1 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2 * new feature: reloading changed shared object files with
3 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
4 * new feature: references to foreign variables and functions
5 can now be compiled and loaded before the corresponding shared
6 object file is loaded, as long as the foreign definitions are
8 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
9 supported. (thanks to Dan Debertin)
10 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
11 just keywords. (thanks to Vincent Arkesteijn)
12 * fixed bug #331: structure-class instances corresponding to
13 DEFSTRUCT forms are now created eagerly.
14 * fixed bug #345: backtraces from calls to undefined functions work
15 on x86 as well. Related bug #61 is now also partially fixed on x86
16 (backtraces from throws to unknown catch tags.)
17 * bug fix: lambda-list parsing is now stricter vrt. order and number
18 of lambda-list keywords.
19 * bug fix: as specified by AMOP, an error is signalled if a
20 class-option appears multiple times in a DEFCLASS form. (reported
22 * bug fix: RANDOM can be compiled when the compiler derives the type
23 of its numeric argument as a disjoint set of small integers.
24 (reported by Paul Dietz)
25 * bug fix: starting a core saved with shared objects loaded when
26 those objects are not available no longer causes threaded SBCL to
27 hang. (reported by Sean Ross)
28 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
30 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
31 argument, which must always be NIL. (reported by Kalle Niemitalo)
32 * bug fix: printing 1.0d+23 no longer results in an error.
33 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
34 * bug fix: structure slot setters preserve evaluation order. (thanks
36 * fixed some bugs related to Unicode integration:
37 ** RUN-PROGRAM can allow its child to take input from a Lisp
38 stream. (reported by Stefan Scholl)
39 ** ASDF-INSTALL successfully downloads .tar.gz files in all
40 locales. (reported by Ken Causey)
41 * fixed some bugs revealed by Paul Dietz' test suite:
42 ** INCF, DECF and REMF evaluate their place form as specified in
44 ** FORMATTER expands ~{ ~} iteration directives with V or #
46 ** FORMATTER deals with the ~@[ ~] conditional directive where the
47 consequent uses no arguments correctly.
48 ** the system has a partial understanding of the (COMPLEX RATIO)
50 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
53 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
54 * new feature: a build-time option (controlled by the :SB-UNICODE
55 keyword feature, enabled by default) for building the system with
56 support for the entire 21-bit character space defined by the
58 * new feature: the system now has rudimentary external-format
59 support; the primary user-visible change at this time is that
60 characters with the high bit set (such as lower-case-e-acute) will
61 print correctly to a terminal in a UTF-8 environment.
62 * minor incompatible change: BASE-CHAR no longer names a class;
63 however, CHARACTER continues to do so, as required by ANSI.
64 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
65 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
66 has been moved to the SB-EXT package (temporarily re-exported from
68 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
69 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
71 * fixed bug #351: better error-handlind and reporting for malformed
73 * fixed bug #350: bignum-printing is now more memory-efficient,
74 allowing printing of very large bignums, eg. (expt 2 10000000).
75 (reported by Bruno Haible)
76 * fixed bug #302: better primitive-type selection for intersection
78 * fixed bug #308: non-graphic characters now all have names, as
79 required. (reported by Bruno Haible)
80 * bug fix: redefining a class with different superclasses now correctly
81 removes it from the direct-subclasses of its previous superclasses.
82 (reported by David Morse)
83 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
84 new value now works. (reported by Bruno Haible)
85 * bug fix: correct canonicalization of multiple non-standard slot
86 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
87 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
88 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
89 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
90 now exists, an signals an error.
91 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
92 during macroexpansion for non-iteration variables as well. (reported
93 by Bruno Haible for CMUCL)
94 * bug fix: Cyclic structures and unprintable objects in compiler
95 messages no longer cause errors. (reported by Bruno Haible)
96 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
97 and READ-FROM-STRING all now return a primary value of NIL if
98 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
99 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
100 (reported by Bruno Haible for CMUCL)
101 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
102 do not designate a condition. (reported by Bruno Haible for
104 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
105 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
106 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
107 permitting resolution in favour of any of the conflicting symbols.
108 (reported by Bruno Haible for CMUCL)
109 * bug fix: EQUAL compiler optimizations is less aggressive on
110 strings which can potentially compare true despite having distinct
111 specialized array element types.
112 * bug fix: unit enumerations can be defined without dividing by
113 zero. (thanks to Vincent Arkesteijn)
114 * FORMAT compile-time argument count checking has been enhanced.
115 (report from Bruno Haible for CMUCL)
116 * a partial workaround for the bug 262: the compiler does not try to
117 inline-expand a local function doing RETURN-FROM from a deleted
118 BLOCK. (thanks to Peter Denno for the bug report and to David
119 Wragg for the simple test case)
120 * fixed some bugs revealed by Paul Dietz' test suite:
121 ** RENAME-PACKAGE allows all package designators as new package
123 ** constraint propagation and lambda variable substitution are
124 more cautious in dealing with partially deleted code.
125 ** compiler sometimes lost reoptimization passes.
126 ** CERROR, given a condition as condition designator, passes its
127 remaining arguments to the continue format control without
129 ** Case-altering FORMAT directives work correctly on non-ASCII
131 ** The REAL type specifier handles bounds outside the floating
132 point ranges without signalling FLOATING-POINT-OVERFLOW.
133 ** Functions with IR1-transformations can create intercomponent
134 references to global functions.
135 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
137 ** FORMAT ~R treats a nil value for its first parameter correctly.
139 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
140 * enhancement: saving cores with foreign code loaded is now
141 supported on x86/NetBSD and sparc/Linux in addition to the previously
143 * bug fix: on some platforms repeated installations caused multiple
144 copies of HTML documentation to be installed -- should not happen
145 any more. (reported by Stefan Scholl)
146 * bug fix: parsing self-recursive alien record types multiple times
147 no longer causes infinite recursion. (reported by Thomas F. Burdick,
148 original patch by Helmut Eller for CMUCL)
149 * bug fix: stack-exhaustion detection works now on NetBSD as well.
150 (thanks to Richard Kreuter)
151 * bug fix: defining classes whose accessors are methods on existing
152 generic functions in other (locked) packages no longer signals
153 bogus package lock violations. (reported by François-René Rideau)
154 * bug fix: special variables as DEFMETHOD parameters no longer have
155 associated bogus type declarations. (reported by David Wragg and
157 * bug fix: read-write consistency on streams of element-type
158 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
159 * bug fix: redefiniton of the only method of a generic function with
160 no DEFGENERIC no longer emits a full WARNING. In addition,
161 redefinition of generic functions with no DEFGENERIC to an
162 incompatible lambda list now signals an error. (thanks to Zach
164 * bug fix: DEFGENERIC now works even when there's a function of the
165 same name in an enclosing lexical environment. (thanks to Zach
167 * fixed compiler failure, caused by instrumenting code during
168 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
169 * optimization: added loop analysis and improved register allocation
170 * fixed some bugs revealed by Paul Dietz' test suite:
171 ** POSITION on displaced vectors with non-zero displacement
172 returns the right answer.
173 ** (SIMPLE-STRING) is a valid type specifier for sequence
175 ** *PRINT-LEVEL* handling for slotless structures is pedantically
177 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
178 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
180 ** The justification version of the FORMAT ~< directive treats
181 non-zero minpad parameter correctly.
183 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
184 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
185 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
186 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
187 the supported interface.
188 * new feature: Single-stepping of code compiled with DEBUG 2 or
189 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
190 * new feature: saving cores with foreign code loaded is now
191 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
192 Timothy Moore's work for CMUCL)
193 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
194 parameters to * instead of NIL if no initform is supplied.
195 (reported by Johan Bockgaard)
196 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
197 correctly when FUNCALL appears as the car of the form. Note:
198 despite this FUNCALL forms are not currently subject to
199 compiler-macro expansion. (port of Raymond Toy's fix for the
200 same from CMUCL, reported by Johan Bockgaard)
201 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
202 (thanks for Teemu Kalvas)
203 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
204 case instead of terminating the iteration (thanks for Julian
205 Squires, Sean Champ and Raymond Toy)
206 * bug fix: incorrect expansion of defgeneric that caused a style
207 warning. (thanks for Zach Beane)
208 * bug fix: slot accessor effective method computation works properly
209 for classes with multiple non-standard applicable methods on
210 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
211 * on x86 compiler supports stack allocation of results of LIST and
212 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
213 CMUCL implementation by Gerd Moellmann)
214 * fixed some bugs revealed by Paul Dietz' test suite:
215 ** FORMAT strings with both the ~<~:;~> form of the justification
216 directive and pretty-printing directives cause an error.
218 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
219 * incompatible change: the internal functions
220 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
221 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
222 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
223 instead of the old functions.
224 * new feature: on platforms where "dladdr" is available foreign
225 function names now appear in backtraces. (based on Helmut Eller's
227 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
228 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
230 * documentation: ASDF manual is now distributed with SBCL.
231 * bug fix: SBCL can now load its contributed modules with REQUIRE
232 even if the system-provided entries have been removed from
233 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
235 * bug fix: backtraces involving undefined functions or assembly
236 routines are more informative. (thanks to Brian Downing)
237 * bug fix: mutually referent alien structures now work correctly.
238 (reported by Rick Taube)
239 * bug fix: structures defined by WITH-ALIEN can be referred to
240 within other definitions in the same WITH-ALIEN.
241 * bug fix: division operators (MOD, TRUNCATE and the like) with
242 constant zero divisors and integer dividends no longer generate
244 * bug fix: provide default methods for INPUT-STREAM-P and
245 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
246 * bug fix: improve the detection and resolution of MOP metacycles
247 (where effective method computation depends on the generic function
248 having its effective method computed). (reported by Bruno Haible)
249 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
250 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
251 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
252 * bug fix: pathnames with relative directory components can be
253 represented relative to default pathnames.
254 * optimization: in taking the GCD of bignums, reduce the two bignums
255 to approximately the same size (using Euclid's algorithm) before
256 applying the more sophisticated binary GCD. (thanks to Juho
258 * optimization: COUNT on bitvectors now operates word-at-a-time.
259 * optimization: ASH with a positive, but not necessarily constant,
260 (leftwards) shift, when in a modular context, is compiled to a
262 * fixed some bugs revealed by Paul Dietz' test suite:
263 ** FORMAT variable parameters ("~V<char>") are defaulted properly
264 if the corresponding argument is NIL.
265 ** FORMAT directives accepting a minpad parameter treat negative
267 ** literal commas as character parameters to FORMAT directives are
269 ** literal spaces directly after ~<Newline> directives within a
270 format-logical-block (~:< ~@:>) do not induce :FILL-style
271 conditional newlines.
272 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
273 as specified: it no longer includes conditional newlines).
274 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
276 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
277 :PER-LINE-PREFIX argument does not evaluate to a string.
279 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
280 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
281 the manual for details; for now, package locks can be disabled by
282 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
283 if no major problems are found then it is likely that they will be
284 compiled in unconditionally.
285 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
286 unsupported operators on all platforms. To load a shared library
287 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
288 object file, link it into a shared library outside of SBCL and
289 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
291 * bug workaround: ROOM T and the GC were not working together
292 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
293 reporting facility aren't preserved. That reporting has been
294 disabled, so now until and unless someone figures out how to make it
295 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
296 * minor incompatible change: as threatened around sbcl-0.8.0, the
297 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
298 The symbols are also exported from SB-PCL for backwards
299 compatibility, but more so than before SB-PCL should be treated as
300 an implementation-internal package.
301 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
303 * the SB-SPROF contrib now works on (most) non-x86 architectures.
304 It is known as of this release not to work on the Alpha, however.
305 * fixed bug #167: errors signalled due to illegal syntax in method
306 bodies are now more legible.
307 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
308 designators and can hence be used with TYPEP.
309 * fixed bug #333: CHECK-TYPE now ensures that the type error
310 signalled, if any, has the right object to be accessed by
311 TYPE-ERROR-DATUM. (reported by Tony Martinez)
312 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
313 5.1.2.3 for multiple-value place subforms. (reported by Kalle
315 * fixed bug #334: programmatic addition of slots using specialized
316 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
317 and :ALLOCATION :CLASS slots.
318 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
319 float exponents. (rereported by Peter Seibel)
320 * fixed a bug: #\Space (and other whitespace characters) are no
321 longer considered to be macro characters in standard syntax by
323 * fixed bug: initialization of condition class metaobjects no longer
324 causes an instance of the condition to be created. (reported by
326 * fixed bug: it is now possible to have slots such that
327 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
328 description is neither :INSTANCE nor :CLASS.
329 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
330 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
331 SLOT-VALUE-USING-CLASS), and no longer causes errors with
332 non-standard SLOT-DEFINITION-ALLOCATION values.
333 * fixed bugs: various race conditions handling exiting threads.
334 CL-PPCRE's thread torture test now passes.
335 * fixed arguably-a-bug: GC time proportional to number of threads in
336 system even when most of them are idle
337 * optimization: improved performance of BIT and SBIT on bit-vectors.
338 * .fasl file incompatibility: The fasl file version number has
339 been incremented because of changes associated with package locks.
341 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
342 * minor incompatible change: the system no longer provides
343 optimization hints (conditions which are TYPEP
344 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
345 mode; these hints are still emitted when higher SPEED optimization
347 * new contrib module: a sampling profiler (profiling by statistical
348 sampling, rather than by instrumenting functions) is available as
349 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
350 * the behaviour of the standard function ED is now customizeable by
351 third parties through a hook variable: see ED's documentation
352 string for information on the protocol.
353 * the compiler no longer emits efficiency notes for (FUNCALL X)
354 when the type of X is uncertain under default optimization
356 * fixed bug 276: mutating a binding of a specialized parameter to a
357 method to something that is not TYPEP the specializer is now
359 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
360 exhibit the required behaviour on the x86 platform. (thanks to
361 Peter van Eynde, Eric Marsden and Bruno Haible)
362 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
363 even for difficult arguments. (reported by Peter Graves)
364 * fixed bug 141a: the backquote printer now descends quoted
366 * fixed another bug in backquote printing: no more destructive
367 modification of the form's list structure. (reported by Brian
369 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
370 move between its address being taken and the call to
371 interrupt_thread, fixing a crashing race condition.
372 * the SB-POSIX contrib implementation has been adjusted so that it
373 no longer exhibits ridiculously poor performance when constructing
374 instances corresponding to C structs.
