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 * bug fix: lambda-list parsing is now stricter vrt. order and number
9 of lambda-list keywords.
10 * fixed some bugs revealed by Paul Dietz' test suite:
11 ** INCF, DECF and REMF evaluate their place form as specified in
14 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
15 * new feature: a build-time option (controlled by the :SB-UNICODE
16 keyword feature, enabled by default) for building the system with
17 support for the entire 21-bit character space defined by the
19 * new feature: the system now has rudimentary external-format
20 support; the primary user-visible change at this time is that
21 characters with the high bit set (such as lower-case-e-acute) will
22 print correctly to a terminal in a UTF-8 environment.
23 * minor incompatible change: BASE-CHAR no longer names a class;
24 however, CHARACTER continues to do so, as required by ANSI.
25 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
26 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
27 has been moved to the SB-EXT package (temporarily re-exported from
29 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
30 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
32 * fixed bug #351: better error-handlind and reporting for malformed
34 * fixed bug #350: bignum-printing is now more memory-efficient,
35 allowing printing of very large bignums, eg. (expt 2 10000000).
36 (reported by Bruno Haible)
37 * fixed bug #302: better primitive-type selection for intersection
39 * fixed bug #308: non-graphic characters now all have names, as
40 required. (reported by Bruno Haible)
41 * bug fix: redefining a class with different superclasses now correctly
42 removes it from the direct-subclasses of its previous superclasses.
43 (reported by David Morse)
44 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
45 new value now works. (reported by Bruno Haible)
46 * bug fix: correct canonicalization of multiple non-standard slot
47 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
48 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
49 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
50 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
51 now exists, an signals an error.
52 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
53 during macroexpansion for non-iteration variables as well. (reported
54 by Bruno Haible for CMUCL)
55 * bug fix: Cyclic structures and unprintable objects in compiler
56 messages no longer cause errors. (reported by Bruno Haible)
57 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
58 and READ-FROM-STRING all now return a primary value of NIL if
59 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
60 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
61 (reported by Bruno Haible for CMUCL)
62 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
63 do not designate a condition. (reported by Bruno Haible for
65 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
66 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
67 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
68 permitting resolution in favour of any of the conflicting symbols.
69 (reported by Bruno Haible for CMUCL)
70 * bug fix: EQUAL compiler optimizations is less aggressive on
71 strings which can potentially compare true despite having distinct
72 specialized array element types.
73 * bug fix: unit enumerations can be defined without dividing by
74 zero. (thanks to Vincent Arkesteijn)
75 * FORMAT compile-time argument count checking has been enhanced.
76 (report from Bruno Haible for CMUCL)
77 * a partial workaround for the bug 262: the compiler does not try to
78 inline-expand a local function doing RETURN-FROM from a deleted
79 BLOCK. (thanks to Peter Denno for the bug report and to David
80 Wragg for the simple test case)
81 * fixed some bugs revealed by Paul Dietz' test suite:
82 ** RENAME-PACKAGE allows all package designators as new package
84 ** constraint propagation and lambda variable substitution are
85 more cautious in dealing with partially deleted code.
86 ** compiler sometimes lost reoptimization passes.
87 ** CERROR, given a condition as condition designator, passes its
88 remaining arguments to the continue format control without
90 ** Case-altering FORMAT directives work correctly on non-ASCII
92 ** The REAL type specifier handles bounds outside the floating
93 point ranges without signalling FLOATING-POINT-OVERFLOW.
94 ** Functions with IR1-transformations can create intercomponent
95 references to global functions.
96 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
98 ** FORMAT ~R treats a nil value for its first parameter correctly.
100 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
101 * enhancement: saving cores with foreign code loaded is now
102 supported on x86/NetBSD and sparc/Linux in addition to the previously
104 * bug fix: on some platforms repeated installations caused multiple
105 copies of HTML documentation to be installed -- should not happen
106 any more. (reported by Stefan Scholl)
107 * bug fix: parsing self-recursive alien record types multiple times
108 no longer causes infinite recursion. (reported by Thomas F. Burdick,
109 original patch by Helmut Eller for CMUCL)
110 * bug fix: stack-exhaustion detection works now on NetBSD as well.
111 (thanks to Richard Kreuter)
112 * bug fix: defining classes whose accessors are methods on existing
113 generic functions in other (locked) packages no longer signals
114 bogus package lock violations. (reported by François-René Rideau)
115 * bug fix: special variables as DEFMETHOD parameters no longer have
116 associated bogus type declarations. (reported by David Wragg and
118 * bug fix: read-write consistency on streams of element-type
119 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
120 * bug fix: redefiniton of the only method of a generic function with
121 no DEFGENERIC no longer emits a full WARNING. In addition,
122 redefinition of generic functions with no DEFGENERIC to an
123 incompatible lambda list now signals an error. (thanks to Zach
125 * bug fix: DEFGENERIC now works even when there's a function of the
126 same name in an enclosing lexical environment. (thanks to Zach
128 * fixed compiler failure, caused by instrumenting code during
129 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
130 * optimization: added loop analysis and improved register allocation
131 * fixed some bugs revealed by Paul Dietz' test suite:
132 ** POSITION on displaced vectors with non-zero displacement
133 returns the right answer.
134 ** (SIMPLE-STRING) is a valid type specifier for sequence
136 ** *PRINT-LEVEL* handling for slotless structures is pedantically
138 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
139 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
141 ** The justification version of the FORMAT ~< directive treats
142 non-zero minpad parameter correctly.
144 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
145 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
146 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
147 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
148 the supported interface.
149 * new feature: Single-stepping of code compiled with DEBUG 2 or
150 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
151 * new feature: saving cores with foreign code loaded is now
152 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
153 Timothy Moore's work for CMUCL)
154 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
155 parameters to * instead of NIL if no initform is supplied.
156 (reported by Johan Bockgaard)
157 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
158 correctly when FUNCALL appears as the car of the form. Note:
159 despite this FUNCALL forms are not currently subject to
160 compiler-macro expansion. (port of Raymond Toy's fix for the
161 same from CMUCL, reported by Johan Bockgaard)
162 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
163 (thanks for Teemu Kalvas)
164 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
165 case instead of terminating the iteration (thanks for Julian
166 Squires, Sean Champ and Raymond Toy)
167 * bug fix: incorrect expansion of defgeneric that caused a style
168 warning. (thanks for Zach Beane)
169 * bug fix: slot accessor effective method computation works properly
170 for classes with multiple non-standard applicable methods on
171 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
172 * on x86 compiler supports stack allocation of results of LIST and
173 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
174 CMUCL implementation by Gerd Moellmann)
175 * fixed some bugs revealed by Paul Dietz' test suite:
176 ** FORMAT strings with both the ~<~:;~> form of the justification
177 directive and pretty-printing directives cause an error.
179 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
180 * incompatible change: the internal functions
181 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
182 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
183 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
184 instead of the old functions.
185 * new feature: on platforms where "dladdr" is available foreign
186 function names now appear in backtraces. (based on Helmut Eller's
188 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
189 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
191 * documentation: ASDF manual is now distributed with SBCL.
192 * bug fix: SBCL can now load its contributed modules with REQUIRE
193 even if the system-provided entries have been removed from
194 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
196 * bug fix: backtraces involving undefined functions or assembly
197 routines are more informative. (thanks to Brian Downing)
198 * bug fix: mutually referent alien structures now work correctly.
199 (reported by Rick Taube)
200 * bug fix: structures defined by WITH-ALIEN can be referred to
201 within other definitions in the same WITH-ALIEN.
202 * bug fix: division operators (MOD, TRUNCATE and the like) with
203 constant zero divisors and integer dividends no longer generate
205 * bug fix: provide default methods for INPUT-STREAM-P and
206 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
207 * bug fix: improve the detection and resolution of MOP metacycles
208 (where effective method computation depends on the generic function
209 having its effective method computed). (reported by Bruno Haible)
210 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
211 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
212 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
213 * bug fix: pathnames with relative directory components can be
214 represented relative to default pathnames.
215 * optimization: in taking the GCD of bignums, reduce the two bignums
216 to approximately the same size (using Euclid's algorithm) before
217 applying the more sophisticated binary GCD. (thanks to Juho
219 * optimization: COUNT on bitvectors now operates word-at-a-time.
220 * optimization: ASH with a positive, but not necessarily constant,
221 (leftwards) shift, when in a modular context, is compiled to a
223 * fixed some bugs revealed by Paul Dietz' test suite:
224 ** FORMAT variable parameters ("~V<char>") are defaulted properly
225 if the corresponding argument is NIL.
226 ** FORMAT directives accepting a minpad parameter treat negative
228 ** literal commas as character parameters to FORMAT directives are
230 ** literal spaces directly after ~<Newline> directives within a
231 format-logical-block (~:< ~@:>) do not induce :FILL-style
232 conditional newlines.
233 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
234 as specified: it no longer includes conditional newlines).
235 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
237 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
238 :PER-LINE-PREFIX argument does not evaluate to a string.
240 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
241 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
242 the manual for details; for now, package locks can be disabled by
243 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
244 if no major problems are found then it is likely that they will be
245 compiled in unconditionally.
246 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
247 unsupported operators on all platforms. To load a shared library
248 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
249 object file, link it into a shared library outside of SBCL and
250 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
252 * bug workaround: ROOM T and the GC were not working together
253 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
254 reporting facility aren't preserved. That reporting has been
255 disabled, so now until and unless someone figures out how to make it
256 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
257 * minor incompatible change: as threatened around sbcl-0.8.0, the
258 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
259 The symbols are also exported from SB-PCL for backwards
260 compatibility, but more so than before SB-PCL should be treated as
261 an implementation-internal package.
262 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
264 * the SB-SPROF contrib now works on (most) non-x86 architectures.
265 It is known as of this release not to work on the Alpha, however.
266 * fixed bug #167: errors signalled due to illegal syntax in method
267 bodies are now more legible.
268 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
269 designators and can hence be used with TYPEP.
270 * fixed bug #333: CHECK-TYPE now ensures that the type error
271 signalled, if any, has the right object to be accessed by
272 TYPE-ERROR-DATUM. (reported by Tony Martinez)
273 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
274 5.1.2.3 for multiple-value place subforms. (reported by Kalle
276 * fixed bug #334: programmatic addition of slots using specialized
277 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
278 and :ALLOCATION :CLASS slots.
