSI
1926 Common Lisp. EVAL still "interprets" a few special cases, but
1927 almost all the interesting cases are handled by creating
1928 a LAMBDA expression, calling COMPILE on it, then calling
1929 FUNCALL on the result.
1930 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1931 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1932 Since the code is newer, there might still be some new bugs
1933 (though not as many as before Martin Atzmueller's fixes:-). But
1934 the new code is substantially simpler and clearer, and hopefully
1935 any remaining bugs will be simpler, less fundamental, and more
1936 fixable then the bugs in the old code.
1937 ** The revised compiler is still a little unsteady on its feet.
1939 *** The debugging information it produces (particularly the names
1940 of FUNCTION objects) is sometimes much less useful than what
1941 the old compiler produced.
1942 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1943 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1944 within a MACROLET) has been temporarily weakened.
1945 ** There are new compiler optimizations for various functions:
1946 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1947 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1948 *** the math functions TRUNCATE, FLOOR, and CEILING
1949 *** the function-of-all-trades COERCE
1950 Mostly these should be transparent, but there's one
1951 potentially-annoying problem (bug 117): when the compiler
1952 inline-expands a function and does type analysis on the result,
1953 it can create control paths which have type mismatches, and
1954 when it can't prove that those control paths aren't taken,
1955 it will issue WARNINGs about the type mismatches. This is
1956 a particular problem in practice for the new sequence functions.
1957 It's not clear how this should be fixed, and for now, a
1958 workaround is given in the entry for 117 in the BUGS file.
1959 ** (Because of the interaction between the two previous items --
1960 occasional inlining problems and new inline expansions -- some
1961 of the new sequence function optimizations won't really kick in
1962 completely until debugging information, and then inlining, are
1963 straightened out in some future version.)
1964 * minor incompatible changes:
1965 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1966 DIRECTORY no longer implicitly promotes NIL slots of its
1967 pathname argument to :WILD. In particular, when you ask for the
1968 contents of a directory (which you used to be able to do without
1969 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1970 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1971 ** changes in behavior that ANSI explicitly defines to be
1972 implementation dependent:
1973 *** The new compiler-only implementation still conforms with ANSI,
1974 but acts a little different than before. Besides the obvious
1975 changes in performance tradeoffs (that the cost per form passed
1976 to EVAL has gone up, and the cost per form executed by EVAL
1977 has gone down), the behavior of the system changes a little
1978 because there are no longer any interpreted function objects.
1979 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1980 e.g. doing COMPILE on the output of interactive DEFUN is
1982 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1984 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1985 has changed from (:CL) to NIL.
1986 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1987 e.g. #\Newline and #\Tab, don't have names specified in the
1988 ANSI Common Lisp standard, is now based on their ASCII symbolic
1989 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1990 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1991 are no longer used for output.
1992 ** changes in internal implementation constants:
1993 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1994 4 million. (If your application spends a lot of time GCing and
1995 you have a lot of RAM, you might want to experiment with
1996 increasing it even more.)
1997 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1998 However, almost all old code should still continue to work without
1999 immediate update, as SB-C-CALL is now a (deprecated) nickname
2001 ** Old operator names in the style DEF-FOO are now deprecated in
2002 favor of new corresponding names DEFINE-FOO, for consistency with
2003 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2004 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2005 internal symbols, but a few supported extensions like
2006 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2007 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2008 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2009 etc. as you get deeper into recursive calls to the debugger
2010 command loop, instead of the old "5]", "5]]", "5]]]"
2011 sequence. (I was motivated to do this when squabbles between
2012 ILISP and SBCL left me very deeply nested in the debugger. In the
2013 short term, this change will probably provoke more ILISP/SBCL
2014 squabbles, but hopefully it will be an improvement in the long run.)
2015 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2016 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2017 function, since declaiming return types involving aliens is
2018 (1) annoyingly messy to do by hand and (2) vital to efficient
2019 compilation of code which calls such functions.
2020 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2021 longer reexported by the SB-EXT package. They're solely useful
2022 for alien code, so it seems more logical that you should get
2023 them from the SB-ALIEN package, not in SB-EXT.
2024 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2025 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2026 features. Instead, the code that they used to control is always
2027 built into the system.
2028 * many other bug fixes
2029 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2030 advantage of the new EVAL-WHEN stuff and to clean them up in
2031 general, and they are now more ANSI-compliant in a number of
2032 ways. Martin Atzmueller is responsible for a lot of this.
2033 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2035 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2036 *** correct ERROR type for various file operations
2037 *** some fixes for Lisp streams
2038 *** DEFMETHOD syntax checking
2039 *** changing old weird representation of debug information as
2040 strings (which, among their other deficiencies, don't transform
2041 correctly when you rename packages, and don't change their
2042 print representation when you change things like *PACKAGE*
2043 and *PRINT-LENGTH*) to symbols and lists of symbols
2044 He also made several improvements and fixed several bugs in DESCRIBE.
2045 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2047 *** misbehavior of WRITE-STRING/WRITE-LINE
2048 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2049 and several other LOOP problems as well
2050 *** DIRECTORY when similar filenames are present
2051 *** DEFGENERIC with :METHOD options
2052 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2053 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2054 *** argument ordering in FIND with :TEST option
2055 *** mishandled package designator argument in APROPOS-LIST
2056 *** various problems in the backquote readmacro
2057 *** a bug in APROPOS
2058 *** probably some others that I'm not describing very well here,
2059 since the CVS log documents them by reference to sbcl-devel
2060 messages, and the SourceForge archives aren't working well.:-(
2061 ** Dan Barlow improved the Alpha port (and is making progress on the
2062 PPC port, for those of you who think different).
2063 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2064 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2065 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2066 continues to work on the SPARC port (for those of you in a position
2067 to look down upon our little PC-compatible boxes from a great height).