376 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
377 * minor incompatible change: the sb-grovel contrib now treats C
378 structures as alien (in the sense of SB-ALIEN) objects rather than
379 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
380 has implications for memory management of client code
381 (sb-grovel-returned objects must now be manually managed) and for
382 type safety (alien objects now have full types).
383 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
384 used to control emission of compiler diagnostics, rather than the
385 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
386 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
387 quality should be considered deprecated.
388 * install.sh now installs the user manual as well
389 * (not quite a new documentable feature, but worth considering in
390 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
391 beginnings of a semantically meaningful condition hierarchy is
392 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
394 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
395 designator as the defaults argument.
396 * fixed bug: Displaced arrays whose displaced-to array has become
397 too small now cause ARRAY-DIMENSION to signal an error, providing
398 for safer bounds-checking. (reported by Bruno Haible)
399 * fixed bug: DEFCLASS slot definitions with identical :READER and
400 :WRITER names now signal a reasonable error. (reported by Thomas
402 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
404 * fixed bug: Invalid slot specification errors now print correctly.
405 (thanks to Zach Beane)
406 * fixed bug 320: Shared to local slot value transfers in class
407 redefinitions now happen corrently from superclasses as
408 well. (reported by Bruno Haible)
409 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
411 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
412 improper lists correctly. (reported by Bruno Haible)
413 * fixed bug 313: source-transform for <fun-name> was erroneously
414 applied to a call of a value of a variable with name <fun-name>.
415 (reported by Antonio Menezes Leitao)
416 * fixed bug 307: The obsolete instance protocol ensures that
417 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
418 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
419 remove dead unknown-values globs from the middle of the stack.
420 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
421 easier to resume long computations after using *BREAK-ON-SIGNALS*
422 to diagnose and fix failures. (thanks to Nikodemus Siivola)
423 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
424 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
425 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
426 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
427 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
429 * fixed a (fairly theoretical) bug in string printing: if
430 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
431 does not have array-element-type equal to the most general string
433 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
434 function name. (thanks to Nikodemus Siivola)
435 * fixed bug: compile-time format string checker failed on
436 non-closed ~{. (reported by Thomas F Burdick)
437 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
438 don't warn on legal loop code involving more than one aggregate
439 boolean. (thanks to Nikodemus Siivola)
440 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
441 in keyword position to MAKE-PACKAGE was not being reported as
443 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
444 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
446 * fixed bugs 280 and 312: the checking for multiple definitions in a
447 file is less likely to become confused by uses of inline
449 * fixed bug: the #S reader macro performs the keyword coercion
450 specified for slot names. (reported by Kalle Niemitalo)
451 * fixed bug: lambda lists may contain symbols whose names start with
452 & but are not lambda-list-keywords; their occurrence triggers a
454 * fixed bug 321: define-method-combination argument lambda lists do
455 not cause invalid code to be generated when &optional or &aux
456 variables have default values. (reported by Bruno Haible)
457 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
458 have CLOS classes; this appears also to have fixed reported
459 instability in stack exhaustion detection.
460 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
461 MUFFLE-WARNING when no associated restart is present is now
463 * optimization: rearranged the expansion of various defining macros
464 so that each expands into only one top-level form in a
465 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
467 * optimization: used a previously little-used slot in symbols to
468 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
470 * fixed some bugs revealed by Paul Dietz' test suite:
471 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
473 ** Fixed an optimization bug related to inheritance of initforms
474 from local to shared slots.
475 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
476 ** CAST optimizer forgot to flush argument derived type.
477 ** print/read consistency on floats is now orders of magnitude
478 more likely. (thanks also to Bruno Haible for a similar report
480 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
481 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
482 for objects of type REAL. Make it so.
483 ** SXHASH is specified (infelicitously) to respect similarity,
484 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
485 it so. (thanks to Markus Ziegler)
486 ** on the Alpha, the compiler succeeds in compiling functions
487 returning a known number of arguments greater than 63.
488 ** fixed handling of invalid NIL arguments in keyword position
490 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
492 ** provided a readably-printable representation for RANDOM-STATE
494 ** ensured that pathnames, where they have a namestring, always
495 print using #P"..." syntax.
497 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
498 * Support for the forthcoming 2.0 version of the NetBSD kernel
499 running on x86 hardware has been added. (thanks to Perry
500 E. Metzger most immediately, and others for their past work)
501 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
502 current release version; SBCL's OpenBSD support had been broken
503 since about the time of OpenBSD's switch to ELF binary format.)
504 * [placeholder for DX summary]
505 ** user code with &REST lists declared dynamic-extent, under high
506 speed or space and low safety and debug optimization policy.
507 * The manual has been converted to Texinfo format and the debugger
508 chapter from the cmucl manual has been added.
509 * A facility has been added to extract documentation strings from
510 sbcl and store them as Texinfo-formatted snippets for inclusion in
511 the manual (via Texinfo's @include directive)
512 * bug fix: compiler emitted division in optimized DEREF. (thanks for
513 the test case to Dave Roberts)
514 * bug fix: multidimensional simple arrays loaded from FASLs had fill
515 pointers. (reported by Sean Ross)
516 * bug fix: PROFILE output is printed nicely even for large numerical
517 values. (thanks to Zach Beane)
518 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
519 greater than 32 handle EOF correctly.
520 * bug fix: on X86 an immediate argument of the IMUL instruction is
521 correctly printed in disassembly. (thanks to Lutz Euler)
522 * bug fix: class slots in redefined classes preserve their old
523 values. (thanks to Bruno Haible and Nikodemus Siivola)
524 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
525 succeeds. (reported by Marco Baringer)
526 * fixed some bugs revealed by Paul Dietz' test suite:
527 ** READ-SEQUENCE now works on ECHO-STREAMs.
528 ** RATIONALIZE works more according to its specification. (thanks
531 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
532 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
533 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
534 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
535 should matter to you only if you rebind the printer control
536 variables and then find you want different bindings in the
537 debugger than in the ordinary execution of your program.)
538 * The runtime build system has been tweaked to support building
539 (on SPARC/SunOS) using a C compiler which invokes Sun's own
540 assembler and linker. (thanks to Nikodemus Siivola)
541 * Unbound, undefined, undeclared variables now trigger full
542 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
543 more often programmer error than deliberate exploitation of undefined
545 * optimization: the hash algorithm for strings has changed to one
546 that is less vulnerable to spurious collisions. (thanks to Juho
548 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
549 needless bounds checking. (thanks to Juho Snellman)
550 * optimization: implemented multiplication as a modular
551 (UNSIGNED-BYTE 32) operation on the PPC backend.
552 * fixed some bugs revealed by Paul Dietz' test suite:
553 ** ADJUST-ARRAY now copies the datum in a zero rank array if
555 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
556 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
557 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
559 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
561 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
563 ** ECHO-STREAMs no longer attempt to echo the end of file value to
564 their output stream on EOF from read.
565 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
566 have been read to end-of-file.
567 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
569 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
570 description of determination of which consecutive characters
572 ** Printing the "Space" character with escaping on now yields "#\\ ",
573 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
574 ** Reading floating-point numbers with *READ-BASE* set to a number
575 less than 10 works correctly.
576 ** Reading floating-point numbers with *READ-BASE* set to a number
577 more than 10 works correctly.
578 ** Printing with *PRINT-READABLY* targets the standard readtable, not
579 the readtable currently in effect.
581 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
582 * minor incompatible change: parsing of namestrings on a physical
583 (Unix) host has changed; numbers after the final #\. in a
584 namestring are no longer interpreted as a version field. This is
585 intented to be largely invisible to the user, except that the
586 meaning of the namestring "*.*.*" has changed: it now refers to a
587 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
588 should usually be replaced by
589 (make-pathname :name :wild :type :wild :version :wild)
590 with the added benefit that this is more likely to be portable.
591 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
592 option to OPEN now signals an error if the file being opened
593 exists; this may have an impact on existing code.
594 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
595 signals received, on the PowerPC platforms (both Linux and
596 Darwin). (thanks to Pierre Mai for pointing out the location of
598 * several fixes on OS X: The system now builds and runs cleanly on
599 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
600 (thanks to Brian Mastenbrook)
601 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
602 second-resolution: integer multiples of 1/3600 between -24 and 24.
603 (thanks to Vincent Arkesteijn)
604 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
605 type when called with 1 argument; PEEK-CHAR checked type of
606 PEEK-TYPE only after having read first character from a
607 stream. (reported by Peter Graves)
608 * bug fix: the garbage collector now has much better locality
609 behaviour, and in particular no longer treats all memory as being
610 exhausted when many small objects point to each other in a deeply
612 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
613 recognized as being TYPEP their class.
614 * bug fix: the PUSHNEW documentation string has been corrected.
615 (thanks to Vincent Arkesteijn)
616 * bug fix: defaulting of the value for the last of an atypically
617 large number of multiple values being bound was not being
618 performed correctly on the Alpha or PPC platforms
619 * optimization: implemented multiplication as a modular
620 (UNSIGNED-BYTE 32) operation on the x86 backend.
621 * optimization: SEARCH on simple-base-strings can now be open-coded.
622 (see also contrib/compiler-extras.lisp for inspiration for
623 teaching the compiler about the Boyer-Moore algorithm).
624 * value, returned by MAX (and MIN) called with several EQUALP, but
625 not EQL, arguments now does not depend on compiler settings.
626 * fixed some bugs revealed by Paul Dietz' test suite:
627 ** in stack analysis liveness information is propagated from
628 non-local entry points.
629 ** pathwise CAST removing failed when the CAST node did not start
631 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
632 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
634 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
635 host is already defined.
636 ** RENAME-FILE works on streams instead of signalling an internal
638 ** PEEK-CHAR uses the current readtable when determining whether
639 or not a character is whitespace.
640 ** MERGE-PATHNAMES handles the case when the pathname does not
641 specify a name while the default-pathname specifies a version.
642 ** Pathnames now stand a better chance of respecting print/read
644 ** Attempting to use standardized file system operators with a
645 pathname with invalid :DIRECTORY components signals a
647 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
649 ** DIRECTORY on logical pathnames is more correct.
650 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
651 signal a TYPE-ERROR if their argument is not a stream.
652 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
653 designator argument does not designate a stream.
654 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
655 examining the synonym.
656 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
658 ** OPEN and WITH-OPEN-STREAM allow opening streams with
659 element-type larger than ([UN]SIGNED-BYTE 32).
661 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
662 * When built with the :SB-FUTEX feature, threaded builds now take
663 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
664 for faster/more reliable mutex and condition variable support.
665 * Incompatible change (but one you probably shouldn't have been using
666 anyway): the interface and code for arbitrating between multiple
667 threads in the same user session has been redesigned.
668 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
669 error when called without an explicit environment argument.
670 (thanks to Nikodemus Siivola)
671 * bug fix: buffered :DIRECTION :IO streams are less likely to become
672 confused about their position. (thanks to Adam Warner and Gerd
674 * bug fix: Pretty printing backquoted forms with unquotations in the
675 argument list position of various code constructs such as LAMBDA
676 now works correctly. (reported by Paul Dietz)
677 * bug fix: Pretty printing unquotations no longer loses all
678 stream position information.
679 * optimization: performance of string output streams is now less
680 poor for multiple small sequence writes.
681 * optimization: performance of CSUBTYPEP in the presence of complex
682 expressions involving CONS and NOT many times has been improved.
683 (reported by Paul Dietz)
684 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
686 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
687 streams for the REPL, for the debugger, and so on. (thanks to
689 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
690 * fixed some bugs revealed by Paul Dietz' test suite:
691 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
692 is ignored if it is too small, rather than propagating through
693 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
694 ** extremely complex negations of CONS types were not being
695 sufficiently canonicalized, leading to inconsistencies in
697 ** VALUES tranformer lost derived type.
699 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
700 * fixed a bootstrapping bug: the build process no longer assumes
701 that the various BOOLE-related constants have the same value in
702 host and target lisps. (noted by Paul Dietz' test suite on an
703 SBCL binary built from CLISP)
704 * The system can now be dynamically linked on the MIPS platform,
705 which enables dynamic loading of foreign code from Lisp. (thanks
706 to Ralf Baechle for discussions on the MIPS ABI)
707 * The system now records debugging information for its own source
708 files in a filesystem-position-independent manner, relative to
709 the "SYS" logical host.
710 * fixed a compiler bug: MV-LET convertion did not check references
711 to the "max args" entry point. (reported by Brian Downing)
712 * tweaked disassembly notes to be less confident about proclaiming
713 some instruction as an LRA. (thanks to Brian Downing)
714 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
715 now each have their own history, command character, and other
716 characteristics. (thanks to David Lichteblau)
717 * fixed some bugs revealed by Paul Dietz' test suite:
718 ** compiler failure in compiling LOGAND expressions including a
720 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
721 shift greater than 32.
722 ** FLUSH-DEST did not mark blocks for type check regeneration.
723 ** HANDLER-CASE failed to accept declarations in handler clauses
724 in some circumstances.
726 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
727 * New code in contrib/sb-introspect (still probably not entirely
728 stable yet) provides some support for smart Lisp development
729 environments like SLIME.
730 * The conditions signalled for errors occurring when loading .fasl
731 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
732 in a way which should help ASDF recover gracefully.
733 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
734 stuff has been cleaned up. If you code contrib/ stuff, this might
735 affect you, and you can look at contrib/README, contrib/STANDARDS,
736 and/or the 0.8.4.27 diff to check.
737 * In full calls the compiler now does not generate checks for declared
738 argument types for all arguments.
739 * various threading fixes
740 ** and some experimental patches which didn't make it into
741 the main tree for this release, but which are shipped in
742 contrib/experimental-thread.patch as a possible fix for some
743 failures (deadlock, spinning...) in GC-intensive multithreaded
745 * fixed PPC build problem (source code incompatibility of different
746 library versions): added offsetof() hackery which attempts to divine
747 where glibc maintainers put uc_mcontext today
748 * fixed bug 282: compiler does not trust type assertions while passing
749 arguments to a full call.
750 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
751 &OPTIONAL VALUES type parameter.
752 * fix bug 214: algorithm for noting rejected templates is now more
753 similar to that of template seletion. (also reported by rydis on
755 * fixed bug 141b: printing backquoted information readably and prettily
756 inserts a space where necessary.