279 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
280 float exponents. (rereported by Peter Seibel)
281 * fixed a bug: #\Space (and other whitespace characters) are no
282 longer considered to be macro characters in standard syntax by
284 * fixed bug: initialization of condition class metaobjects no longer
285 causes an instance of the condition to be created. (reported by
287 * fixed bug: it is now possible to have slots such that
288 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
289 description is neither :INSTANCE nor :CLASS.
290 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
291 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
292 SLOT-VALUE-USING-CLASS), and no longer causes errors with
293 non-standard SLOT-DEFINITION-ALLOCATION values.
294 * fixed bugs: various race conditions handling exiting threads.
295 CL-PPCRE's thread torture test now passes.
296 * fixed arguably-a-bug: GC time proportional to number of threads in
297 system even when most of them are idle
298 * optimization: improved performance of BIT and SBIT on bit-vectors.
299 * .fasl file incompatibility: The fasl file version number has
300 been incremented because of changes associated with package locks.
302 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
303 * minor incompatible change: the system no longer provides
304 optimization hints (conditions which are TYPEP
305 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
306 mode; these hints are still emitted when higher SPEED optimization
308 * new contrib module: a sampling profiler (profiling by statistical
309 sampling, rather than by instrumenting functions) is available as
310 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
311 * the behaviour of the standard function ED is now customizeable by
312 third parties through a hook variable: see ED's documentation
313 string for information on the protocol.
314 * the compiler no longer emits efficiency notes for (FUNCALL X)
315 when the type of X is uncertain under default optimization
317 * fixed bug 276: mutating a binding of a specialized parameter to a
318 method to something that is not TYPEP the specializer is now
320 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
321 exhibit the required behaviour on the x86 platform. (thanks to
322 Peter van Eynde, Eric Marsden and Bruno Haible)
323 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
324 even for difficult arguments. (reported by Peter Graves)
325 * fixed bug 141a: the backquote printer now descends quoted
327 * fixed another bug in backquote printing: no more destructive
328 modification of the form's list structure. (reported by Brian
330 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
331 move between its address being taken and the call to
332 interrupt_thread, fixing a crashing race condition.
333 * the SB-POSIX contrib implementation has been adjusted so that it
334 no longer exhibits ridiculously poor performance when constructing
335 instances corresponding to C structs.
337 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
338 * minor incompatible change: the sb-grovel contrib now treats C
339 structures as alien (in the sense of SB-ALIEN) objects rather than
340 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
341 has implications for memory management of client code
342 (sb-grovel-returned objects must now be manually managed) and for
343 type safety (alien objects now have full types).
344 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
345 used to control emission of compiler diagnostics, rather than the
346 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
347 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
348 quality should be considered deprecated.
349 * install.sh now installs the user manual as well
350 * (not quite a new documentable feature, but worth considering in
351 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
352 beginnings of a semantically meaningful condition hierarchy is
353 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
355 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
356 designator as the defaults argument.
357 * fixed bug: Displaced arrays whose displaced-to array has become
358 too small now cause ARRAY-DIMENSION to signal an error, providing
359 for safer bounds-checking. (reported by Bruno Haible)
360 * fixed bug: DEFCLASS slot definitions with identical :READER and
361 :WRITER names now signal a reasonable error. (reported by Thomas
363 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
365 * fixed bug: Invalid slot specification errors now print correctly.
366 (thanks to Zach Beane)
367 * fixed bug 320: Shared to local slot value transfers in class
368 redefinitions now happen corrently from superclasses as
369 well. (reported by Bruno Haible)
370 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
372 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
373 improper lists correctly. (reported by Bruno Haible)
374 * fixed bug 313: source-transform for <fun-name> was erroneously
375 applied to a call of a value of a variable with name <fun-name>.
376 (reported by Antonio Menezes Leitao)
377 * fixed bug 307: The obsolete instance protocol ensures that
378 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
379 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
380 remove dead unknown-values globs from the middle of the stack.
381 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
382 easier to resume long computations after using *BREAK-ON-SIGNALS*
383 to diagnose and fix failures. (thanks to Nikodemus Siivola)
384 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
385 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
386 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
387 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
388 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
390 * fixed a (fairly theoretical) bug in string printing: if
391 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
392 does not have array-element-type equal to the most general string
394 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
395 function name. (thanks to Nikodemus Siivola)
396 * fixed bug: compile-time format string checker failed on
397 non-closed ~{. (reported by Thomas F Burdick)
398 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
399 don't warn on legal loop code involving more than one aggregate
400 boolean. (thanks to Nikodemus Siivola)
401 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
402 in keyword position to MAKE-PACKAGE was not being reported as
404 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
405 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
407 * fixed bugs 280 and 312: the checking for multiple definitions in a
408 file is less likely to become confused by uses of inline
410 * fixed bug: the #S reader macro performs the keyword coercion
411 specified for slot names. (reported by Kalle Niemitalo)
412 * fixed bug: lambda lists may contain symbols whose names start with
413 & but are not lambda-list-keywords; their occurrence triggers a
415 * fixed bug 321: define-method-combination argument lambda lists do
416 not cause invalid code to be generated when &optional or &aux
417 variables have default values. (reported by Bruno Haible)
418 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
419 have CLOS classes; this appears also to have fixed reported
420 instability in stack exhaustion detection.
421 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
422 MUFFLE-WARNING when no associated restart is present is now
424 * optimization: rearranged the expansion of various defining macros
425 so that each expands into only one top-level form in a
426 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
428 * optimization: used a previously little-used slot in symbols to
429 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
431 * fixed some bugs revealed by Paul Dietz' test suite:
432 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
434 ** Fixed an optimization bug related to inheritance of initforms
435 from local to shared slots.
436 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
437 ** CAST optimizer forgot to flush argument derived type.
438 ** print/read consistency on floats is now orders of magnitude
439 more likely. (thanks also to Bruno Haible for a similar report
441 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
442 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
443 for objects of type REAL. Make it so.
444 ** SXHASH is specified (infelicitously) to respect similarity,
445 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
446 it so. (thanks to Markus Ziegler)
447 ** on the Alpha, the compiler succeeds in compiling functions
448 returning a known number of arguments greater than 63.
449 ** fixed handling of invalid NIL arguments in keyword position
451 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
453 ** provided a readably-printable representation for RANDOM-STATE
455 ** ensured that pathnames, where they have a namestring, always
456 print using #P"..." syntax.
458 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
459 * Support for the forthcoming 2.0 version of the NetBSD kernel
460 running on x86 hardware has been added. (thanks to Perry
461 E. Metzger most immediately, and others for their past work)
462 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
463 current release version; SBCL's OpenBSD support had been broken
464 since about the time of OpenBSD's switch to ELF binary format.)
465 * [placeholder for DX summary]
466 ** user code with &REST lists declared dynamic-extent, under high
467 speed or space and low safety and debug optimization policy.
468 * The manual has been converted to Texinfo format and the debugger
469 chapter from the cmucl manual has been added.
470 * A facility has been added to extract documentation strings from
471 sbcl and store them as Texinfo-formatted snippets for inclusion in
472 the manual (via Texinfo's @include directive)
473 * bug fix: compiler emitted division in optimized DEREF. (thanks for
474 the test case to Dave Roberts)
475 * bug fix: multidimensional simple arrays loaded from FASLs had fill
476 pointers. (reported by Sean Ross)
477 * bug fix: PROFILE output is printed nicely even for large numerical
478 values. (thanks to Zach Beane)
479 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
480 greater than 32 handle EOF correctly.
481 * bug fix: on X86 an immediate argument of the IMUL instruction is
482 correctly printed in disassembly. (thanks to Lutz Euler)
483 * bug fix: class slots in redefined classes preserve their old
484 values. (thanks to Bruno Haible and Nikodemus Siivola)
485 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
486 succeeds. (reported by Marco Baringer)
487 * fixed some bugs revealed by Paul Dietz' test suite:
488 ** READ-SEQUENCE now works on ECHO-STREAMs.
489 ** RATIONALIZE works more according to its specification. (thanks
492 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
493 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
494 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
495 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
496 should matter to you only if you rebind the printer control
497 variables and then find you want different bindings in the
498 debugger than in the ordinary execution of your program.)
499 * The runtime build system has been tweaked to support building
500 (on SPARC/SunOS) using a C compiler which invokes Sun's own
501 assembler and linker. (thanks to Nikodemus Siivola)
502 * Unbound, undefined, undeclared variables now trigger full
503 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
504 more often programmer error than deliberate exploitation of undefined
506 * optimization: the hash algorithm for strings has changed to one
507 that is less vulnerable to spurious collisions. (thanks to Juho
509 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
510 needless bounds checking. (thanks to Juho Snellman)
511 * optimization: implemented multiplication as a modular
512 (UNSIGNED-BYTE 32) operation on the PPC backend.
513 * fixed some bugs revealed by Paul Dietz' test suite:
514 ** ADJUST-ARRAY now copies the datum in a zero rank array if
516 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
517 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
518 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
520 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
522 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
524 ** ECHO-STREAMs no longer attempt to echo the end of file value to
525 their output stream on EOF from read.
526 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
527 have been read to end-of-file.
528 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
530 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
531 description of determination of which consecutive characters
533 ** Printing the "Space" character with escaping on now yields "#\\ ",
534 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
535 ** Reading floating-point numbers with *READ-BASE* set to a number
536 less than 10 works correctly.
537 ** Reading floating-point numbers with *READ-BASE* set to a number
538 more than 10 works correctly.
539 ** Printing with *PRINT-READABLY* targets the standard readtable, not
540 the readtable currently in effect.
542 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
543 * minor incompatible change: parsing of namestrings on a physical
544 (Unix) host has changed; numbers after the final #\. in a
545 namestring are no longer interpreted as a version field. This is
546 intented to be largely invisible to the user, except that the
547 meaning of the namestring "*.*.*" has changed: it now refers to a
548 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
549 should usually be replaced by
550 (make-pathname :name :wild :type :wild :version :wild)
551 with the added benefit that this is more likely to be portable.
552 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
553 option to OPEN now signals an error if the file being opened
554 exists; this may have an impact on existing code.
555 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
556 signals received, on the PowerPC platforms (both Linux and
557 Darwin). (thanks to Pierre Mai for pointing out the location of
559 * several fixes on OS X: The system now builds and runs cleanly on
560 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
561 (thanks to Brian Mastenbrook)
562 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
563 second-resolution: integer multiples of 1/3600 between -24 and 24.