2068 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2069 and uses that copy, rather than the current dynamic value, when
2070 it's trying to decide whether to truncate output. Thus e.g.
2071 (let ((*print-lines* 50))
2072 (pprint-logical-block (stream nil)
2074 (let ((*print-lines* 8))
2075 (print (aref possiblybigthings i) stream)))))
2076 should now truncate the logical block only at 50 lines, instead of
2077 often truncating it at 8 lines, as it did before.
2078 * The doc/cmucl/ directory, containing old CMU CL documentation
2079 from the time of the fork, is no longer part of the base system.
2080 SourceForge has shut down its anonymous FTP service, and with it
2081 my original plan for distributing the old CMU CL documentation
2082 there. For now, if you need these files you can download an old
2083 SBCL source release and extract them from it.
2084 * The fasl file version number changed again, for dozens of reasons,
2085 some of which are apparent above.
2087 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2088 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2089 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2090 patch, so that DISASSEMBLE output is much nicer.
2091 * The code in the SB-PROFILE package now seems reasonably stable.
2092 I still haven't decided what the final interface should look like
2093 (I'd like PROFILE to interact cleanly with TRACE, since both
2094 facilities use function encapsulation) but if you have a need
2095 for profiling now, you can probably use it successfully with
2096 the current CMU-CL-style interface.
2097 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2098 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2099 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2100 link now returns the dangling link itself, and for similar
2101 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2102 link itself. (In these cases the old code signalled an error and
2103 looped endlessly, respectively.) Thus, DIRECTORY now works even
2104 in the presence of dangling and cyclic symbolic links.
2105 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2106 is now a supported extension again, since the consensus on
2107 sbcl-devel was that it can be useful for ordinary development
2108 work, not just for debugging SBCL itself.
2109 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2110 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2111 that currently-defined functions might be redefined later with
2112 different return types.
2113 * Hash tables can be printed readably, as inspired by CMU CL code
2114 of Eric Marsden and SBCL code of Martin Atzmueller.
2115 * better error handling in CLOS method combination, thanks to
2116 Martin Atzmueller porting Pierre Mai's CMU CL patches
2117 * more overflow fixes for >16Mbyte I/O buffers
2118 * A bug in READ has been fixed, so that now a single Ctrl-D
2119 character suffices to cause end-of-file on character streams.
2120 In particular, now you only need one Ctrl-D at the command
2121 line (not two) to exit SBCL.
2122 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2124 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2125 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2126 does the right thing.
2127 * The implementation of some type tests, especially for CONDITION
2128 types, is now tidier and maybe faster, due to CMU CL code
2129 originally by Douglas Crosher, ported by Martin Atzmueller.
2130 * Some math functions have been fixed, and there are new
2131 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2132 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2133 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2134 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2135 likely to be built into sbcl-0.7.0.)
2136 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2137 CL, ported to SBCL by Martin Atzmueller)
2138 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2139 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2140 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2141 * There's a new slam.sh hack to shorten the edit/compile/debug
2142 cycle for low-level changes to SBCL itself, and a new
2143 :SB-AFTER-XC-CORE target feature to control the generation of
2144 the after-xc.core file needed by slam.sh.
2145 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2146 COMPILE-FILE is no longer supported, so that now every function
2147 gets an entry point, so that block compilation looks a little
2148 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2149 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2150 returns the number of bytes consed since the system started,
2151 rather than the number consed since the first time the function
2152 was called. (The new definition parallels ANSI functions like
2153 CL:GET-INTERNAL-RUN-TIME.)
2154 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2155 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2156 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2157 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2158 * Fasl file version numbers are now independent of the target CPU,
2159 since historically most system changes which required version
2160 number changes have affected all CPUs equally. Similarly,
2161 the byte fasl file version is now equal to the ordinary
2164 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2165 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2166 is no longer recognized. I apologize for this, because it was
2167 listed in SB-EXT as a supported extension, but I found that
2168 its existing behavior was poorly specified, as well as incorrectly
2169 specified, and it looked like too much of a mess to straighten it
2170 out. I have enough on my hands trying to get ANSI stuff to work..
2171 * many patches ported from CMU CL by Martin Atzmueller, with
2172 half a dozen bug fixes in pretty-printing and the debugger, and
2173 half a dozen others elsewhere
2174 * fixed bug 13: Floating point infinities are now supported again.
2175 They might still be a little bit flaky, but thanks to bug reports
2176 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2177 as flaky as they were.
2178 * The --noprogrammer command line option is now supported. (Its
2179 behavior is slightly different in detail from what the old man
2180 page claimed it would do, but it's still appropriate under the
2181 same circumstances that the man page talks about.)