757 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
758 (CHARACTER 10) are now reported as errors, rather than propagated
759 as unknown types. (reported by piso on #lisp)
760 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
761 with values NIL and :ERROR. (thanks to Milan Zamazal)
762 * fixed bug 191c: CLOS now does proper keyword argument checking as
763 described in CLHS 7.6.5 and 7.6.5.1.
764 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
765 longer raise an error; further, using a list as a for-as-arithmetic
766 counter now raises a meaningful error.
767 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
768 now understood by sequence creation functions such as MAKE-SEQUENCE
770 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
771 a STRING-INPUT-STREAM.
772 * compiler enhancement: SIGNUM is now better able to derive the type
774 * type declarations inside WITH-SLOTS are checked. (reported by
776 * fixed some bugs revealed by Paul Dietz' test suite:
777 ** incorrect optimization of TRUNCATE for a positive first
778 argument and negative second.
779 ** compiler failure in let-convertion during flushing dead code.
780 ** compiler failure while deriving type of TRUNCATE on an
781 interval, containing 0.
782 ** ASH of a negative bignum by a negative bignum count now returns
784 ** intersection of CONS types now canonicalizes properly, fixing
785 inconsistencies in SUBTYPEP.
787 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
788 * incompatible change: The --disable-debugger command line
789 option now clobbers the debugger at a more fundamental
790 level, by redefining #'INVOKE-DEBUGGER instead of by
791 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
792 is specified by ANSI to ignore *DEBUGGER-HOOK* and
793 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
794 enter the debugger REPL and then suffer recursive errors
795 because *DEBUG-IO* is also messed up in --disable-debugger mode;
796 while under the new system, BREAK in --disable-debugger mode
797 terminates the system just as an unhandled error would.
798 * fixed compiler performance when processing loops with a step >1;
799 * bug fix: DOCUMENTATION now retrieves generic function
800 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
801 support has been systematized, and now supports the methods
802 specified by ANSI, along with a default method and a method for
803 slot documentation. (reported by Nathan Froyd)
804 * bug fix: effective methods associated with a generic function are
805 no longer cached over a change of that generic function's method
806 combination. (reported by Andreas Fuchs)
807 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
808 names a non-existent directory. (thanks to Andreas Fuchs)
809 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
810 the PPC platform, lead to stack corruption; this has been fixed.
811 (reported by Rainer Joswig)
812 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
813 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
814 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
815 arithmetic in general, is now much more reliable on the PPC
817 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
818 platform now returns the right answer.
819 * optimization: restored some effective method precomputation in
820 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
821 precomputation is now tunable.
822 * optimization: compiler-internal data structure use has been
823 reviewed, and changes have been made that should improve the
824 performance of the compiler by about 20%.
825 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
826 simple-base-strings and simple-bit-vectors is improved.
827 * optimization: the optimization of 32-bit logical and arithmetic
828 functions introduced in version 0.8.3 on the x86 has been
829 implemented on the mips, ppc and sparc platforms; an
830 implementation of the same facility, but for 64-bit arithmetic,
831 has been added for the alpha.
832 * microoptimization: the compiler is better able to make use of the
833 x86 LEA instruction for multiplication by constants.
834 * bug fix: in some situations compiler did not report usage of
835 generic arithmetic in (SPEED 3) policy.
836 * bug 145b fix: compiler used wrong type specifier while converting
837 MEMBER-types to numeric.
838 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
840 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
841 index argument now works properly on the Alpha platform.
842 * bug fix: floating point exception treatment on the Alpha platform
844 * bug fix: FILE-POSITION works much better on string input and
845 output streams. (thanks to Nikodemus Siivola)
846 * bug fix: many threading/garbage collection symptoms sorted.
847 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
848 might be pseudo-atomic.
849 * internal change: Stopping for GC is now done with signals not
850 ptrace. GC is now done in whichever thread wanted it, instead of
852 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
854 * bug fix: result form in DO is not contained in the implicit
856 * incompatible change: ICR structure is changed; the value part of
857 CONTINUATION is now called LVAR; corresponding functions are
858 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
859 * added type deriver for ISQRT (thanks to Robert E. Brown).
860 * bug fix: better support for loading from the command line when an
861 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
863 * fixed some bugs revealed by Paul Dietz' test suite:
864 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
865 ** ROUND and FROUND now give the right answer when given very
866 small float arguments.
867 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
869 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
870 ** (GCD 0 <negative-integer>) returned <negative-integer>.
871 ** LCM should return a non-negative integer.
872 ** PARSE-INTEGER returned the index of a terminator instead of the
873 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
874 ** PARSE-INTEGER returned an incorrect index being applied to a
876 ** LCM with two arguments of 0 returns 0 rather than signalling
878 ** unsigned addition of a 32-bit constant with the high bit set no
879 longer causes an internal compiler error on the x86.
880 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
881 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
882 with negative last argument.
883 ** byte specifiers with zero size and position no longer cause
884 an error during type derivation.
885 ** bignum multiplication on the Alpha platform now returns the
887 * porting: The system now builds on SuSE AMD64, although it still
888 generates a 32-bit binary.
889 * .fasl file incompatibility: The fasl file version number has
890 been incremented (because of the changes to internal compiler
891 data structures referred to above).
893 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
894 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
895 more accurately, on the Darwin kernel running on PowerPC hardware.
896 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
897 * Compiler code deletion notes now signal a condition of type
898 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
899 an associated MUFFLE-WARNING restart.
900 * The compiler now performs limited argument count validation of
901 constant format strings in FORMAT, and where appropriate in ERROR,
902 CERROR and WARN. (thanks to Gerd Moellmann)
903 * New ASDF-INSTALL contrib can be used for automatic download and
904 installation of third-party Lisp code from CCLAN or other sites
906 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
907 which forces another thread to execute a function supplied by the
909 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
910 now accept and act upon their :ELEMENT-TYPE keyword argument.
911 (reported by Martin Atzmueller, Edi Weitz)
912 * bug fix: FILE-POSITION now accepts position designators up to
913 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
914 is the greater. (thanks to Patrik Nordebo)
915 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
916 Kalle Olavi Niemitalo)
917 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
918 printable. (reported by Eric Marsden)
919 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
920 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
921 * x86 bug fix in control stack exhaustion checking: now shows backtrace
922 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
923 (thanks to Stig Sandoe)
924 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
925 UNREAD-CHAR now correctly includes the unread character in the
926 target sequence. (thanks to Gerd Moellmann)
927 * bug fix in threaded builds: the system can now be suspended and
928 resumed by shell job control with minimal disruption.
929 * bug fixes in times and timezones >2038AD
930 * better handling of "where is GNU make?" problem in build scripts
931 (thanks to Nikodemus Siivola)
932 * new optimization: inside a named function any reference to a
933 function with the same name is considered to be a self-reference;
934 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
935 optimization quality.
936 * new optimization on x86: logical functions and + now have
937 optimized (UNSIGNED-BYTE 32) versions, which are automatically
938 used when the result is truncated to 32 bits.
939 * VALUES declaration is partially enabled.
940 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
941 * fixed some bugs revealed by Paul Dietz' test suite:
942 ** The system now obeys the constraint imposed by
943 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
944 types form a lattice under type intersection.
945 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
946 ** ASSOC now ignores NIL elements in an alist.
947 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
948 and (1+ MOST-POSITIVE-FIXNUM) answers.
949 ** The addition of a method with invalid qualifiers to a generic
950 function does not cause an error to be signalled immediately;
951 a warning is signalled, and the error is generated only on
952 calling the generic function.
953 * changed .fasl file version number, in part to add type codes for
954 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
955 obscure ANSI requirements
957 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
958 * fixed bug 148: failure to inline-expand a local function left
959 garbage, confusing the compiler.
960 * fixed bugs 3cd: structure slot readers perform type check if the
961 slot can have an invalid value (i.e. it is either not initialized
962 or can be written with a less specific slot writer).
963 * bug fix: the compiler now traps array references to elements off
964 the end of an array; previously, the bounds checking in some
965 circumstances could go off-by-one.
966 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
968 * type declarations for array element types now obey the description
969 on the CLHS page "Declaration TYPE", as per discussions on
970 sbcl-help around 2003-05-08. This means that a declaration
971 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
972 declaration, all references to BAR will be asserted or assumed
973 (with THE, so dependent on compiler policy) to involve objects of
974 type FOO. Note that no such declaration is implied in
975 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
976 * declared types of functions from the "Conditions"
977 chapter. (reported by Paul Dietz)
978 * bug fix: CERROR accepts a function as its first argument.
979 * bug fix: NTH an NTHCDR accept a bignum as index
980 arguments. (reported by Adam Warner)
981 * optimization: character compare routines now optimize comparing
982 against a constant character. (reported by Gilbert Baumann)
983 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
984 argument now works properly on the MIPS platform.
985 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
986 * repeated evaluation of the same DEFSTRUCT, a slot of which is
987 declared to have a functional type, does not cause an error
989 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
990 many :DEFAULT-INITARGS. (reported by Istvan Marko)
991 * fixed bug: if last continuation of a deleted block has a
992 destination, this destination should be deleted too. (reported by
994 * fixed a bug in the bootstrap process: the host compiler's values
995 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
996 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
997 test case from Patrik Nordebo)
998 * improved the ability of the disassembler on the PPC platform to
999 provide helpful disassembly notes.
1000 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1001 the class in more cases than previously.
1002 * bug fix: FILE-POSITION now understands :START and :END for
1003 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1004 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1005 8) data. (thanks to David Lichteblau)
1006 * bug fix: it is possible to add a method to a generic function
1007 without lambda list.
1008 * bug fix: reader failed to signal END-OF-FILE inside an
1009 object representation. (reported by Nikodemus Siivola)
1010 * fixed some bugs revealed by Paul Dietz' test suite:
1011 ** LAST and [N]BUTLAST should accept a bignum.
1012 ** condition slot accessors are methods.
1013 ** (VECTOR NIL) is a subtype of STRING.
1015 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1016 * minor incompatible change: some nonsensical specialized lambda
1017 lists (used in DEFMETHOD) which were previously ignored now signal
1019 * minor incompatible change: the system is now aware of the types of
1020 variables in the COMMON-LISP package, and will signal errors for
1021 most violations of these type constraints (where previously they
1022 were silently accepted).
1023 * minor incompatible change: COMPILE-FILE now uses the freedom
1024 afforded (ANSI 3.2.2.3) to use derived function types for
1025 functions defined in the same file. This also permits the system
1026 to warn on static type mismatches and function
1027 redefinition. (Currently it does not work with high DEBUG level.)
1028 * minor incompatible change: VALUES declaration is disabled.
1029 * When issuing notes, the compiler now signals a condition of type
1030 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1031 restart for use in user handlers. It is expected that the
1032 COMPILER-NOTE condition will eventually become a condition
1033 supertype to a hierarchy of note types, which will then be
1034 handleable in a similar fashion. However, at the moment, no such
1035 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1036 but it's an implementation detail, not a classification for the
1038 * Changes in type checking closed the following bugs:
1039 ** type checking of unused values (192b, 194d, 203);
1040 ** template selection based on unsafe type assertions (192c, 236);
1041 ** type checking in branches (194bc).
1042 * A short form of VALUES type specifier has ANSI meaning (it has
1043 increased the number of situations when SBCL cannot perform type
1045 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1046 keywords or constants is permissible.
1047 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1048 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1049 classes. (thanks to Antonio Martinez)
1050 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1051 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1052 argument to a pretty-print stream. Instead, it leaves any such
1053 operations to DESCRIBE-OBJECT methods.
1054 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1056 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1057 lambda lists. (thanks to David Lichteblau)
1058 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1059 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1060 (thanks to Gerd Moellmann)
1061 * fixed bug 246: increased compilation speed of long
1062 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1064 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1066 * DEFCONSTANT now throws a condition of type
1067 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1068 constant to a non-EQL value; CONTINUE and ABORT restarts
1069 respectively change and preserve the value.
1070 * fixed bug 63: The code walker, part of the implementation of CLOS,
1071 is now better at handling symbol macros.
1072 * bug fix: There is no longer an internal implementation type named
1073 CL:LENGTH. (reported by Raymond Toy)
1074 * bug fix: In macro-like defining macros/special operators the
1075 implicit block does not enclose the lambda list.
1076 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1077 enclosed in parentheses) are not suitable as type specifiers, and
1078 their use properly signals an error now.
1079 * bug fix: An argument count mismatch for a type specifier in code
1080 being compiled no longer causes an unhandled error at compile
1081 time, but signals a compile-time warning.
1082 * fixed simple vector readable printing
1083 * bug fix: DESCRIBE takes more care over whether the class
1084 precedence list slot of a class is bound before accessing it.
1085 (reported by Markus Krummenacker)
1086 * bug fix: FORMATTER can successfully compile pretty-printer format
1087 strings which use variants of the ~* directive inside.
1088 * bug fix: SEARCH now applies its TEST predicate to the elements of
1089 the arguments in the correct order. (thanks to Wolfhard Buss)
1090 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1091 in type inference. (thanks to Robert E. Brown)
1092 * bug fix: Reading in symbols with an explicit package name of ""
1093 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1094 * fixed some bugs revealed by Paul Dietz' test suite:
1095 ** NIL is now allowed as a structure slot name.
1096 ** Arbitrary numbers, not just REALs, are allowed in certain
1097 circumstances in LOOP for-as-arithmetic clauses.
1098 ** Multiple class redefinitions before slot access no longer
1099 causes a type error.
1100 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1101 association between the name and a class.
1102 ** Generic functions with non-standard method-combination and over
1103 five methods all of which return constants no longer return NIL
1104 after the first few invocations. (thanks to Gerd Moellmann)
1105 ** CALL-NEXT-METHOD with no arguments now passes the original
1106 values of the arguments, even in the presence of assignment.
1107 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1109 ** DISASSEMBLE works with closures and funcallable instances.
1110 ** ADD-METHOD now returns the generic function, not the new method.
1111 ** FIND-METHOD signals an error if the lengths of the specializers
1112 is incompatible with the generic function, even if the ERRORP
1114 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1115 which its argument is a member.
1116 ** DEFCLASS only redefines the class named by its class-name
1117 argument if that name is the proper name of the class;
1118 otherwise, it creates a new class.