564 (thanks to Vincent Arkesteijn)
565 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
566 type when called with 1 argument; PEEK-CHAR checked type of
567 PEEK-TYPE only after having read first character from a
568 stream. (reported by Peter Graves)
569 * bug fix: the garbage collector now has much better locality
570 behaviour, and in particular no longer treats all memory as being
571 exhausted when many small objects point to each other in a deeply
573 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
574 recognized as being TYPEP their class.
575 * bug fix: the PUSHNEW documentation string has been corrected.
576 (thanks to Vincent Arkesteijn)
577 * bug fix: defaulting of the value for the last of an atypically
578 large number of multiple values being bound was not being
579 performed correctly on the Alpha or PPC platforms
580 * optimization: implemented multiplication as a modular
581 (UNSIGNED-BYTE 32) operation on the x86 backend.
582 * optimization: SEARCH on simple-base-strings can now be open-coded.
583 (see also contrib/compiler-extras.lisp for inspiration for
584 teaching the compiler about the Boyer-Moore algorithm).
585 * value, returned by MAX (and MIN) called with several EQUALP, but
586 not EQL, arguments now does not depend on compiler settings.
587 * fixed some bugs revealed by Paul Dietz' test suite:
588 ** in stack analysis liveness information is propagated from
589 non-local entry points.
590 ** pathwise CAST removing failed when the CAST node did not start
592 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
593 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
595 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
596 host is already defined.
597 ** RENAME-FILE works on streams instead of signalling an internal
599 ** PEEK-CHAR uses the current readtable when determining whether
600 or not a character is whitespace.
601 ** MERGE-PATHNAMES handles the case when the pathname does not
602 specify a name while the default-pathname specifies a version.
603 ** Pathnames now stand a better chance of respecting print/read
605 ** Attempting to use standardized file system operators with a
606 pathname with invalid :DIRECTORY components signals a
608 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
610 ** DIRECTORY on logical pathnames is more correct.
611 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
612 signal a TYPE-ERROR if their argument is not a stream.
613 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
614 designator argument does not designate a stream.
615 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
616 examining the synonym.
617 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
619 ** OPEN and WITH-OPEN-STREAM allow opening streams with
620 element-type larger than ([UN]SIGNED-BYTE 32).
622 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
623 * When built with the :SB-FUTEX feature, threaded builds now take
624 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
625 for faster/more reliable mutex and condition variable support.
626 * Incompatible change (but one you probably shouldn't have been using
627 anyway): the interface and code for arbitrating between multiple
628 threads in the same user session has been redesigned.
629 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
630 error when called without an explicit environment argument.
631 (thanks to Nikodemus Siivola)
632 * bug fix: buffered :DIRECTION :IO streams are less likely to become
633 confused about their position. (thanks to Adam Warner and Gerd
635 * bug fix: Pretty printing backquoted forms with unquotations in the
636 argument list position of various code constructs such as LAMBDA
637 now works correctly. (reported by Paul Dietz)
638 * bug fix: Pretty printing unquotations no longer loses all
639 stream position information.
640 * optimization: performance of string output streams is now less
641 poor for multiple small sequence writes.
642 * optimization: performance of CSUBTYPEP in the presence of complex
643 expressions involving CONS and NOT many times has been improved.
644 (reported by Paul Dietz)
645 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
647 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
648 streams for the REPL, for the debugger, and so on. (thanks to
650 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
651 * fixed some bugs revealed by Paul Dietz' test suite:
652 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
653 is ignored if it is too small, rather than propagating through
654 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
655 ** extremely complex negations of CONS types were not being
656 sufficiently canonicalized, leading to inconsistencies in
658 ** VALUES tranformer lost derived type.
660 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
661 * fixed a bootstrapping bug: the build process no longer assumes
662 that the various BOOLE-related constants have the same value in
663 host and target lisps. (noted by Paul Dietz' test suite on an
664 SBCL binary built from CLISP)
665 * The system can now be dynamically linked on the MIPS platform,
666 which enables dynamic loading of foreign code from Lisp. (thanks
667 to Ralf Baechle for discussions on the MIPS ABI)
668 * The system now records debugging information for its own source
669 files in a filesystem-position-independent manner, relative to
670 the "SYS" logical host.
671 * fixed a compiler bug: MV-LET convertion did not check references
672 to the "max args" entry point. (reported by Brian Downing)
673 * tweaked disassembly notes to be less confident about proclaiming
674 some instruction as an LRA. (thanks to Brian Downing)
675 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
676 now each have their own history, command character, and other
677 characteristics. (thanks to David Lichteblau)
678 * fixed some bugs revealed by Paul Dietz' test suite:
679 ** compiler failure in compiling LOGAND expressions including a
681 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
682 shift greater than 32.
683 ** FLUSH-DEST did not mark blocks for type check regeneration.
684 ** HANDLER-CASE failed to accept declarations in handler clauses
685 in some circumstances.
687 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
688 * New code in contrib/sb-introspect (still probably not entirely
689 stable yet) provides some support for smart Lisp development
690 environments like SLIME.
691 * The conditions signalled for errors occurring when loading .fasl
692 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
693 in a way which should help ASDF recover gracefully.
694 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
695 stuff has been cleaned up. If you code contrib/ stuff, this might
696 affect you, and you can look at contrib/README, contrib/STANDARDS,
697 and/or the 0.8.4.27 diff to check.
698 * In full calls the compiler now does not generate checks for declared
699 argument types for all arguments.
700 * various threading fixes
701 ** and some experimental patches which didn't make it into
702 the main tree for this release, but which are shipped in
703 contrib/experimental-thread.patch as a possible fix for some
704 failures (deadlock, spinning...) in GC-intensive multithreaded
706 * fixed PPC build problem (source code incompatibility of different
707 library versions): added offsetof() hackery which attempts to divine
708 where glibc maintainers put uc_mcontext today
709 * fixed bug 282: compiler does not trust type assertions while passing
710 arguments to a full call.
711 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
712 &OPTIONAL VALUES type parameter.
713 * fix bug 214: algorithm for noting rejected templates is now more
714 similar to that of template seletion. (also reported by rydis on
716 * fixed bug 141b: printing backquoted information readably and prettily
717 inserts a space where necessary.
718 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
719 (CHARACTER 10) are now reported as errors, rather than propagated
720 as unknown types. (reported by piso on #lisp)
721 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
722 with values NIL and :ERROR. (thanks to Milan Zamazal)
723 * fixed bug 191c: CLOS now does proper keyword argument checking as
724 described in CLHS 7.6.5 and 7.6.5.1.
725 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
726 longer raise an error; further, using a list as a for-as-arithmetic
727 counter now raises a meaningful error.
728 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
729 now understood by sequence creation functions such as MAKE-SEQUENCE
731 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
732 a STRING-INPUT-STREAM.
733 * compiler enhancement: SIGNUM is now better able to derive the type
735 * type declarations inside WITH-SLOTS are checked. (reported by
737 * fixed some bugs revealed by Paul Dietz' test suite:
738 ** incorrect optimization of TRUNCATE for a positive first
739 argument and negative second.
740 ** compiler failure in let-convertion during flushing dead code.
741 ** compiler failure while deriving type of TRUNCATE on an
742 interval, containing 0.
743 ** ASH of a negative bignum by a negative bignum count now returns
745 ** intersection of CONS types now canonicalizes properly, fixing
746 inconsistencies in SUBTYPEP.
748 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
749 * incompatible change: The --disable-debugger command line
750 option now clobbers the debugger at a more fundamental
751 level, by redefining #'INVOKE-DEBUGGER instead of by
752 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
753 is specified by ANSI to ignore *DEBUGGER-HOOK* and
754 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
755 enter the debugger REPL and then suffer recursive errors
756 because *DEBUG-IO* is also messed up in --disable-debugger mode;
757 while under the new system, BREAK in --disable-debugger mode
758 terminates the system just as an unhandled error would.
759 * fixed compiler performance when processing loops with a step >1;
760 * bug fix: DOCUMENTATION now retrieves generic function
761 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
762 support has been systematized, and now supports the methods
763 specified by ANSI, along with a default method and a method for
764 slot documentation. (reported by Nathan Froyd)
765 * bug fix: effective methods associated with a generic function are
766 no longer cached over a change of that generic function's method
767 combination. (reported by Andreas Fuchs)
768 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
769 names a non-existent directory. (thanks to Andreas Fuchs)
770 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
771 the PPC platform, lead to stack corruption; this has been fixed.
772 (reported by Rainer Joswig)
773 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
774 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
775 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
776 arithmetic in general, is now much more reliable on the PPC
778 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
779 platform now returns the right answer.
780 * optimization: restored some effective method precomputation in
781 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
782 precomputation is now tunable.
783 * optimization: compiler-internal data structure use has been
784 reviewed, and changes have been made that should improve the
785 performance of the compiler by about 20%.
786 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
787 simple-base-strings and simple-bit-vectors is improved.
788 * optimization: the optimization of 32-bit logical and arithmetic
789 functions introduced in version 0.8.3 on the x86 has been
790 implemented on the mips, ppc and sparc platforms; an
791 implementation of the same facility, but for 64-bit arithmetic,
792 has been added for the alpha.
793 * microoptimization: the compiler is better able to make use of the
794 x86 LEA instruction for multiplication by constants.
795 * bug fix: in some situations compiler did not report usage of
796 generic arithmetic in (SPEED 3) policy.
797 * bug 145b fix: compiler used wrong type specifier while converting
798 MEMBER-types to numeric.
799 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
801 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
802 index argument now works properly on the Alpha platform.
803 * bug fix: floating point exception treatment on the Alpha platform
805 * bug fix: FILE-POSITION works much better on string input and
806 output streams. (thanks to Nikodemus Siivola)
807 * bug fix: many threading/garbage collection symptoms sorted.
808 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
809 might be pseudo-atomic.
810 * internal change: Stopping for GC is now done with signals not
811 ptrace. GC is now done in whichever thread wanted it, instead of
813 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
815 * bug fix: result form in DO is not contained in the implicit
817 * incompatible change: ICR structure is changed; the value part of
818 CONTINUATION is now called LVAR; corresponding functions are
819 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
820 * added type deriver for ISQRT (thanks to Robert E. Brown).
821 * bug fix: better support for loading from the command line when an
822 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
824 * fixed some bugs revealed by Paul Dietz' test suite:
825 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
826 ** ROUND and FROUND now give the right answer when given very
827 small float arguments.
828 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
830 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
831 ** (GCD 0 <negative-integer>) returned <negative-integer>.
832 ** LCM should return a non-negative integer.