2182 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2183 are now supported, and enabled by default. Thus, the compiler can
2184 handle many floating point and complex operations much less
2185 inefficiently. (Thus e.g. you can implement a complex FFT
2187 * The compiler now detects type mismatches between DECLAIM FTYPE
2188 and DEFUN better, and implements CHECK-TYPE more correctly, and
2189 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2190 thanks to patches from Martin Atzmueller.
2191 * various fixes to make the cross-compiler more portable to
2192 ANSI-conforming-but-different cross-compilation hosts (notably
2193 Lispworks for Windows, following bug reports from Arthur Lemmens)
2194 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2195 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2196 fixed thanks to Pierre Mai's CMU CL patches.
2197 * fixes to keep the system from overflowing internal counters when
2198 it tries to use i/o buffers larger than 16M bytes
2199 * fixed bug 45a: Various internal functions required to support
2200 complex special functions have been merged from CMU CL sources.
2201 (When I was first setting up SBCL, I misunderstood a compile-time
2202 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2203 * improved support for type intersection and union, fixing bug 12
2204 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2205 more obscure bugs as well
2206 * some steps toward byte-compiling non-performance-critical
2207 parts of the system, courtesy of patches from Martin Atzmueller
2208 * Christophe Rhodes has made some debian packages of sbcl at
2209 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2210 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2211 stable, but are nonetheless usable. When he's ready, I'd be happy
2212 to add them to the SourceForge "File Releases" section. (And if
2213 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2214 * new fasl file format version number (because of changes in
2215 internal representation of (OR ..) types to accommodate the new
2216 support for (AND ..) types, among other things)
2218 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2219 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2221 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2223 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2224 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2225 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2226 * DESCRIBE now works on structure objects again.
2227 * Most function call argument type mismatches are now handled as
2228 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2229 know whether the function will be redefined before the call is
2230 executed. (The compiler could flag local calls with full WARNINGs,
2231 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2232 it doesn't keep track of enough information to know whether calls
2233 are local in this sense.)
2234 * Compiler output is now more verbose, with messages truncated
2235 later than before. (There should be some supported way for users
2236 to override the default verbosity, but I haven't decided how to
2237 provide it yet, so this behavior is still controlled by the internal
2238 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2239 src/compiler/ir1util.lisp.)
2240 * Fasl file format version numbers have increased again, because
2241 support for the Gray streams extension changes the layout of the
2242 system's STREAM objects.
2243 * The Gray subclassable streams extension now works, thanks to a
2244 patch from Martin Atzmueller.
2245 * The full LOAD-FOREIGN extension (not just the primitive
2246 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2247 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2248 but like most other programs, it defaults to copying the Unix
2249 environment from the original process instead of starting the
2250 new process in an empty environment.
2251 * Extensions which manipulate the Unix environment now support
2252 an :ENVIRONMENT keyword option which doesn't smash case or
2253 do other bad things. The CMU-CL-style :ENV option is retained
2254 for porting convenience.
2255 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2256 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2258 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2260 * A patch from Martin Atzmueller seems to have solved the SIGINT
2261 problem, and as far as we know, signal-handling now works cleanly.
2262 (If you find any new bugs, please report them!)
2263 * The system no longer defaults Lisp source file names to types
2264 ".l", ".cl", or ".lsp", but only to ".lisp".
2265 * The compiler no longer uses special default file extensions for
2266 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2267 seems to expect a single default extension for all compiled code,
2268 and there's no compelling reason to try to stretch the standard
2269 to allow two different extensions.) Instead, byte-compiled files
2270 default to the same extension as native-compiled files.
2271 * Fasl file format version numbers have increased again, because
2272 a rearrangement of internal implementation packages made some
2273 dumped symbols in old fasl files unreadable in new cores.
2274 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2275 many fewer weird special cases.
2276 * Bug #17 (differing COMPILE-FILE behavior between logical and
2277 physical pathnames) has been fixed, and some related misbehavior too,
2278 thanks to a patch from Martin Atzmueller.
2279 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2280 by Tim Moore, ported to SBCL by Martin Atzmueller.
2281 * Martin Atzmueller fixed several filesystem-related problems,
2282 including bug #36, in part by porting CMU CL patches, which were
2283 written in part by Paul Werkowski.
2284 * More compiler warnings in src/runtime/ are gone, thanks to
2285 more patches from Martin Atzmueller.
2286 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2289 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2291 * DESCRIBE now works on CONDITION objects.
2292 * The debugger now handles errors which arise when trying to print
2293 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2295 * The build system now uses an additional file, customize-target-features.lisp,
2296 to allow local modifications to the target *FEATURES* list. (The point of
2297 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2298 debugging features enabled, without having to worry about propagating it
2299 into everyone's system when I do a "cvs update".) When no
2300 customize-target-features.lisp file exists, the target *FEATURES* list
2301 should be constructed the same way as before.
2302 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2303 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2304 warn about behavior which is undefined under the ANSI spec. Note: This
2305 is specified by ANSI, but it's not very popular with programmers.
2306 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2307 macro in the SBCL sources for an example of a workaround which you
2308 might use to make portable ANSI-standard code which does what you want.
