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 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
5 supported. (thanks to Dan Debertin)
6 * fixed bug #331: structure-class instances corresponding to
7 DEFSTRUCT forms are now created eagerly.
8 * fixed bug #345: backtraces from calls to undefined functions work
9 on x86 as well. Related bug #61 is now also partially fixed on x86
10 (backtraces from throws to unknown catch tags.)
11 * bug fix: lambda-list parsing is now stricter vrt. order and number
12 of lambda-list keywords.
13 * bug fix: as specified by AMOP, an error is signalled if a
14 class-option appears multiple times in a DEFCLASS form. (reported
16 * bug fix: RANDOM can be compiled when the compiler derives the type
17 of its numeric argument as a disjoint set of small integers.
18 (reported by Paul Dietz)
19 * fixed some bugs revealed by Paul Dietz' test suite:
20 ** INCF, DECF and REMF evaluate their place form as specified in
22 ** FORMATTER expands ~{ ~} iteration directives with V or #
24 ** FORMATTER deals with the ~@[ ~] conditional directive where the
25 consequent uses no arguments correctly.
26 ** the system has a partial understanding of the (COMPLEX RATIO)
29 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
30 * new feature: a build-time option (controlled by the :SB-UNICODE
31 keyword feature, enabled by default) for building the system with
32 support for the entire 21-bit character space defined by the
34 * new feature: the system now has rudimentary external-format
35 support; the primary user-visible change at this time is that
36 characters with the high bit set (such as lower-case-e-acute) will
37 print correctly to a terminal in a UTF-8 environment.
38 * minor incompatible change: BASE-CHAR no longer names a class;
39 however, CHARACTER continues to do so, as required by ANSI.
40 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
41 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
42 has been moved to the SB-EXT package (temporarily re-exported from
44 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
45 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
47 * fixed bug #351: better error-handlind and reporting for malformed
49 * fixed bug #350: bignum-printing is now more memory-efficient,
50 allowing printing of very large bignums, eg. (expt 2 10000000).
51 (reported by Bruno Haible)
52 * fixed bug #302: better primitive-type selection for intersection
54 * fixed bug #308: non-graphic characters now all have names, as
55 required. (reported by Bruno Haible)
56 * bug fix: redefining a class with different superclasses now correctly
57 removes it from the direct-subclasses of its previous superclasses.
58 (reported by David Morse)
59 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
60 new value now works. (reported by Bruno Haible)
61 * bug fix: correct canonicalization of multiple non-standard slot
62 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
63 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
64 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
65 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
66 now exists, an signals an error.
67 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
68 during macroexpansion for non-iteration variables as well. (reported
69 by Bruno Haible for CMUCL)
70 * bug fix: Cyclic structures and unprintable objects in compiler
71 messages no longer cause errors. (reported by Bruno Haible)
72 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
73 and READ-FROM-STRING all now return a primary value of NIL if
74 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
75 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
76 (reported by Bruno Haible for CMUCL)
77 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
78 do not designate a condition. (reported by Bruno Haible for
80 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
81 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
82 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
83 permitting resolution in favour of any of the conflicting symbols.
84 (reported by Bruno Haible for CMUCL)
85 * bug fix: EQUAL compiler optimizations is less aggressive on
86 strings which can potentially compare true despite having distinct
87 specialized array element types.
88 * bug fix: unit enumerations can be defined without dividing by
89 zero. (thanks to Vincent Arkesteijn)
90 * FORMAT compile-time argument count checking has been enhanced.
91 (report from Bruno Haible for CMUCL)
92 * a partial workaround for the bug 262: the compiler does not try to
93 inline-expand a local function doing RETURN-FROM from a deleted
94 BLOCK. (thanks to Peter Denno for the bug report and to David
95 Wragg for the simple test case)
96 * fixed some bugs revealed by Paul Dietz' test suite:
97 ** RENAME-PACKAGE allows all package designators as new package
99 ** constraint propagation and lambda variable substitution are
100 more cautious in dealing with partially deleted code.
101 ** compiler sometimes lost reoptimization passes.
102 ** CERROR, given a condition as condition designator, passes its
103 remaining arguments to the continue format control without
105 ** Case-altering FORMAT directives work correctly on non-ASCII
107 ** The REAL type specifier handles bounds outside the floating
108 point ranges without signalling FLOATING-POINT-OVERFLOW.
109 ** Functions with IR1-transformations can create intercomponent
110 references to global functions.
111 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
113 ** FORMAT ~R treats a nil value for its first parameter correctly.
115 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
116 * enhancement: saving cores with foreign code loaded is now
117 supported on x86/NetBSD and sparc/Linux in addition to the previously
119 * bug fix: on some platforms repeated installations caused multiple
120 copies of HTML documentation to be installed -- should not happen
121 any more. (reported by Stefan Scholl)
122 * bug fix: parsing self-recursive alien record types multiple times
123 no longer causes infinite recursion. (reported by Thomas F. Burdick,
124 original patch by Helmut Eller for CMUCL)
125 * bug fix: stack-exhaustion detection works now on NetBSD as well.
126 (thanks to Richard Kreuter)
127 * bug fix: defining classes whose accessors are methods on existing
128 generic functions in other (locked) packages no longer signals
129 bogus package lock violations. (reported by François-René Rideau)
130 * bug fix: special variables as DEFMETHOD parameters no longer have
131 associated bogus type declarations. (reported by David Wragg and
133 * bug fix: read-write consistency on streams of element-type
134 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
135 * bug fix: redefiniton of the only method of a generic function with
136 no DEFGENERIC no longer emits a full WARNING. In addition,
137 redefinition of generic functions with no DEFGENERIC to an
138 incompatible lambda list now signals an error. (thanks to Zach
140 * bug fix: DEFGENERIC now works even when there's a function of the
141 same name in an enclosing lexical environment. (thanks to Zach
143 * fixed compiler failure, caused by instrumenting code during
144 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
145 * optimization: added loop analysis and improved register allocation
146 * fixed some bugs revealed by Paul Dietz' test suite:
147 ** POSITION on displaced vectors with non-zero displacement
148 returns the right answer.
149 ** (SIMPLE-STRING) is a valid type specifier for sequence
151 ** *PRINT-LEVEL* handling for slotless structures is pedantically
153 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
154 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
156 ** The justification version of the FORMAT ~< directive treats
157 non-zero minpad parameter correctly.
159 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
160 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
161 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
162 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
163 the supported interface.
164 * new feature: Single-stepping of code compiled with DEBUG 2 or
165 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
166 * new feature: saving cores with foreign code loaded is now
167 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
168 Timothy Moore's work for CMUCL)
169 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
170 parameters to * instead of NIL if no initform is supplied.
171 (reported by Johan Bockgaard)
172 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
173 correctly when FUNCALL appears as the car of the form. Note:
174 despite this FUNCALL forms are not currently subject to
175 compiler-macro expansion. (port of Raymond Toy's fix for the
176 same from CMUCL, reported by Johan Bockgaard)
177 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
178 (thanks for Teemu Kalvas)
179 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
180 case instead of terminating the iteration (thanks for Julian
181 Squires, Sean Champ and Raymond Toy)
182 * bug fix: incorrect expansion of defgeneric that caused a style
183 warning. (thanks for Zach Beane)
184 * bug fix: slot accessor effective method computation works properly
185 for classes with multiple non-standard applicable methods on
186 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
187 * on x86 compiler supports stack allocation of results of LIST and
188 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
189 CMUCL implementation by Gerd Moellmann)
190 * fixed some bugs revealed by Paul Dietz' test suite:
191 ** FORMAT strings with both the ~<~:;~> form of the justification
192 directive and pretty-printing directives cause an error.
194 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
195 * incompatible change: the internal functions
196 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
197 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
198 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
199 instead of the old functions.
200 * new feature: on platforms where "dladdr" is available foreign
201 function names now appear in backtraces. (based on Helmut Eller's
203 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
204 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
206 * documentation: ASDF manual is now distributed with SBCL.
207 * bug fix: SBCL can now load its contributed modules with REQUIRE
208 even if the system-provided entries have been removed from
209 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
211 * bug fix: backtraces involving undefined functions or assembly
212 routines are more informative. (thanks to Brian Downing)
213 * bug fix: mutually referent alien structures now work correctly.
214 (reported by Rick Taube)
215 * bug fix: structures defined by WITH-ALIEN can be referred to
216 within other definitions in the same WITH-ALIEN.
217 * bug fix: division operators (MOD, TRUNCATE and the like) with
218 constant zero divisors and integer dividends no longer generate
220 * bug fix: provide default methods for INPUT-STREAM-P and
221 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
222 * bug fix: improve the detection and resolution of MOP metacycles
223 (where effective method computation depends on the generic function
224 having its effective method computed). (reported by Bruno Haible)
225 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
226 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
227 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
228 * bug fix: pathnames with relative directory components can be
229 represented relative to default pathnames.
230 * optimization: in taking the GCD of bignums, reduce the two bignums
231 to approximately the same size (using Euclid's algorithm) before
232 applying the more sophisticated binary GCD. (thanks to Juho
234 * optimization: COUNT on bitvectors now operates word-at-a-time.
235 * optimization: ASH with a positive, but not necessarily constant,
236 (leftwards) shift, when in a modular context, is compiled to a
238 * fixed some bugs revealed by Paul Dietz' test suite:
239 ** FORMAT variable parameters ("~V<char>") are defaulted properly
240 if the corresponding argument is NIL.
241 ** FORMAT directives accepting a minpad parameter treat negative
243 ** literal commas as character parameters to FORMAT directives are
245 ** literal spaces directly after ~<Newline> directives within a
246 format-logical-block (~:< ~@:>) do not induce :FILL-style
247 conditional newlines.
248 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
249 as specified: it no longer includes conditional newlines).
250 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
252 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
253 :PER-LINE-PREFIX argument does not evaluate to a string.
255 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
256 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
257 the manual for details; for now, package locks can be disabled by
258 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
259 if no major problems are found then it is likely that they will be
260 compiled in unconditionally.
261 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
262 unsupported operators on all platforms. To load a shared library
263 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
264 object file, link it into a shared library outside of SBCL and
265 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
267 * bug workaround: ROOM T and the GC were not working together
268 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
269 reporting facility aren't preserved. That reporting has been
270 disabled, so now until and unless someone figures out how to make it
271 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
272 * minor incompatible change: as threatened around sbcl-0.8.0, the
273 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
274 The symbols are also exported from SB-PCL for backwards
275 compatibility, but more so than before SB-PCL should be treated as
276 an implementation-internal package.
277 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
279 * the SB-SPROF contrib now works on (most) non-x86 architectures.
280 It is known as of this release not to work on the Alpha, however.
281 * fixed bug #167: errors signalled due to illegal syntax in method
282 bodies are now more legible.
283 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
284 designators and can hence be used with TYPEP.
285 * fixed bug #333: CHECK-TYPE now ensures that the type error
286 signalled, if any, has the right object to be accessed by
287 TYPE-ERROR-DATUM. (reported by Tony Martinez)
288 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
289 5.1.2.3 for multiple-value place subforms. (reported by Kalle
291 * fixed bug #334: programmatic addition of slots using specialized
292 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
293 and :ALLOCATION :CLASS slots.
294 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
295 float exponents. (rereported by Peter Seibel)
296 * fixed a bug: #\Space (and other whitespace characters) are no
297 longer considered to be macro characters in standard syntax by
299 * fixed bug: initialization of condition class metaobjects no longer
300 causes an instance of the condition to be created. (reported by
302 * fixed bug: it is now possible to have slots such that
303 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
304 description is neither :INSTANCE nor :CLASS.
305 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
306 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
307 SLOT-VALUE-USING-CLASS), and no longer causes errors with
308 non-standard SLOT-DEFINITION-ALLOCATION values.
309 * fixed bugs: various race conditions handling exiting threads.
310 CL-PPCRE's thread torture test now passes.
311 * fixed arguably-a-bug: GC time proportional to number of threads in
312 system even when most of them are idle
313 * optimization: improved performance of BIT and SBIT on bit-vectors.
314 * .fasl file incompatibility: The fasl file version number has
315 been incremented because of changes associated with package locks.
317 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
318 * minor incompatible change: the system no longer provides
319 optimization hints (conditions which are TYPEP
320 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
321 mode; these hints are still emitted when higher SPEED optimization
323 * new contrib module: a sampling profiler (profiling by statistical
324 sampling, rather than by instrumenting functions) is available as
325 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
326 * the behaviour of the standard function ED is now customizeable by
327 third parties through a hook variable: see ED's documentation
328 string for information on the protocol.
329 * the compiler no longer emits efficiency notes for (FUNCALL X)
330 when the type of X is uncertain under default optimization
332 * fixed bug 276: mutating a binding of a specialized parameter to a
333 method to something that is not TYPEP the specializer is now
335 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
336 exhibit the required behaviour on the x86 platform. (thanks to
337 Peter van Eynde, Eric Marsden and Bruno Haible)
338 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
339 even for difficult arguments. (reported by Peter Graves)
340 * fixed bug 141a: the backquote printer now descends quoted
342 * fixed another bug in backquote printing: no more destructive
343 modification of the form's list structure. (reported by Brian
345 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
346 move between its address being taken and the call to
347 interrupt_thread, fixing a crashing race condition.
348 * the SB-POSIX contrib implementation has been adjusted so that it
349 no longer exhibits ridiculously poor performance when constructing
350 instances corresponding to C structs.
352 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
353 * minor incompatible change: the sb-grovel contrib now treats C
354 structures as alien (in the sense of SB-ALIEN) objects rather than
355 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
356 has implications for memory management of client code
357 (sb-grovel-returned objects must now be manually managed) and for
358 type safety (alien objects now have full types).
359 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
360 used to control emission of compiler diagnostics, rather than the
361 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
362 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
363 quality should be considered deprecated.
364 * install.sh now installs the user manual as well
365 * (not quite a new documentable feature, but worth considering in
366 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
367 beginnings of a semantically meaningful condition hierarchy is
368 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
370 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
371 designator as the defaults argument.