833 ** PARSE-INTEGER returned the index of a terminator instead of the
834 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
835 ** PARSE-INTEGER returned an incorrect index being applied to a
837 ** LCM with two arguments of 0 returns 0 rather than signalling
839 ** unsigned addition of a 32-bit constant with the high bit set no
840 longer causes an internal compiler error on the x86.
841 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
842 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
843 with negative last argument.
844 ** byte specifiers with zero size and position no longer cause
845 an error during type derivation.
846 ** bignum multiplication on the Alpha platform now returns the
848 * porting: The system now builds on SuSE AMD64, although it still
849 generates a 32-bit binary.
850 * .fasl file incompatibility: The fasl file version number has
851 been incremented (because of the changes to internal compiler
852 data structures referred to above).
854 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
855 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
856 more accurately, on the Darwin kernel running on PowerPC hardware.
857 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
858 * Compiler code deletion notes now signal a condition of type
859 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
860 an associated MUFFLE-WARNING restart.
861 * The compiler now performs limited argument count validation of
862 constant format strings in FORMAT, and where appropriate in ERROR,
863 CERROR and WARN. (thanks to Gerd Moellmann)
864 * New ASDF-INSTALL contrib can be used for automatic download and
865 installation of third-party Lisp code from CCLAN or other sites
867 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
868 which forces another thread to execute a function supplied by the
870 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
871 now accept and act upon their :ELEMENT-TYPE keyword argument.
872 (reported by Martin Atzmueller, Edi Weitz)
873 * bug fix: FILE-POSITION now accepts position designators up to
874 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
875 is the greater. (thanks to Patrik Nordebo)
876 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
877 Kalle Olavi Niemitalo)
878 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
879 printable. (reported by Eric Marsden)
880 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
881 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
882 * x86 bug fix in control stack exhaustion checking: now shows backtrace
883 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
884 (thanks to Stig Sandoe)
885 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
886 UNREAD-CHAR now correctly includes the unread character in the
887 target sequence. (thanks to Gerd Moellmann)
888 * bug fix in threaded builds: the system can now be suspended and
889 resumed by shell job control with minimal disruption.
890 * bug fixes in times and timezones >2038AD
891 * better handling of "where is GNU make?" problem in build scripts
892 (thanks to Nikodemus Siivola)
893 * new optimization: inside a named function any reference to a
894 function with the same name is considered to be a self-reference;
895 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
896 optimization quality.
897 * new optimization on x86: logical functions and + now have
898 optimized (UNSIGNED-BYTE 32) versions, which are automatically
899 used when the result is truncated to 32 bits.
900 * VALUES declaration is partially enabled.
901 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
902 * fixed some bugs revealed by Paul Dietz' test suite:
903 ** The system now obeys the constraint imposed by
904 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
905 types form a lattice under type intersection.
906 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
907 ** ASSOC now ignores NIL elements in an alist.
908 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
909 and (1+ MOST-POSITIVE-FIXNUM) answers.
910 ** The addition of a method with invalid qualifiers to a generic
911 function does not cause an error to be signalled immediately;
912 a warning is signalled, and the error is generated only on
913 calling the generic function.
914 * changed .fasl file version number, in part to add type codes for
915 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
916 obscure ANSI requirements
918 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
919 * fixed bug 148: failure to inline-expand a local function left
920 garbage, confusing the compiler.
921 * fixed bugs 3cd: structure slot readers perform type check if the
922 slot can have an invalid value (i.e. it is either not initialized
923 or can be written with a less specific slot writer).
924 * bug fix: the compiler now traps array references to elements off
925 the end of an array; previously, the bounds checking in some
926 circumstances could go off-by-one.
927 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
929 * type declarations for array element types now obey the description
930 on the CLHS page "Declaration TYPE", as per discussions on
931 sbcl-help around 2003-05-08. This means that a declaration
932 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
933 declaration, all references to BAR will be asserted or assumed
934 (with THE, so dependent on compiler policy) to involve objects of
935 type FOO. Note that no such declaration is implied in
936 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
937 * declared types of functions from the "Conditions"
938 chapter. (reported by Paul Dietz)
939 * bug fix: CERROR accepts a function as its first argument.
940 * bug fix: NTH an NTHCDR accept a bignum as index
941 arguments. (reported by Adam Warner)
942 * optimization: character compare routines now optimize comparing
943 against a constant character. (reported by Gilbert Baumann)
944 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
945 argument now works properly on the MIPS platform.
946 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
947 * repeated evaluation of the same DEFSTRUCT, a slot of which is
948 declared to have a functional type, does not cause an error
950 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
951 many :DEFAULT-INITARGS. (reported by Istvan Marko)
952 * fixed bug: if last continuation of a deleted block has a
953 destination, this destination should be deleted too. (reported by
955 * fixed a bug in the bootstrap process: the host compiler's values
956 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
957 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
958 test case from Patrik Nordebo)
959 * improved the ability of the disassembler on the PPC platform to
960 provide helpful disassembly notes.
961 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
962 the class in more cases than previously.
963 * bug fix: FILE-POSITION now understands :START and :END for
964 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
965 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
966 8) data. (thanks to David Lichteblau)
967 * bug fix: it is possible to add a method to a generic function
969 * bug fix: reader failed to signal END-OF-FILE inside an
970 object representation. (reported by Nikodemus Siivola)
971 * fixed some bugs revealed by Paul Dietz' test suite:
972 ** LAST and [N]BUTLAST should accept a bignum.
973 ** condition slot accessors are methods.
974 ** (VECTOR NIL) is a subtype of STRING.
976 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
977 * minor incompatible change: some nonsensical specialized lambda
978 lists (used in DEFMETHOD) which were previously ignored now signal
980 * minor incompatible change: the system is now aware of the types of
981 variables in the COMMON-LISP package, and will signal errors for
982 most violations of these type constraints (where previously they
983 were silently accepted).
984 * minor incompatible change: COMPILE-FILE now uses the freedom
985 afforded (ANSI 3.2.2.3) to use derived function types for
986 functions defined in the same file. This also permits the system
987 to warn on static type mismatches and function
988 redefinition. (Currently it does not work with high DEBUG level.)
989 * minor incompatible change: VALUES declaration is disabled.
990 * When issuing notes, the compiler now signals a condition of type
991 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
992 restart for use in user handlers. It is expected that the
993 COMPILER-NOTE condition will eventually become a condition
994 supertype to a hierarchy of note types, which will then be
995 handleable in a similar fashion. However, at the moment, no such
996 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
997 but it's an implementation detail, not a classification for the
999 * Changes in type checking closed the following bugs:
1000 ** type checking of unused values (192b, 194d, 203);
1001 ** template selection based on unsafe type assertions (192c, 236);
1002 ** type checking in branches (194bc).
1003 * A short form of VALUES type specifier has ANSI meaning (it has
1004 increased the number of situations when SBCL cannot perform type
1006 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1007 keywords or constants is permissible.
1008 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1009 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1010 classes. (thanks to Antonio Martinez)
1011 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1012 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1013 argument to a pretty-print stream. Instead, it leaves any such
1014 operations to DESCRIBE-OBJECT methods.
1015 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1017 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1018 lambda lists. (thanks to David Lichteblau)
1019 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1020 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1021 (thanks to Gerd Moellmann)
1022 * fixed bug 246: increased compilation speed of long
1023 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1025 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1027 * DEFCONSTANT now throws a condition of type
1028 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1029 constant to a non-EQL value; CONTINUE and ABORT restarts
1030 respectively change and preserve the value.
1031 * fixed bug 63: The code walker, part of the implementation of CLOS,
1032 is now better at handling symbol macros.
1033 * bug fix: There is no longer an internal implementation type named
1034 CL:LENGTH. (reported by Raymond Toy)
1035 * bug fix: In macro-like defining macros/special operators the
1036 implicit block does not enclose the lambda list.
1037 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1038 enclosed in parentheses) are not suitable as type specifiers, and
1039 their use properly signals an error now.
1040 * bug fix: An argument count mismatch for a type specifier in code
1041 being compiled no longer causes an unhandled error at compile
1042 time, but signals a compile-time warning.
1043 * fixed simple vector readable printing
1044 * bug fix: DESCRIBE takes more care over whether the class
1045 precedence list slot of a class is bound before accessing it.
1046 (reported by Markus Krummenacker)
1047 * bug fix: FORMATTER can successfully compile pretty-printer format
1048 strings which use variants of the ~* directive inside.
1049 * bug fix: SEARCH now applies its TEST predicate to the elements of
1050 the arguments in the correct order. (thanks to Wolfhard Buss)
1051 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1052 in type inference. (thanks to Robert E. Brown)
1053 * bug fix: Reading in symbols with an explicit package name of ""
1054 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1055 * fixed some bugs revealed by Paul Dietz' test suite:
1056 ** NIL is now allowed as a structure slot name.
1057 ** Arbitrary numbers, not just REALs, are allowed in certain
1058 circumstances in LOOP for-as-arithmetic clauses.
1059 ** Multiple class redefinitions before slot access no longer
1060 causes a type error.
1061 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1062 association between the name and a class.
1063 ** Generic functions with non-standard method-combination and over
1064 five methods all of which return constants no longer return NIL
1065 after the first few invocations. (thanks to Gerd Moellmann)
1066 ** CALL-NEXT-METHOD with no arguments now passes the original
1067 values of the arguments, even in the presence of assignment.
1068 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1070 ** DISASSEMBLE works with closures and funcallable instances.
1071 ** ADD-METHOD now returns the generic function, not the new method.
1072 ** FIND-METHOD signals an error if the lengths of the specializers
1073 is incompatible with the generic function, even if the ERRORP
1075 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1076 which its argument is a member.
1077 ** DEFCLASS only redefines the class named by its class-name
1078 argument if that name is the proper name of the class;
1079 otherwise, it creates a new class.
1080 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1081 of the UNBOUND-SLOT condition to the name of the slot.
1082 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1083 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1084 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1085 SLOT-MAKUNBOUND in the specified fashion.
1087 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1088 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1089 cross-compilation host. As a consequence, we can now bootstrap our
1090 way up to SBCL starting with a bare gcc toolchain and human-readable
1091 source code (first the source to CLISP, then the source to SBCL).
1092 * A contributed module containing a partial implementation of the
1093 simple-streams interface has been included. (thanks to Rudi
1095 * A contributed module implementing the RFC1321 Message Digest
1096 Algorithm, known as MD5, has been included.
1097 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1098 no longer has any effect, as the code controlled by this feature
1099 has been deleted. (As far as we know, no-one has ever built using
1100 this feature, and its semantics were confused in any case).