2309 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2310 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2311 when it's used as a non-toplevel form. (This is required in order
2312 to implement the DEFCONSTANT-EQX macro.)
2313 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2314 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2315 funny things, due to bugs in the implementation of EVAL-WHEN
2316 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2317 non-toplevel DEFCONSTANTs are uncommon.)
2318 * The core file version number and fasl file version number have been
2319 incremented, because the old noncompliant DEFCONSTANT behavior involved
2320 calling functions which no longer exist, and because I also took the
2321 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2322 * fixed bug 1 (error handling before read-eval-print loop starts), and
2323 redid debugger restarts and related debugger commands somewhat while
2325 ** The QUIT debugger command is gone, since it did something
2326 rather different than the SB-EXT:QUIT command, and since it never
2327 worked properly outside the main toplevel read/eval/print loop.
2328 Invoking the new TOPLEVEL restart provides the same functionality.
2329 ** The GO debugger command is also gone, since you can just invoke
2330 the CONTINUE restart directly instead.
2331 ** The TOP debugger command is also gone, since it's redundant with the
2332 FRAME 0 command, and since it interfered with abbreviations for the
2334 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2336 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2337 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2338 some porting work by Martin Atzmueller)
2339 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2340 with a boatload of patches which he ported to SBCL. Now that those
2342 ** The system tries to make sure that its low-priority messages
2343 are prefixed by semicolons, to help people who like to use
2344 syntax highlighting in their ILISP buffer. (This patch
2345 was originally due to Raymond Toy.)
2346 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2347 patches originally written by Raymond Toy.
2348 ** The compiler understands coercion between single-value and
2349 multiple-VALUES type expressions better, getting rid of some very
2350 weird behavior, thanks to patches originally by Robert MacLachlan
2351 and Douglas Crosher.
2352 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2353 lambda lists, thanks to a patch originally by Pierre Mai.
2354 ** The system no longer bogusly warns about "abbreviated type
2356 ** The compiler gets less confused by inlining and RETURN-FROM,
2357 thanks to some patches originally by Tim Moore.
2358 ** The system no longer hangs when dumping circular lists to fasl
2359 files, thanks to a patch originally from Douglas Crosher.
2360 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2361 undefined function error.
2362 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2363 in structure slot initforms) for now, documented workaround instead:-|
2364 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2365 * fixed bug 5: added stubs for various Gray stream functions called
2366 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2367 installed, at least appropriate type errors are generated
2368 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2369 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2370 consistently in DEFMETHOD forms.
2371 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2372 it doesn't seem to affect SBCL after all
2373 * The C runtime system now builds with better optimization and many
2374 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2376 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2378 * The system is now under CVS at SourceForge (instead of the
2379 CVS repository on my home machine).
2380 * The new signal handling code has been tweaked to treat register
2381 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2382 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2383 downstream, e.g. in debug-int.lisp, has implicit dependencies
2384 on the unsignedness of integer representation of machine words,
2385 and that caused the system to bomb out with infinite regress
2386 when trying to recover from type errors involving signed values,
2387 e.g. (BUTLAST '(1 2 3) -1).)
2388 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2389 (This was one of the bugs Peter Van Eynde reported back in July.)
2390 * The system now uses code inspired by Colin Walters' O(N)
2391 implementation of MAP (from the cmucl-imp@cons.org mailing
2392 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2393 inline the MAP operation, and there is more than one
2394 sequence argument to the MAP call (so that it can't just
2395 do ETYPECASE once and for all based on the type of the
2396 single sequence argument). (The old non-inline implementation
2397 of the general M-argument sequence-of-length-N case required
2398 O(M*N*N) time when any of the sequence arguments were LISTs.)
2399 * The QUIT :UNIX-CODE keyword argument has been renamed to
2400 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2401 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2403 * Raymond Wiker's patches to port dynamic loading from Linux to
2404 FreeBSD have been added.
2405 * The BUGS file is now more nearly up to date, thanks in large part
2406 to Martin Atzmueller's review of it.
2407 * The debugger now flushes standard output streams before it begins
2408 its output ("debugger invoked" and so forth).
2409 * The core version number and fasl file version number have both
2410 been incremented, because of incompatible changes in the layout
2412 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2413 used to not be called for a saved Lisp image.)
2414 * Martin Atzmueller's version of a patch to fix a compiler crash,
2415 as posted on sbcl-devel 13 September 2000, has been installed.
2416 * Instead of installing Martin Atzmueller's patch for the
2417 compiler transform for SUBSEQ, I deleted the compiler transform,
2418 and transforms for some similar consing operations.
2419 * A bug in signal handling which kept TRACE from working on OpenBSD
2421 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2422 away by constant folding
2423 * The system now defines its address space constants in one place
2424 (in the Lisp sources), and propagates them automatically elsewhere
2425 (through GENESIS and the sbcl.h file). Therefore, patching the
2426 address map is less unnecessarily tedious and error-prone. The
2427 Lisp names of address space constants have also been systematized.
2428 * CVS tags like dollar-Header-dollar have been removed from
2429 the sources, because they have never saved me trouble and
2430 they've been source of trouble working with patches and other
2431 diff-related operations.