372 * fixed bug: Displaced arrays whose displaced-to array has become
373 too small now cause ARRAY-DIMENSION to signal an error, providing
374 for safer bounds-checking. (reported by Bruno Haible)
375 * fixed bug: DEFCLASS slot definitions with identical :READER and
376 :WRITER names now signal a reasonable error. (reported by Thomas
378 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
380 * fixed bug: Invalid slot specification errors now print correctly.
381 (thanks to Zach Beane)
382 * fixed bug 320: Shared to local slot value transfers in class
383 redefinitions now happen corrently from superclasses as
384 well. (reported by Bruno Haible)
385 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
387 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
388 improper lists correctly. (reported by Bruno Haible)
389 * fixed bug 313: source-transform for <fun-name> was erroneously
390 applied to a call of a value of a variable with name <fun-name>.
391 (reported by Antonio Menezes Leitao)
392 * fixed bug 307: The obsolete instance protocol ensures that
393 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
394 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
395 remove dead unknown-values globs from the middle of the stack.
396 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
397 easier to resume long computations after using *BREAK-ON-SIGNALS*
398 to diagnose and fix failures. (thanks to Nikodemus Siivola)
399 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
400 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
401 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
402 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
403 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
405 * fixed a (fairly theoretical) bug in string printing: if
406 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
407 does not have array-element-type equal to the most general string
409 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
410 function name. (thanks to Nikodemus Siivola)
411 * fixed bug: compile-time format string checker failed on
412 non-closed ~{. (reported by Thomas F Burdick)
413 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
414 don't warn on legal loop code involving more than one aggregate
415 boolean. (thanks to Nikodemus Siivola)
416 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
417 in keyword position to MAKE-PACKAGE was not being reported as
419 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
420 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
422 * fixed bugs 280 and 312: the checking for multiple definitions in a
423 file is less likely to become confused by uses of inline
425 * fixed bug: the #S reader macro performs the keyword coercion
426 specified for slot names. (reported by Kalle Niemitalo)
427 * fixed bug: lambda lists may contain symbols whose names start with
428 & but are not lambda-list-keywords; their occurrence triggers a
430 * fixed bug 321: define-method-combination argument lambda lists do
431 not cause invalid code to be generated when &optional or &aux
432 variables have default values. (reported by Bruno Haible)
433 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
434 have CLOS classes; this appears also to have fixed reported
435 instability in stack exhaustion detection.
436 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
437 MUFFLE-WARNING when no associated restart is present is now
439 * optimization: rearranged the expansion of various defining macros
440 so that each expands into only one top-level form in a
441 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
443 * optimization: used a previously little-used slot in symbols to
444 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
446 * fixed some bugs revealed by Paul Dietz' test suite:
447 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
449 ** Fixed an optimization bug related to inheritance of initforms
450 from local to shared slots.
451 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
452 ** CAST optimizer forgot to flush argument derived type.
453 ** print/read consistency on floats is now orders of magnitude
454 more likely. (thanks also to Bruno Haible for a similar report
456 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
457 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
458 for objects of type REAL. Make it so.
459 ** SXHASH is specified (infelicitously) to respect similarity,
460 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
461 it so. (thanks to Markus Ziegler)
462 ** on the Alpha, the compiler succeeds in compiling functions
463 returning a known number of arguments greater than 63.
464 ** fixed handling of invalid NIL arguments in keyword position
466 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
468 ** provided a readably-printable representation for RANDOM-STATE
470 ** ensured that pathnames, where they have a namestring, always
471 print using #P"..." syntax.
473 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
474 * Support for the forthcoming 2.0 version of the NetBSD kernel
475 running on x86 hardware has been added. (thanks to Perry
476 E. Metzger most immediately, and others for their past work)
477 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
478 current release version; SBCL's OpenBSD support had been broken
479 since about the time of OpenBSD's switch to ELF binary format.)
480 * [placeholder for DX summary]
481 ** user code with &REST lists declared dynamic-extent, under high
482 speed or space and low safety and debug optimization policy.
483 * The manual has been converted to Texinfo format and the debugger
484 chapter from the cmucl manual has been added.
485 * A facility has been added to extract documentation strings from
486 sbcl and store them as Texinfo-formatted snippets for inclusion in
487 the manual (via Texinfo's @include directive)
488 * bug fix: compiler emitted division in optimized DEREF. (thanks for
489 the test case to Dave Roberts)
490 * bug fix: multidimensional simple arrays loaded from FASLs had fill
491 pointers. (reported by Sean Ross)
492 * bug fix: PROFILE output is printed nicely even for large numerical
493 values. (thanks to Zach Beane)
494 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
495 greater than 32 handle EOF correctly.
496 * bug fix: on X86 an immediate argument of the IMUL instruction is
497 correctly printed in disassembly. (thanks to Lutz Euler)
498 * bug fix: class slots in redefined classes preserve their old
499 values. (thanks to Bruno Haible and Nikodemus Siivola)
500 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
501 succeeds. (reported by Marco Baringer)
502 * fixed some bugs revealed by Paul Dietz' test suite:
503 ** READ-SEQUENCE now works on ECHO-STREAMs.
504 ** RATIONALIZE works more according to its specification. (thanks
507 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
508 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
509 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
510 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
511 should matter to you only if you rebind the printer control
512 variables and then find you want different bindings in the
513 debugger than in the ordinary execution of your program.)
514 * The runtime build system has been tweaked to support building
515 (on SPARC/SunOS) using a C compiler which invokes Sun's own
516 assembler and linker. (thanks to Nikodemus Siivola)
517 * Unbound, undefined, undeclared variables now trigger full
518 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
519 more often programmer error than deliberate exploitation of undefined
521 * optimization: the hash algorithm for strings has changed to one
522 that is less vulnerable to spurious collisions. (thanks to Juho
524 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
525 needless bounds checking. (thanks to Juho Snellman)
526 * optimization: implemented multiplication as a modular
527 (UNSIGNED-BYTE 32) operation on the PPC backend.
528 * fixed some bugs revealed by Paul Dietz' test suite:
529 ** ADJUST-ARRAY now copies the datum in a zero rank array if
531 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
532 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
533 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
535 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
537 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
539 ** ECHO-STREAMs no longer attempt to echo the end of file value to
540 their output stream on EOF from read.
541 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
542 have been read to end-of-file.
543 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
545 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
546 description of determination of which consecutive characters
548 ** Printing the "Space" character with escaping on now yields "#\\ ",
549 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
550 ** Reading floating-point numbers with *READ-BASE* set to a number
551 less than 10 works correctly.
552 ** Reading floating-point numbers with *READ-BASE* set to a number
553 more than 10 works correctly.
554 ** Printing with *PRINT-READABLY* targets the standard readtable, not
555 the readtable currently in effect.
557 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
558 * minor incompatible change: parsing of namestrings on a physical
559 (Unix) host has changed; numbers after the final #\. in a
560 namestring are no longer interpreted as a version field. This is
561 intented to be largely invisible to the user, except that the
562 meaning of the namestring "*.*.*" has changed: it now refers to a
563 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
564 should usually be replaced by
565 (make-pathname :name :wild :type :wild :version :wild)
566 with the added benefit that this is more likely to be portable.
567 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
568 option to OPEN now signals an error if the file being opened
569 exists; this may have an impact on existing code.
570 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
571 signals received, on the PowerPC platforms (both Linux and
572 Darwin). (thanks to Pierre Mai for pointing out the location of
574 * several fixes on OS X: The system now builds and runs cleanly on
575 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
576 (thanks to Brian Mastenbrook)
577 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
578 second-resolution: integer multiples of 1/3600 between -24 and 24.
579 (thanks to Vincent Arkesteijn)
580 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
581 type when called with 1 argument; PEEK-CHAR checked type of
582 PEEK-TYPE only after having read first character from a
583 stream. (reported by Peter Graves)
584 * bug fix: the garbage collector now has much better locality
585 behaviour, and in particular no longer treats all memory as being
586 exhausted when many small objects point to each other in a deeply
588 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
589 recognized as being TYPEP their class.
590 * bug fix: the PUSHNEW documentation string has been corrected.
591 (thanks to Vincent Arkesteijn)
592 * bug fix: defaulting of the value for the last of an atypically
593 large number of multiple values being bound was not being
594 performed correctly on the Alpha or PPC platforms
595 * optimization: implemented multiplication as a modular
596 (UNSIGNED-BYTE 32) operation on the x86 backend.
597 * optimization: SEARCH on simple-base-strings can now be open-coded.
598 (see also contrib/compiler-extras.lisp for inspiration for
599 teaching the compiler about the Boyer-Moore algorithm).
600 * value, returned by MAX (and MIN) called with several EQUALP, but
601 not EQL, arguments now does not depend on compiler settings.
602 * fixed some bugs revealed by Paul Dietz' test suite:
603 ** in stack analysis liveness information is propagated from
604 non-local entry points.
605 ** pathwise CAST removing failed when the CAST node did not start
607 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
608 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
610 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
611 host is already defined.
612 ** RENAME-FILE works on streams instead of signalling an internal
614 ** PEEK-CHAR uses the current readtable when determining whether
615 or not a character is whitespace.
616 ** MERGE-PATHNAMES handles the case when the pathname does not
617 specify a name while the default-pathname specifies a version.
618 ** Pathnames now stand a better chance of respecting print/read
620 ** Attempting to use standardized file system operators with a
621 pathname with invalid :DIRECTORY components signals a
623 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
625 ** DIRECTORY on logical pathnames is more correct.
626 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
627 signal a TYPE-ERROR if their argument is not a stream.
628 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
629 designator argument does not designate a stream.
630 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
631 examining the synonym.
632 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
634 ** OPEN and WITH-OPEN-STREAM allow opening streams with
635 element-type larger than ([UN]SIGNED-BYTE 32).
637 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
638 * When built with the :SB-FUTEX feature, threaded builds now take
639 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
640 for faster/more reliable mutex and condition variable support.
641 * Incompatible change (but one you probably shouldn't have been using
642 anyway): the interface and code for arbitrating between multiple
643 threads in the same user session has been redesigned.
644 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
645 error when called without an explicit environment argument.
646 (thanks to Nikodemus Siivola)
647 * bug fix: buffered :DIRECTION :IO streams are less likely to become
648 confused about their position. (thanks to Adam Warner and Gerd
650 * bug fix: Pretty printing backquoted forms with unquotations in the
651 argument list position of various code constructs such as LAMBDA
652 now works correctly. (reported by Paul Dietz)
653 * bug fix: Pretty printing unquotations no longer loses all
654 stream position information.
655 * optimization: performance of string output streams is now less
656 poor for multiple small sequence writes.
657 * optimization: performance of CSUBTYPEP in the presence of complex
658 expressions involving CONS and NOT many times has been improved.
659 (reported by Paul Dietz)
660 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
662 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
663 streams for the REPL, for the debugger, and so on. (thanks to
665 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
666 * fixed some bugs revealed by Paul Dietz' test suite:
667 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
668 is ignored if it is too small, rather than propagating through
669 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
670 ** extremely complex negations of CONS types were not being
671 sufficiently canonicalized, leading to inconsistencies in
673 ** VALUES tranformer lost derived type.
675 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
676 * fixed a bootstrapping bug: the build process no longer assumes
677 that the various BOOLE-related constants have the same value in
678 host and target lisps. (noted by Paul Dietz' test suite on an
679 SBCL binary built from CLISP)
680 * The system can now be dynamically linked on the MIPS platform,
681 which enables dynamic loading of foreign code from Lisp. (thanks
682 to Ralf Baechle for discussions on the MIPS ABI)
683 * The system now records debugging information for its own source
684 files in a filesystem-position-independent manner, relative to
685 the "SYS" logical host.
686 * fixed a compiler bug: MV-LET convertion did not check references
687 to the "max args" entry point. (reported by Brian Downing)
688 * tweaked disassembly notes to be less confident about proclaiming
689 some instruction as an LRA. (thanks to Brian Downing)
690 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
691 now each have their own history, command character, and other
692 characteristics. (thanks to David Lichteblau)
693 * fixed some bugs revealed by Paul Dietz' test suite:
694 ** compiler failure in compiling LOGAND expressions including a
696 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
697 shift greater than 32.
698 ** FLUSH-DEST did not mark blocks for type check regeneration.
699 ** HANDLER-CASE failed to accept declarations in handler clauses
700 in some circumstances.
702 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
703 * New code in contrib/sb-introspect (still probably not entirely
704 stable yet) provides some support for smart Lisp development
705 environments like SLIME.
706 * The conditions signalled for errors occurring when loading .fasl
707 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
708 in a way which should help ASDF recover gracefully.
709 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
710 stuff has been cleaned up. If you code contrib/ stuff, this might
711 affect you, and you can look at contrib/README, contrib/STANDARDS,
712 and/or the 0.8.4.27 diff to check.
713 * In full calls the compiler now does not generate checks for declared
714 argument types for all arguments.
715 * various threading fixes
716 ** and some experimental patches which didn't make it into
717 the main tree for this release, but which are shipped in
718 contrib/experimental-thread.patch as a possible fix for some
719 failures (deadlock, spinning...) in GC-intensive multithreaded
721 * fixed PPC build problem (source code incompatibility of different
722 library versions): added offsetof() hackery which attempts to divine
723 where glibc maintainers put uc_mcontext today
724 * fixed bug 282: compiler does not trust type assertions while passing
725 arguments to a full call.
726 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
727 &OPTIONAL VALUES type parameter.
728 * fix bug 214: algorithm for noting rejected templates is now more
729 similar to that of template seletion. (also reported by rydis on
731 * fixed bug 141b: printing backquoted information readably and prettily
732 inserts a space where necessary.
733 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
734 (CHARACTER 10) are now reported as errors, rather than propagated
735 as unknown types. (reported by piso on #lisp)
736 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
737 with values NIL and :ERROR. (thanks to Milan Zamazal)
738 * fixed bug 191c: CLOS now does proper keyword argument checking as
739 described in CLHS 7.6.5 and 7.6.5.1.
740 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
741 longer raise an error; further, using a list as a for-as-arithmetic
742 counter now raises a meaningful error.
743 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
744 now understood by sequence creation functions such as MAKE-SEQUENCE
746 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
747 a STRING-INPUT-STREAM.