1101 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1102 work on conditions (as required by the ANSI specification),
1103 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1104 expected behaviour on conditions. Users should note, however,
1105 that such behaviour is not required by the ANSI specification,
1106 and so use of this behaviour may render their code unportable.
1107 * fixed some bugs revealed by Paul Dietz' test suite:
1108 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1110 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1111 ** SXHASH on condition objects no longer returns NIL.
1112 ** :ALLOCATION :CLASS slots are better treated; their values are
1113 updated on class redefinition, and initforms inherited from
1114 superclasses are applied.
1115 ** REMOVE-METHOD returns its generic function argument even when
1116 no method was removed.
1117 ** SHARED-INITIALIZE now initializes the values of the requested
1118 slots, including those with :ALLOCATION :CLASS.
1119 ** ALLOCATE-INSTANCE now works on structure classes defined via
1120 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1122 ** SLOT-EXISTS-P now works on conditions, as well as structures
1124 ** MAKE-LOAD-FORM now has the required methods on
1125 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1126 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1127 keyword, and now implements the SLOT-NAMES argument.
1128 ** methods with &OPTIONAL arguments no longer allow too many
1129 arguments to be passed in the call without error.
1130 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1131 option is consistent with the required arguments of the generic
1132 function lambda list.
1133 * bug fix: REQUIRE accepts a string designator. (Thanks to
1135 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1136 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1137 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1138 * bug fix: The compiler now checks for duplicated variables in macro
1140 * bug fix: SETQ on globals returns the correct value.
1141 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1142 not a COMPILER-ERROR (followed by some other strange error on
1143 choosing the CONTINUE restart).
1144 * bug fix: make.sh and friends are now more consistent in the way that
1145 they look for GNU "make".
1147 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1148 * experimental native threads support (on x86 Linux >=2.4 only).
1149 This is not compiled in by default: you need to add :SB-THREAD to
1150 the target features. See the "Beyond ANSI" chapter of the manual
1152 * fix for longstanding nonANSIism: The old distinction between
1153 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1154 The return value from CL:FIND-CLASS is now a CLOS class, and
1155 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1156 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1157 * An interface to the MetaObject Protocol, as described in Kiczales,
1158 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1159 MIT Press, 1991, is available from the SB-MOP package.
1160 * incompatible change: the SB-PCL package should now be considered
1161 a private implementation detail, and no longer a semi-private MOP
1163 * minor incompatible change: due to rearrangement for threads, the
1164 control stack and binding stack are now allocated at arbitrary
1165 addresses instead of being hardcoded per-port. Users affected by
1166 this probably have to be doing advanced things with shared
1167 libraries, and will know who they are.
1168 * minor incompatible change: Previously, all --eval forms used were
1169 processed with READ before any of them were processed with EVAL.
1170 Now each --eval form is processed with both READ and EVAL before
1171 the next --eval form is processed. (Thus package operations like
1172 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1173 work as the user might reasonably expect.)
1174 * minor incompatible change: *STANDARD-INPUT* is now only an
1175 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1177 * minor incompatible change: Y-OR-N-P is now character-oriented,
1178 not line oriented. Also, YES-OR-NO-P now works without errors.
1179 (thanks to Antonio Martinez)
1180 * sb-aclrepl module improvements: an integrated inspector, added
1181 repl features, and a bug fix to :trace command.
1182 * Known functions, which cannot be open coded by the backend, are
1183 considered to be able to check types of their arguments. (fixing
1184 a bug report by Nathan J. Froyd)
1185 * fixed a bug in computing method discriminating functions: It is
1186 now possible to define methods specialized on classes which have
1187 forward-referenced superclasses. (thanks to Gerd Moellmann)
1188 * fixed evaluation order in optional entries (reported by Gilbert
1190 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1191 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1192 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1193 required-by-AMOP NAME argument, as well as CLASS and
1194 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1195 * fixed bug 20: DEFMETHOD can define methods using names that are
1196 not the proper names of classes to designate class specializers.
1197 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1198 Unix file descriptors, instead of blowing up. (thanks to Antonio
1200 * Garbage collection refactoring: user-visible change is that a
1201 call to the GC function during WITHOUT-GCING will not do garbage
1202 collection until the end of the WITHOUT-GCING. If you were doing
1203 this you were probably losing anyway.
1204 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1205 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1206 TYPEP the latter but not the former.
1207 * The compiler issues a full WARNING for calls to undefined functions
1208 with names from the CL package.
1209 * MAP-INTO for a vector destination is open coded. (reported by
1210 Brian Downing on c.l.l)
1211 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1212 documentation string.
1213 * fixed some bugs revealed by Paul Dietz' test suite:
1214 ** COPY-ALIST now signals an error if its argument is a dotted
1216 ** Condition slots are now accessed more correctly in the presence
1217 of multiple initargs for a given slot.
1218 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1219 exclude restarts of the same name associated with a different
1221 ** DEFCLASS of forward-referenced classes with another
1222 forward-referenced class in the superclasses list no longer
1224 ** Condition slots are now initialized once each, not multiple
1225 times. (thanks to Gerd Moellmann)
1226 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1228 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1229 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1230 arguments contain duplicated elements.
1231 ** RESTART-CASE understands local macros.
1232 ** RESTART-CASE associates exactly its own restarts with a condition.
1233 ** ENDP in safe mode checks its argument to be of type LIST.
1234 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1235 in question is unbound.
1236 ** Optimized MAKE-INSTANCE functions no longer cause internal
1237 assertion failures in the presence of duplicate initargs.
1238 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1239 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1240 propagated even in the ctor optimized implementation of
1242 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1244 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1245 * a better implementation of SXHASH on (simple) bit vectors,
1246 measured both in execution speed and in distribution of results
1247 over the positive fixnums, has been installed. Likewise, a better
1248 implementation of EQUAL for simple bit vectors is now available.
1249 * fixed CEILING optimization for a divisor of form 2^k.
1250 * fixed bug 240 (emitting extra style warnings "using the lexical
1251 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1252 by Antonio Martinez)
1253 * fixed SXHASH, giving different results for NIL depending on type
1254 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1255 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1256 variable. (found by Rolf Wester)
1257 * SBCL does not ignore type declarations for special
1258 variables. (reported by rif on c.l.l 2003-03-05)
1259 * some bug fixes in contrib/sb-aclrepl/
1260 * fixed some bugs revealed by Paul Dietz' test suite:
1261 ** a bug in the CONS type specifier, whereby the CAR and CDR
1262 types got intertwined, has been fixed;
1263 ** the type system is now able to reason about the interaction
1264 between INTEGER and RATIO types more completely;
1265 ** APPEND, [N]REVERSE and NRECONC check that those their
1266 arguments, which must be proper lists, are really so;
1267 ** An array specialized to be unable to hold elements has been
1268 implemented, as required -- yes, really -- by ANSI;
1269 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1270 on malformed property lists;
1272 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1273 * incompatible packaging change: in line with Unix convention,
1274 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1275 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1276 * REQUIRE and PROVIDE are now optionally capable of doing something
1277 useful. See the documentation string for REQUIRE.
1278 * infrastructure for a managed SBCL contrib system: contributed
1279 modules in this release include:
1280 ** the ASDF system definition facility;
1281 ** an interface to the BSD Sockets API;
1282 ** an ACL-like convenience interface to the repl;
1283 (thanks to Kevin Rosenberg)
1284 ** an implementation of ROTATE-BYTE, with efficient implementation
1286 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1287 gives a closer approximation to the right answer than previously.
1288 (thanks to Raymond Toy)
1289 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1290 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1291 optional environment arguments, as required by ANSI.
1292 * fixed bugs in other functions taking environment objects, allowing
1293 calls with an explicit NIL environment argument to be compiled
1295 * fixed bug 228: primary return values from
1296 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1297 COMPILE or FUNCTION.
1298 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1299 :TYPE will no longer signal errors on innocuous objects.
1300 * fixed bug 231b: SETQ is better at respecting type declarations in
1301 the lexical environment.
1302 * fixed a bug in DEFCLASS: classes named by symbols with no or
1303 unprintable packages can now be defined.
1304 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1305 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1306 * fixed compiler failure related to checking types of functions.
1307 (reported by Robert E. Brown)
1308 * the compiler is now much more consistent in its error-checking
1309 treatment of bounding index arguments to sequence functions: in
1310 (SAFETY 3) code, errors will be signalled in almost all cases if
1311 invalid sequence bounding indices are passed to functions defined
1312 by ANSI to operate on sequences.
1313 * fixed a bug in the build procedure: documentation of SBCL-specific
1314 packages is now preserved and available in the final Lisp image.
1315 * lifted FDEFINITION lookup out of loops in the implementation of
1316 many list operations. (thanks to Robert E. Brown)
1317 * fixed a bug in the reader: the #n# reader macro now works for
1318 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1319 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1320 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1321 E. Brown and Tony Martinez respectively)
1322 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1323 SLOT-BOUNDP: in particular, it is now able to optimize them much
1324 better, and is now not vulnerable to having packages renamed.
1325 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1326 always called when a slot is not present in an instance. (thanks
1328 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1329 longer signal an error on unreading a character following EOF on
1330 the previous constituent stream. (thanks to Tony Martinez)
1331 * fixed some bugs revealed by Paul Dietz' test suite:
1332 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1333 not just nonnegative fixnums;
1334 ** the logical bit-array operators such as BIT-AND now accept an
1335 explicit NIL for their "opt-arg" argument (to indicate a
1336 freshly-consed result bit-array);
1337 ** ELT now signals an error on an invalid sequence index in safe
1339 ** the type system is now cleverer about negations of numeric
1340 types, and consequently understands the BIGNUM and RATIO types
1342 ** the type system is now cleverer about the interaction between
1343 INTEGER and RATIO types: while bugs still remain, many more
1344 cases are accurately computed;
1345 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1346 if it is in the last clause;
1347 ** CONSTANTLY now correctly returns a side-effect-free function in
1349 ** DECLARE is no longer treated as a special-operator; in
1350 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1351 * incremented fasl file version number due to the change in the
1352 DEFSTRUCT-SLOT-DESCRIPTION structure.
1354 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1355 * minor incompatible change: code processed by the "interpreter" or
1356 EVAL now has a compilation optimization policy of (DEBUG 2)
1357 (changed from (DEBUG 1)) to improve debuggability of interactive
1358 development, and to allow the use of the debug RETURN command in
1360 * an experimental implementation of the RETURN command for the
1361 debugger has been included. (thanks to Frederik Kuivinen)
1362 * fixed bug 62: constraints were not propagated into a loop.