2432 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2433 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2435 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2437 * The system has been ported to OpenBSD.
2438 * The system now compiles with a simple "sh make.sh" on the systems
2439 that it's supported on. I.e., now you no longer need to tweak
2440 text in the target-features.lisp-expr and symlinks in src/runtime/
2441 by hand, the make.sh takes care of it for you.
2442 * The system is no longer so grossly inefficient when compiling code
2443 involving vectors implemented as general (not simple) vectors (VECTOR T),
2444 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2445 which dares to use the various sequence functions on non-simple
2446 vectors, takes less of a performance hit.
2447 * There is now a primitive type predicate VECTOR-T-P
2448 to test for the (VECTOR T) type, so that e.g.
2449 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2450 can now be compiled with some semblance of efficiency. (The old code
2451 turned the type declaration into a full call to %TYPEP at runtime!)
2452 * AREF on (VECTOR T) is still not fast, since it's still compiled
2453 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2454 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2455 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2456 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2457 without making full calls to SUBTYPEP at runtime.
2458 (Some analogous efficiency issues for non-simple vectors specialized to
2459 element types other than T, or for non-simple multidimensional arrays,
2460 have not been addressed. They could almost certainly be handled the
2461 same way if anyone is motivated to do so.)
2462 * The changes in array handling break binary compatibility, so
2463 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2464 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2465 instead of (VALUES T T).
2466 * By following the instructions that Dan Barlow posted to sbcl-devel
2467 on 2 July 2000, I was able to enable primitive dynamic object
2468 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2469 functionality is not implemented (since it calls ld to resolve
2470 library references automatically, requiring RUN-PROGRAM for its
2471 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2472 try to resolve library references) is now supported.
2473 * The system now flushes the standard output streams when it terminates,
2474 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2475 them at several other probably-convenient times, e.g. in each pass of
2476 the toplevel read-eval-print loop, and after evaluating a form given
2477 as an "--eval" command-line option. (These changes were motivated by a
2478 discussion of stream flushing issues on cmucl-imp in August 2000.)
2479 * The source transform for TYPEP of array types no longer assumes
2480 that an array whose element type is a not-yet-defined type
2481 is implemented as an array of T, but instead punts, so that the
2482 type will be interpreted at runtime.
2483 * There is now some support for cross-compiling in make.sh: each of
2484 the phases of make.sh has its own script. (This should be transparent
2485 to people doing ordinary, non-cross-compile builds.)
2486 * Since my laptop doesn't have hundreds of megabytes of memory like
2487 my desktop machine, I became more motivated to do some items on
2488 my to-do list in order to reduce the size of the system a little:
2489 ** Arrange for various needed-only-at-cold-init things to be
2490 uninterned after cold init. To support this, those things have
2491 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2492 symbols with such names are now uninterned after cold init).
2493 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2494 fasl files for cold load.
2495 ** Remove the old compiler structure pooling code (which used to
2496 be conditional on the target feature :SB-ALLOC) completely.
2497 ** Redo the representation of some data in cold init to be more compact.
2498 (I also looked into supporting byte compiled code at bootstrap time,
2499 which would probably reduce the size of the system a lot, but that
2500 looked too complicated, so I punted for now.)
2501 * The maximum signal nesting depth in the src/runtime/ support code has
2502 been reduced from 4096 to 256. (I don't know any reason for the very
2503 large old value. If the new smaller value turns out to break something,
2504 I'll probably just bump it back up.)
2505 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2507 * Many, many bugs reported by Peter Van Eynde have been added to
2508 the BUGS list; some have even been fixed.
2509 * While enabling dynamic object file loading, I tried to make the
2510 code easier to understand, renaming various functions and variables
2511 with less ambiguous names, and changing some function calling
2512 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2513 * While trying to figure out how to do the OpenBSD port, I tried to
2514 clean up some of the code in src/runtime/. In particular, I dropped
2515 support for non-POSIX signal handling, added various comments,
2516 tweaked the code to reduce the number of compilation warnings, and
2517 renamed some files to increase consistency.
2518 * To support the new automatic configuration functionality in make.sh,
2519 the source file target-features.lisp-expr has been replaced with the
2520 source file base-target-features.lisp-expr and the machine-generated
2521 file local-target-features.lisp-expr.
2522 * fixed a stupid quoting error in make.sh so that using CMU CL
2523 "lisp -batch" as cross-compilation host works again
2525 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2527 * DESCRIBE no longer tries to call itself recursively to describe
2528 bound/fbound values, so that it no longer fails on symbols which are
2529 bound to themselves (like keywords, T, and NIL).
2530 * DESCRIBE now works on generic functions.
2531 * The printer now prints less-screwed-up representations of closures
2532 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2533 * A private symbol is used instead of the :EMPTY keyword previously
2534 used to mark empty slots in hash tables. Thus
2535 (DEFVAR *HT* (MAKE-HASH-TABLE))
2536 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2537 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2538 now does what ANSI says that it should. (You can still get
2539 similar noncompliant behavior if bang on the hash table
2540 implementation with all the symbols you get back from
2541 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2542 This breaks binary compatibility, since tests for equality to
2543 :EMPTY are wired into things like the macroexpansion of
2544 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2546 * There's now a minimal placeholder implementation for CL:STEP,
2547 as required by ANSI.