748 * compiler enhancement: SIGNUM is now better able to derive the type
750 * type declarations inside WITH-SLOTS are checked. (reported by
752 * fixed some bugs revealed by Paul Dietz' test suite:
753 ** incorrect optimization of TRUNCATE for a positive first
754 argument and negative second.
755 ** compiler failure in let-convertion during flushing dead code.
756 ** compiler failure while deriving type of TRUNCATE on an
757 interval, containing 0.
758 ** ASH of a negative bignum by a negative bignum count now returns
760 ** intersection of CONS types now canonicalizes properly, fixing
761 inconsistencies in SUBTYPEP.
763 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
764 * incompatible change: The --disable-debugger command line
765 option now clobbers the debugger at a more fundamental
766 level, by redefining #'INVOKE-DEBUGGER instead of by
767 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
768 is specified by ANSI to ignore *DEBUGGER-HOOK* and
769 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
770 enter the debugger REPL and then suffer recursive errors
771 because *DEBUG-IO* is also messed up in --disable-debugger mode;
772 while under the new system, BREAK in --disable-debugger mode
773 terminates the system just as an unhandled error would.
774 * fixed compiler performance when processing loops with a step >1;
775 * bug fix: DOCUMENTATION now retrieves generic function
776 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
777 support has been systematized, and now supports the methods
778 specified by ANSI, along with a default method and a method for
779 slot documentation. (reported by Nathan Froyd)
780 * bug fix: effective methods associated with a generic function are
781 no longer cached over a change of that generic function's method
782 combination. (reported by Andreas Fuchs)
783 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
784 names a non-existent directory. (thanks to Andreas Fuchs)
785 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
786 the PPC platform, lead to stack corruption; this has been fixed.
787 (reported by Rainer Joswig)
788 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
789 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
790 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
791 arithmetic in general, is now much more reliable on the PPC
793 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
794 platform now returns the right answer.
795 * optimization: restored some effective method precomputation in
796 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
797 precomputation is now tunable.
798 * optimization: compiler-internal data structure use has been
799 reviewed, and changes have been made that should improve the
800 performance of the compiler by about 20%.
801 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
802 simple-base-strings and simple-bit-vectors is improved.
803 * optimization: the optimization of 32-bit logical and arithmetic
804 functions introduced in version 0.8.3 on the x86 has been
805 implemented on the mips, ppc and sparc platforms; an
806 implementation of the same facility, but for 64-bit arithmetic,
807 has been added for the alpha.
808 * microoptimization: the compiler is better able to make use of the
809 x86 LEA instruction for multiplication by constants.
810 * bug fix: in some situations compiler did not report usage of
811 generic arithmetic in (SPEED 3) policy.
812 * bug 145b fix: compiler used wrong type specifier while converting
813 MEMBER-types to numeric.
814 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
816 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
817 index argument now works properly on the Alpha platform.
818 * bug fix: floating point exception treatment on the Alpha platform
820 * bug fix: FILE-POSITION works much better on string input and
821 output streams. (thanks to Nikodemus Siivola)
822 * bug fix: many threading/garbage collection symptoms sorted.
823 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
824 might be pseudo-atomic.
825 * internal change: Stopping for GC is now done with signals not
826 ptrace. GC is now done in whichever thread wanted it, instead of
828 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
830 * bug fix: result form in DO is not contained in the implicit
832 * incompatible change: ICR structure is changed; the value part of
833 CONTINUATION is now called LVAR; corresponding functions are
834 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
835 * added type deriver for ISQRT (thanks to Robert E. Brown).
836 * bug fix: better support for loading from the command line when an
837 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
839 * fixed some bugs revealed by Paul Dietz' test suite:
840 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
841 ** ROUND and FROUND now give the right answer when given very
842 small float arguments.
843 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
845 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
846 ** (GCD 0 <negative-integer>) returned <negative-integer>.
847 ** LCM should return a non-negative integer.
848 ** PARSE-INTEGER returned the index of a terminator instead of the
849 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
850 ** PARSE-INTEGER returned an incorrect index being applied to a
852 ** LCM with two arguments of 0 returns 0 rather than signalling
854 ** unsigned addition of a 32-bit constant with the high bit set no
855 longer causes an internal compiler error on the x86.
856 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
857 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
858 with negative last argument.
859 ** byte specifiers with zero size and position no longer cause
860 an error during type derivation.
861 ** bignum multiplication on the Alpha platform now returns the
863 * porting: The system now builds on SuSE AMD64, although it still
864 generates a 32-bit binary.
865 * .fasl file incompatibility: The fasl file version number has
866 been incremented (because of the changes to internal compiler
867 data structures referred to above).
869 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
870 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
871 more accurately, on the Darwin kernel running on PowerPC hardware.
872 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
873 * Compiler code deletion notes now signal a condition of type
874 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
875 an associated MUFFLE-WARNING restart.
876 * The compiler now performs limited argument count validation of
877 constant format strings in FORMAT, and where appropriate in ERROR,
878 CERROR and WARN. (thanks to Gerd Moellmann)
879 * New ASDF-INSTALL contrib can be used for automatic download and
880 installation of third-party Lisp code from CCLAN or other sites
882 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
883 which forces another thread to execute a function supplied by the
885 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
886 now accept and act upon their :ELEMENT-TYPE keyword argument.
887 (reported by Martin Atzmueller, Edi Weitz)
888 * bug fix: FILE-POSITION now accepts position designators up to
889 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
890 is the greater. (thanks to Patrik Nordebo)
891 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
892 Kalle Olavi Niemitalo)
893 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
894 printable. (reported by Eric Marsden)
895 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
896 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
897 * x86 bug fix in control stack exhaustion checking: now shows backtrace
898 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
899 (thanks to Stig Sandoe)
900 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
901 UNREAD-CHAR now correctly includes the unread character in the
902 target sequence. (thanks to Gerd Moellmann)
903 * bug fix in threaded builds: the system can now be suspended and
904 resumed by shell job control with minimal disruption.
905 * bug fixes in times and timezones >2038AD
906 * better handling of "where is GNU make?" problem in build scripts
907 (thanks to Nikodemus Siivola)
908 * new optimization: inside a named function any reference to a
909 function with the same name is considered to be a self-reference;
910 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
911 optimization quality.
912 * new optimization on x86: logical functions and + now have
913 optimized (UNSIGNED-BYTE 32) versions, which are automatically
914 used when the result is truncated to 32 bits.
915 * VALUES declaration is partially enabled.
916 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
917 * fixed some bugs revealed by Paul Dietz' test suite:
918 ** The system now obeys the constraint imposed by
919 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
920 types form a lattice under type intersection.
921 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
922 ** ASSOC now ignores NIL elements in an alist.
923 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
924 and (1+ MOST-POSITIVE-FIXNUM) answers.
925 ** The addition of a method with invalid qualifiers to a generic
926 function does not cause an error to be signalled immediately;
927 a warning is signalled, and the error is generated only on
928 calling the generic function.
929 * changed .fasl file version number, in part to add type codes for
930 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
931 obscure ANSI requirements
933 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
934 * fixed bug 148: failure to inline-expand a local function left
935 garbage, confusing the compiler.
936 * fixed bugs 3cd: structure slot readers perform type check if the
937 slot can have an invalid value (i.e. it is either not initialized
938 or can be written with a less specific slot writer).
939 * bug fix: the compiler now traps array references to elements off
940 the end of an array; previously, the bounds checking in some
941 circumstances could go off-by-one.
942 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
944 * type declarations for array element types now obey the description
945 on the CLHS page "Declaration TYPE", as per discussions on
946 sbcl-help around 2003-05-08. This means that a declaration
947 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
948 declaration, all references to BAR will be asserted or assumed
949 (with THE, so dependent on compiler policy) to involve objects of
950 type FOO. Note that no such declaration is implied in
951 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
952 * declared types of functions from the "Conditions"
953 chapter. (reported by Paul Dietz)
954 * bug fix: CERROR accepts a function as its first argument.
955 * bug fix: NTH an NTHCDR accept a bignum as index
956 arguments. (reported by Adam Warner)
957 * optimization: character compare routines now optimize comparing
958 against a constant character. (reported by Gilbert Baumann)
959 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
960 argument now works properly on the MIPS platform.
961 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
962 * repeated evaluation of the same DEFSTRUCT, a slot of which is
963 declared to have a functional type, does not cause an error
965 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
966 many :DEFAULT-INITARGS. (reported by Istvan Marko)
967 * fixed bug: if last continuation of a deleted block has a
968 destination, this destination should be deleted too. (reported by
970 * fixed a bug in the bootstrap process: the host compiler's values
971 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
972 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
973 test case from Patrik Nordebo)
974 * improved the ability of the disassembler on the PPC platform to
975 provide helpful disassembly notes.
976 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
977 the class in more cases than previously.
978 * bug fix: FILE-POSITION now understands :START and :END for
979 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
980 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
981 8) data. (thanks to David Lichteblau)
982 * bug fix: it is possible to add a method to a generic function
984 * bug fix: reader failed to signal END-OF-FILE inside an
985 object representation. (reported by Nikodemus Siivola)
986 * fixed some bugs revealed by Paul Dietz' test suite:
987 ** LAST and [N]BUTLAST should accept a bignum.
988 ** condition slot accessors are methods.
989 ** (VECTOR NIL) is a subtype of STRING.
991 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
992 * minor incompatible change: some nonsensical specialized lambda
993 lists (used in DEFMETHOD) which were previously ignored now signal
995 * minor incompatible change: the system is now aware of the types of
996 variables in the COMMON-LISP package, and will signal errors for
997 most violations of these type constraints (where previously they
998 were silently accepted).
999 * minor incompatible change: COMPILE-FILE now uses the freedom
1000 afforded (ANSI 3.2.2.3) to use derived function types for
1001 functions defined in the same file. This also permits the system
1002 to warn on static type mismatches and function
1003 redefinition. (Currently it does not work with high DEBUG level.)
1004 * minor incompatible change: VALUES declaration is disabled.
1005 * When issuing notes, the compiler now signals a condition of type
1006 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1007 restart for use in user handlers. It is expected that the
1008 COMPILER-NOTE condition will eventually become a condition
1009 supertype to a hierarchy of note types, which will then be
1010 handleable in a similar fashion. However, at the moment, no such
1011 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1012 but it's an implementation detail, not a classification for the
1014 * Changes in type checking closed the following bugs:
1015 ** type checking of unused values (192b, 194d, 203);
1016 ** template selection based on unsafe type assertions (192c, 236);
1017 ** type checking in branches (194bc).
1018 * A short form of VALUES type specifier has ANSI meaning (it has
1019 increased the number of situations when SBCL cannot perform type
1021 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1022 keywords or constants is permissible.
1023 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1024 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1025 classes. (thanks to Antonio Martinez)
1026 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1027 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1028 argument to a pretty-print stream. Instead, it leaves any such
1029 operations to DESCRIBE-OBJECT methods.
1030 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1032 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1033 lambda lists. (thanks to David Lichteblau)
1034 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1035 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1036 (thanks to Gerd Moellmann)
1037 * fixed bug 246: increased compilation speed of long
1038 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1040 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1042 * DEFCONSTANT now throws a condition of type
1043 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1044 constant to a non-EQL value; CONTINUE and ABORT restarts
1045 respectively change and preserve the value.
1046 * fixed bug 63: The code walker, part of the implementation of CLOS,
1047 is now better at handling symbol macros.
1048 * bug fix: There is no longer an internal implementation type named
1049 CL:LENGTH. (reported by Raymond Toy)
1050 * bug fix: In macro-like defining macros/special operators the
1051 implicit block does not enclose the lambda list.
1052 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1053 enclosed in parentheses) are not suitable as type specifiers, and
1054 their use properly signals an error now.
1055 * bug fix: An argument count mismatch for a type specifier in code
1056 being compiled no longer causes an unhandled error at compile
1057 time, but signals a compile-time warning.
1058 * fixed simple vector readable printing
1059 * bug fix: DESCRIBE takes more care over whether the class
1060 precedence list slot of a class is bound before accessing it.
1061 (reported by Markus Krummenacker)
1062 * bug fix: FORMATTER can successfully compile pretty-printer format
1063 strings which use variants of the ~* directive inside.
1064 * bug fix: SEARCH now applies its TEST predicate to the elements of
1065 the arguments in the correct order. (thanks to Wolfhard Buss)
1066 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1067 in type inference. (thanks to Robert E. Brown)
1068 * bug fix: Reading in symbols with an explicit package name of ""
1069 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1070 * fixed some bugs revealed by Paul Dietz' test suite:
1071 ** NIL is now allowed as a structure slot name.
1072 ** Arbitrary numbers, not just REALs, are allowed in certain
1073 circumstances in LOOP for-as-arithmetic clauses.
1074 ** Multiple class redefinitions before slot access no longer
1075 causes a type error.
1076 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1077 association between the name and a class.
1078 ** Generic functions with non-standard method-combination and over
1079 five methods all of which return constants no longer return NIL
1080 after the first few invocations. (thanks to Gerd Moellmann)
1081 ** CALL-NEXT-METHOD with no arguments now passes the original
1082 values of the arguments, even in the presence of assignment.
1083 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1085 ** DISASSEMBLE works with closures and funcallable instances.
1086 ** ADD-METHOD now returns the generic function, not the new method.
1087 ** FIND-METHOD signals an error if the lengths of the specializers
1088 is incompatible with the generic function, even if the ERRORP
1090 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1091 which its argument is a member.
1092 ** DEFCLASS only redefines the class named by its class-name
1093 argument if that name is the proper name of the class;
1094 otherwise, it creates a new class.
1095 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1096 of the UNBOUND-SLOT condition to the name of the slot.
1097 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1098 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1099 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1100 SLOT-MAKUNBOUND in the specified fashion.
1102 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1103 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1104 cross-compilation host. As a consequence, we can now bootstrap our
1105 way up to SBCL starting with a bare gcc toolchain and human-readable
1106 source code (first the source to CLISP, then the source to SBCL).
1107 * A contributed module containing a partial implementation of the
1108 simple-streams interface has been included. (thanks to Rudi
1110 * A contributed module implementing the RFC1321 Message Digest
1111 Algorithm, known as MD5, has been included.
1112 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1113 no longer has any effect, as the code controlled by this feature
1114 has been deleted. (As far as we know, no-one has ever built using
1115 this feature, and its semantics were confused in any case).