1363 * fixed bug in embedded calls of SORT (reported and investigated by
1365 * fixed some bugs revealed by Paul Dietz' test suite:
1366 ** printing and reading of arrays with some dimensions having
1367 length 0 (thanks to Gerd Moellmann);
1368 ** BOA constructor with &AUX argument without a default value does
1369 not cause a type error;
1370 ** CONSTANTP now returns true for all self-evaluating objects.
1372 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1373 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1374 accessors that are related by inheritance, as specified in the
1375 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1376 Valtteri Vuorikoski)
1377 * The compiler is now able to inline functions that were defined in
1378 a complex lexical environment (e.g. inside a MACROLET).
1379 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1381 * Support for the upcoming FreeBSD-5.0 release has been included.
1382 (thanks to Dag-Erling Smorgrav)
1383 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1384 effect when it is not in a toplevel context.
1385 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1386 stand a better chance of being correct. (thanks to Gerd
1388 * fixed bug in COERCE, which now signals an error on coercing a
1389 rational to a bounded real type which excludes the expected
1391 * The compiler is now able to derive types more accurately from the
1392 COERCE and COMPILE functions.
1393 * fixed bug 223: functional binding is considered to be constant
1394 only for symbols in the CL package.
1395 * fixed bug 231: SETQ did not check the type of a variable being set
1396 (reported by Robert E. Brown)
1397 * A new optimization for MAKE-INSTANCE has been included, fixing
1398 various bugs (including relating to :ALLOCATION :CLASS slots and
1399 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1401 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1402 ** As required by ANSI, LOOP now disallows anonymous collection
1403 clauses such as COLLECT I in conjunction with aggregate boolean
1404 clauses such as THEREIS (= I 1);
1405 ** LOOP now signals an error when any variable is reused in the
1406 same loop (including the potentially useful construct analogous
1407 to WITH A = 1 WITH A = (1+ A);
1408 ** IT is only a special loop symbol within the first clause of a
1409 conditional loop clause;
1410 ** LOOP with a typed iteration variable over a hashtable now
1411 signals a type error iff it should.
1412 * fixed some other bugs revealed by Paul Dietz' test suite:
1413 ** FILE-STREAM now names the class previously known as FD-STREAM;
1414 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1415 argument) no longer signals an error;
1416 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1417 of the slot symbol, rather than using the current package
1418 ((:CONC-NAME "") continues to intern the slot's name in the
1420 * incremented fasl file version number, because of the incompatible
1421 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1422 of the new implementation of DEFINE-COMPILER-MACRO.
1424 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1425 * Support for building SBCL for MIPS platforms running in
1426 little-endian mode has now been checked in, and basic
1427 functionality on said platforms verified.
1428 * minor incompatible change: PCL now records the pathname of a file
1429 in which methods and the like are defined, rather than its
1431 * minor incompatible change: TRUENAME now considers the truename of
1432 a file naming a directory to be the pathname with :DIRECTORY
1433 component indicating that directory.
1434 * minor incompatible change: a NAMED clause in the extended form of
1435 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1436 reason for the previous behaviour is unclear.
1437 * more systematization and improvement of CLOS and MOP conformance
1438 in PCL (thanks to Gerd Moellman and Pierre Mai):
1439 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1441 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1442 primary methods with no specializers;
1443 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1445 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1446 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1447 FORWARD-REFERENCED-CLASSes; error reporting on
1448 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1450 ** SXHASH on CLOS instances now uses a slot internal to the
1451 instance to return different numbers on distinct instances,
1452 while preserving the same return value through invocations of
1454 ** DEFMETHOD signals errors when methods with longer incongruent
1455 lambda lists are added to generic functions;
1456 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1457 CLASS, as specified in AMOP;
1458 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1459 on the order returned by the primary method for classes of
1460 class STANDARD-CLASS;
1461 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1462 * fixed some bugs shown by Paul Dietz' test suite:
1463 ** DOLIST puts its body in TAGBODY;
1464 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1466 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1467 value producing form;
1468 ** if more variables are given to PROGV than values, extra
1469 variables are bound and made to have no value;
1470 ** NSUBSTITUTE on list arguments gets the right answer with
1472 ** ELT signals an error of type TYPE-ERROR when the index argument
1473 is not a valid sequence index;
1474 ** LOOP signals (at macroexpansion time) an error of type
1475 PROGRAM-ERROR when duplicate variable names are found;
1476 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1477 ** FUNCALL of special-operators now cause an error of type
1479 ** PSETQ now works as required in the presence of side-effecting
1480 symbol-macro places;
1481 ** NCONC accepts any object as its last argument;
1482 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1484 ** loop-for-as-package does not require a package to be explicitely
1486 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1488 * fixed bug 166: compiler preserves "there is a way to go"
1489 invariant when deleting code.
1490 * fixed bug 172: macro lambda lists with required arguments after
1491 &REST arguments now cause an error to be signalled. (thanks to
1493 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1494 bug. (thanks to Matthew Danish)
1495 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1497 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1498 arguments are lexically rebound. (thanks to Gerd Moellmann and
1500 * fixed bug 194: error messages are now more informative when there
1501 is no primary method applicable in a call to a generic
1502 function. (thanks to Gerd Moellmann)
1503 * fixed bug in command line argument checking (thanks to Julian
1505 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1506 (thanks to Matthew Danish)
1507 * incremented fasl file version number, because of the
1508 SXHASH-related changes in the layout of CLOS data structures
1510 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1511 * minor incompatible change: The runtime (the Unix executable named
1512 "sbcl") is now much pickier about the .core files it will load.
1513 Essentially it now requires .core files to descend from the same
1514 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1515 as the runtime does. (The intent is to prevent the crashes which
1516 can occur, and which can even be reported as mysterious failures,
1517 when people patch the sources or change the build parameters
1518 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1519 sbcl and .core files.)
1520 * fixed bug: VALUES-LIST is no longer optimized away.
1521 * fixed bug 142: The FFI conversion of C string values to Lisp
1522 string values no longer conses excessively. (thanks to Nathan
1523 Froyd porting Raymond Toy's fix to CMU CL)
1524 * began to systematize and improve MOP conformance in PCL (thanks to
1525 Nathan Froyd, Gerd Moellman and Pierre Mai):
1526 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1528 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1529 ** FINALIZE-INHERITANCE is now called on class finalization;
1530 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1531 argument precedence order.
1532 * fixed bug 202: The compiler no longer fails on functions whose
1533 derived types contradict their declared type.
1534 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1535 so it can be non-toplevel.
1536 * The fasl file version number has changed (because of the new
1537 implementation of DEFMACRO).
1538 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1539 safe code, that any length requirement by their type-specifier
1540 argument is valid. The exceptions to this are described in bug
1542 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1543 functions now check the types of their inputs as required by ANSI.
1544 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1545 introduced symbol is DECLAREd to be SPECIAL.
1546 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1547 * fixed bug: :COUNT argument to sequence functions may be negative
1548 * fixed bug: body of DO-SYMBOLS may contain declarations
1549 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1550 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1551 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1552 symbol macro only once
1553 * fixed printing of call frame when argument list is unavailable
1554 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1555 * compiler no longer signals WARNING on unknown keyword
1558 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1559 * A beta-quality port to the mips architecture running Linux,
1560 based on the old CMUCL backend, has been made. It has been tested
1561 on a big-endian kernel, and works sufficiently well to be able to
1562 rebuild itself; it has not been tested in little-endian mode.
1563 * fixed an inconsistency between gencgc.c and purify.c which made
1564 dumping/loading .core files unreliable
1565 * fixed bug 120a: The compiler now deals correctly with IFs where
1566 the consequent is the same as the alternative, instead of
1567 misderiving the return type. (thanks to Alexey Dejneka)
1568 * fixed bug 113: Logical pathnames are now dumpable (the logical
1569 host is resolved at load-time, throwing an error if it is not
1571 * fixed bug 174: FORMAT's error message is slightly clearer when a
1572 non-printing character is used in a format directive.
1573 * fixed several bugs in compiler checking of type declarations, i.e.
1574 violations of the Python "declarations are assertions" principle
1575 (thanks to Alexey Dejneka)
1576 * fixed several bugs in PCL's error checking (thanks to Gerd
1578 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1580 * fixed bug in compilation of functions as first class values
1581 (thanks to Antonio Martinez-Shotton)
1582 * The compiler's handling TYPE-ERRORs which it can prove will
1583 inevitably happen at runtime has been cleaned up and corrected
1584 in several ways. (thanks to Alexey Dejneka)
1585 * improved argument type checking for various basic arithmetic
1586 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1587 lavished on them in the past that they can be compiled in many
1588 ways in different special cases
1589 * fixed bug 181: compiler checks validity of user supplied type
1591 * cleaned up code flushing in optimization: Function calls which
1592 should signal errors for safety purposes (e.g. which ANSI says
1593 should signal errors when their arguments are of incorrect type)
1594 are no longer optimized away.
1595 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1596 * incremented fasl file version number, because changes in the
1597 implementation of sequence functions like COERCE caused
1598 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1599 in old inline expansions) to become undefined. (Actually these
1600 changes were later undone, so we might very well be binary
1601 compatible with 0.7.7 after all, but leaving the version number
1602 incremented seemed like the simplest and most conservative
1605 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1606 * An alpha-quality port to the parisc architecture running Linux,
1607 based on the old CMUCL backend, has been made. This, even more so
1608 than the other backends, should be considered still a work in
1609 progress; known problems include that the Linux kernel in 64-bit
1610 mode does not propagate the correct sigcontext structure to
1611 userspace, and consequently SBCL on a parisc64 kernel will not
1613 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1614 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1615 suppressed.) Also now that the compiler is looking at declarations
1616 in the environment, it checks optimization declarations as well,
1617 and suppresses inlining when (> DEBUG SPEED).
1618 * More fixes have been made to treatment of floating point exception
1619 treatment and other Unix signals. In particular, floating point
1620 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1621 * The detection and handling of control stack exhaustion (infinite
1622 or very deeply nested recursion) has changed. Stack exhaustion
1623 detection is now done by write-protecting pages at the OS level
1624 and applies at all optimization settings; when found, a
1625 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1626 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1627 be caught using IGNORE-ERRORS.
1628 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1629 correctly when transforming recursive LABELS functions to LETs.
1630 (thanks to Alexey Dejneka)
1631 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1632 that are names of constants or global variables.