2548 * An obscure bug in the interaction of the normal compiler, the byte
2549 compiler, inlining, and structure predicates has been patched
2550 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2551 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2553 * Missing ordinary arguments in a macro call are now detected even
2554 when the macro lambda list contains &KEY or &REST.
2555 * The debugger no longer complains about encountering the top of the
2556 stack when you type "FRAME 0" to explicitly instruct it to go to
2557 the top of the stack. And it now prints the frame you request even
2558 if it's the current frame (instead of saying "You are here.").
2559 * As specified by ANSI, the system now always prints keywords
2560 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2562 * The default initial SIZE of HASH-TABLEs is now smaller.
2563 * Type information from CLOS class dispatch is now propagated
2564 into DEFMETHOD bodies, so that e.g.
2565 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2567 is now basically equivalent to
2568 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2569 (DECLARE (TYPE SINGLE-FLOAT X))
2571 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2572 operation, without having to do run-time type dispatch.
2573 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2574 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2575 * Since I don't seem to be making big file reorganizations very often
2576 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2577 is overflowing my ability to conveniently back them up), I've finally
2578 checked the system into CVS. (The CVS repository is on my home system,
2579 not at SourceForge -- putting it on SourceForge might come later.)
2580 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2581 high-level GC-NOTIFY-FOO functions send their output. (There's
2582 still very little control of where low-level verbose GC functions
2583 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2584 less than it used to -- the GC-NOTIFY-FOO functions are now under
2585 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2586 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2587 in only one place in the source code, and propagates it automatically
2588 everywhere that it's needed. Thus e.g. when I bump the version from
2589 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2590 * The C source files now include boilerplate legalese and documentation
2591 at the head of each file (just as the Lisp source files already did).
2592 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2593 to his page will be replaced with a link to his new CLiki service.
2595 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2597 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2598 * The build process now looks for GNU make under the default name "gmake",
2599 instead of "make" as it used to. If GNU make is not available as "gmake"
2600 on your system, you can change this default behavior by setting the
2601 GNUMAKE environment variable.
2602 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2603 can build without error under CMU CL.
2605 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2607 * There is now a partial SBCL user manual (with some new text and some
2608 text cribbed from the CMU CL manual).
2609 * The beginnings of a profiler have been added (starting with the
2610 CMU CL profiler and simplifying and cleaning up). Eventually the
2611 main interface should be through the TRACE macro, but for now,
2612 it's still accessed through vaguely CMU-CL-style functions and macros
2613 exported from the package SB-PROFILE.
2614 * Some problems left over from porting CMU CL to the new
2615 cross-compilation bootstrap process have been cleaned up:
2616 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2617 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2618 part when getting rid of the compiler *BACKEND* stuff.)
2619 ** The value of *NULL-TYPE* was screwed up, because it was
2620 being initialized before the type system knew the final
2621 definition of the 'NULL type. This screwed up several key
2622 optimizations in the compiler, causing inefficiency in all sorts
2623 of places. (I found it because I wanted to understand why
2624 GET-INTERNAL-RUN-TIME was consing.)
2625 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2626 PROCLAIM FTYPE information. Unfortunately this broke binary
2627 compatibility again, since now the forms output by DEFGENERIC
2628 to refer to functions which didn't exist in 0.6.3.
2629 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2630 can use the new (as of 0.6.3) transform for SOME into MAP into
2632 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2633 values to (INTEGER 0 1000000), so that the time code will no longer
2634 occasionally get blown up by Linux returning 1000000 microseconds
2635 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2636 its output conform to the ANSI spec. (Alas, this makes its output
2637 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2638 * A full call to MAP NIL with a single sequence argument no longer conses.
2639 * fixes to problems pointed out by Martin Atzmueller:
2640 * The manual page no longer talks about multiprocessing as though
2641 it were currently supported.
2642 * The ILISP support patches have been removed from the distribution,
2643 because as of version 5.10.1, ILISP now supports SBCL without us
2644 having to maintain patches.
2645 * added a modified version of Raymond Toy's recent CMU CL patch for
2646 EQUALP comparison of HASH-TABLE
2648 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2650 * The system still can't cross-compile itself with
2651 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2652 entails), but at least it can compile more of itself that way
2653 than it used to be able to, and various buglets which were uncovered
2654 by trying to cross-compile itself that way have now been fixed.
2655 * This release breaks binary compatibility again. This time
2656 at least I've incremented the FASL file format version to 2, so that the
2657 problem can be detected reliably instead of just causing weird errors.
2658 * various new style warnings:
2659 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2660 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2661 ** using the lexical binding of a variable named in the *FOO* style
2662 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2663 as specified by ANSI.
2664 * *RANDOM-STATE* is no longer automatically initialized from
2665 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2666 default behavior of the system is to repeat its behavior every time
2667 it's run. If you'd like to change this behavior, you can always
2668 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2669 old convention there was no comparably easy way to get the system to
2670 repeat its behavior every time it was run.
2671 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2672 FTYPE declarations has been removed, in favor of their ANSI
2673 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2674 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2675 the types of their sequence arguments can be determined at compile time.