1119 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1120 of the UNBOUND-SLOT condition to the name of the slot.
1121 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1122 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1123 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1124 SLOT-MAKUNBOUND in the specified fashion.
1126 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1127 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1128 cross-compilation host. As a consequence, we can now bootstrap our
1129 way up to SBCL starting with a bare gcc toolchain and human-readable
1130 source code (first the source to CLISP, then the source to SBCL).
1131 * A contributed module containing a partial implementation of the
1132 simple-streams interface has been included. (thanks to Rudi
1134 * A contributed module implementing the RFC1321 Message Digest
1135 Algorithm, known as MD5, has been included.
1136 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1137 no longer has any effect, as the code controlled by this feature
1138 has been deleted. (As far as we know, no-one has ever built using
1139 this feature, and its semantics were confused in any case).
1140 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1141 work on conditions (as required by the ANSI specification),
1142 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1143 expected behaviour on conditions. Users should note, however,
1144 that such behaviour is not required by the ANSI specification,
1145 and so use of this behaviour may render their code unportable.
1146 * fixed some bugs revealed by Paul Dietz' test suite:
1147 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1149 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1150 ** SXHASH on condition objects no longer returns NIL.
1151 ** :ALLOCATION :CLASS slots are better treated; their values are
1152 updated on class redefinition, and initforms inherited from
1153 superclasses are applied.
1154 ** REMOVE-METHOD returns its generic function argument even when
1155 no method was removed.
1156 ** SHARED-INITIALIZE now initializes the values of the requested
1157 slots, including those with :ALLOCATION :CLASS.
1158 ** ALLOCATE-INSTANCE now works on structure classes defined via
1159 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1161 ** SLOT-EXISTS-P now works on conditions, as well as structures
1163 ** MAKE-LOAD-FORM now has the required methods on
1164 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1165 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1166 keyword, and now implements the SLOT-NAMES argument.
1167 ** methods with &OPTIONAL arguments no longer allow too many
1168 arguments to be passed in the call without error.
1169 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1170 option is consistent with the required arguments of the generic
1171 function lambda list.
1172 * bug fix: REQUIRE accepts a string designator. (Thanks to
1174 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1175 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1176 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1177 * bug fix: The compiler now checks for duplicated variables in macro
1179 * bug fix: SETQ on globals returns the correct value.
1180 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1181 not a COMPILER-ERROR (followed by some other strange error on
1182 choosing the CONTINUE restart).
1183 * bug fix: make.sh and friends are now more consistent in the way that
1184 they look for GNU "make".
1186 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1187 * experimental native threads support (on x86 Linux >=2.4 only).
1188 This is not compiled in by default: you need to add :SB-THREAD to
1189 the target features. See the "Beyond ANSI" chapter of the manual
1191 * fix for longstanding nonANSIism: The old distinction between
1192 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1193 The return value from CL:FIND-CLASS is now a CLOS class, and
1194 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1195 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1196 * An interface to the MetaObject Protocol, as described in Kiczales,
1197 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1198 MIT Press, 1991, is available from the SB-MOP package.
1199 * incompatible change: the SB-PCL package should now be considered
1200 a private implementation detail, and no longer a semi-private MOP
1202 * minor incompatible change: due to rearrangement for threads, the
1203 control stack and binding stack are now allocated at arbitrary
1204 addresses instead of being hardcoded per-port. Users affected by
1205 this probably have to be doing advanced things with shared
1206 libraries, and will know who they are.
1207 * minor incompatible change: Previously, all --eval forms used were
1208 processed with READ before any of them were processed with EVAL.
1209 Now each --eval form is processed with both READ and EVAL before
1210 the next --eval form is processed. (Thus package operations like
1211 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1212 work as the user might reasonably expect.)
1213 * minor incompatible change: *STANDARD-INPUT* is now only an
1214 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1216 * minor incompatible change: Y-OR-N-P is now character-oriented,
1217 not line oriented. Also, YES-OR-NO-P now works without errors.
1218 (thanks to Antonio Martinez)
1219 * sb-aclrepl module improvements: an integrated inspector, added
1220 repl features, and a bug fix to :trace command.
1221 * Known functions, which cannot be open coded by the backend, are
1222 considered to be able to check types of their arguments. (fixing
1223 a bug report by Nathan J. Froyd)
1224 * fixed a bug in computing method discriminating functions: It is
1225 now possible to define methods specialized on classes which have
1226 forward-referenced superclasses. (thanks to Gerd Moellmann)
1227 * fixed evaluation order in optional entries (reported by Gilbert
1229 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1230 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1231 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1232 required-by-AMOP NAME argument, as well as CLASS and
1233 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1234 * fixed bug 20: DEFMETHOD can define methods using names that are
1235 not the proper names of classes to designate class specializers.
1236 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1237 Unix file descriptors, instead of blowing up. (thanks to Antonio
1239 * Garbage collection refactoring: user-visible change is that a
1240 call to the GC function during WITHOUT-GCING will not do garbage
1241 collection until the end of the WITHOUT-GCING. If you were doing
1242 this you were probably losing anyway.
1243 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1244 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1245 TYPEP the latter but not the former.
1246 * The compiler issues a full WARNING for calls to undefined functions
1247 with names from the CL package.
1248 * MAP-INTO for a vector destination is open coded. (reported by
1249 Brian Downing on c.l.l)
1250 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1251 documentation string.
1252 * fixed some bugs revealed by Paul Dietz' test suite:
1253 ** COPY-ALIST now signals an error if its argument is a dotted
1255 ** Condition slots are now accessed more correctly in the presence
1256 of multiple initargs for a given slot.
1257 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1258 exclude restarts of the same name associated with a different
1260 ** DEFCLASS of forward-referenced classes with another
1261 forward-referenced class in the superclasses list no longer
1263 ** Condition slots are now initialized once each, not multiple
1264 times. (thanks to Gerd Moellmann)
1265 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1267 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1268 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1269 arguments contain duplicated elements.
1270 ** RESTART-CASE understands local macros.
1271 ** RESTART-CASE associates exactly its own restarts with a condition.
1272 ** ENDP in safe mode checks its argument to be of type LIST.
1273 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1274 in question is unbound.
1275 ** Optimized MAKE-INSTANCE functions no longer cause internal
1276 assertion failures in the presence of duplicate initargs.
1277 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1278 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1279 propagated even in the ctor optimized implementation of
1281 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1283 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1284 * a better implementation of SXHASH on (simple) bit vectors,
1285 measured both in execution speed and in distribution of results
1286 over the positive fixnums, has been installed. Likewise, a better
1287 implementation of EQUAL for simple bit vectors is now available.
1288 * fixed CEILING optimization for a divisor of form 2^k.
1289 * fixed bug 240 (emitting extra style warnings "using the lexical
1290 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1291 by Antonio Martinez)
1292 * fixed SXHASH, giving different results for NIL depending on type
1293 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1294 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1295 variable. (found by Rolf Wester)
1296 * SBCL does not ignore type declarations for special
1297 variables. (reported by rif on c.l.l 2003-03-05)
1298 * some bug fixes in contrib/sb-aclrepl/
1299 * fixed some bugs revealed by Paul Dietz' test suite:
1300 ** a bug in the CONS type specifier, whereby the CAR and CDR
1301 types got intertwined, has been fixed;
1302 ** the type system is now able to reason about the interaction
1303 between INTEGER and RATIO types more completely;
1304 ** APPEND, [N]REVERSE and NRECONC check that those their
1305 arguments, which must be proper lists, are really so;
1306 ** An array specialized to be unable to hold elements has been
1307 implemented, as required -- yes, really -- by ANSI;
1308 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1309 on malformed property lists;
1311 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1312 * incompatible packaging change: in line with Unix convention,
1313 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1314 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1315 * REQUIRE and PROVIDE are now optionally capable of doing something
1316 useful. See the documentation string for REQUIRE.
1317 * infrastructure for a managed SBCL contrib system: contributed
1318 modules in this release include:
1319 ** the ASDF system definition facility;
1320 ** an interface to the BSD Sockets API;
1321 ** an ACL-like convenience interface to the repl;
1322 (thanks to Kevin Rosenberg)
1323 ** an implementation of ROTATE-BYTE, with efficient implementation
1325 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1326 gives a closer approximation to the right answer than previously.
1327 (thanks to Raymond Toy)
1328 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1329 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1330 optional environment arguments, as required by ANSI.
1331 * fixed bugs in other functions taking environment objects, allowing
1332 calls with an explicit NIL environment argument to be compiled
1334 * fixed bug 228: primary return values from
1335 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1336 COMPILE or FUNCTION.
1337 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1338 :TYPE will no longer signal errors on innocuous objects.
1339 * fixed bug 231b: SETQ is better at respecting type declarations in
1340 the lexical environment.
1341 * fixed a bug in DEFCLASS: classes named by symbols with no or
1342 unprintable packages can now be defined.
1343 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1344 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1345 * fixed compiler failure related to checking types of functions.
1346 (reported by Robert E. Brown)
1347 * the compiler is now much more consistent in its error-checking
1348 treatment of bounding index arguments to sequence functions: in
1349 (SAFETY 3) code, errors will be signalled in almost all cases if
1350 invalid sequence bounding indices are passed to functions defined
1351 by ANSI to operate on sequences.
1352 * fixed a bug in the build procedure: documentation of SBCL-specific
1353 packages is now preserved and available in the final Lisp image.
1354 * lifted FDEFINITION lookup out of loops in the implementation of
1355 many list operations. (thanks to Robert E. Brown)
1356 * fixed a bug in the reader: the #n# reader macro now works for
1357 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1358 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1359 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1360 E. Brown and Tony Martinez respectively)
1361 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1362 SLOT-BOUNDP: in particular, it is now able to optimize them much
1363 better, and is now not vulnerable to having packages renamed.
1364 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1365 always called when a slot is not present in an instance. (thanks
1367 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1368 longer signal an error on unreading a character following EOF on
1369 the previous constituent stream. (thanks to Tony Martinez)
1370 * fixed some bugs revealed by Paul Dietz' test suite:
1371 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1372 not just nonnegative fixnums;
1373 ** the logical bit-array operators such as BIT-AND now accept an
1374 explicit NIL for their "opt-arg" argument (to indicate a
1375 freshly-consed result bit-array);
1376 ** ELT now signals an error on an invalid sequence index in safe
1378 ** the type system is now cleverer about negations of numeric
1379 types, and consequently understands the BIGNUM and RATIO types
1381 ** the type system is now cleverer about the interaction between
1382 INTEGER and RATIO types: while bugs still remain, many more
1383 cases are accurately computed;
1384 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1385 if it is in the last clause;
1386 ** CONSTANTLY now correctly returns a side-effect-free function in
1388 ** DECLARE is no longer treated as a special-operator; in
1389 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1390 * incremented fasl file version number due to the change in the
1391 DEFSTRUCT-SLOT-DESCRIPTION structure.
1393 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1394 * minor incompatible change: code processed by the "interpreter" or
1395 EVAL now has a compilation optimization policy of (DEBUG 2)
1396 (changed from (DEBUG 1)) to improve debuggability of interactive
1397 development, and to allow the use of the debug RETURN command in
1399 * an experimental implementation of the RETURN command for the
1400 debugger has been included. (thanks to Frederik Kuivinen)
1401 * fixed bug 62: constraints were not propagated into a loop.
1402 * fixed bug in embedded calls of SORT (reported and investigated by
1404 * fixed some bugs revealed by Paul Dietz' test suite:
1405 ** printing and reading of arrays with some dimensions having
1406 length 0 (thanks to Gerd Moellmann);
1407 ** BOA constructor with &AUX argument without a default value does
1408 not cause a type error;
1409 ** CONSTANTP now returns true for all self-evaluating objects.
1411 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1412 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1413 accessors that are related by inheritance, as specified in the
1414 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1415 Valtteri Vuorikoski)
1416 * The compiler is now able to inline functions that were defined in
1417 a complex lexical environment (e.g. inside a MACROLET).
1418 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1420 * Support for the upcoming FreeBSD-5.0 release has been included.
1421 (thanks to Dag-Erling Smorgrav)
1422 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1423 effect when it is not in a toplevel context.
1424 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1425 stand a better chance of being correct. (thanks to Gerd
1427 * fixed bug in COERCE, which now signals an error on coercing a
1428 rational to a bounded real type which excludes the expected
1430 * The compiler is now able to derive types more accurately from the
1431 COERCE and COMPILE functions.
1432 * fixed bug 223: functional binding is considered to be constant
1433 only for symbols in the CL package.
1434 * fixed bug 231: SETQ did not check the type of a variable being set
1435 (reported by Robert E. Brown)
1436 * A new optimization for MAKE-INSTANCE has been included, fixing
1437 various bugs (including relating to :ALLOCATION :CLASS slots and
1438 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1440 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1441 ** As required by ANSI, LOOP now disallows anonymous collection
1442 clauses such as COLLECT I in conjunction with aggregate boolean
1443 clauses such as THEREIS (= I 1);
1444 ** LOOP now signals an error when any variable is reused in the
1445 same loop (including the potentially useful construct analogous
1446 to WITH A = 1 WITH A = (1+ A);
1447 ** IT is only a special loop symbol within the first clause of a
1448 conditional loop clause;
1449 ** LOOP with a typed iteration variable over a hashtable now
1450 signals a type error iff it should.
1451 * fixed some other bugs revealed by Paul Dietz' test suite:
1452 ** FILE-STREAM now names the class previously known as FD-STREAM;
1453 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1454 argument) no longer signals an error;
1455 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1456 of the slot symbol, rather than using the current package
1457 ((:CONC-NAME "") continues to intern the slot's name in the
1459 * incremented fasl file version number, because of the incompatible
1460 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1461 of the new implementation of DEFINE-COMPILER-MACRO.
1463 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1464 * Support for building SBCL for MIPS platforms running in
1465 little-endian mode has now been checked in, and basic
1466 functionality on said platforms verified.
1467 * minor incompatible change: PCL now records the pathname of a file
1468 in which methods and the like are defined, rather than its
1470 * minor incompatible change: TRUENAME now considers the truename of
1471 a file naming a directory to be the pathname with :DIRECTORY
1472 component indicating that directory.
1473 * minor incompatible change: a NAMED clause in the extended form of
1474 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1475 reason for the previous behaviour is unclear.