1116 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1117 work on conditions (as required by the ANSI specification),
1118 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1119 expected behaviour on conditions. Users should note, however,
1120 that such behaviour is not required by the ANSI specification,
1121 and so use of this behaviour may render their code unportable.
1122 * fixed some bugs revealed by Paul Dietz' test suite:
1123 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1125 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1126 ** SXHASH on condition objects no longer returns NIL.
1127 ** :ALLOCATION :CLASS slots are better treated; their values are
1128 updated on class redefinition, and initforms inherited from
1129 superclasses are applied.
1130 ** REMOVE-METHOD returns its generic function argument even when
1131 no method was removed.
1132 ** SHARED-INITIALIZE now initializes the values of the requested
1133 slots, including those with :ALLOCATION :CLASS.
1134 ** ALLOCATE-INSTANCE now works on structure classes defined via
1135 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1137 ** SLOT-EXISTS-P now works on conditions, as well as structures
1139 ** MAKE-LOAD-FORM now has the required methods on
1140 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1141 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1142 keyword, and now implements the SLOT-NAMES argument.
1143 ** methods with &OPTIONAL arguments no longer allow too many
1144 arguments to be passed in the call without error.
1145 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1146 option is consistent with the required arguments of the generic
1147 function lambda list.
1148 * bug fix: REQUIRE accepts a string designator. (Thanks to
1150 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1151 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1152 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1153 * bug fix: The compiler now checks for duplicated variables in macro
1155 * bug fix: SETQ on globals returns the correct value.
1156 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1157 not a COMPILER-ERROR (followed by some other strange error on
1158 choosing the CONTINUE restart).
1159 * bug fix: make.sh and friends are now more consistent in the way that
1160 they look for GNU "make".
1162 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1163 * experimental native threads support (on x86 Linux >=2.4 only).
1164 This is not compiled in by default: you need to add :SB-THREAD to
1165 the target features. See the "Beyond ANSI" chapter of the manual
1167 * fix for longstanding nonANSIism: The old distinction between
1168 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1169 The return value from CL:FIND-CLASS is now a CLOS class, and
1170 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1171 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1172 * An interface to the MetaObject Protocol, as described in Kiczales,
1173 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1174 MIT Press, 1991, is available from the SB-MOP package.
1175 * incompatible change: the SB-PCL package should now be considered
1176 a private implementation detail, and no longer a semi-private MOP
1178 * minor incompatible change: due to rearrangement for threads, the
1179 control stack and binding stack are now allocated at arbitrary
1180 addresses instead of being hardcoded per-port. Users affected by
1181 this probably have to be doing advanced things with shared
1182 libraries, and will know who they are.
1183 * minor incompatible change: Previously, all --eval forms used were
1184 processed with READ before any of them were processed with EVAL.
1185 Now each --eval form is processed with both READ and EVAL before
1186 the next --eval form is processed. (Thus package operations like
1187 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1188 work as the user might reasonably expect.)
1189 * minor incompatible change: *STANDARD-INPUT* is now only an
1190 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1192 * minor incompatible change: Y-OR-N-P is now character-oriented,
1193 not line oriented. Also, YES-OR-NO-P now works without errors.
1194 (thanks to Antonio Martinez)
1195 * sb-aclrepl module improvements: an integrated inspector, added
1196 repl features, and a bug fix to :trace command.
1197 * Known functions, which cannot be open coded by the backend, are
1198 considered to be able to check types of their arguments. (fixing
1199 a bug report by Nathan J. Froyd)
1200 * fixed a bug in computing method discriminating functions: It is
1201 now possible to define methods specialized on classes which have
1202 forward-referenced superclasses. (thanks to Gerd Moellmann)
1203 * fixed evaluation order in optional entries (reported by Gilbert
1205 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1206 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1207 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1208 required-by-AMOP NAME argument, as well as CLASS and
1209 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1210 * fixed bug 20: DEFMETHOD can define methods using names that are
1211 not the proper names of classes to designate class specializers.
1212 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1213 Unix file descriptors, instead of blowing up. (thanks to Antonio
1215 * Garbage collection refactoring: user-visible change is that a
1216 call to the GC function during WITHOUT-GCING will not do garbage
1217 collection until the end of the WITHOUT-GCING. If you were doing
1218 this you were probably losing anyway.
1219 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1220 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1221 TYPEP the latter but not the former.
1222 * The compiler issues a full WARNING for calls to undefined functions
1223 with names from the CL package.
1224 * MAP-INTO for a vector destination is open coded. (reported by
1225 Brian Downing on c.l.l)
1226 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1227 documentation string.
1228 * fixed some bugs revealed by Paul Dietz' test suite:
1229 ** COPY-ALIST now signals an error if its argument is a dotted
1231 ** Condition slots are now accessed more correctly in the presence
1232 of multiple initargs for a given slot.
1233 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1234 exclude restarts of the same name associated with a different
1236 ** DEFCLASS of forward-referenced classes with another
1237 forward-referenced class in the superclasses list no longer
1239 ** Condition slots are now initialized once each, not multiple
1240 times. (thanks to Gerd Moellmann)
1241 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1243 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1244 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1245 arguments contain duplicated elements.
1246 ** RESTART-CASE understands local macros.
1247 ** RESTART-CASE associates exactly its own restarts with a condition.
1248 ** ENDP in safe mode checks its argument to be of type LIST.
1249 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1250 in question is unbound.
1251 ** Optimized MAKE-INSTANCE functions no longer cause internal
1252 assertion failures in the presence of duplicate initargs.
1253 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1254 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1255 propagated even in the ctor optimized implementation of
1257 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1259 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1260 * a better implementation of SXHASH on (simple) bit vectors,
1261 measured both in execution speed and in distribution of results
1262 over the positive fixnums, has been installed. Likewise, a better
1263 implementation of EQUAL for simple bit vectors is now available.
1264 * fixed CEILING optimization for a divisor of form 2^k.
1265 * fixed bug 240 (emitting extra style warnings "using the lexical
1266 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1267 by Antonio Martinez)
1268 * fixed SXHASH, giving different results for NIL depending on type
1269 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1270 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1271 variable. (found by Rolf Wester)
1272 * SBCL does not ignore type declarations for special
1273 variables. (reported by rif on c.l.l 2003-03-05)
1274 * some bug fixes in contrib/sb-aclrepl/
1275 * fixed some bugs revealed by Paul Dietz' test suite:
1276 ** a bug in the CONS type specifier, whereby the CAR and CDR
1277 types got intertwined, has been fixed;
1278 ** the type system is now able to reason about the interaction
1279 between INTEGER and RATIO types more completely;
1280 ** APPEND, [N]REVERSE and NRECONC check that those their
1281 arguments, which must be proper lists, are really so;
1282 ** An array specialized to be unable to hold elements has been
1283 implemented, as required -- yes, really -- by ANSI;
1284 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1285 on malformed property lists;
1287 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1288 * incompatible packaging change: in line with Unix convention,
1289 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1290 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1291 * REQUIRE and PROVIDE are now optionally capable of doing something
1292 useful. See the documentation string for REQUIRE.
1293 * infrastructure for a managed SBCL contrib system: contributed
1294 modules in this release include:
1295 ** the ASDF system definition facility;
1296 ** an interface to the BSD Sockets API;
1297 ** an ACL-like convenience interface to the repl;
1298 (thanks to Kevin Rosenberg)
1299 ** an implementation of ROTATE-BYTE, with efficient implementation
1301 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1302 gives a closer approximation to the right answer than previously.
1303 (thanks to Raymond Toy)
1304 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1305 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1306 optional environment arguments, as required by ANSI.
1307 * fixed bugs in other functions taking environment objects, allowing
1308 calls with an explicit NIL environment argument to be compiled
1310 * fixed bug 228: primary return values from
1311 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1312 COMPILE or FUNCTION.
1313 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1314 :TYPE will no longer signal errors on innocuous objects.
1315 * fixed bug 231b: SETQ is better at respecting type declarations in
1316 the lexical environment.
1317 * fixed a bug in DEFCLASS: classes named by symbols with no or
1318 unprintable packages can now be defined.
1319 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1320 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1321 * fixed compiler failure related to checking types of functions.
1322 (reported by Robert E. Brown)
1323 * the compiler is now much more consistent in its error-checking
1324 treatment of bounding index arguments to sequence functions: in
1325 (SAFETY 3) code, errors will be signalled in almost all cases if
1326 invalid sequence bounding indices are passed to functions defined
1327 by ANSI to operate on sequences.
1328 * fixed a bug in the build procedure: documentation of SBCL-specific
1329 packages is now preserved and available in the final Lisp image.
1330 * lifted FDEFINITION lookup out of loops in the implementation of
1331 many list operations. (thanks to Robert E. Brown)
1332 * fixed a bug in the reader: the #n# reader macro now works for
1333 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1334 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1335 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1336 E. Brown and Tony Martinez respectively)
1337 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1338 SLOT-BOUNDP: in particular, it is now able to optimize them much
1339 better, and is now not vulnerable to having packages renamed.
1340 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1341 always called when a slot is not present in an instance. (thanks
1343 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1344 longer signal an error on unreading a character following EOF on
1345 the previous constituent stream. (thanks to Tony Martinez)
1346 * fixed some bugs revealed by Paul Dietz' test suite:
1347 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1348 not just nonnegative fixnums;
1349 ** the logical bit-array operators such as BIT-AND now accept an
1350 explicit NIL for their "opt-arg" argument (to indicate a
1351 freshly-consed result bit-array);
1352 ** ELT now signals an error on an invalid sequence index in safe
1354 ** the type system is now cleverer about negations of numeric
1355 types, and consequently understands the BIGNUM and RATIO types
1357 ** the type system is now cleverer about the interaction between
1358 INTEGER and RATIO types: while bugs still remain, many more
1359 cases are accurately computed;
1360 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1361 if it is in the last clause;
1362 ** CONSTANTLY now correctly returns a side-effect-free function in
1364 ** DECLARE is no longer treated as a special-operator; in
1365 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1366 * incremented fasl file version number due to the change in the
1367 DEFSTRUCT-SLOT-DESCRIPTION structure.
1369 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1370 * minor incompatible change: code processed by the "interpreter" or
1371 EVAL now has a compilation optimization policy of (DEBUG 2)
1372 (changed from (DEBUG 1)) to improve debuggability of interactive
1373 development, and to allow the use of the debug RETURN command in
1375 * an experimental implementation of the RETURN command for the
1376 debugger has been included. (thanks to Frederik Kuivinen)
1377 * fixed bug 62: constraints were not propagated into a loop.
1378 * fixed bug in embedded calls of SORT (reported and investigated by
1380 * fixed some bugs revealed by Paul Dietz' test suite:
1381 ** printing and reading of arrays with some dimensions having
1382 length 0 (thanks to Gerd Moellmann);
1383 ** BOA constructor with &AUX argument without a default value does
1384 not cause a type error;
1385 ** CONSTANTP now returns true for all self-evaluating objects.
1387 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1388 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1389 accessors that are related by inheritance, as specified in the
1390 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1391 Valtteri Vuorikoski)
1392 * The compiler is now able to inline functions that were defined in
1393 a complex lexical environment (e.g. inside a MACROLET).
1394 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1396 * Support for the upcoming FreeBSD-5.0 release has been included.
1397 (thanks to Dag-Erling Smorgrav)
1398 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1399 effect when it is not in a toplevel context.
1400 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1401 stand a better chance of being correct. (thanks to Gerd
1403 * fixed bug in COERCE, which now signals an error on coercing a
1404 rational to a bounded real type which excludes the expected
1406 * The compiler is now able to derive types more accurately from the
1407 COERCE and COMPILE functions.
1408 * fixed bug 223: functional binding is considered to be constant
1409 only for symbols in the CL package.
1410 * fixed bug 231: SETQ did not check the type of a variable being set
1411 (reported by Robert E. Brown)
1412 * A new optimization for MAKE-INSTANCE has been included, fixing
1413 various bugs (including relating to :ALLOCATION :CLASS slots and
1414 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1416 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1417 ** As required by ANSI, LOOP now disallows anonymous collection
1418 clauses such as COLLECT I in conjunction with aggregate boolean
1419 clauses such as THEREIS (= I 1);
1420 ** LOOP now signals an error when any variable is reused in the
1421 same loop (including the potentially useful construct analogous
1422 to WITH A = 1 WITH A = (1+ A);
1423 ** IT is only a special loop symbol within the first clause of a
1424 conditional loop clause;
1425 ** LOOP with a typed iteration variable over a hashtable now
1426 signals a type error iff it should.
1427 * fixed some other bugs revealed by Paul Dietz' test suite:
1428 ** FILE-STREAM now names the class previously known as FD-STREAM;
1429 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1430 argument) no longer signals an error;
1431 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1432 of the slot symbol, rather than using the current package
1433 ((:CONC-NAME "") continues to intern the slot's name in the
1435 * incremented fasl file version number, because of the incompatible
1436 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1437 of the new implementation of DEFINE-COMPILER-MACRO.
1439 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1440 * Support for building SBCL for MIPS platforms running in
1441 little-endian mode has now been checked in, and basic
1442 functionality on said platforms verified.
1443 * minor incompatible change: PCL now records the pathname of a file
1444 in which methods and the like are defined, rather than its
1446 * minor incompatible change: TRUENAME now considers the truename of
1447 a file naming a directory to be the pathname with :DIRECTORY
1448 component indicating that directory.
1449 * minor incompatible change: a NAMED clause in the extended form of
1450 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1451 reason for the previous behaviour is unclear.