2676 This is done through a new open code expansion for MAP which eliminates
2677 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2678 argument types can be determined at compile time.
2679 * The optimizer now transforms COERCE into an identity operation when it
2680 can prove that the coerced object is already of the correct type. (This
2681 can be a win for machine generated code, including the output of other
2682 optimization transforms, such as the MAP transform above.)
2683 * Credit information has been moved from source file headers into CREDITS.
2684 * Source file headers have been made more standard.
2685 * The CASE macro now compiles without complaining even when it has
2688 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2690 * (Note that the way that the PCL macroexpansions were rewritten
2691 to accommodate the change in DEFGENERIC below breaks binary
2692 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2693 not run under sbcl-0.6.2. Once we get out of alpha releases,
2694 i.e. hit release 1.0.0, we'll probably try to maintain binary
2695 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2696 and sbcl-1.4.4. Until then, however, it might be fairly common
2697 for maintenance releases to break binary compatibility.)
2698 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2700 * The reporting of unhandled signals has been changed to print some
2701 explanatory text as well as the report form. (Previously only
2702 the report form was printed.)
2703 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2704 it defines, so that the compiler no longer issues undefined function
2705 warnings for compiled-but-not-yet-loaded generic functions.
2706 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2707 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2708 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2709 section of the ANSI spec are supported.
2710 * The "" nickname for the "KEYWORD" package has been removed.
2711 The reader still handles symbol tokens which begin with a package marker
2712 as keywords, but it doesn't expose its mechanism for doing so in the
2713 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2714 * The system now issues STYLE-WARNINGs for contradictory TYPE
2715 proclamations. (Warnings for contradictory FTYPE proclamations would
2716 be nice too, but those can't be done usefully unless the type system
2717 is made smarter about FUNCTION types.)
2718 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2719 systematized, so that "*target-*.lisp is supposed to exist only on the
2720 target and imply that there's a related file which exists on the
2721 host, and *host-*.lisp is supposed to exist only on the host and imply
2722 that there's a related file which exists on the target. This involves a
2723 lot of renaming. Hopefully the acute confusion caused by the renaming
2724 will be justified by the reduction in chronic confusion..
2725 ** runtime-type.lisp -> early-target-type.lisp
2726 ** target-type.lisp -> late-target-type.lisp
2727 ** early-host-format.lisp -> early-format.lisp
2728 ** late-host-format.lisp -> late-format.lisp
2729 ** host-error.lisp -> misc-error.lisp
2730 ** early-error.lisp -> early-target-error.lisp
2731 ** late-error.lisp -> late-target-error.lisp
2732 ** host-defboot.lisp -> early-defboot.lisp
2733 ** code/misc.lisp -> code/target-misc.lisp
2734 ** code/host-misc.lisp -> code/misc.lisp
2735 ** code/numbers.lisp -> code/target-numbers.lisp
2736 ** code/early-numbers.lisp -> numbers.lisp
2737 ** early-host-type.lisp -> early-type.lisp
2738 ** late-host-type.lisp -> late-type.lisp
2739 ** host-typep.lisp -> typep.lisp
2740 ** load.lisp -> target-load.lisp
2741 ** host-load.lisp -> load.lisp
2742 ** host-disassem.lisp -> disassem.lisp
2743 ** host-insts.lisp -> insts.lisp
2744 ** byte-comp.lisp -> target-byte-comp.lisp
2745 ** host-byte-comp.lisp -> byte-comp.lisp
2746 ** host-signal.lisp -> signal.lisp
2747 ** host-defstruct.lisp -> defstruct.lisp
2748 ** late-target-type.lisp -> deftypes-for-target.lisp
2749 Furthermore, several other previously target-only files foo.lisp (e.g.
2750 hash-table.lisp and random.lisp) have been split into a target-and-host
2751 foo.lisp file and a target-only target-foo.lisp file, with their key type
2752 definitions in the target-and-host part, so that the cross-compiler will
2753 know more about target types.
2754 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2755 gone away. In their place are various *BACKEND-FOO* variables
2756 corresponding to the slots of the old structure.
2757 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2758 into the target SBCL has been fixed.
2759 * The chill.lisp system for loading cold code into a running SBCL
2761 * Support for the CMU CL "scavenger hook" extension has been removed.
2762 (It was undocumented and unused in the CMU CL sources that SBCL was
2763 derived from, and stale in sbcl-0.6.1.)
2764 * Various errors in the cross-compiler type system were detected
2765 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2766 (enabling various consistency checks). Many of them were fixed,
2767 but some hard problems remain, so the compiler is back to
2768 running without *TYPE-SYSTEM-INITIALIZED* for now.
2769 * As part of the cross-compiler type system cleanup, I implemented
2770 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2771 * I have started adding UNCROSS calls throughout the type system
2772 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2773 on cross-compiler input files will be able to go away, and various
2775 * CONSTANTP now returns true for quoted forms (as explicitly required
2778 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2780 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2781 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2782 (MAKE-LIST -1). In the longer term, it ought to become true
2783 that declarations are assertions even at SAFETY 1. For now, it's not
2784 quite true even at SAFETY 3, but it's at least more nearly true..