1476 * more systematization and improvement of CLOS and MOP conformance
1477 in PCL (thanks to Gerd Moellman and Pierre Mai):
1478 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1480 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1481 primary methods with no specializers;
1482 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1484 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1485 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1486 FORWARD-REFERENCED-CLASSes; error reporting on
1487 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1489 ** SXHASH on CLOS instances now uses a slot internal to the
1490 instance to return different numbers on distinct instances,
1491 while preserving the same return value through invocations of
1493 ** DEFMETHOD signals errors when methods with longer incongruent
1494 lambda lists are added to generic functions;
1495 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1496 CLASS, as specified in AMOP;
1497 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1498 on the order returned by the primary method for classes of
1499 class STANDARD-CLASS;
1500 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1501 * fixed some bugs shown by Paul Dietz' test suite:
1502 ** DOLIST puts its body in TAGBODY;
1503 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1505 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1506 value producing form;
1507 ** if more variables are given to PROGV than values, extra
1508 variables are bound and made to have no value;
1509 ** NSUBSTITUTE on list arguments gets the right answer with
1511 ** ELT signals an error of type TYPE-ERROR when the index argument
1512 is not a valid sequence index;
1513 ** LOOP signals (at macroexpansion time) an error of type
1514 PROGRAM-ERROR when duplicate variable names are found;
1515 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1516 ** FUNCALL of special-operators now cause an error of type
1518 ** PSETQ now works as required in the presence of side-effecting
1519 symbol-macro places;
1520 ** NCONC accepts any object as its last argument;
1521 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1523 ** loop-for-as-package does not require a package to be explicitely
1525 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1527 * fixed bug 166: compiler preserves "there is a way to go"
1528 invariant when deleting code.
1529 * fixed bug 172: macro lambda lists with required arguments after
1530 &REST arguments now cause an error to be signalled. (thanks to
1532 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1533 bug. (thanks to Matthew Danish)
1534 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1536 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1537 arguments are lexically rebound. (thanks to Gerd Moellmann and
1539 * fixed bug 194: error messages are now more informative when there
1540 is no primary method applicable in a call to a generic
1541 function. (thanks to Gerd Moellmann)
1542 * fixed bug in command line argument checking (thanks to Julian
1544 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1545 (thanks to Matthew Danish)
1546 * incremented fasl file version number, because of the
1547 SXHASH-related changes in the layout of CLOS data structures
1549 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1550 * minor incompatible change: The runtime (the Unix executable named
1551 "sbcl") is now much pickier about the .core files it will load.
1552 Essentially it now requires .core files to descend from the same
1553 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1554 as the runtime does. (The intent is to prevent the crashes which
1555 can occur, and which can even be reported as mysterious failures,
1556 when people patch the sources or change the build parameters
1557 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1558 sbcl and .core files.)
1559 * fixed bug: VALUES-LIST is no longer optimized away.
1560 * fixed bug 142: The FFI conversion of C string values to Lisp
1561 string values no longer conses excessively. (thanks to Nathan
1562 Froyd porting Raymond Toy's fix to CMU CL)
1563 * began to systematize and improve MOP conformance in PCL (thanks to
1564 Nathan Froyd, Gerd Moellman and Pierre Mai):
1565 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1567 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1568 ** FINALIZE-INHERITANCE is now called on class finalization;
1569 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1570 argument precedence order.
1571 * fixed bug 202: The compiler no longer fails on functions whose
1572 derived types contradict their declared type.
1573 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1574 so it can be non-toplevel.
1575 * The fasl file version number has changed (because of the new
1576 implementation of DEFMACRO).
1577 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1578 safe code, that any length requirement by their type-specifier
1579 argument is valid. The exceptions to this are described in bug
1581 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1582 functions now check the types of their inputs as required by ANSI.
1583 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1584 introduced symbol is DECLAREd to be SPECIAL.
1585 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1586 * fixed bug: :COUNT argument to sequence functions may be negative
1587 * fixed bug: body of DO-SYMBOLS may contain declarations
1588 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1589 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1590 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1591 symbol macro only once
1592 * fixed printing of call frame when argument list is unavailable
1593 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1594 * compiler no longer signals WARNING on unknown keyword
1597 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1598 * A beta-quality port to the mips architecture running Linux,
1599 based on the old CMUCL backend, has been made. It has been tested
1600 on a big-endian kernel, and works sufficiently well to be able to
1601 rebuild itself; it has not been tested in little-endian mode.
1602 * fixed an inconsistency between gencgc.c and purify.c which made
1603 dumping/loading .core files unreliable
1604 * fixed bug 120a: The compiler now deals correctly with IFs where
1605 the consequent is the same as the alternative, instead of
1606 misderiving the return type. (thanks to Alexey Dejneka)
1607 * fixed bug 113: Logical pathnames are now dumpable (the logical
1608 host is resolved at load-time, throwing an error if it is not
1610 * fixed bug 174: FORMAT's error message is slightly clearer when a
1611 non-printing character is used in a format directive.
1612 * fixed several bugs in compiler checking of type declarations, i.e.
1613 violations of the Python "declarations are assertions" principle
1614 (thanks to Alexey Dejneka)
1615 * fixed several bugs in PCL's error checking (thanks to Gerd
1617 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1619 * fixed bug in compilation of functions as first class values
1620 (thanks to Antonio Martinez-Shotton)
1621 * The compiler's handling TYPE-ERRORs which it can prove will
1622 inevitably happen at runtime has been cleaned up and corrected
1623 in several ways. (thanks to Alexey Dejneka)
1624 * improved argument type checking for various basic arithmetic
1625 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1626 lavished on them in the past that they can be compiled in many
1627 ways in different special cases
1628 * fixed bug 181: compiler checks validity of user supplied type
1630 * cleaned up code flushing in optimization: Function calls which
1631 should signal errors for safety purposes (e.g. which ANSI says
1632 should signal errors when their arguments are of incorrect type)
1633 are no longer optimized away.
1634 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1635 * incremented fasl file version number, because changes in the
1636 implementation of sequence functions like COERCE caused
1637 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1638 in old inline expansions) to become undefined. (Actually these
1639 changes were later undone, so we might very well be binary
1640 compatible with 0.7.7 after all, but leaving the version number
1641 incremented seemed like the simplest and most conservative
1644 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1645 * An alpha-quality port to the parisc architecture running Linux,
1646 based on the old CMUCL backend, has been made. This, even more so
1647 than the other backends, should be considered still a work in
1648 progress; known problems include that the Linux kernel in 64-bit
1649 mode does not propagate the correct sigcontext structure to
1650 userspace, and consequently SBCL on a parisc64 kernel will not
1652 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1653 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1654 suppressed.) Also now that the compiler is looking at declarations
1655 in the environment, it checks optimization declarations as well,
1656 and suppresses inlining when (> DEBUG SPEED).
1657 * More fixes have been made to treatment of floating point exception
1658 treatment and other Unix signals. In particular, floating point
1659 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1660 * The detection and handling of control stack exhaustion (infinite
1661 or very deeply nested recursion) has changed. Stack exhaustion
1662 detection is now done by write-protecting pages at the OS level
1663 and applies at all optimization settings; when found, a
1664 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1665 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1666 be caught using IGNORE-ERRORS.
1667 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1668 correctly when transforming recursive LABELS functions to LETs.
1669 (thanks to Alexey Dejneka)
1670 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1671 that are names of constants or global variables.
1672 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1673 alien routines with docstrings.
1674 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1675 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1677 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1678 (thanks to Eric Marsden)
1679 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1680 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1681 * Bug fix: The SPARC backend can now compile functions involving
1682 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1683 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1684 to a functions expecting a PATHNAME-DESIGNATOR.
1685 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1686 lambda lists. (thanks to Alexey Dejneka)
1687 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1688 functions so that EXPORTing the name of the class doesn't cause
1689 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1690 to Antonio Martinez for reporting this)
1691 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1692 constant-folded without failing an assertion. (thanks to Einar
1693 Floystad Dorum for reporting this)
1694 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1695 types are now dealt with more correctly by the compiler.
1696 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1697 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1698 behaviour analogous to RENAME-FILE. This puts its behaviour more
1699 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1701 * The fasl file version number has changed again. (because of the
1702 bug fix involving the names of PCL MAKE-INSTANCE functions)
1704 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1705 * bug fix: Floating point exceptions are treated much more
1706 consistently on the x86/Linux and PPC/Linux platforms.
1707 * Array initialization with :INITIAL-ELEMENT is now much faster for
1708 cases when the compiler cannot open code the array creation, but
1709 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1710 array accesses have also seen a speed increase.
1711 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1712 specified. (This was at the root of some bad interactions between
1713 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1715 * bug fix: Internal error arguments for undefined functions are now
1716 computed correctly on the PPC/Linux platform.
1717 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1718 Raymond Toy's patch for CMU CL)
1719 * Support for the Solaris 9 operating environment has been included
1720 (thanks to Daniel Merritt)
1721 * A very ugly but hopefully complete draft of the missing FFI chapter
1722 of the manual has been created by reformatting the corresponding
1723 CMU CL manual chapter into (currently very ugly and incoherent)
1724 DocBook and bringing it up to date for SBCL behavior. Thus, the
1725 manual is now essentially complete, at least by my extreme
1726 once-and-only-once standards, whereby it's acceptable to refer to
1727 the doc strings of SB-EXT functions as the primary documentation.
1728 * The fasl file version number has changed again, due to cleanup of
1729 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1730 is no longer a static symbol.)
1732 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1733 * SBCL now builds with OpenMCL (version 0.12) as the
1734 cross-compilation host; also, more progress has been made toward
1735 bootstrapping under CLISP.
1736 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1738 * bug 158 fixed: The compiler can now deal with integer loop
1739 increments different from 1; fixing this turned out also to fix
1741 * bug 169 fixed: no more bogus warnings about using lexical bindings
1742 despite the presence of perfectly good SPECIAL declarations (thanks
1743 to David Lichteblau)
1744 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1745 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1746 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1748 * bug 180 fixed: Method combination specifications no longer ignore
1749 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1750 * bug fix: Structure type predicate functions now check their argument
1751 count as they should.
1752 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1753 correctly. (thanks to Pierre Mai)
1754 * minor incompatible change: The --noprogrammer option is deprecated
1755 in favor of the new --disable-debugger option, which is very similar.
1756 (The major difference is that it takes effect at a slightly different
1757 time at startup, causing handling of errors in --sysinit and
1758 --userinit files will be affected differently.) The
1759 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1760 been added to allow this functionality to be controlled from ordinary
1761 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1762 maintainers, who might want to run non-interactive scripts to
1763 build SBCL cores which will later be used interactively.)
1764 * minor incompatible change: The LOAD function no longer, when given
1765 a wild pathname to load, loads all files matching that pathname.
1766 Instead, an error of type FILE-ERROR is signalled.
1768 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1769 * bug 147 fixed: The compiler preserves its block link/count
1770 invariants more correctly now so that it doesn't crash. (thanks
1772 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1774 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1775 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1776 no-value-for-this-character as NIL (as specified by ANSI).
1777 * HOST-NAMESTRING on physical pathnames now returns a string that is
1778 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1779 (thanks to Christophe Rhodes)
1780 * The Alpha port handles icache flushing more correctly. (thanks to
1782 * More progress has been made toward bootstrapping under CLISP. (thanks
1783 to Christophe Rhodes)
1784 * The fasl file format has changed again, because dynamic loading
1785 on OpenBSD (which has non-ELF object files) motivated some cleanups
1786 in the way that foreign symbols are transformed and passed around.
1787 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1788 is no longer treated as whitespace by the reader, but instead as
1789 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1790 |A
\7fB|, instead of A as it used to.
1792 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1793 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1794 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1795 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1796 this as of 0.7.1.45, but was left out of the previous news section
1797 (thanks to Dan Barlow)
1798 * SBCL now runs on the Solaris operating system on SPARC architectures
1799 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1800 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1801 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1802 * SPARC backend cleanups, allowing builds of cores optimized for V8
1803 and V9 SPARCS, and also emission of code targeted to a particular
1804 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1806 * SBCL is closer to bootstrapping under CLISP, thanks to various
1807 fixes by Christophe Rhodes.
1808 * The fasl file format has changed again, to allow the compiler's
1809 INFO database to support symbol macros.
1810 * The user manual (in doc/) is formatted into HTML more nicely.
1811 (thanks to coreythomas)
1812 * The system is smarter about SUBTYPEP relationships, especially
1813 those involving NOT types (including types such as ATOM which are
1814 represented internally using NOT types). Thus SUBTYPEP is less
1815 likely to return (VALUES NIL NIL) in general, and in particular
1816 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1818 * The fasl file format has changed again, because the internal
1819 representation of types now includes a new slot to support the new
1820 SUBTYPEP-of-NOT-types logic.
1821 * (not a change in the main branch of SBCL, but a related prototype
1822 which can hopefully be merged into the main branch of SBCL in the
1823 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1824 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1825 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1826 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1827 dispatch macro character combinations. (thanks to Alexey Dejneka)
1828 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1829 that superficially look like logical namestrings correctly.
1830 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1832 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1833 default supplied-p) for &optional and &key arguments. (thanks to
1836 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1837 * incompatible change: The compiler is now less aggressive about
1838 tail call optimization, doing it only when (> SPACE DEBUG) or
1839 (> SPEED DEBUG). (This is an incompatible change because there are
1840 programs which relied on the old CMU-CL-style behavior to optimize
1841 away their unbounded recursion which will now die of stack overflow.)
1842 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1843 for non-GENCGC systems has been increased to 20M (since that
1844 seems much closer to the likely performance optimum for modern
1845 systems than the old 4M value was)
1846 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1847 and *DEBUG-PRINT-LEVEL*
1848 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1849 of CMU CL's support for SPARC, and various endianness and other
1850 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1851 * new syntactic sugar for the Unix command line: --load foo.bar is now
1852 an alternate notation for --eval '(load "foo.bar")'.