1452 * more systematization and improvement of CLOS and MOP conformance
1453 in PCL (thanks to Gerd Moellman and Pierre Mai):
1454 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1456 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1457 primary methods with no specializers;
1458 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1460 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1461 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1462 FORWARD-REFERENCED-CLASSes; error reporting on
1463 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1465 ** SXHASH on CLOS instances now uses a slot internal to the
1466 instance to return different numbers on distinct instances,
1467 while preserving the same return value through invocations of
1469 ** DEFMETHOD signals errors when methods with longer incongruent
1470 lambda lists are added to generic functions;
1471 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1472 CLASS, as specified in AMOP;
1473 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1474 on the order returned by the primary method for classes of
1475 class STANDARD-CLASS;
1476 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1477 * fixed some bugs shown by Paul Dietz' test suite:
1478 ** DOLIST puts its body in TAGBODY;
1479 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1481 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1482 value producing form;
1483 ** if more variables are given to PROGV than values, extra
1484 variables are bound and made to have no value;
1485 ** NSUBSTITUTE on list arguments gets the right answer with
1487 ** ELT signals an error of type TYPE-ERROR when the index argument
1488 is not a valid sequence index;
1489 ** LOOP signals (at macroexpansion time) an error of type
1490 PROGRAM-ERROR when duplicate variable names are found;
1491 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1492 ** FUNCALL of special-operators now cause an error of type
1494 ** PSETQ now works as required in the presence of side-effecting
1495 symbol-macro places;
1496 ** NCONC accepts any object as its last argument;
1497 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1499 ** loop-for-as-package does not require a package to be explicitely
1501 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1503 * fixed bug 166: compiler preserves "there is a way to go"
1504 invariant when deleting code.
1505 * fixed bug 172: macro lambda lists with required arguments after
1506 &REST arguments now cause an error to be signalled. (thanks to
1508 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1509 bug. (thanks to Matthew Danish)
1510 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1512 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1513 arguments are lexically rebound. (thanks to Gerd Moellmann and
1515 * fixed bug 194: error messages are now more informative when there
1516 is no primary method applicable in a call to a generic
1517 function. (thanks to Gerd Moellmann)
1518 * fixed bug in command line argument checking (thanks to Julian
1520 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1521 (thanks to Matthew Danish)
1522 * incremented fasl file version number, because of the
1523 SXHASH-related changes in the layout of CLOS data structures
1525 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1526 * minor incompatible change: The runtime (the Unix executable named
1527 "sbcl") is now much pickier about the .core files it will load.
1528 Essentially it now requires .core files to descend from the same
1529 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1530 as the runtime does. (The intent is to prevent the crashes which
1531 can occur, and which can even be reported as mysterious failures,
1532 when people patch the sources or change the build parameters
1533 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1534 sbcl and .core files.)
1535 * fixed bug: VALUES-LIST is no longer optimized away.
1536 * fixed bug 142: The FFI conversion of C string values to Lisp
1537 string values no longer conses excessively. (thanks to Nathan
1538 Froyd porting Raymond Toy's fix to CMU CL)
1539 * began to systematize and improve MOP conformance in PCL (thanks to
1540 Nathan Froyd, Gerd Moellman and Pierre Mai):
1541 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1543 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1544 ** FINALIZE-INHERITANCE is now called on class finalization;
1545 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1546 argument precedence order.
1547 * fixed bug 202: The compiler no longer fails on functions whose
1548 derived types contradict their declared type.
1549 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1550 so it can be non-toplevel.
1551 * The fasl file version number has changed (because of the new
1552 implementation of DEFMACRO).
1553 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1554 safe code, that any length requirement by their type-specifier
1555 argument is valid. The exceptions to this are described in bug
1557 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1558 functions now check the types of their inputs as required by ANSI.
1559 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1560 introduced symbol is DECLAREd to be SPECIAL.
1561 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1562 * fixed bug: :COUNT argument to sequence functions may be negative
1563 * fixed bug: body of DO-SYMBOLS may contain declarations
1564 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1565 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1566 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1567 symbol macro only once
1568 * fixed printing of call frame when argument list is unavailable
1569 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1570 * compiler no longer signals WARNING on unknown keyword
1573 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1574 * A beta-quality port to the mips architecture running Linux,
1575 based on the old CMUCL backend, has been made. It has been tested
1576 on a big-endian kernel, and works sufficiently well to be able to
1577 rebuild itself; it has not been tested in little-endian mode.
1578 * fixed an inconsistency between gencgc.c and purify.c which made
1579 dumping/loading .core files unreliable
1580 * fixed bug 120a: The compiler now deals correctly with IFs where
1581 the consequent is the same as the alternative, instead of
1582 misderiving the return type. (thanks to Alexey Dejneka)
1583 * fixed bug 113: Logical pathnames are now dumpable (the logical
1584 host is resolved at load-time, throwing an error if it is not
1586 * fixed bug 174: FORMAT's error message is slightly clearer when a
1587 non-printing character is used in a format directive.
1588 * fixed several bugs in compiler checking of type declarations, i.e.
1589 violations of the Python "declarations are assertions" principle
1590 (thanks to Alexey Dejneka)
1591 * fixed several bugs in PCL's error checking (thanks to Gerd
1593 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1595 * fixed bug in compilation of functions as first class values
1596 (thanks to Antonio Martinez-Shotton)
1597 * The compiler's handling TYPE-ERRORs which it can prove will
1598 inevitably happen at runtime has been cleaned up and corrected
1599 in several ways. (thanks to Alexey Dejneka)
1600 * improved argument type checking for various basic arithmetic
1601 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1602 lavished on them in the past that they can be compiled in many
1603 ways in different special cases
1604 * fixed bug 181: compiler checks validity of user supplied type
1606 * cleaned up code flushing in optimization: Function calls which
1607 should signal errors for safety purposes (e.g. which ANSI says
1608 should signal errors when their arguments are of incorrect type)
1609 are no longer optimized away.
1610 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1611 * incremented fasl file version number, because changes in the
1612 implementation of sequence functions like COERCE caused
1613 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1614 in old inline expansions) to become undefined. (Actually these
1615 changes were later undone, so we might very well be binary
1616 compatible with 0.7.7 after all, but leaving the version number
1617 incremented seemed like the simplest and most conservative
1620 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1621 * An alpha-quality port to the parisc architecture running Linux,
1622 based on the old CMUCL backend, has been made. This, even more so
1623 than the other backends, should be considered still a work in
1624 progress; known problems include that the Linux kernel in 64-bit
1625 mode does not propagate the correct sigcontext structure to
1626 userspace, and consequently SBCL on a parisc64 kernel will not
1628 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1629 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1630 suppressed.) Also now that the compiler is looking at declarations
1631 in the environment, it checks optimization declarations as well,
1632 and suppresses inlining when (> DEBUG SPEED).
1633 * More fixes have been made to treatment of floating point exception
1634 treatment and other Unix signals. In particular, floating point
1635 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1636 * The detection and handling of control stack exhaustion (infinite
1637 or very deeply nested recursion) has changed. Stack exhaustion
1638 detection is now done by write-protecting pages at the OS level
1639 and applies at all optimization settings; when found, a
1640 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1641 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1642 be caught using IGNORE-ERRORS.
1643 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1644 correctly when transforming recursive LABELS functions to LETs.
1645 (thanks to Alexey Dejneka)
1646 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1647 that are names of constants or global variables.
1648 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1649 alien routines with docstrings.
1650 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1651 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1653 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1654 (thanks to Eric Marsden)
1655 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1656 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1657 * Bug fix: The SPARC backend can now compile functions involving
1658 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1659 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1660 to a functions expecting a PATHNAME-DESIGNATOR.
1661 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1662 lambda lists. (thanks to Alexey Dejneka)
1663 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1664 functions so that EXPORTing the name of the class doesn't cause
1665 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1666 to Antonio Martinez for reporting this)
1667 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1668 constant-folded without failing an assertion. (thanks to Einar
1669 Floystad Dorum for reporting this)
1670 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1671 types are now dealt with more correctly by the compiler.
1672 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1673 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1674 behaviour analogous to RENAME-FILE. This puts its behaviour more
1675 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1677 * The fasl file version number has changed again. (because of the
1678 bug fix involving the names of PCL MAKE-INSTANCE functions)
1680 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1681 * bug fix: Floating point exceptions are treated much more
1682 consistently on the x86/Linux and PPC/Linux platforms.
1683 * Array initialization with :INITIAL-ELEMENT is now much faster for
1684 cases when the compiler cannot open code the array creation, but
1685 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1686 array accesses have also seen a speed increase.
1687 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1688 specified. (This was at the root of some bad interactions between
1689 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1691 * bug fix: Internal error arguments for undefined functions are now
1692 computed correctly on the PPC/Linux platform.
1693 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1694 Raymond Toy's patch for CMU CL)
1695 * Support for the Solaris 9 operating environment has been included
1696 (thanks to Daniel Merritt)
1697 * A very ugly but hopefully complete draft of the missing FFI chapter
1698 of the manual has been created by reformatting the corresponding
1699 CMU CL manual chapter into (currently very ugly and incoherent)
1700 DocBook and bringing it up to date for SBCL behavior. Thus, the
1701 manual is now essentially complete, at least by my extreme
1702 once-and-only-once standards, whereby it's acceptable to refer to
1703 the doc strings of SB-EXT functions as the primary documentation.
1704 * The fasl file version number has changed again, due to cleanup of
1705 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1706 is no longer a static symbol.)
1708 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1709 * SBCL now builds with OpenMCL (version 0.12) as the
1710 cross-compilation host; also, more progress has been made toward
1711 bootstrapping under CLISP.
1712 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1714 * bug 158 fixed: The compiler can now deal with integer loop
1715 increments different from 1; fixing this turned out also to fix
1717 * bug 169 fixed: no more bogus warnings about using lexical bindings
1718 despite the presence of perfectly good SPECIAL declarations (thanks
1719 to David Lichteblau)
1720 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1721 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1722 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1724 * bug 180 fixed: Method combination specifications no longer ignore
1725 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1726 * bug fix: Structure type predicate functions now check their argument
1727 count as they should.
1728 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1729 correctly. (thanks to Pierre Mai)
1730 * minor incompatible change: The --noprogrammer option is deprecated
1731 in favor of the new --disable-debugger option, which is very similar.
1732 (The major difference is that it takes effect at a slightly different
1733 time at startup, causing handling of errors in --sysinit and
1734 --userinit files will be affected differently.) The
1735 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1736 been added to allow this functionality to be controlled from ordinary
1737 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1738 maintainers, who might want to run non-interactive scripts to
1739 build SBCL cores which will later be used interactively.)
1740 * minor incompatible change: The LOAD function no longer, when given
1741 a wild pathname to load, loads all files matching that pathname.
1742 Instead, an error of type FILE-ERROR is signalled.
1744 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1745 * bug 147 fixed: The compiler preserves its block link/count
1746 invariants more correctly now so that it doesn't crash. (thanks
1748 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1750 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1751 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1752 no-value-for-this-character as NIL (as specified by ANSI).
1753 * HOST-NAMESTRING on physical pathnames now returns a string that is
1754 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1755 (thanks to Christophe Rhodes)
1756 * The Alpha port handles icache flushing more correctly. (thanks to
1758 * More progress has been made toward bootstrapping under CLISP. (thanks
1759 to Christophe Rhodes)
1760 * The fasl file format has changed again, because dynamic loading
1761 on OpenBSD (which has non-ELF object files) motivated some cleanups
1762 in the way that foreign symbols are transformed and passed around.
1763 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1764 is no longer treated as whitespace by the reader, but instead as
1765 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1766 |A
\7fB|, instead of A as it used to.
1768 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1769 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1770 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1771 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1772 this as of 0.7.1.45, but was left out of the previous news section
1773 (thanks to Dan Barlow)
1774 * SBCL now runs on the Solaris operating system on SPARC architectures
1775 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1776 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1777 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1778 * SPARC backend cleanups, allowing builds of cores optimized for V8
1779 and V9 SPARCS, and also emission of code targeted to a particular
1780 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1782 * SBCL is closer to bootstrapping under CLISP, thanks to various
1783 fixes by Christophe Rhodes.
1784 * The fasl file format has changed again, to allow the compiler's
1785 INFO database to support symbol macros.
1786 * The user manual (in doc/) is formatted into HTML more nicely.
1787 (thanks to coreythomas)
1788 * The system is smarter about SUBTYPEP relationships, especially
1789 those involving NOT types (including types such as ATOM which are
1790 represented internally using NOT types). Thus SUBTYPEP is less
1791 likely to return (VALUES NIL NIL) in general, and in particular
1792 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1794 * The fasl file format has changed again, because the internal
1795 representation of types now includes a new slot to support the new
1796 SUBTYPEP-of-NOT-types logic.
1797 * (not a change in the main branch of SBCL, but a related prototype
1798 which can hopefully be merged into the main branch of SBCL in the
1799 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1800 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1801 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1802 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1803 dispatch macro character combinations. (thanks to Alexey Dejneka)
1804 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1805 that superficially look like logical namestrings correctly.
1806 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1808 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1809 default supplied-p) for &optional and &key arguments. (thanks to
1812 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1813 * incompatible change: The compiler is now less aggressive about
1814 tail call optimization, doing it only when (> SPACE DEBUG) or
1815 (> SPEED DEBUG). (This is an incompatible change because there are
1816 programs which relied on the old CMU-CL-style behavior to optimize
1817 away their unbounded recursion which will now die of stack overflow.)
1818 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1819 for non-GENCGC systems has been increased to 20M (since that
1820 seems much closer to the likely performance optimum for modern
1821 systems than the old 4M value was)
1822 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1823 and *DEBUG-PRINT-LEVEL*
1824 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1825 of CMU CL's support for SPARC, and various endianness and other
1826 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1827 * new syntactic sugar for the Unix command line: --load foo.bar is now
1828 an alternate notation for --eval '(load "foo.bar")'.