2785 (Note that this change seems to increases the size of the system by
2786 O(5%) and to decrease the speed of the compiler by 20% or more.)
2787 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2788 for the problem of printing dozens of lines of distracting information
2789 about low-level system machinery as part of the top stack frame
2790 on entry to the debugger when an undefined function was called.
2791 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2792 is not reset to COMMON-LISP-USER.
2793 * Compilation of stuff related to dyncount.lisp has been made conditional
2794 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2795 smaller. The various dyncount-related symbols have been moved into
2796 a new "SB-DYNCOUNT" package.
2797 * tty-inspect.lisp has been renamed to inspect.lisp.
2798 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2799 feature has gone away. (When we eventually port to other flavors of
2800 libc and/or Unix, we'll try to make the differences between flavors
2801 invisible at the user level.)
2802 * Various other *FEATURES* tags, and/or their associated conditionals,
2803 have been removed if obsolescent, or given better documentation, or
2804 sometimes given more-mnemonic names.
2806 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2808 * tidied up "make.sh" script
2809 * tidied up system directory structure
2810 * better "clean.sh" behavior
2811 * added doc/FOR-CMUCL-DEVELOPERS
2812 * many many small tweaks to output format, e.g. removing possibly-confusing
2813 trailing #\. character in DESCRIBE-INSTANCE
2814 * (EQUALP #\A 'A) no longer signals an error.
2815 * new hashing code, including EQUALP hashing
2816 * tidied up Lisp initialization and toplevel
2817 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2818 * command line argument processing
2819 * added POSIX-GETENV function to deal with Unix-ish environment variables
2820 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2821 terminating SBCL on EOF
2822 * non-verbose GC by default
2823 * There is no more "sbcl" shell script; the sbcl file is now the C
2824 runtime executable (just like CMU CL).
2825 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2827 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2828 symbol and package information as Lisp native symbol and package objects
2829 instead of strings naming symbols and strings naming packages. This way,
2830 whenever packages are renamed (as in warm init), debug information is
2831 transformed along with everything else.
2832 * tweaked the optimization policy declarations which control the building
2833 of SBCL itself. Now, among other things, the system no longer saves
2834 source location debugging information. (This helps two problems at once
2835 by reducing SBCL size and by keeping SBCL from trying to look for its
2836 sources -- which may not exist -- when reporting errors.)
2837 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2838 debugging and testing purposes
2839 * cleaned up printing, making the printer call PRINT-OBJECT for
2840 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2841 methods, giving nearly-ANSI behavior
2842 * converted almost all special variables to use *FOO* naming convention
2843 * deleted PARSE-TIME functionality, since it can be done portably
2844 * moved some files out of cold init into warm init
2845 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2847 * regularized formatting of source files
2848 * added an install.sh script
2849 * fixed ridiculous memory usage of cross-compiler by making
2850 compiler/alloc.lisp not try to do pooling unless it can hook
2851 itself into the GC of the cross-compilation host. Now the system
2852 builds nicely on my old laptop.
2853 * added :SB-ALLOC in target-features.lisp-expr
2854 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2855 was not implemented)
2856 * re-did condition handling and note reporting in the compiler. Notes
2857 are no longer handled by signalling conditions. Style warnings
2858 and warnings are handled more correctly and reported in such a way
2859 that it's easy to find one or the other in your output (so that you
2860 can e.g. figure out which of many problems caused COMPILE-FILE to
2862 * changed the severity of several compiler warnings from full WARNING
2863 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2864 compiler note reporting so that it doesn't use the condition system
2865 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2866 COMPILE-FILE command)
2867 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2868 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2869 no longer supported, since their implementation was deeply intertwingled
2870 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2871 * removed bogus "support" for compiler macros named (SETF FOO), and
2872 removed the compiler macro for SETF INFO (but only after making a fool
2873 of myself on the cmucl-imp mailing list by posting a bogus patch for
2874 DEFINE-COMPILER-MACRO..)
2875 * Compiled files containing forms which have side effects on the Lisp
2876 reader (such as DEFPACKAGE forms) are now handled more correctly.
2877 (Compiler queuing of top level lambdas has been suppressed by setting
2878 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2879 * deleted various currently-unused source files, e.g. gengc.lisp. They
2880 may be added back at some point e.g. when porting to other architectures,
2881 but until they are it's distracting to distribute them and to try to
2883 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2884 there were so many of them they're just distractions, and UNCROSS is
2885 known to be able to handle the current sources
2886 * moved PROFILE functionality into TRACE, so that it will be clear
2887 how the wrapping and unwrapping of functions when you profile them
2888 interacts with the wrapping and unwrapping of functions when you
2889 trace them. (Actually, the functionality isn't there yet, but at least
2890 the interface specification is there. Hopefully, the functionality will
2891 arrive with some future maintenance release.)
2892 * removed host-oops.lisp
2893 * changed signature of QUIT function to allow UNIX-CODE argument
2894 * fixed READ-SEQUENCE bug
2895 * tweaked verbose GC output so that it looks more like the progress
2896 output that ANSI specifies for functions like LOAD
2897 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2898 * added <http://sbcl.sourceforge.com> to the banner information printed by