1854 ** The system now detects stack overflow and handles it gracefully,
1855 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1856 optimization settings. (This is a good thing in general, and
1857 its introduction in this version should be particularly timely
1858 for anyone whose code fails because of suppression of tail
1860 ** The system now hunts for the C variable "environ" in a more
1861 devious way, to avoid segfaults when the C library version differs
1862 between compile time and run time. (thanks to Christophe Rhodes)
1863 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1864 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1865 ** The compiler no longer issues bogus style warnings for undefined
1866 classes in the same source file as the DEFCLASSes which defined
1867 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1869 ** fixes in CONDITION class precedence list for undefined function
1870 errors (thanks to Alexei Dejneka)
1871 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1872 correctly. (thanks to Dan Barlow)
1873 ** portability fixes aiming at bootstrapping under CLISP (thanks
1874 to Dave McDonald and Christophe Rhodes)
1875 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1876 ** fixes in type translation and and type inference (thanks to
1878 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1879 ** various fixes in the new ports (thanks to Dan Barlow)
1880 * several changes related to debugging:
1881 ** suppression of tail recursion, as noted above
1882 ** stack overflow detection, as noted above
1883 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1884 is now the default. (For some time encapsulation has been more
1885 reliable than the breakpoint-based :ENCAPSULATE NIL
1886 implementation, at least on X86 systems; and I just noticed that
1887 encapsulation also seems closer to the spirit of the ANSI
1890 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1892 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1893 up properly again. (There was a packaging bug in 0.7.0 which
1894 left their definitions in SB-SYS::LOAD-FOREIGN and
1895 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1896 vital for most things which interface to C-level interfaces,
1897 like extensions working with sockets or databases or
1898 Perl-compatible regexes or whatever, and the need to fix
1899 this bug is the main reason that 0.7.1 was released so
1901 ** DEFGENERIC is now choosier about the methods it redefines, so that
1902 reLOADing a previously-LOADed file containing DEFGENERICs does
1903 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1904 cycle now works as it should. (thanks to Alexey Dejneka)
1905 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1906 fixed by Christophe Rhodes. (He actually submitted this patch
1907 months ago, and I delayed until after 0.7.0.)
1908 ** Bug 111 (internal compiler confusion about runtime checks on
1909 FUNCTION types) was fixed by Alexey Dejneka.
1910 * Some internal cleanups (getting rid of variables which aren't
1911 needed now that the byte interpreter is gone) caused the fasl
1912 file format number to change again.
1914 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1915 * major incompatible change: The default fasl file extension, i.e. the
1916 default extension for files produced by COMPILE-FILE, has changed
1917 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1919 ** There are many changes in the implementation of the compiler.
1920 SBCL is now essentially a compiler-only implementation of ANSI
1921 Common Lisp. EVAL still "interprets" a few special cases, but
1922 almost all the interesting cases are handled by creating
1923 a LAMBDA expression, calling COMPILE on it, then calling
1924 FUNCALL on the result.
1925 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1926 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1927 Since the code is newer, there might still be some new bugs
1928 (though not as many as before Martin Atzmueller's fixes:-). But
1929 the new code is substantially simpler and clearer, and hopefully
1930 any remaining bugs will be simpler, less fundamental, and more
1931 fixable then the bugs in the old code.
1932 ** The revised compiler is still a little unsteady on its feet.
1934 *** The debugging information it produces (particularly the names
1935 of FUNCTION objects) is sometimes much less useful than what
1936 the old compiler produced.
1937 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1938 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1939 within a MACROLET) has been temporarily weakened.
1940 ** There are new compiler optimizations for various functions:
1941 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1942 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1943 *** the math functions TRUNCATE, FLOOR, and CEILING
1944 *** the function-of-all-trades COERCE
1945 Mostly these should be transparent, but there's one
1946 potentially-annoying problem (bug 117): when the compiler
1947 inline-expands a function and does type analysis on the result,
1948 it can create control paths which have type mismatches, and
1949 when it can't prove that those control paths aren't taken,
1950 it will issue WARNINGs about the type mismatches. This is
1951 a particular problem in practice for the new sequence functions.
1952 It's not clear how this should be fixed, and for now, a
1953 workaround is given in the entry for 117 in the BUGS file.
1954 ** (Because of the interaction between the two previous items --
1955 occasional inlining problems and new inline expansions -- some
1956 of the new sequence function optimizations won't really kick in
1957 completely until debugging information, and then inlining, are
1958 straightened out in some future version.)
1959 * minor incompatible changes:
1960 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1961 DIRECTORY no longer implicitly promotes NIL slots of its
1962 pathname argument to :WILD. In particular, when you ask for the
1963 contents of a directory (which you used to be able to do without
1964 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1965 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1966 ** changes in behavior that ANSI explicitly defines to be
1967 implementation dependent:
1968 *** The new compiler-only implementation still conforms with ANSI,
1969 but acts a little different than before. Besides the obvious
1970 changes in performance tradeoffs (that the cost per form passed
1971 to EVAL has gone up, and the cost per form executed by EVAL
1972 has gone down), the behavior of the system changes a little
1973 because there are no longer any interpreted function objects.
1974 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1975 e.g. doing COMPILE on the output of interactive DEFUN is
1977 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1979 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1980 has changed from (:CL) to NIL.
1981 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1982 e.g. #\Newline and #\Tab, don't have names specified in the
1983 ANSI Common Lisp standard, is now based on their ASCII symbolic
1984 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1985 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1986 are no longer used for output.
1987 ** changes in internal implementation constants:
1988 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1989 4 million. (If your application spends a lot of time GCing and
1990 you have a lot of RAM, you might want to experiment with
1991 increasing it even more.)
1992 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1993 However, almost all old code should still continue to work without
1994 immediate update, as SB-C-CALL is now a (deprecated) nickname
1996 ** Old operator names in the style DEF-FOO are now deprecated in
1997 favor of new corresponding names DEFINE-FOO, for consistency with
1998 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1999 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2000 internal symbols, but a few supported extensions like
2001 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2002 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2003 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2004 etc. as you get deeper into recursive calls to the debugger
2005 command loop, instead of the old "5]", "5]]", "5]]]"
2006 sequence. (I was motivated to do this when squabbles between
2007 ILISP and SBCL left me very deeply nested in the debugger. In the
2008 short term, this change will probably provoke more ILISP/SBCL
2009 squabbles, but hopefully it will be an improvement in the long run.)
2010 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2011 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2012 function, since declaiming return types involving aliens is
2013 (1) annoyingly messy to do by hand and (2) vital to efficient
2014 compilation of code which calls such functions.
2015 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2016 longer reexported by the SB-EXT package. They're solely useful
2017 for alien code, so it seems more logical that you should get
2018 them from the SB-ALIEN package, not in SB-EXT.
2019 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2020 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2021 features. Instead, the code that they used to control is always
2022 built into the system.
2023 * many other bug fixes
2024 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2025 advantage of the new EVAL-WHEN stuff and to clean them up in
2026 general, and they are now more ANSI-compliant in a number of
2027 ways. Martin Atzmueller is responsible for a lot of this.
2028 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2030 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2031 *** correct ERROR type for various file operations
2032 *** some fixes for Lisp streams
2033 *** DEFMETHOD syntax checking
2034 *** changing old weird representation of debug information as
2035 strings (which, among their other deficiencies, don't transform
2036 correctly when you rename packages, and don't change their
2037 print representation when you change things like *PACKAGE*
2038 and *PRINT-LENGTH*) to symbols and lists of symbols
2039 He also made several improvements and fixed several bugs in DESCRIBE.
2040 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2042 *** misbehavior of WRITE-STRING/WRITE-LINE
2043 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2044 and several other LOOP problems as well
2045 *** DIRECTORY when similar filenames are present
2046 *** DEFGENERIC with :METHOD options
2047 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2048 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2049 *** argument ordering in FIND with :TEST option
2050 *** mishandled package designator argument in APROPOS-LIST
2051 *** various problems in the backquote readmacro
2052 *** a bug in APROPOS
2053 *** probably some others that I'm not describing very well here,
2054 since the CVS log documents them by reference to sbcl-devel
2055 messages, and the SourceForge archives aren't working well.:-(
2056 ** Dan Barlow improved the Alpha port (and is making progress on the
2057 PPC port, for those of you who think different).
2058 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2059 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2060 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2061 continues to work on the SPARC port (for those of you in a position
2062 to look down upon our little PC-compatible boxes from a great height).
2063 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2064 and uses that copy, rather than the current dynamic value, when
2065 it's trying to decide whether to truncate output. Thus e.g.
2066 (let ((*print-lines* 50))
2067 (pprint-logical-block (stream nil)
2069 (let ((*print-lines* 8))
2070 (print (aref possiblybigthings i) stream)))))
2071 should now truncate the logical block only at 50 lines, instead of
2072 often truncating it at 8 lines, as it did before.
2073 * The doc/cmucl/ directory, containing old CMU CL documentation
2074 from the time of the fork, is no longer part of the base system.
2075 SourceForge has shut down its anonymous FTP service, and with it
2076 my original plan for distributing the old CMU CL documentation
2077 there. For now, if you need these files you can download an old
2078 SBCL source release and extract them from it.
2079 * The fasl file version number changed again, for dozens of reasons,
2080 some of which are apparent above.
2082 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2083 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2084 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2085 patch, so that DISASSEMBLE output is much nicer.
2086 * The code in the SB-PROFILE package now seems reasonably stable.
2087 I still haven't decided what the final interface should look like
2088 (I'd like PROFILE to interact cleanly with TRACE, since both
2089 facilities use function encapsulation) but if you have a need
2090 for profiling now, you can probably use it successfully with
2091 the current CMU-CL-style interface.
2092 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2093 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2094 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2095 link now returns the dangling link itself, and for similar
2096 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2097 link itself. (In these cases the old code signalled an error and
2098 looped endlessly, respectively.) Thus, DIRECTORY now works even
2099 in the presence of dangling and cyclic symbolic links.
2100 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2101 is now a supported extension again, since the consensus on
2102 sbcl-devel was that it can be useful for ordinary development
2103 work, not just for debugging SBCL itself.
2104 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2105 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2106 that currently-defined functions might be redefined later with
2107 different return types.
2108 * Hash tables can be printed readably, as inspired by CMU CL code
2109 of Eric Marsden and SBCL code of Martin Atzmueller.
2110 * better error handling in CLOS method combination, thanks to
2111 Martin Atzmueller porting Pierre Mai's CMU CL patches
2112 * more overflow fixes for >16Mbyte I/O buffers
2113 * A bug in READ has been fixed, so that now a single Ctrl-D
2114 character suffices to cause end-of-file on character streams.
2115 In particular, now you only need one Ctrl-D at the command
2116 line (not two) to exit SBCL.
2117 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2119 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2120 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2121 does the right thing.
2122 * The implementation of some type tests, especially for CONDITION
2123 types, is now tidier and maybe faster, due to CMU CL code
2124 originally by Douglas Crosher, ported by Martin Atzmueller.
2125 * Some math functions have been fixed, and there are new
2126 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2127 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2128 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2129 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2130 likely to be built into sbcl-0.7.0.)
2131 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2132 CL, ported to SBCL by Martin Atzmueller)
2133 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2134 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2135 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2136 * There's a new slam.sh hack to shorten the edit/compile/debug
2137 cycle for low-level changes to SBCL itself, and a new
2138 :SB-AFTER-XC-CORE target feature to control the generation of
2139 the after-xc.core file needed by slam.sh.
2140 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2141 COMPILE-FILE is no longer supported, so that now every function
2142 gets an entry point, so that block compilation looks a little
2143 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2144 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2145 returns the number of bytes consed since the system started,
2146 rather than the number consed since the first time the function
2147 was called. (The new definition parallels ANSI functions like
2148 CL:GET-INTERNAL-RUN-TIME.)
2149 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2150 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2151 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2152 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2153 * Fasl file version numbers are now independent of the target CPU,
2154 since historically most system changes which required version
2155 number changes have affected all CPUs equally. Similarly,
2156 the byte fasl file version is now equal to the ordinary
2159 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2160 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2161 is no longer recognized. I apologize for this, because it was
2162 listed in SB-EXT as a supported extension, but I found that
2163 its existing behavior was poorly specified, as well as incorrectly
2164 specified, and it looked like too much of a mess to straighten it
2165 out. I have enough on my hands trying to get ANSI stuff to work..
2166 * many patches ported from CMU CL by Martin Atzmueller, with
2167 half a dozen bug fixes in pretty-printing and the debugger, and
2168 half a dozen others elsewhere
2169 * fixed bug 13: Floating point infinities are now supported again.
2170 They might still be a little bit flaky, but thanks to bug reports
2171 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2172 as flaky as they were.
2173 * The --noprogrammer command line option is now supported. (Its
2174 behavior is slightly different in detail from what the old man
2175 page claimed it would do, but it's still appropriate under the
2176 same circumstances that the man page talks about.)
2177 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2178 are now supported, and enabled by default. Thus, the compiler can
2179 handle many floating point and complex operations much less
2180 inefficiently. (Thus e.g. you can implement a complex FFT
2182 * The compiler now detects type mismatches between DECLAIM FTYPE
2183 and DEFUN better, and implements CHECK-TYPE more correctly, and
2184 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2185 thanks to patches from Martin Atzmueller.
2186 * various fixes to make the cross-compiler more portable to
2187 ANSI-conforming-but-different cross-compilation hosts (notably
2188 Lispworks for Windows, following bug reports from Arthur Lemmens)
2189 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2190 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2191 fixed thanks to Pierre Mai's CMU CL patches.
2192 * fixes to keep the system from overflowing internal counters when
2193 it tries to use i/o buffers larger than 16M bytes
2194 * fixed bug 45a: Various internal functions required to support
2195 complex special functions have been merged from CMU CL sources.
2196 (When I was first setting up SBCL, I misunderstood a compile-time
2197 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2198 * improved support for type intersection and union, fixing bug 12
2199 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2200 more obscure bugs as well
2201 * some steps toward byte-compiling non-performance-critical
2202 parts of the system, courtesy of patches from Martin Atzmueller
2203 * Christophe Rhodes has made some debian packages of sbcl at
2204 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2205 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2206 stable, but are nonetheless usable. When he's ready, I'd be happy
2207 to add them to the SourceForge "File Releases" section. (And if
2208 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2209 * new fasl file format version number (because of changes in
2210 internal representation of (OR ..) types to accommodate the new
2211 support for (AND ..) types, among other things)
2213 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2214 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2216 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2218 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2219 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2220 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2221 * DESCRIBE now works on structure objects again.