1830 ** The system now detects stack overflow and handles it gracefully,
1831 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1832 optimization settings. (This is a good thing in general, and
1833 its introduction in this version should be particularly timely
1834 for anyone whose code fails because of suppression of tail
1836 ** The system now hunts for the C variable "environ" in a more
1837 devious way, to avoid segfaults when the C library version differs
1838 between compile time and run time. (thanks to Christophe Rhodes)
1839 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1840 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1841 ** The compiler no longer issues bogus style warnings for undefined
1842 classes in the same source file as the DEFCLASSes which defined
1843 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1845 ** fixes in CONDITION class precedence list for undefined function
1846 errors (thanks to Alexei Dejneka)
1847 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1848 correctly. (thanks to Dan Barlow)
1849 ** portability fixes aiming at bootstrapping under CLISP (thanks
1850 to Dave McDonald and Christophe Rhodes)
1851 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1852 ** fixes in type translation and and type inference (thanks to
1854 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1855 ** various fixes in the new ports (thanks to Dan Barlow)
1856 * several changes related to debugging:
1857 ** suppression of tail recursion, as noted above
1858 ** stack overflow detection, as noted above
1859 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1860 is now the default. (For some time encapsulation has been more
1861 reliable than the breakpoint-based :ENCAPSULATE NIL
1862 implementation, at least on X86 systems; and I just noticed that
1863 encapsulation also seems closer to the spirit of the ANSI
1866 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1868 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1869 up properly again. (There was a packaging bug in 0.7.0 which
1870 left their definitions in SB-SYS::LOAD-FOREIGN and
1871 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1872 vital for most things which interface to C-level interfaces,
1873 like extensions working with sockets or databases or
1874 Perl-compatible regexes or whatever, and the need to fix
1875 this bug is the main reason that 0.7.1 was released so
1877 ** DEFGENERIC is now choosier about the methods it redefines, so that
1878 reLOADing a previously-LOADed file containing DEFGENERICs does
1879 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1880 cycle now works as it should. (thanks to Alexey Dejneka)
1881 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1882 fixed by Christophe Rhodes. (He actually submitted this patch
1883 months ago, and I delayed until after 0.7.0.)
1884 ** Bug 111 (internal compiler confusion about runtime checks on
1885 FUNCTION types) was fixed by Alexey Dejneka.
1886 * Some internal cleanups (getting rid of variables which aren't
1887 needed now that the byte interpreter is gone) caused the fasl
1888 file format number to change again.
1890 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1891 * major incompatible change: The default fasl file extension, i.e. the
1892 default extension for files produced by COMPILE-FILE, has changed
1893 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1895 ** There are many changes in the implementation of the compiler.
1896 SBCL is now essentially a compiler-only implementation of ANSI
1897 Common Lisp. EVAL still "interprets" a few special cases, but
1898 almost all the interesting cases are handled by creating
1899 a LAMBDA expression, calling COMPILE on it, then calling
1900 FUNCALL on the result.
1901 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1902 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1903 Since the code is newer, there might still be some new bugs
1904 (though not as many as before Martin Atzmueller's fixes:-). But
1905 the new code is substantially simpler and clearer, and hopefully
1906 any remaining bugs will be simpler, less fundamental, and more
1907 fixable then the bugs in the old code.
1908 ** The revised compiler is still a little unsteady on its feet.
1910 *** The debugging information it produces (particularly the names
1911 of FUNCTION objects) is sometimes much less useful than what
1912 the old compiler produced.
1913 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1914 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1915 within a MACROLET) has been temporarily weakened.
1916 ** There are new compiler optimizations for various functions:
1917 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1918 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1919 *** the math functions TRUNCATE, FLOOR, and CEILING
1920 *** the function-of-all-trades COERCE
1921 Mostly these should be transparent, but there's one
1922 potentially-annoying problem (bug 117): when the compiler
1923 inline-expands a function and does type analysis on the result,
1924 it can create control paths which have type mismatches, and
1925 when it can't prove that those control paths aren't taken,
1926 it will issue WARNINGs about the type mismatches. This is
1927 a particular problem in practice for the new sequence functions.
1928 It's not clear how this should be fixed, and for now, a
1929 workaround is given in the entry for 117 in the BUGS file.
1930 ** (Because of the interaction between the two previous items --
1931 occasional inlining problems and new inline expansions -- some
1932 of the new sequence function optimizations won't really kick in
1933 completely until debugging information, and then inlining, are
1934 straightened out in some future version.)
1935 * minor incompatible changes:
1936 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1937 DIRECTORY no longer implicitly promotes NIL slots of its
1938 pathname argument to :WILD. In particular, when you ask for the
1939 contents of a directory (which you used to be able to do without
1940 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1941 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1942 ** changes in behavior that ANSI explicitly defines to be
1943 implementation dependent:
1944 *** The new compiler-only implementation still conforms with ANSI,
1945 but acts a little different than before. Besides the obvious
1946 changes in performance tradeoffs (that the cost per form passed
1947 to EVAL has gone up, and the cost per form executed by EVAL
1948 has gone down), the behavior of the system changes a little
1949 because there are no longer any interpreted function objects.
1950 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1951 e.g. doing COMPILE on the output of interactive DEFUN is
1953 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1955 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1956 has changed from (:CL) to NIL.
1957 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1958 e.g. #\Newline and #\Tab, don't have names specified in the
1959 ANSI Common Lisp standard, is now based on their ASCII symbolic
1960 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1961 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1962 are no longer used for output.
1963 ** changes in internal implementation constants:
1964 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1965 4 million. (If your application spends a lot of time GCing and
1966 you have a lot of RAM, you might want to experiment with
1967 increasing it even more.)
1968 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1969 However, almost all old code should still continue to work without
1970 immediate update, as SB-C-CALL is now a (deprecated) nickname
1972 ** Old operator names in the style DEF-FOO are now deprecated in
1973 favor of new corresponding names DEFINE-FOO, for consistency with
1974 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1975 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1976 internal symbols, but a few supported extensions like
1977 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1978 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1979 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1980 etc. as you get deeper into recursive calls to the debugger
1981 command loop, instead of the old "5]", "5]]", "5]]]"
1982 sequence. (I was motivated to do this when squabbles between
1983 ILISP and SBCL left me very deeply nested in the debugger. In the
1984 short term, this change will probably provoke more ILISP/SBCL
1985 squabbles, but hopefully it will be an improvement in the long run.)
1986 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1987 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1988 function, since declaiming return types involving aliens is
1989 (1) annoyingly messy to do by hand and (2) vital to efficient
1990 compilation of code which calls such functions.
1991 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1992 longer reexported by the SB-EXT package. They're solely useful
1993 for alien code, so it seems more logical that you should get
1994 them from the SB-ALIEN package, not in SB-EXT.
1995 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1996 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1997 features. Instead, the code that they used to control is always
1998 built into the system.
1999 * many other bug fixes
2000 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2001 advantage of the new EVAL-WHEN stuff and to clean them up in
2002 general, and they are now more ANSI-compliant in a number of
2003 ways. Martin Atzmueller is responsible for a lot of this.
2004 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2006 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2007 *** correct ERROR type for various file operations
2008 *** some fixes for Lisp streams
2009 *** DEFMETHOD syntax checking
2010 *** changing old weird representation of debug information as
2011 strings (which, among their other deficiencies, don't transform
2012 correctly when you rename packages, and don't change their
2013 print representation when you change things like *PACKAGE*
2014 and *PRINT-LENGTH*) to symbols and lists of symbols
2015 He also made several improvements and fixed several bugs in DESCRIBE.
2016 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2018 *** misbehavior of WRITE-STRING/WRITE-LINE
2019 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2020 and several other LOOP problems as well
2021 *** DIRECTORY when similar filenames are present
2022 *** DEFGENERIC with :METHOD options
2023 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2024 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2025 *** argument ordering in FIND with :TEST option
2026 *** mishandled package designator argument in APROPOS-LIST
2027 *** various problems in the backquote readmacro
2028 *** a bug in APROPOS
2029 *** probably some others that I'm not describing very well here,
2030 since the CVS log documents them by reference to sbcl-devel
2031 messages, and the SourceForge archives aren't working well.:-(
2032 ** Dan Barlow improved the Alpha port (and is making progress on the
2033 PPC port, for those of you who think different).
2034 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2035 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2036 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2037 continues to work on the SPARC port (for those of you in a position
2038 to look down upon our little PC-compatible boxes from a great height).
2039 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2040 and uses that copy, rather than the current dynamic value, when
2041 it's trying to decide whether to truncate output. Thus e.g.
2042 (let ((*print-lines* 50))
2043 (pprint-logical-block (stream nil)
2045 (let ((*print-lines* 8))
2046 (print (aref possiblybigthings i) stream)))))
2047 should now truncate the logical block only at 50 lines, instead of
2048 often truncating it at 8 lines, as it did before.
2049 * The doc/cmucl/ directory, containing old CMU CL documentation
2050 from the time of the fork, is no longer part of the base system.
2051 SourceForge has shut down its anonymous FTP service, and with it
2052 my original plan for distributing the old CMU CL documentation
2053 there. For now, if you need these files you can download an old
2054 SBCL source release and extract them from it.
2055 * The fasl file version number changed again, for dozens of reasons,
2056 some of which are apparent above.
2058 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2059 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2060 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2061 patch, so that DISASSEMBLE output is much nicer.
2062 * The code in the SB-PROFILE package now seems reasonably stable.
2063 I still haven't decided what the final interface should look like
2064 (I'd like PROFILE to interact cleanly with TRACE, since both
2065 facilities use function encapsulation) but if you have a need
2066 for profiling now, you can probably use it successfully with
2067 the current CMU-CL-style interface.
2068 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2069 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2070 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2071 link now returns the dangling link itself, and for similar
2072 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2073 link itself. (In these cases the old code signalled an error and
2074 looped endlessly, respectively.) Thus, DIRECTORY now works even
2075 in the presence of dangling and cyclic symbolic links.
2076 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2077 is now a supported extension again, since the consensus on
2078 sbcl-devel was that it can be useful for ordinary development
2079 work, not just for debugging SBCL itself.
2080 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2081 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2082 that currently-defined functions might be redefined later with
2083 different return types.
2084 * Hash tables can be printed readably, as inspired by CMU CL code
2085 of Eric Marsden and SBCL code of Martin Atzmueller.
2086 * better error handling in CLOS method combination, thanks to
2087 Martin Atzmueller porting Pierre Mai's CMU CL patches
2088 * more overflow fixes for >16Mbyte I/O buffers
2089 * A bug in READ has been fixed, so that now a single Ctrl-D
2090 character suffices to cause end-of-file on character streams.
2091 In particular, now you only need one Ctrl-D at the command
2092 line (not two) to exit SBCL.
2093 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2095 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2096 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2097 does the right thing.
2098 * The implementation of some type tests, especially for CONDITION
2099 types, is now tidier and maybe faster, due to CMU CL code
2100 originally by Douglas Crosher, ported by Martin Atzmueller.
2101 * Some math functions have been fixed, and there are new
2102 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2103 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2104 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2105 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2106 likely to be built into sbcl-0.7.0.)
2107 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2108 CL, ported to SBCL by Martin Atzmueller)
2109 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2110 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2111 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2112 * There's a new slam.sh hack to shorten the edit/compile/debug
2113 cycle for low-level changes to SBCL itself, and a new
2114 :SB-AFTER-XC-CORE target feature to control the generation of
2115 the after-xc.core file needed by slam.sh.
2116 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2117 COMPILE-FILE is no longer supported, so that now every function
2118 gets an entry point, so that block compilation looks a little
2119 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2120 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2121 returns the number of bytes consed since the system started,
2122 rather than the number consed since the first time the function
2123 was called. (The new definition parallels ANSI functions like
2124 CL:GET-INTERNAL-RUN-TIME.)
2125 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2126 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2127 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2128 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2129 * Fasl file version numbers are now independent of the target CPU,
2130 since historically most system changes which required version
2131 number changes have affected all CPUs equally. Similarly,
2132 the byte fasl file version is now equal to the ordinary
2135 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2136 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2137 is no longer recognized. I apologize for this, because it was
2138 listed in SB-EXT as a supported extension, but I found that
2139 its existing behavior was poorly specified, as well as incorrectly
2140 specified, and it looked like too much of a mess to straighten it
2141 out. I have enough on my hands trying to get ANSI stuff to work..
2142 * many patches ported from CMU CL by Martin Atzmueller, with
2143 half a dozen bug fixes in pretty-printing and the debugger, and
2144 half a dozen others elsewhere
2145 * fixed bug 13: Floating point infinities are now supported again.
2146 They might still be a little bit flaky, but thanks to bug reports
2147 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2148 as flaky as they were.
2149 * The --noprogrammer command line option is now supported. (Its
2150 behavior is slightly different in detail from what the old man
2151 page claimed it would do, but it's still appropriate under the
2152 same circumstances that the man page talks about.)
2153 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2154 are now supported, and enabled by default. Thus, the compiler can
2155 handle many floating point and complex operations much less
2156 inefficiently. (Thus e.g. you can implement a complex FFT
2158 * The compiler now detects type mismatches between DECLAIM FTYPE
2159 and DEFUN better, and implements CHECK-TYPE more correctly, and
2160 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2161 thanks to patches from Martin Atzmueller.
2162 * various fixes to make the cross-compiler more portable to
2163 ANSI-conforming-but-different cross-compilation hosts (notably
2164 Lispworks for Windows, following bug reports from Arthur Lemmens)
2165 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2166 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2167 fixed thanks to Pierre Mai's CMU CL patches.
2168 * fixes to keep the system from overflowing internal counters when
2169 it tries to use i/o buffers larger than 16M bytes
2170 * fixed bug 45a: Various internal functions required to support
2171 complex special functions have been merged from CMU CL sources.
2172 (When I was first setting up SBCL, I misunderstood a compile-time
2173 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2174 * improved support for type intersection and union, fixing bug 12
2175 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2176 more obscure bugs as well
2177 * some steps toward byte-compiling non-performance-critical
2178 parts of the system, courtesy of patches from Martin Atzmueller
2179 * Christophe Rhodes has made some debian packages of sbcl at
2180 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2181 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2182 stable, but are nonetheless usable. When he's ready, I'd be happy
2183 to add them to the SourceForge "File Releases" section. (And if
2184 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2185 * new fasl file format version number (because of changes in
2186 internal representation of (OR ..) types to accommodate the new
2187 support for (AND ..) types, among other things)
2189 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2190 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2192 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2194 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2195 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2196 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2197 * DESCRIBE now works on structure objects again.
2198 * Most function call argument type mismatches are now handled as
2199 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2200 know whether the function will be redefined before the call is
2201 executed. (The compiler could flag local calls with full WARNINGs,
2202 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2203 it doesn't keep track of enough information to know whether calls
2204 are local in this sense.)