1633 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1634 alien routines with docstrings.
1635 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1636 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1638 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1639 (thanks to Eric Marsden)
1640 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1641 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1642 * Bug fix: The SPARC backend can now compile functions involving
1643 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1644 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1645 to a functions expecting a PATHNAME-DESIGNATOR.
1646 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1647 lambda lists. (thanks to Alexey Dejneka)
1648 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1649 functions so that EXPORTing the name of the class doesn't cause
1650 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1651 to Antonio Martinez for reporting this)
1652 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1653 constant-folded without failing an assertion. (thanks to Einar
1654 Floystad Dorum for reporting this)
1655 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1656 types are now dealt with more correctly by the compiler.
1657 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1658 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1659 behaviour analogous to RENAME-FILE. This puts its behaviour more
1660 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1662 * The fasl file version number has changed again. (because of the
1663 bug fix involving the names of PCL MAKE-INSTANCE functions)
1665 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1666 * bug fix: Floating point exceptions are treated much more
1667 consistently on the x86/Linux and PPC/Linux platforms.
1668 * Array initialization with :INITIAL-ELEMENT is now much faster for
1669 cases when the compiler cannot open code the array creation, but
1670 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1671 array accesses have also seen a speed increase.
1672 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1673 specified. (This was at the root of some bad interactions between
1674 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1676 * bug fix: Internal error arguments for undefined functions are now
1677 computed correctly on the PPC/Linux platform.
1678 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1679 Raymond Toy's patch for CMU CL)
1680 * Support for the Solaris 9 operating environment has been included
1681 (thanks to Daniel Merritt)
1682 * A very ugly but hopefully complete draft of the missing FFI chapter
1683 of the manual has been created by reformatting the corresponding
1684 CMU CL manual chapter into (currently very ugly and incoherent)
1685 DocBook and bringing it up to date for SBCL behavior. Thus, the
1686 manual is now essentially complete, at least by my extreme
1687 once-and-only-once standards, whereby it's acceptable to refer to
1688 the doc strings of SB-EXT functions as the primary documentation.
1689 * The fasl file version number has changed again, due to cleanup of
1690 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1691 is no longer a static symbol.)
1693 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1694 * SBCL now builds with OpenMCL (version 0.12) as the
1695 cross-compilation host; also, more progress has been made toward
1696 bootstrapping under CLISP.
1697 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1699 * bug 158 fixed: The compiler can now deal with integer loop
1700 increments different from 1; fixing this turned out also to fix
1702 * bug 169 fixed: no more bogus warnings about using lexical bindings
1703 despite the presence of perfectly good SPECIAL declarations (thanks
1704 to David Lichteblau)
1705 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1706 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1707 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1709 * bug 180 fixed: Method combination specifications no longer ignore
1710 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1711 * bug fix: Structure type predicate functions now check their argument
1712 count as they should.
1713 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1714 correctly. (thanks to Pierre Mai)
1715 * minor incompatible change: The --noprogrammer option is deprecated
1716 in favor of the new --disable-debugger option, which is very similar.
1717 (The major difference is that it takes effect at a slightly different
1718 time at startup, causing handling of errors in --sysinit and
1719 --userinit files will be affected differently.) The
1720 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1721 been added to allow this functionality to be controlled from ordinary
1722 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1723 maintainers, who might want to run non-interactive scripts to
1724 build SBCL cores which will later be used interactively.)
1725 * minor incompatible change: The LOAD function no longer, when given
1726 a wild pathname to load, loads all files matching that pathname.
1727 Instead, an error of type FILE-ERROR is signalled.
1729 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1730 * bug 147 fixed: The compiler preserves its block link/count
1731 invariants more correctly now so that it doesn't crash. (thanks
1733 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1735 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1736 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1737 no-value-for-this-character as NIL (as specified by ANSI).
1738 * HOST-NAMESTRING on physical pathnames now returns a string that is
1739 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1740 (thanks to Christophe Rhodes)
1741 * The Alpha port handles icache flushing more correctly. (thanks to
1743 * More progress has been made toward bootstrapping under CLISP. (thanks
1744 to Christophe Rhodes)
1745 * The fasl file format has changed again, because dynamic loading
1746 on OpenBSD (which has non-ELF object files) motivated some cleanups
1747 in the way that foreign symbols are transformed and passed around.
1748 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1749 is no longer treated as whitespace by the reader, but instead as
1750 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1751 |A
\7fB|, instead of A as it used to.
1753 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1754 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1755 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1756 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1757 this as of 0.7.1.45, but was left out of the previous news section
1758 (thanks to Dan Barlow)
1759 * SBCL now runs on the Solaris operating system on SPARC architectures
1760 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1761 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1762 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1763 * SPARC backend cleanups, allowing builds of cores optimized for V8
1764 and V9 SPARCS, and also emission of code targeted to a particular
1765 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1767 * SBCL is closer to bootstrapping under CLISP, thanks to various
1768 fixes by Christophe Rhodes.
1769 * The fasl file format has changed again, to allow the compiler's
1770 INFO database to support symbol macros.
1771 * The user manual (in doc/) is formatted into HTML more nicely.
1772 (thanks to coreythomas)
1773 * The system is smarter about SUBTYPEP relationships, especially
1774 those involving NOT types (including types such as ATOM which are
1775 represented internally using NOT types). Thus SUBTYPEP is less
1776 likely to return (VALUES NIL NIL) in general, and in particular
1777 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1779 * The fasl file format has changed again, because the internal
1780 representation of types now includes a new slot to support the new
1781 SUBTYPEP-of-NOT-types logic.
1782 * (not a change in the main branch of SBCL, but a related prototype
1783 which can hopefully be merged into the main branch of SBCL in the
1784 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1785 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1786 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1787 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1788 dispatch macro character combinations. (thanks to Alexey Dejneka)
1789 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1790 that superficially look like logical namestrings correctly.
1791 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1793 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1794 default supplied-p) for &optional and &key arguments. (thanks to
1797 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1798 * incompatible change: The compiler is now less aggressive about
1799 tail call optimization, doing it only when (> SPACE DEBUG) or
1800 (> SPEED DEBUG). (This is an incompatible change because there are
1801 programs which relied on the old CMU-CL-style behavior to optimize
1802 away their unbounded recursion which will now die of stack overflow.)
1803 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1804 for non-GENCGC systems has been increased to 20M (since that
1805 seems much closer to the likely performance optimum for modern
1806 systems than the old 4M value was)
1807 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1808 and *DEBUG-PRINT-LEVEL*
1809 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1810 of CMU CL's support for SPARC, and various endianness and other
1811 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1812 * new syntactic sugar for the Unix command line: --load foo.bar is now
1813 an alternate notation for --eval '(load "foo.bar")'.
1815 ** The system now detects stack overflow and handles it gracefully,
1816 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1817 optimization settings. (This is a good thing in general, and
1818 its introduction in this version should be particularly timely
1819 for anyone whose code fails because of suppression of tail
1821 ** The system now hunts for the C variable "environ" in a more
1822 devious way, to avoid segfaults when the C library version differs
1823 between compile time and run time. (thanks to Christophe Rhodes)
1824 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1825 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1826 ** The compiler no longer issues bogus style warnings for undefined
1827 classes in the same source file as the DEFCLASSes which defined
1828 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1830 ** fixes in CONDITION class precedence list for undefined function
1831 errors (thanks to Alexei Dejneka)
1832 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1833 correctly. (thanks to Dan Barlow)
1834 ** portability fixes aiming at bootstrapping under CLISP (thanks
1835 to Dave McDonald and Christophe Rhodes)
1836 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1837 ** fixes in type translation and and type inference (thanks to
1839 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1840 ** various fixes in the new ports (thanks to Dan Barlow)
1841 * several changes related to debugging:
1842 ** suppression of tail recursion, as noted above
1843 ** stack overflow detection, as noted above
1844 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1845 is now the default. (For some time encapsulation has been more
1846 reliable than the breakpoint-based :ENCAPSULATE NIL
1847 implementation, at least on X86 systems; and I just noticed that
1848 encapsulation also seems closer to the spirit of the ANSI
1851 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1853 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1854 up properly again. (There was a packaging bug in 0.7.0 which
1855 left their definitions in SB-SYS::LOAD-FOREIGN and
1856 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1857 vital for most things which interface to C-level interfaces,
1858 like extensions working with sockets or databases or
1859 Perl-compatible regexes or whatever, and the need to fix
1860 this bug is the main reason that 0.7.1 was released so
1862 ** DEFGENERIC is now choosier about the methods it redefines, so that
1863 reLOADing a previously-LOADed file containing DEFGENERICs does
1864 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1865 cycle now works as it should. (thanks to Alexey Dejneka)
1866 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1867 fixed by Christophe Rhodes. (He actually submitted this patch
1868 months ago, and I delayed until after 0.7.0.)
1869 ** Bug 111 (internal compiler confusion about runtime checks on
1870 FUNCTION types) was fixed by Alexey Dejneka.
1871 * Some internal cleanups (getting rid of variables which aren't
1872 needed now that the byte interpreter is gone) caused the fasl
1873 file format number to change again.
1875 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1876 * major incompatible change: The default fasl file extension, i.e. the
1877 default extension for files produced by COMPILE-FILE, has changed
1878 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1880 ** There are many changes in the implementation of the compiler.
1881 SBCL is now essentially a compiler-only implementation of ANSI
1882 Common Lisp. EVAL still "interprets" a few special cases, but
1883 almost all the interesting cases are handled by creating
1884 a LAMBDA expression, calling COMPILE on it, then calling
1885 FUNCALL on the result.
1886 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1887 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1888 Since the code is newer, there might still be some new bugs
1889 (though not as many as before Martin Atzmueller's fixes:-). But
1890 the new code is substantially simpler and clearer, and hopefully
1891 any remaining bugs will be simpler, less fundamental, and more
1892 fixable then the bugs in the old code.
1893 ** The revised compiler is still a little unsteady on its feet.
1895 *** The debugging information it produces (particularly the names
1896 of FUNCTION objects) is sometimes much less useful than what
1897 the old compiler produced.
1898 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1899 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1900 within a MACROLET) has been temporarily weakened.
1901 ** There are new compiler optimizations for various functions:
1902 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1903 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1904 *** the math functions TRUNCATE, FLOOR, and CEILING
1905 *** the function-of-all-trades COERCE
1906 Mostly these should be transparent, but there's one
1907 potentially-annoying problem (bug 117): when the compiler
1908 inline-expands a function and does type analysis on the result,
1909 it can create control paths which have type mismatches, and
1910 when it can't prove that those control paths aren't taken,
1911 it will issue WARNINGs about the type mismatches. This is
1912 a particular problem in practice for the new sequence functions.