2222 * Most function call argument type mismatches are now handled as
2223 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2224 know whether the function will be redefined before the call is
2225 executed. (The compiler could flag local calls with full WARNINGs,
2226 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2227 it doesn't keep track of enough information to know whether calls
2228 are local in this sense.)
2229 * Compiler output is now more verbose, with messages truncated
2230 later than before. (There should be some supported way for users
2231 to override the default verbosity, but I haven't decided how to
2232 provide it yet, so this behavior is still controlled by the internal
2233 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2234 src/compiler/ir1util.lisp.)
2235 * Fasl file format version numbers have increased again, because
2236 support for the Gray streams extension changes the layout of the
2237 system's STREAM objects.
2238 * The Gray subclassable streams extension now works, thanks to a
2239 patch from Martin Atzmueller.
2240 * The full LOAD-FOREIGN extension (not just the primitive
2241 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2242 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2243 but like most other programs, it defaults to copying the Unix
2244 environment from the original process instead of starting the
2245 new process in an empty environment.
2246 * Extensions which manipulate the Unix environment now support
2247 an :ENVIRONMENT keyword option which doesn't smash case or
2248 do other bad things. The CMU-CL-style :ENV option is retained
2249 for porting convenience.
2250 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2251 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2253 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2255 * A patch from Martin Atzmueller seems to have solved the SIGINT
2256 problem, and as far as we know, signal-handling now works cleanly.
2257 (If you find any new bugs, please report them!)
2258 * The system no longer defaults Lisp source file names to types
2259 ".l", ".cl", or ".lsp", but only to ".lisp".
2260 * The compiler no longer uses special default file extensions for
2261 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2262 seems to expect a single default extension for all compiled code,
2263 and there's no compelling reason to try to stretch the standard
2264 to allow two different extensions.) Instead, byte-compiled files
2265 default to the same extension as native-compiled files.
2266 * Fasl file format version numbers have increased again, because
2267 a rearrangement of internal implementation packages made some
2268 dumped symbols in old fasl files unreadable in new cores.
2269 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2270 many fewer weird special cases.
2271 * Bug #17 (differing COMPILE-FILE behavior between logical and
2272 physical pathnames) has been fixed, and some related misbehavior too,
2273 thanks to a patch from Martin Atzmueller.
2274 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2275 by Tim Moore, ported to SBCL by Martin Atzmueller.
2276 * Martin Atzmueller fixed several filesystem-related problems,
2277 including bug #36, in part by porting CMU CL patches, which were
2278 written in part by Paul Werkowski.
2279 * More compiler warnings in src/runtime/ are gone, thanks to
2280 more patches from Martin Atzmueller.
2281 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2284 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2286 * DESCRIBE now works on CONDITION objects.
2287 * The debugger now handles errors which arise when trying to print
2288 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2290 * The build system now uses an additional file, customize-target-features.lisp,
2291 to allow local modifications to the target *FEATURES* list. (The point of
2292 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2293 debugging features enabled, without having to worry about propagating it
2294 into everyone's system when I do a "cvs update".) When no
2295 customize-target-features.lisp file exists, the target *FEATURES* list
2296 should be constructed the same way as before.
2297 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2298 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2299 warn about behavior which is undefined under the ANSI spec. Note: This
2300 is specified by ANSI, but it's not very popular with programmers.
2301 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2302 macro in the SBCL sources for an example of a workaround which you
2303 might use to make portable ANSI-standard code which does what you want.
2304 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2305 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2306 when it's used as a non-toplevel form. (This is required in order
2307 to implement the DEFCONSTANT-EQX macro.)
2308 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2309 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2310 funny things, due to bugs in the implementation of EVAL-WHEN
2311 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2312 non-toplevel DEFCONSTANTs are uncommon.)
2313 * The core file version number and fasl file version number have been
2314 incremented, because the old noncompliant DEFCONSTANT behavior involved
2315 calling functions which no longer exist, and because I also took the
2316 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2317 * fixed bug 1 (error handling before read-eval-print loop starts), and
2318 redid debugger restarts and related debugger commands somewhat while
2320 ** The QUIT debugger command is gone, since it did something
2321 rather different than the SB-EXT:QUIT command, and since it never
2322 worked properly outside the main toplevel read/eval/print loop.
2323 Invoking the new TOPLEVEL restart provides the same functionality.
2324 ** The GO debugger command is also gone, since you can just invoke
2325 the CONTINUE restart directly instead.
2326 ** The TOP debugger command is also gone, since it's redundant with the
2327 FRAME 0 command, and since it interfered with abbreviations for the
2329 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2331 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2332 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2333 some porting work by Martin Atzmueller)
2334 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2335 with a boatload of patches which he ported to SBCL. Now that those
2337 ** The system tries to make sure that its low-priority messages
2338 are prefixed by semicolons, to help people who like to use
2339 syntax highlighting in their ILISP buffer. (This patch
2340 was originally due to Raymond Toy.)
2341 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2342 patches originally written by Raymond Toy.
2343 ** The compiler understands coercion between single-value and
2344 multiple-VALUES type expressions better, getting rid of some very
2345 weird behavior, thanks to patches originally by Robert MacLachlan
2346 and Douglas Crosher.
2347 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2348 lambda lists, thanks to a patch originally by Pierre Mai.
2349 ** The system no longer bogusly warns about "abbreviated type
2351 ** The compiler gets less confused by inlining and RETURN-FROM,
2352 thanks to some patches originally by Tim Moore.
2353 ** The system no longer hangs when dumping circular lists to fasl
2354 files, thanks to a patch originally from Douglas Crosher.
2355 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2356 undefined function error.
2357 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2358 in structure slot initforms) for now, documented workaround instead:-|
2359 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2360 * fixed bug 5: added stubs for various Gray stream functions called
2361 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2362 installed, at least appropriate type errors are generated
2363 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2364 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2365 consistently in DEFMETHOD forms.
2366 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2367 it doesn't seem to affect SBCL after all
2368 * The C runtime system now builds with better optimization and many
2369 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2371 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2373 * The system is now under CVS at SourceForge (instead of the
2374 CVS repository on my home machine).
2375 * The new signal handling code has been tweaked to treat register
2376 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2377 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2378 downstream, e.g. in debug-int.lisp, has implicit dependencies
2379 on the unsignedness of integer representation of machine words,
2380 and that caused the system to bomb out with infinite regress
2381 when trying to recover from type errors involving signed values,
2382 e.g. (BUTLAST '(1 2 3) -1).)
2383 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2384 (This was one of the bugs Peter Van Eynde reported back in July.)
2385 * The system now uses code inspired by Colin Walters' O(N)
2386 implementation of MAP (from the cmucl-imp@cons.org mailing
2387 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2388 inline the MAP operation, and there is more than one
2389 sequence argument to the MAP call (so that it can't just
2390 do ETYPECASE once and for all based on the type of the
2391 single sequence argument). (The old non-inline implementation
2392 of the general M-argument sequence-of-length-N case required
2393 O(M*N*N) time when any of the sequence arguments were LISTs.)
2394 * The QUIT :UNIX-CODE keyword argument has been renamed to
2395 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2396 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2398 * Raymond Wiker's patches to port dynamic loading from Linux to
2399 FreeBSD have been added.
2400 * The BUGS file is now more nearly up to date, thanks in large part
2401 to Martin Atzmueller's review of it.
2402 * The debugger now flushes standard output streams before it begins
2403 its output ("debugger invoked" and so forth).
2404 * The core version number and fasl file version number have both
2405 been incremented, because of incompatible changes in the layout
2407 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2408 used to not be called for a saved Lisp image.)
2409 * Martin Atzmueller's version of a patch to fix a compiler crash,
2410 as posted on sbcl-devel 13 September 2000, has been installed.
2411 * Instead of installing Martin Atzmueller's patch for the
2412 compiler transform for SUBSEQ, I deleted the compiler transform,
2413 and transforms for some similar consing operations.
2414 * A bug in signal handling which kept TRACE from working on OpenBSD
2416 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2417 away by constant folding
2418 * The system now defines its address space constants in one place
2419 (in the Lisp sources), and propagates them automatically elsewhere
2420 (through GENESIS and the sbcl.h file). Therefore, patching the
2421 address map is less unnecessarily tedious and error-prone. The
2422 Lisp names of address space constants have also been systematized.
2423 * CVS tags like dollar-Header-dollar have been removed from
2424 the sources, because they have never saved me trouble and
2425 they've been source of trouble working with patches and other
2426 diff-related operations.
2427 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2428 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2430 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2432 * The system has been ported to OpenBSD.
2433 * The system now compiles with a simple "sh make.sh" on the systems
2434 that it's supported on. I.e., now you no longer need to tweak
2435 text in the target-features.lisp-expr and symlinks in src/runtime/
2436 by hand, the make.sh takes care of it for you.
2437 * The system is no longer so grossly inefficient when compiling code
2438 involving vectors implemented as general (not simple) vectors (VECTOR T),
2439 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2440 which dares to use the various sequence functions on non-simple
2441 vectors, takes less of a performance hit.
2442 * There is now a primitive type predicate VECTOR-T-P
2443 to test for the (VECTOR T) type, so that e.g.
2444 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2445 can now be compiled with some semblance of efficiency. (The old code
2446 turned the type declaration into a full call to %TYPEP at runtime!)
2447 * AREF on (VECTOR T) is still not fast, since it's still compiled
2448 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2449 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2450 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2451 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2452 without making full calls to SUBTYPEP at runtime.
2453 (Some analogous efficiency issues for non-simple vectors specialized to
2454 element types other than T, or for non-simple multidimensional arrays,
2455 have not been addressed. They could almost certainly be handled the
2456 same way if anyone is motivated to do so.)
2457 * The changes in array handling break binary compatibility, so
2458 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2459 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2460 instead of (VALUES T T).
2461 * By following the instructions that Dan Barlow posted to sbcl-devel
2462 on 2 July 2000, I was able to enable primitive dynamic object
2463 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2464 functionality is not implemented (since it calls ld to resolve
2465 library references automatically, requiring RUN-PROGRAM for its
2466 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2467 try to resolve library references) is now supported.
2468 * The system now flushes the standard output streams when it terminates,
2469 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2470 them at several other probably-convenient times, e.g. in each pass of
2471 the toplevel read-eval-print loop, and after evaluating a form given
2472 as an "--eval" command-line option. (These changes were motivated by a
2473 discussion of stream flushing issues on cmucl-imp in August 2000.)
2474 * The source transform for TYPEP of array types no longer assumes
2475 that an array whose element type is a not-yet-defined type
2476 is implemented as an array of T, but instead punts, so that the
2477 type will be interpreted at runtime.
2478 * There is now some support for cross-compiling in make.sh: each of
2479 the phases of make.sh has its own script. (This should be transparent
2480 to people doing ordinary, non-cross-compile builds.)
2481 * Since my laptop doesn't have hundreds of megabytes of memory like
2482 my desktop machine, I became more motivated to do some items on
2483 my to-do list in order to reduce the size of the system a little:
2484 ** Arrange for various needed-only-at-cold-init things to be
2485 uninterned after cold init. To support this, those things have
2486 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2487 symbols with such names are now uninterned after cold init).
2488 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2489 fasl files for cold load.
2490 ** Remove the old compiler structure pooling code (which used to
2491 be conditional on the target feature :SB-ALLOC) completely.
2492 ** Redo the representation of some data in cold init to be more compact.
2493 (I also looked into supporting byte compiled code at bootstrap time,
2494 which would probably reduce the size of the system a lot, but that
2495 looked too complicated, so I punted for now.)
2496 * The maximum signal nesting depth in the src/runtime/ support code has
2497 been reduced from 4096 to 256. (I don't know any reason for the very
2498 large old value. If the new smaller value turns out to break something,
2499 I'll probably just bump it back up.)
2500 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2502 * Many, many bugs reported by Peter Van Eynde have been added to
2503 the BUGS list; some have even been fixed.
2504 * While enabling dynamic object file loading, I tried to make the
2505 code easier to understand, renaming various functions and variables
2506 with less ambiguous names, and changing some function calling
2507 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2508 * While trying to figure out how to do the OpenBSD port, I tried to
2509 clean up some of the code in src/runtime/. In particular, I dropped
2510 support for non-POSIX signal handling, added various comments,
2511 tweaked the code to reduce the number of compilation warnings, and
2512 renamed some files to increase consistency.
2513 * To support the new automatic configuration functionality in make.sh,
2514 the source file target-features.lisp-expr has been replaced with the
2515 source file base-target-features.lisp-expr and the machine-generated
2516 file local-target-features.lisp-expr.
2517 * fixed a stupid quoting error in make.sh so that using CMU CL
2518 "lisp -batch" as cross-compilation host works again
2520 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2522 * DESCRIBE no longer tries to call itself recursively to describe
2523 bound/fbound values, so that it no longer fails on symbols which are
2524 bound to themselves (like keywords, T, and NIL).
2525 * DESCRIBE now works on generic functions.
2526 * The printer now prints less-screwed-up representations of closures
2527 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2528 * A private symbol is used instead of the :EMPTY keyword previously
2529 used to mark empty slots in hash tables. Thus
2530 (DEFVAR *HT* (MAKE-HASH-TABLE))
2531 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2532 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2533 now does what ANSI says that it should. (You can still get
2534 similar noncompliant behavior if bang on the hash table
2535 implementation with all the symbols you get back from
2536 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2537 This breaks binary compatibility, since tests for equality to
2538 :EMPTY are wired into things like the macroexpansion of
2539 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2541 * There's now a minimal placeholder implementation for CL:STEP,
2542 as required by ANSI.
2543 * An obscure bug in the interaction of the normal compiler, the byte
2544 compiler, inlining, and structure predicates has been patched
2545 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2546 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2548 * Missing ordinary arguments in a macro call are now detected even
2549 when the macro lambda list contains &KEY or &REST.
2550 * The debugger no longer complains about encountering the top of the
2551 stack when you type "FRAME 0" to explicitly instruct it to go to
2552 the top of the stack. And it now prints the frame you request even
2553 if it's the current frame (instead of saying "You are here.").
2554 * As specified by ANSI, the system now always prints keywords
2555 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2557 * The default initial SIZE of HASH-TABLEs is now smaller.
2558 * Type information from CLOS class dispatch is now propagated
2559 into DEFMETHOD bodies, so that e.g.