2205 * Compiler output is now more verbose, with messages truncated
2206 later than before. (There should be some supported way for users
2207 to override the default verbosity, but I haven't decided how to
2208 provide it yet, so this behavior is still controlled by the internal
2209 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2210 src/compiler/ir1util.lisp.)
2211 * Fasl file format version numbers have increased again, because
2212 support for the Gray streams extension changes the layout of the
2213 system's STREAM objects.
2214 * The Gray subclassable streams extension now works, thanks to a
2215 patch from Martin Atzmueller.
2216 * The full LOAD-FOREIGN extension (not just the primitive
2217 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2218 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2219 but like most other programs, it defaults to copying the Unix
2220 environment from the original process instead of starting the
2221 new process in an empty environment.
2222 * Extensions which manipulate the Unix environment now support
2223 an :ENVIRONMENT keyword option which doesn't smash case or
2224 do other bad things. The CMU-CL-style :ENV option is retained
2225 for porting convenience.
2226 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2227 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2229 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2231 * A patch from Martin Atzmueller seems to have solved the SIGINT
2232 problem, and as far as we know, signal-handling now works cleanly.
2233 (If you find any new bugs, please report them!)
2234 * The system no longer defaults Lisp source file names to types
2235 ".l", ".cl", or ".lsp", but only to ".lisp".
2236 * The compiler no longer uses special default file extensions for
2237 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2238 seems to expect a single default extension for all compiled code,
2239 and there's no compelling reason to try to stretch the standard
2240 to allow two different extensions.) Instead, byte-compiled files
2241 default to the same extension as native-compiled files.
2242 * Fasl file format version numbers have increased again, because
2243 a rearrangement of internal implementation packages made some
2244 dumped symbols in old fasl files unreadable in new cores.
2245 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2246 many fewer weird special cases.
2247 * Bug #17 (differing COMPILE-FILE behavior between logical and
2248 physical pathnames) has been fixed, and some related misbehavior too,
2249 thanks to a patch from Martin Atzmueller.
2250 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2251 by Tim Moore, ported to SBCL by Martin Atzmueller.
2252 * Martin Atzmueller fixed several filesystem-related problems,
2253 including bug #36, in part by porting CMU CL patches, which were
2254 written in part by Paul Werkowski.
2255 * More compiler warnings in src/runtime/ are gone, thanks to
2256 more patches from Martin Atzmueller.
2257 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2260 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2262 * DESCRIBE now works on CONDITION objects.
2263 * The debugger now handles errors which arise when trying to print
2264 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2266 * The build system now uses an additional file, customize-target-features.lisp,
2267 to allow local modifications to the target *FEATURES* list. (The point of
2268 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2269 debugging features enabled, without having to worry about propagating it
2270 into everyone's system when I do a "cvs update".) When no
2271 customize-target-features.lisp file exists, the target *FEATURES* list
2272 should be constructed the same way as before.
2273 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2274 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2275 warn about behavior which is undefined under the ANSI spec. Note: This
2276 is specified by ANSI, but it's not very popular with programmers.
2277 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2278 macro in the SBCL sources for an example of a workaround which you
2279 might use to make portable ANSI-standard code which does what you want.
2280 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2281 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2282 when it's used as a non-toplevel form. (This is required in order
2283 to implement the DEFCONSTANT-EQX macro.)
2284 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2285 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2286 funny things, due to bugs in the implementation of EVAL-WHEN
2287 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2288 non-toplevel DEFCONSTANTs are uncommon.)
2289 * The core file version number and fasl file version number have been
2290 incremented, because the old noncompliant DEFCONSTANT behavior involved
2291 calling functions which no longer exist, and because I also took the
2292 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2293 * fixed bug 1 (error handling before read-eval-print loop starts), and
2294 redid debugger restarts and related debugger commands somewhat while
2296 ** The QUIT debugger command is gone, since it did something
2297 rather different than the SB-EXT:QUIT command, and since it never
2298 worked properly outside the main toplevel read/eval/print loop.
2299 Invoking the new TOPLEVEL restart provides the same functionality.
2300 ** The GO debugger command is also gone, since you can just invoke
2301 the CONTINUE restart directly instead.
2302 ** The TOP debugger command is also gone, since it's redundant with the
2303 FRAME 0 command, and since it interfered with abbreviations for the
2305 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2307 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2308 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2309 some porting work by Martin Atzmueller)
2310 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2311 with a boatload of patches which he ported to SBCL. Now that those
2313 ** The system tries to make sure that its low-priority messages
2314 are prefixed by semicolons, to help people who like to use
2315 syntax highlighting in their ILISP buffer. (This patch
2316 was originally due to Raymond Toy.)
2317 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2318 patches originally written by Raymond Toy.
2319 ** The compiler understands coercion between single-value and
2320 multiple-VALUES type expressions better, getting rid of some very
2321 weird behavior, thanks to patches originally by Robert MacLachlan
2322 and Douglas Crosher.
2323 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2324 lambda lists, thanks to a patch originally by Pierre Mai.
2325 ** The system no longer bogusly warns about "abbreviated type
2327 ** The compiler gets less confused by inlining and RETURN-FROM,
2328 thanks to some patches originally by Tim Moore.
2329 ** The system no longer hangs when dumping circular lists to fasl
2330 files, thanks to a patch originally from Douglas Crosher.
2331 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2332 undefined function error.
2333 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2334 in structure slot initforms) for now, documented workaround instead:-|
2335 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2336 * fixed bug 5: added stubs for various Gray stream functions called
2337 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2338 installed, at least appropriate type errors are generated
2339 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2340 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2341 consistently in DEFMETHOD forms.
2342 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2343 it doesn't seem to affect SBCL after all
2344 * The C runtime system now builds with better optimization and many
2345 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2347 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2349 * The system is now under CVS at SourceForge (instead of the
2350 CVS repository on my home machine).
2351 * The new signal handling code has been tweaked to treat register
2352 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2353 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2354 downstream, e.g. in debug-int.lisp, has implicit dependencies
2355 on the unsignedness of integer representation of machine words,
2356 and that caused the system to bomb out with infinite regress
2357 when trying to recover from type errors involving signed values,
2358 e.g. (BUTLAST '(1 2 3) -1).)
2359 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2360 (This was one of the bugs Peter Van Eynde reported back in July.)
2361 * The system now uses code inspired by Colin Walters' O(N)
2362 implementation of MAP (from the cmucl-imp@cons.org mailing
2363 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2364 inline the MAP operation, and there is more than one
2365 sequence argument to the MAP call (so that it can't just
2366 do ETYPECASE once and for all based on the type of the
2367 single sequence argument). (The old non-inline implementation
2368 of the general M-argument sequence-of-length-N case required
2369 O(M*N*N) time when any of the sequence arguments were LISTs.)
2370 * The QUIT :UNIX-CODE keyword argument has been renamed to
2371 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2372 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2374 * Raymond Wiker's patches to port dynamic loading from Linux to
2375 FreeBSD have been added.
2376 * The BUGS file is now more nearly up to date, thanks in large part
2377 to Martin Atzmueller's review of it.
2378 * The debugger now flushes standard output streams before it begins
2379 its output ("debugger invoked" and so forth).
2380 * The core version number and fasl file version number have both
2381 been incremented, because of incompatible changes in the layout
2383 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2384 used to not be called for a saved Lisp image.)
2385 * Martin Atzmueller's version of a patch to fix a compiler crash,
2386 as posted on sbcl-devel 13 September 2000, has been installed.
2387 * Instead of installing Martin Atzmueller's patch for the
2388 compiler transform for SUBSEQ, I deleted the compiler transform,
2389 and transforms for some similar consing operations.
2390 * A bug in signal handling which kept TRACE from working on OpenBSD
2392 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2393 away by constant folding
2394 * The system now defines its address space constants in one place
2395 (in the Lisp sources), and propagates them automatically elsewhere
2396 (through GENESIS and the sbcl.h file). Therefore, patching the
2397 address map is less unnecessarily tedious and error-prone. The
2398 Lisp names of address space constants have also been systematized.
2399 * CVS tags like dollar-Header-dollar have been removed from
2400 the sources, because they have never saved me trouble and
2401 they've been source of trouble working with patches and other
2402 diff-related operations.
2403 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2404 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2406 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2408 * The system has been ported to OpenBSD.
2409 * The system now compiles with a simple "sh make.sh" on the systems
2410 that it's supported on. I.e., now you no longer need to tweak
2411 text in the target-features.lisp-expr and symlinks in src/runtime/
2412 by hand, the make.sh takes care of it for you.
2413 * The system is no longer so grossly inefficient when compiling code
2414 involving vectors implemented as general (not simple) vectors (VECTOR T),
2415 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2416 which dares to use the various sequence functions on non-simple
2417 vectors, takes less of a performance hit.
2418 * There is now a primitive type predicate VECTOR-T-P
2419 to test for the (VECTOR T) type, so that e.g.
2420 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2421 can now be compiled with some semblance of efficiency. (The old code
2422 turned the type declaration into a full call to %TYPEP at runtime!)
2423 * AREF on (VECTOR T) is still not fast, since it's still compiled
2424 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2425 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2426 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2427 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2428 without making full calls to SUBTYPEP at runtime.
2429 (Some analogous efficiency issues for non-simple vectors specialized to
2430 element types other than T, or for non-simple multidimensional arrays,
2431 have not been addressed. They could almost certainly be handled the
2432 same way if anyone is motivated to do so.)
2433 * The changes in array handling break binary compatibility, so
2434 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2435 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2436 instead of (VALUES T T).
2437 * By following the instructions that Dan Barlow posted to sbcl-devel
2438 on 2 July 2000, I was able to enable primitive dynamic object
2439 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2440 functionality is not implemented (since it calls ld to resolve
2441 library references automatically, requiring RUN-PROGRAM for its
2442 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2443 try to resolve library references) is now supported.
2444 * The system now flushes the standard output streams when it terminates,
2445 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2446 them at several other probably-convenient times, e.g. in each pass of
2447 the toplevel read-eval-print loop, and after evaluating a form given
2448 as an "--eval" command-line option. (These changes were motivated by a
2449 discussion of stream flushing issues on cmucl-imp in August 2000.)
2450 * The source transform for TYPEP of array types no longer assumes
2451 that an array whose element type is a not-yet-defined type
2452 is implemented as an array of T, but instead punts, so that the
2453 type will be interpreted at runtime.
2454 * There is now some support for cross-compiling in make.sh: each of
2455 the phases of make.sh has its own script. (This should be transparent
2456 to people doing ordinary, non-cross-compile builds.)
2457 * Since my laptop doesn't have hundreds of megabytes of memory like
2458 my desktop machine, I became more motivated to do some items on
2459 my to-do list in order to reduce the size of the system a little:
2460 ** Arrange for various needed-only-at-cold-init things to be
2461 uninterned after cold init. To support this, those things have
2462 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2463 symbols with such names are now uninterned after cold init).
2464 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2465 fasl files for cold load.
2466 ** Remove the old compiler structure pooling code (which used to
2467 be conditional on the target feature :SB-ALLOC) completely.
2468 ** Redo the representation of some data in cold init to be more compact.
2469 (I also looked into supporting byte compiled code at bootstrap time,
2470 which would probably reduce the size of the system a lot, but that
2471 looked too complicated, so I punted for now.)
2472 * The maximum signal nesting depth in the src/runtime/ support code has
2473 been reduced from 4096 to 256. (I don't know any reason for the very
2474 large old value. If the new smaller value turns out to break something,
2475 I'll probably just bump it back up.)
2476 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2478 * Many, many bugs reported by Peter Van Eynde have been added to
2479 the BUGS list; some have even been fixed.
2480 * While enabling dynamic object file loading, I tried to make the
2481 code easier to understand, renaming various functions and variables
2482 with less ambiguous names, and changing some function calling
2483 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2484 * While trying to figure out how to do the OpenBSD port, I tried to
2485 clean up some of the code in src/runtime/. In particular, I dropped
2486 support for non-POSIX signal handling, added various comments,
2487 tweaked the code to reduce the number of compilation warnings, and
2488 renamed some files to increase consistency.
2489 * To support the new automatic configuration functionality in make.sh,
2490 the source file target-features.lisp-expr has been replaced with the
2491 source file base-target-features.lisp-expr and the machine-generated
2492 file local-target-features.lisp-expr.
2493 * fixed a stupid quoting error in make.sh so that using CMU CL
2494 "lisp -batch" as cross-compilation host works again
2496 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2498 * DESCRIBE no longer tries to call itself recursively to describe
2499 bound/fbound values, so that it no longer fails on symbols which are
2500 bound to themselves (like keywords, T, and NIL).
2501 * DESCRIBE now works on generic functions.
2502 * The printer now prints less-screwed-up representations of closures
2503 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2504 * A private symbol is used instead of the :EMPTY keyword previously
2505 used to mark empty slots in hash tables. Thus
2506 (DEFVAR *HT* (MAKE-HASH-TABLE))
2507 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2508 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2509 now does what ANSI says that it should. (You can still get
2510 similar noncompliant behavior if bang on the hash table
2511 implementation with all the symbols you get back from
2512 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2513 This breaks binary compatibility, since tests for equality to
2514 :EMPTY are wired into things like the macroexpansion of
2515 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2517 * There's now a minimal placeholder implementation for CL:STEP,
2518 as required by ANSI.
2519 * An obscure bug in the interaction of the normal compiler, the byte
2520 compiler, inlining, and structure predicates has been patched
2521 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2522 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2524 * Missing ordinary arguments in a macro call are now detected even
2525 when the macro lambda list contains &KEY or &REST.
2526 * The debugger no longer complains about encountering the top of the
2527 stack when you type "FRAME 0" to explicitly instruct it to go to
2528 the top of the stack. And it now prints the frame you request even
2529 if it's the current frame (instead of saying "You are here.").
2530 * As specified by ANSI, the system now always prints keywords
2531 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2533 * The default initial SIZE of HASH-TABLEs is now smaller.
2534 * Type information from CLOS class dispatch is now propagated
2535 into DEFMETHOD bodies, so that e.g.
2536 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2538 is now basically equivalent to
2539 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2540 (DECLARE (TYPE SINGLE-FLOAT X))
2542 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2543 operation, without having to do run-time type dispatch.