2560 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2562 is now basically equivalent to
2563 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2564 (DECLARE (TYPE SINGLE-FLOAT X))
2566 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2567 operation, without having to do run-time type dispatch.
2568 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2569 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2570 * Since I don't seem to be making big file reorganizations very often
2571 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2572 is overflowing my ability to conveniently back them up), I've finally
2573 checked the system into CVS. (The CVS repository is on my home system,
2574 not at SourceForge -- putting it on SourceForge might come later.)
2575 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2576 high-level GC-NOTIFY-FOO functions send their output. (There's
2577 still very little control of where low-level verbose GC functions
2578 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2579 less than it used to -- the GC-NOTIFY-FOO functions are now under
2580 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2581 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2582 in only one place in the source code, and propagates it automatically
2583 everywhere that it's needed. Thus e.g. when I bump the version from
2584 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2585 * The C source files now include boilerplate legalese and documentation
2586 at the head of each file (just as the Lisp source files already did).
2587 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2588 to his page will be replaced with a link to his new CLiki service.
2590 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2592 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2593 * The build process now looks for GNU make under the default name "gmake",
2594 instead of "make" as it used to. If GNU make is not available as "gmake"
2595 on your system, you can change this default behavior by setting the
2596 GNUMAKE environment variable.
2597 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2598 can build without error under CMU CL.
2600 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2602 * There is now a partial SBCL user manual (with some new text and some
2603 text cribbed from the CMU CL manual).
2604 * The beginnings of a profiler have been added (starting with the
2605 CMU CL profiler and simplifying and cleaning up). Eventually the
2606 main interface should be through the TRACE macro, but for now,
2607 it's still accessed through vaguely CMU-CL-style functions and macros
2608 exported from the package SB-PROFILE.
2609 * Some problems left over from porting CMU CL to the new
2610 cross-compilation bootstrap process have been cleaned up:
2611 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2612 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2613 part when getting rid of the compiler *BACKEND* stuff.)
2614 ** The value of *NULL-TYPE* was screwed up, because it was
2615 being initialized before the type system knew the final
2616 definition of the 'NULL type. This screwed up several key
2617 optimizations in the compiler, causing inefficiency in all sorts
2618 of places. (I found it because I wanted to understand why
2619 GET-INTERNAL-RUN-TIME was consing.)
2620 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2621 PROCLAIM FTYPE information. Unfortunately this broke binary
2622 compatibility again, since now the forms output by DEFGENERIC
2623 to refer to functions which didn't exist in 0.6.3.
2624 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2625 can use the new (as of 0.6.3) transform for SOME into MAP into
2627 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2628 values to (INTEGER 0 1000000), so that the time code will no longer
2629 occasionally get blown up by Linux returning 1000000 microseconds
2630 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2631 its output conform to the ANSI spec. (Alas, this makes its output
2632 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2633 * A full call to MAP NIL with a single sequence argument no longer conses.
2634 * fixes to problems pointed out by Martin Atzmueller:
2635 * The manual page no longer talks about multiprocessing as though
2636 it were currently supported.
2637 * The ILISP support patches have been removed from the distribution,
2638 because as of version 5.10.1, ILISP now supports SBCL without us
2639 having to maintain patches.
2640 * added a modified version of Raymond Toy's recent CMU CL patch for
2641 EQUALP comparison of HASH-TABLE
2643 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2645 * The system still can't cross-compile itself with
2646 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2647 entails), but at least it can compile more of itself that way
2648 than it used to be able to, and various buglets which were uncovered
2649 by trying to cross-compile itself that way have now been fixed.
2650 * This release breaks binary compatibility again. This time
2651 at least I've incremented the FASL file format version to 2, so that the
2652 problem can be detected reliably instead of just causing weird errors.
2653 * various new style warnings:
2654 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2655 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2656 ** using the lexical binding of a variable named in the *FOO* style
2657 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2658 as specified by ANSI.
2659 * *RANDOM-STATE* is no longer automatically initialized from
2660 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2661 default behavior of the system is to repeat its behavior every time
2662 it's run. If you'd like to change this behavior, you can always
2663 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2664 old convention there was no comparably easy way to get the system to
2665 repeat its behavior every time it was run.
2666 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2667 FTYPE declarations has been removed, in favor of their ANSI
2668 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2669 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2670 the types of their sequence arguments can be determined at compile time.
2671 This is done through a new open code expansion for MAP which eliminates
2672 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2673 argument types can be determined at compile time.
2674 * The optimizer now transforms COERCE into an identity operation when it
2675 can prove that the coerced object is already of the correct type. (This
2676 can be a win for machine generated code, including the output of other
2677 optimization transforms, such as the MAP transform above.)
2678 * Credit information has been moved from source file headers into CREDITS.
2679 * Source file headers have been made more standard.
2680 * The CASE macro now compiles without complaining even when it has
2683 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2685 * (Note that the way that the PCL macroexpansions were rewritten
2686 to accommodate the change in DEFGENERIC below breaks binary
2687 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2688 not run under sbcl-0.6.2. Once we get out of alpha releases,
2689 i.e. hit release 1.0.0, we'll probably try to maintain binary
2690 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2691 and sbcl-1.4.4. Until then, however, it might be fairly common
2692 for maintenance releases to break binary compatibility.)
2693 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2695 * The reporting of unhandled signals has been changed to print some
2696 explanatory text as well as the report form. (Previously only
2697 the report form was printed.)
2698 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2699 it defines, so that the compiler no longer issues undefined function
2700 warnings for compiled-but-not-yet-loaded generic functions.
2701 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2702 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2703 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2704 section of the ANSI spec are supported.
2705 * The "" nickname for the "KEYWORD" package has been removed.
2706 The reader still handles symbol tokens which begin with a package marker
2707 as keywords, but it doesn't expose its mechanism for doing so in the
2708 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2709 * The system now issues STYLE-WARNINGs for contradictory TYPE
2710 proclamations. (Warnings for contradictory FTYPE proclamations would
2711 be nice too, but those can't be done usefully unless the type system
2712 is made smarter about FUNCTION types.)
2713 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2714 systematized, so that "*target-*.lisp is supposed to exist only on the
2715 target and imply that there's a related file which exists on the
2716 host, and *host-*.lisp is supposed to exist only on the host and imply
2717 that there's a related file which exists on the target. This involves a
2718 lot of renaming. Hopefully the acute confusion caused by the renaming
2719 will be justified by the reduction in chronic confusion..
2720 ** runtime-type.lisp -> early-target-type.lisp
2721 ** target-type.lisp -> late-target-type.lisp
2722 ** early-host-format.lisp -> early-format.lisp
2723 ** late-host-format.lisp -> late-format.lisp
2724 ** host-error.lisp -> misc-error.lisp
2725 ** early-error.lisp -> early-target-error.lisp
2726 ** late-error.lisp -> late-target-error.lisp
2727 ** host-defboot.lisp -> early-defboot.lisp
2728 ** code/misc.lisp -> code/target-misc.lisp
2729 ** code/host-misc.lisp -> code/misc.lisp
2730 ** code/numbers.lisp -> code/target-numbers.lisp
2731 ** code/early-numbers.lisp -> numbers.lisp
2732 ** early-host-type.lisp -> early-type.lisp
2733 ** late-host-type.lisp -> late-type.lisp
2734 ** host-typep.lisp -> typep.lisp
2735 ** load.lisp -> target-load.lisp
2736 ** host-load.lisp -> load.lisp
2737 ** host-disassem.lisp -> disassem.lisp
2738 ** host-insts.lisp -> insts.lisp
2739 ** byte-comp.lisp -> target-byte-comp.lisp
2740 ** host-byte-comp.lisp -> byte-comp.lisp
2741 ** host-signal.lisp -> signal.lisp
2742 ** host-defstruct.lisp -> defstruct.lisp
2743 ** late-target-type.lisp -> deftypes-for-target.lisp
2744 Furthermore, several other previously target-only files foo.lisp (e.g.
2745 hash-table.lisp and random.lisp) have been split into a target-and-host
2746 foo.lisp file and a target-only target-foo.lisp file, with their key type
2747 definitions in the target-and-host part, so that the cross-compiler will
2748 know more about target types.
2749 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2750 gone away. In their place are various *BACKEND-FOO* variables
2751 corresponding to the slots of the old structure.
2752 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2753 into the target SBCL has been fixed.
2754 * The chill.lisp system for loading cold code into a running SBCL
2756 * Support for the CMU CL "scavenger hook" extension has been removed.
2757 (It was undocumented and unused in the CMU CL sources that SBCL was
2758 derived from, and stale in sbcl-0.6.1.)
2759 * Various errors in the cross-compiler type system were detected
2760 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2761 (enabling various consistency checks). Many of them were fixed,
2762 but some hard problems remain, so the compiler is back to
2763 running without *TYPE-SYSTEM-INITIALIZED* for now.
2764 * As part of the cross-compiler type system cleanup, I implemented
2765 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2766 * I have started adding UNCROSS calls throughout the type system
2767 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2768 on cross-compiler input files will be able to go away, and various
2770 * CONSTANTP now returns true for quoted forms (as explicitly required
2773 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2775 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2776 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2777 (MAKE-LIST -1). In the longer term, it ought to become true
2778 that declarations are assertions even at SAFETY 1. For now, it's not
2779 quite true even at SAFETY 3, but it's at least more nearly true..
2780 (Note that this change seems to increases the size of the system by
2781 O(5%) and to decrease the speed of the compiler by 20% or more.)
2782 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2783 for the problem of printing dozens of lines of distracting information
2784 about low-level system machinery as part of the top stack frame
2785 on entry to the debugger when an undefined function was called.
2786 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2787 is not reset to COMMON-LISP-USER.
2788 * Compilation of stuff related to dyncount.lisp has been made conditional
2789 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2790 smaller. The various dyncount-related symbols have been moved into
2791 a new "SB-DYNCOUNT" package.
2792 * tty-inspect.lisp has been renamed to inspect.lisp.
2793 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2794 feature has gone away. (When we eventually port to other flavors of
2795 libc and/or Unix, we'll try to make the differences between flavors
2796 invisible at the user level.)
2797 * Various other *FEATURES* tags, and/or their associated conditionals,
2798 have been removed if obsolescent, or given better documentation, or
2799 sometimes given more-mnemonic names.
2801 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2803 * tidied up "make.sh" script
2804 * tidied up system directory structure
2805 * better "clean.sh" behavior
2806 * added doc/FOR-CMUCL-DEVELOPERS
2807 * many many small tweaks to output format, e.g. removing possibly-confusing
2808 trailing #\. character in DESCRIBE-INSTANCE
2809 * (EQUALP #\A 'A) no longer signals an error.
2810 * new hashing code, including EQUALP hashing
2811 * tidied up Lisp initialization and toplevel
2812 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2813 * command line argument processing
2814 * added POSIX-GETENV function to deal with Unix-ish environment variables
2815 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2816 terminating SBCL on EOF
2817 * non-verbose GC by default
2818 * There is no more "sbcl" shell script; the sbcl file is now the C
2819 runtime executable (just like CMU CL).
2820 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2822 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2823 symbol and package information as Lisp native symbol and package objects
2824 instead of strings naming symbols and strings naming packages. This way,
2825 whenever packages are renamed (as in warm init), debug information is
2826 transformed along with everything else.
2827 * tweaked the optimization policy declarations which control the building
2828 of SBCL itself. Now, among other things, the system no longer saves
2829 source location debugging information. (This helps two problems at once
2830 by reducing SBCL size and by keeping SBCL from trying to look for its
2831 sources -- which may not exist -- when reporting errors.)
2832 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2833 debugging and testing purposes
2834 * cleaned up printing, making the printer call PRINT-OBJECT for
2835 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2836 methods, giving nearly-ANSI behavior
2837 * converted almost all special variables to use *FOO* naming convention
2838 * deleted PARSE-TIME functionality, since it can be done portably
2839 * moved some files out of cold init into warm init
2840 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2842 * regularized formatting of source files
2843 * added an install.sh script
2844 * fixed ridiculous memory usage of cross-compiler by making
2845 compiler/alloc.lisp not try to do pooling unless it can hook
2846 itself into the GC of the cross-compilation host. Now the system
2847 builds nicely on my old laptop.
2848 * added :SB-ALLOC in target-features.lisp-expr
2849 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2850 was not implemented)
2851 * re-did condition handling and note reporting in the compiler. Notes
2852 are no longer handled by signalling conditions. Style warnings
2853 and warnings are handled more correctly and reported in such a way
2854 that it's easy to find one or the other in your output (so that you
2855 can e.g. figure out which of many problems caused COMPILE-FILE to
2857 * changed the severity of several compiler warnings from full WARNING
2858 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2859 compiler note reporting so that it doesn't use the condition system
2860 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2861 COMPILE-FILE command)
2862 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2863 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2864 no longer supported, since their implementation was deeply intertwingled
2865 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2866 * removed bogus "support" for compiler macros named (SETF FOO), and
2867 removed the compiler macro for SETF INFO (but only after making a fool
2868 of myself on the cmucl-imp mailing list by posting a bogus patch for
2869 DEFINE-COMPILER-MACRO..)
2870 * Compiled files containing forms which have side effects on the Lisp
2871 reader (such as DEFPACKAGE forms) are now handled more correctly.
2872 (Compiler queuing of top level lambdas has been suppressed by setting
2873 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2874 * deleted various currently-unused source files, e.g. gengc.lisp. They
2875 may be added back at some point e.g. when porting to other architectures,
2876 but until they are it's distracting to distribute them and to try to
2878 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2879 there were so many of them they're just distractions, and UNCROSS is
2880 known to be able to handle the current sources
2881 * moved PROFILE functionality into TRACE, so that it will be clear
2882 how the wrapping and unwrapping of functions when you profile them
2883 interacts with the wrapping and unwrapping of functions when you
2884 trace them. (Actually, the functionality isn't there yet, but at least
2885 the interface specification is there. Hopefully, the functionality will
2886 arrive with some future maintenance release.)
2887 * removed host-oops.lisp
2888 * changed signature of QUIT function to allow UNIX-CODE argument
2889 * fixed READ-SEQUENCE bug
2890 * tweaked verbose GC output so that it looks more like the progress
2891 output that ANSI specifies for functions like LOAD
2892 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2893 * added <http://sbcl.sourceforge.com> to the banner information printed by