2544 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2545 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2546 * Since I don't seem to be making big file reorganizations very often
2547 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2548 is overflowing my ability to conveniently back them up), I've finally
2549 checked the system into CVS. (The CVS repository is on my home system,
2550 not at SourceForge -- putting it on SourceForge might come later.)
2551 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2552 high-level GC-NOTIFY-FOO functions send their output. (There's
2553 still very little control of where low-level verbose GC functions
2554 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2555 less than it used to -- the GC-NOTIFY-FOO functions are now under
2556 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2557 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2558 in only one place in the source code, and propagates it automatically
2559 everywhere that it's needed. Thus e.g. when I bump the version from
2560 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2561 * The C source files now include boilerplate legalese and documentation
2562 at the head of each file (just as the Lisp source files already did).
2563 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2564 to his page will be replaced with a link to his new CLiki service.
2566 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2568 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2569 * The build process now looks for GNU make under the default name "gmake",
2570 instead of "make" as it used to. If GNU make is not available as "gmake"
2571 on your system, you can change this default behavior by setting the
2572 GNUMAKE environment variable.
2573 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2574 can build without error under CMU CL.
2576 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2578 * There is now a partial SBCL user manual (with some new text and some
2579 text cribbed from the CMU CL manual).
2580 * The beginnings of a profiler have been added (starting with the
2581 CMU CL profiler and simplifying and cleaning up). Eventually the
2582 main interface should be through the TRACE macro, but for now,
2583 it's still accessed through vaguely CMU-CL-style functions and macros
2584 exported from the package SB-PROFILE.
2585 * Some problems left over from porting CMU CL to the new
2586 cross-compilation bootstrap process have been cleaned up:
2587 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2588 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2589 part when getting rid of the compiler *BACKEND* stuff.)
2590 ** The value of *NULL-TYPE* was screwed up, because it was
2591 being initialized before the type system knew the final
2592 definition of the 'NULL type. This screwed up several key
2593 optimizations in the compiler, causing inefficiency in all sorts
2594 of places. (I found it because I wanted to understand why
2595 GET-INTERNAL-RUN-TIME was consing.)
2596 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2597 PROCLAIM FTYPE information. Unfortunately this broke binary
2598 compatibility again, since now the forms output by DEFGENERIC
2599 to refer to functions which didn't exist in 0.6.3.
2600 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2601 can use the new (as of 0.6.3) transform for SOME into MAP into
2603 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2604 values to (INTEGER 0 1000000), so that the time code will no longer
2605 occasionally get blown up by Linux returning 1000000 microseconds
2606 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2607 its output conform to the ANSI spec. (Alas, this makes its output
2608 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2609 * A full call to MAP NIL with a single sequence argument no longer conses.
2610 * fixes to problems pointed out by Martin Atzmueller:
2611 * The manual page no longer talks about multiprocessing as though
2612 it were currently supported.
2613 * The ILISP support patches have been removed from the distribution,
2614 because as of version 5.10.1, ILISP now supports SBCL without us
2615 having to maintain patches.
2616 * added a modified version of Raymond Toy's recent CMU CL patch for
2617 EQUALP comparison of HASH-TABLE
2619 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2621 * The system still can't cross-compile itself with
2622 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2623 entails), but at least it can compile more of itself that way
2624 than it used to be able to, and various buglets which were uncovered
2625 by trying to cross-compile itself that way have now been fixed.
2626 * This release breaks binary compatibility again. This time
2627 at least I've incremented the FASL file format version to 2, so that the
2628 problem can be detected reliably instead of just causing weird errors.
2629 * various new style warnings:
2630 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2631 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2632 ** using the lexical binding of a variable named in the *FOO* style
2633 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2634 as specified by ANSI.
2635 * *RANDOM-STATE* is no longer automatically initialized from
2636 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2637 default behavior of the system is to repeat its behavior every time
2638 it's run. If you'd like to change this behavior, you can always
2639 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2640 old convention there was no comparably easy way to get the system to
2641 repeat its behavior every time it was run.
2642 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2643 FTYPE declarations has been removed, in favor of their ANSI
2644 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2645 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2646 the types of their sequence arguments can be determined at compile time.
2647 This is done through a new open code expansion for MAP which eliminates
2648 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2649 argument types can be determined at compile time.
2650 * The optimizer now transforms COERCE into an identity operation when it
2651 can prove that the coerced object is already of the correct type. (This
2652 can be a win for machine generated code, including the output of other
2653 optimization transforms, such as the MAP transform above.)
2654 * Credit information has been moved from source file headers into CREDITS.
2655 * Source file headers have been made more standard.
2656 * The CASE macro now compiles without complaining even when it has
2659 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2661 * (Note that the way that the PCL macroexpansions were rewritten
2662 to accommodate the change in DEFGENERIC below breaks binary
2663 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2664 not run under sbcl-0.6.2. Once we get out of alpha releases,
2665 i.e. hit release 1.0.0, we'll probably try to maintain binary
2666 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2667 and sbcl-1.4.4. Until then, however, it might be fairly common
2668 for maintenance releases to break binary compatibility.)
2669 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2671 * The reporting of unhandled signals has been changed to print some
2672 explanatory text as well as the report form. (Previously only
2673 the report form was printed.)
2674 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2675 it defines, so that the compiler no longer issues undefined function
2676 warnings for compiled-but-not-yet-loaded generic functions.
2677 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2678 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2679 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2680 section of the ANSI spec are supported.
2681 * The "" nickname for the "KEYWORD" package has been removed.
2682 The reader still handles symbol tokens which begin with a package marker
2683 as keywords, but it doesn't expose its mechanism for doing so in the
2684 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2685 * The system now issues STYLE-WARNINGs for contradictory TYPE
2686 proclamations. (Warnings for contradictory FTYPE proclamations would
2687 be nice too, but those can't be done usefully unless the type system
2688 is made smarter about FUNCTION types.)
2689 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2690 systematized, so that "*target-*.lisp is supposed to exist only on the
2691 target and imply that there's a related file which exists on the
2692 host, and *host-*.lisp is supposed to exist only on the host and imply
2693 that there's a related file which exists on the target. This involves a
2694 lot of renaming. Hopefully the acute confusion caused by the renaming
2695 will be justified by the reduction in chronic confusion..
2696 ** runtime-type.lisp -> early-target-type.lisp
2697 ** target-type.lisp -> late-target-type.lisp
2698 ** early-host-format.lisp -> early-format.lisp
2699 ** late-host-format.lisp -> late-format.lisp
2700 ** host-error.lisp -> misc-error.lisp
2701 ** early-error.lisp -> early-target-error.lisp
2702 ** late-error.lisp -> late-target-error.lisp
2703 ** host-defboot.lisp -> early-defboot.lisp
2704 ** code/misc.lisp -> code/target-misc.lisp
2705 ** code/host-misc.lisp -> code/misc.lisp
2706 ** code/numbers.lisp -> code/target-numbers.lisp
2707 ** code/early-numbers.lisp -> numbers.lisp
2708 ** early-host-type.lisp -> early-type.lisp
2709 ** late-host-type.lisp -> late-type.lisp
2710 ** host-typep.lisp -> typep.lisp
2711 ** load.lisp -> target-load.lisp
2712 ** host-load.lisp -> load.lisp
2713 ** host-disassem.lisp -> disassem.lisp
2714 ** host-insts.lisp -> insts.lisp
2715 ** byte-comp.lisp -> target-byte-comp.lisp
2716 ** host-byte-comp.lisp -> byte-comp.lisp
2717 ** host-signal.lisp -> signal.lisp
2718 ** host-defstruct.lisp -> defstruct.lisp
2719 ** late-target-type.lisp -> deftypes-for-target.lisp
2720 Furthermore, several other previously target-only files foo.lisp (e.g.
2721 hash-table.lisp and random.lisp) have been split into a target-and-host
2722 foo.lisp file and a target-only target-foo.lisp file, with their key type
2723 definitions in the target-and-host part, so that the cross-compiler will
2724 know more about target types.
2725 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2726 gone away. In their place are various *BACKEND-FOO* variables
2727 corresponding to the slots of the old structure.
2728 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2729 into the target SBCL has been fixed.
2730 * The chill.lisp system for loading cold code into a running SBCL
2732 * Support for the CMU CL "scavenger hook" extension has been removed.
2733 (It was undocumented and unused in the CMU CL sources that SBCL was
2734 derived from, and stale in sbcl-0.6.1.)
2735 * Various errors in the cross-compiler type system were detected
2736 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2737 (enabling various consistency checks). Many of them were fixed,
2738 but some hard problems remain, so the compiler is back to
2739 running without *TYPE-SYSTEM-INITIALIZED* for now.
2740 * As part of the cross-compiler type system cleanup, I implemented
2741 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2742 * I have started adding UNCROSS calls throughout the type system
2743 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2744 on cross-compiler input files will be able to go away, and various
2746 * CONSTANTP now returns true for quoted forms (as explicitly required
2749 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2751 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2752 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2753 (MAKE-LIST -1). In the longer term, it ought to become true
2754 that declarations are assertions even at SAFETY 1. For now, it's not
2755 quite true even at SAFETY 3, but it's at least more nearly true..
2756 (Note that this change seems to increases the size of the system by
2757 O(5%) and to decrease the speed of the compiler by 20% or more.)
2758 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2759 for the problem of printing dozens of lines of distracting information
2760 about low-level system machinery as part of the top stack frame
2761 on entry to the debugger when an undefined function was called.
2762 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2763 is not reset to COMMON-LISP-USER.
2764 * Compilation of stuff related to dyncount.lisp has been made conditional
2765 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2766 smaller. The various dyncount-related symbols have been moved into
2767 a new "SB-DYNCOUNT" package.
2768 * tty-inspect.lisp has been renamed to inspect.lisp.
2769 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2770 feature has gone away. (When we eventually port to other flavors of
2771 libc and/or Unix, we'll try to make the differences between flavors
2772 invisible at the user level.)
2773 * Various other *FEATURES* tags, and/or their associated conditionals,
2774 have been removed if obsolescent, or given better documentation, or
2775 sometimes given more-mnemonic names.
2777 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2779 * tidied up "make.sh" script
2780 * tidied up system directory structure
2781 * better "clean.sh" behavior
2782 * added doc/FOR-CMUCL-DEVELOPERS
2783 * many many small tweaks to output format, e.g. removing possibly-confusing
2784 trailing #\. character in DESCRIBE-INSTANCE
2785 * (EQUALP #\A 'A) no longer signals an error.
2786 * new hashing code, including EQUALP hashing
2787 * tidied up Lisp initialization and toplevel
2788 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2789 * command line argument processing
2790 * added POSIX-GETENV function to deal with Unix-ish environment variables
2791 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2792 terminating SBCL on EOF
2793 * non-verbose GC by default
2794 * There is no more "sbcl" shell script; the sbcl file is now the C
2795 runtime executable (just like CMU CL).
2796 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2798 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2799 symbol and package information as Lisp native symbol and package objects
2800 instead of strings naming symbols and strings naming packages. This way,
2801 whenever packages are renamed (as in warm init), debug information is
2802 transformed along with everything else.
2803 * tweaked the optimization policy declarations which control the building
2804 of SBCL itself. Now, among other things, the system no longer saves
2805 source location debugging information. (This helps two problems at once
2806 by reducing SBCL size and by keeping SBCL from trying to look for its
2807 sources -- which may not exist -- when reporting errors.)
2808 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2809 debugging and testing purposes
2810 * cleaned up printing, making the printer call PRINT-OBJECT for
2811 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2812 methods, giving nearly-ANSI behavior
2813 * converted almost all special variables to use *FOO* naming convention
2814 * deleted PARSE-TIME functionality, since it can be done portably
2815 * moved some files out of cold init into warm init
2816 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2818 * regularized formatting of source files
2819 * added an install.sh script
2820 * fixed ridiculous memory usage of cross-compiler by making
2821 compiler/alloc.lisp not try to do pooling unless it can hook
2822 itself into the GC of the cross-compilation host. Now the system
2823 builds nicely on my old laptop.
2824 * added :SB-ALLOC in target-features.lisp-expr
2825 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2826 was not implemented)
2827 * re-did condition handling and note reporting in the compiler. Notes
2828 are no longer handled by signalling conditions. Style warnings
2829 and warnings are handled more correctly and reported in such a way
2830 that it's easy to find one or the other in your output (so that you
2831 can e.g. figure out which of many problems caused COMPILE-FILE to
2833 * changed the severity of several compiler warnings from full WARNING
2834 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2835 compiler note reporting so that it doesn't use the condition system
2836 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2837 COMPILE-FILE command)
2838 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2839 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2840 no longer supported, since their implementation was deeply intertwingled
2841 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2842 * removed bogus "support" for compiler macros named (SETF FOO), and
2843 removed the compiler macro for SETF INFO (but only after making a fool
2844 of myself on the cmucl-imp mailing list by posting a bogus patch for
2845 DEFINE-COMPILER-MACRO..)
2846 * Compiled files containing forms which have side effects on the Lisp
2847 reader (such as DEFPACKAGE forms) are now handled more correctly.
2848 (Compiler queuing of top level lambdas has been suppressed by setting
2849 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2850 * deleted various currently-unused source files, e.g. gengc.lisp. They
2851 may be added back at some point e.g. when porting to other architectures,
2852 but until they are it's distracting to distribute them and to try to
2854 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2855 there were so many of them they're just distractions, and UNCROSS is
2856 known to be able to handle the current sources
2857 * moved PROFILE functionality into TRACE, so that it will be clear
2858 how the wrapping and unwrapping of functions when you profile them
2859 interacts with the wrapping and unwrapping of functions when you
2860 trace them. (Actually, the functionality isn't there yet, but at least
2861 the interface specification is there. Hopefully, the functionality will
2862 arrive with some future maintenance release.)
2863 * removed host-oops.lisp
2864 * changed signature of QUIT function to allow UNIX-CODE argument
2865 * fixed READ-SEQUENCE bug
2866 * tweaked verbose GC output so that it looks more like the progress
2867 output that ANSI specifies for functions like LOAD
2868 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2869 * added <http://sbcl.sourceforge.com> to the banner information printed by