1913 It's not clear how this should be fixed, and for now, a
1914 workaround is given in the entry for 117 in the BUGS file.
1915 ** (Because of the interaction between the two previous items --
1916 occasional inlining problems and new inline expansions -- some
1917 of the new sequence function optimizations won't really kick in
1918 completely until debugging information, and then inlining, are
1919 straightened out in some future version.)
1920 * minor incompatible changes:
1921 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1922 DIRECTORY no longer implicitly promotes NIL slots of its
1923 pathname argument to :WILD. In particular, when you ask for the
1924 contents of a directory (which you used to be able to do without
1925 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1926 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1927 ** changes in behavior that ANSI explicitly defines to be
1928 implementation dependent:
1929 *** The new compiler-only implementation still conforms with ANSI,
1930 but acts a little different than before. Besides the obvious
1931 changes in performance tradeoffs (that the cost per form passed
1932 to EVAL has gone up, and the cost per form executed by EVAL
1933 has gone down), the behavior of the system changes a little
1934 because there are no longer any interpreted function objects.
1935 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1936 e.g. doing COMPILE on the output of interactive DEFUN is
1938 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1940 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1941 has changed from (:CL) to NIL.
1942 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1943 e.g. #\Newline and #\Tab, don't have names specified in the
1944 ANSI Common Lisp standard, is now based on their ASCII symbolic
1945 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1946 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1947 are no longer used for output.
1948 ** changes in internal implementation constants:
1949 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1950 4 million. (If your application spends a lot of time GCing and
1951 you have a lot of RAM, you might want to experiment with
1952 increasing it even more.)
1953 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1954 However, almost all old code should still continue to work without
1955 immediate update, as SB-C-CALL is now a (deprecated) nickname
1957 ** Old operator names in the style DEF-FOO are now deprecated in
1958 favor of new corresponding names DEFINE-FOO, for consistency with
1959 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1960 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1961 internal symbols, but a few supported extensions like
1962 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1963 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1964 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1965 etc. as you get deeper into recursive calls to the debugger
1966 command loop, instead of the old "5]", "5]]", "5]]]"
1967 sequence. (I was motivated to do this when squabbles between
1968 ILISP and SBCL left me very deeply nested in the debugger. In the
1969 short term, this change will probably provoke more ILISP/SBCL
1970 squabbles, but hopefully it will be an improvement in the long run.)
1971 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1972 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1973 function, since declaiming return types involving aliens is
1974 (1) annoyingly messy to do by hand and (2) vital to efficient
1975 compilation of code which calls such functions.
1976 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1977 longer reexported by the SB-EXT package. They're solely useful
1978 for alien code, so it seems more logical that you should get
1979 them from the SB-ALIEN package, not in SB-EXT.
1980 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1981 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1982 features. Instead, the code that they used to control is always
1983 built into the system.
1984 * many other bug fixes
1985 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1986 advantage of the new EVAL-WHEN stuff and to clean them up in
1987 general, and they are now more ANSI-compliant in a number of
1988 ways. Martin Atzmueller is responsible for a lot of this.
1989 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1991 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1992 *** correct ERROR type for various file operations
1993 *** some fixes for Lisp streams
1994 *** DEFMETHOD syntax checking
1995 *** changing old weird representation of debug information as
1996 strings (which, among their other deficiencies, don't transform
1997 correctly when you rename packages, and don't change their
1998 print representation when you change things like *PACKAGE*
1999 and *PRINT-LENGTH*) to symbols and lists of symbols
2000 He also made several improvements and fixed several bugs in DESCRIBE.
2001 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2003 *** misbehavior of WRITE-STRING/WRITE-LINE
2004 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2005 and several other LOOP problems as well
2006 *** DIRECTORY when similar filenames are present
2007 *** DEFGENERIC with :METHOD options
2008 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2009 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2010 *** argument ordering in FIND with :TEST option
2011 *** mishandled package designator argument in APROPOS-LIST
2012 *** various problems in the backquote readmacro
2013 *** a bug in APROPOS
2014 *** probably some others that I'm not describing very well here,
2015 since the CVS log documents them by reference to sbcl-devel
2016 messages, and the SourceForge archives aren't working well.:-(
2017 ** Dan Barlow improved the Alpha port (and is making progress on the
2018 PPC port, for those of you who think different).
2019 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2020 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2021 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2022 continues to work on the SPARC port (for those of you in a position
2023 to look down upon our little PC-compatible boxes from a great height).
2024 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2025 and uses that copy, rather than the current dynamic value, when
2026 it's trying to decide whether to truncate output. Thus e.g.
2027 (let ((*print-lines* 50))
2028 (pprint-logical-block (stream nil)
2030 (let ((*print-lines* 8))
2031 (print (aref possiblybigthings i) stream)))))
2032 should now truncate the logical block only at 50 lines, instead of
2033 often truncating it at 8 lines, as it did before.
2034 * The doc/cmucl/ directory, containing old CMU CL documentation
2035 from the time of the fork, is no longer part of the base system.
2036 SourceForge has shut down its anonymous FTP service, and with it
2037 my original plan for distributing the old CMU CL documentation
2038 there. For now, if you need these files you can download an old
2039 SBCL source release and extract them from it.
2040 * The fasl file version number changed again, for dozens of reasons,
2041 some of which are apparent above.
2043 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2044 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2045 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2046 patch, so that DISASSEMBLE output is much nicer.
2047 * The code in the SB-PROFILE package now seems reasonably stable.
2048 I still haven't decided what the final interface should look like
2049 (I'd like PROFILE to interact cleanly with TRACE, since both
2050 facilities use function encapsulation) but if you have a need
2051 for profiling now, you can probably use it successfully with
2052 the current CMU-CL-style interface.
2053 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2054 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2055 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2056 link now returns the dangling link itself, and for similar
2057 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2058 link itself. (In these cases the old code signalled an error and
2059 looped endlessly, respectively.) Thus, DIRECTORY now works even
2060 in the presence of dangling and cyclic symbolic links.
2061 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2062 is now a supported extension again, since the consensus on
2063 sbcl-devel was that it can be useful for ordinary development
2064 work, not just for debugging SBCL itself.
2065 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2066 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2067 that currently-defined functions might be redefined later with
2068 different return types.
2069 * Hash tables can be printed readably, as inspired by CMU CL code
2070 of Eric Marsden and SBCL code of Martin Atzmueller.
2071 * better error handling in CLOS method combination, thanks to
2072 Martin Atzmueller porting Pierre Mai's CMU CL patches
2073 * more overflow fixes for >16Mbyte I/O buffers
2074 * A bug in READ has been fixed, so that now a single Ctrl-D
2075 character suffices to cause end-of-file on character streams.
2076 In particular, now you only need one Ctrl-D at the command
2077 line (not two) to exit SBCL.
2078 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2080 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2081 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2082 does the right thing.
2083 * The implementation of some type tests, especially for CONDITION
2084 types, is now tidier and maybe faster, due to CMU CL code
2085 originally by Douglas Crosher, ported by Martin Atzmueller.
2086 * Some math functions have been fixed, and there are new
2087 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2088 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2089 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2090 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2091 likely to be built into sbcl-0.7.0.)
2092 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2093 CL, ported to SBCL by Martin Atzmueller)
2094 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2095 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2096 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2097 * There's a new slam.sh hack to shorten the edit/compile/debug
2098 cycle for low-level changes to SBCL itself, and a new
2099 :SB-AFTER-XC-CORE target feature to control the generation of
2100 the after-xc.core file needed by slam.sh.
2101 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2102 COMPILE-FILE is no longer supported, so that now every function
2103 gets an entry point, so that block compilation looks a little
2104 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2105 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2106 returns the number of bytes consed since the system started,
2107 rather than the number consed since the first time the function
2108 was called. (The new definition parallels ANSI functions like
2109 CL:GET-INTERNAL-RUN-TIME.)
2110 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2111 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2112 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2113 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2114 * Fasl file version numbers are now independent of the target CPU,
2115 since historically most system changes which required version
2116 number changes have affected all CPUs equally. Similarly,
2117 the byte fasl file version is now equal to the ordinary
2120 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2121 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2122 is no longer recognized. I apologize for this, because it was
2123 listed in SB-EXT as a supported extension, but I found that
2124 its existing behavior was poorly specified, as well as incorrectly
2125 specified, and it looked like too much of a mess to straighten it
2126 out. I have enough on my hands trying to get ANSI stuff to work..
2127 * many patches ported from CMU CL by Martin Atzmueller, with
2128 half a dozen bug fixes in pretty-printing and the debugger, and
2129 half a dozen others elsewhere
2130 * fixed bug 13: Floating point infinities are now supported again.
2131 They might still be a little bit flaky, but thanks to bug reports
2132 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2133 as flaky as they were.
2134 * The --noprogrammer command line option is now supported. (Its
2135 behavior is slightly different in detail from what the old man
2136 page claimed it would do, but it's still appropriate under the
2137 same circumstances that the man page talks about.)
2138 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2139 are now supported, and enabled by default. Thus, the compiler can
2140 handle many floating point and complex operations much less
2141 inefficiently. (Thus e.g. you can implement a complex FFT
2143 * The compiler now detects type mismatches between DECLAIM FTYPE
2144 and DEFUN better, and implements CHECK-TYPE more correctly, and
2145 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2146 thanks to patches from Martin Atzmueller.
2147 * various fixes to make the cross-compiler more portable to
2148 ANSI-conforming-but-different cross-compilation hosts (notably
2149 Lispworks for Windows, following bug reports from Arthur Lemmens)
2150 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2151 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2152 fixed thanks to Pierre Mai's CMU CL patches.
2153 * fixes to keep the system from overflowing internal counters when
2154 it tries to use i/o buffers larger than 16M bytes
2155 * fixed bug 45a: Various internal functions required to support
2156 complex special functions have been merged from CMU CL sources.
2157 (When I was first setting up SBCL, I misunderstood a compile-time
2158 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2159 * improved support for type intersection and union, fixing bug 12
2160 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2161 more obscure bugs as well
2162 * some steps toward byte-compiling non-performance-critical
2163 parts of the system, courtesy of patches from Martin Atzmueller
2164 * Christophe Rhodes has made some debian packages of sbcl at
2165 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2166 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2167 stable, but are nonetheless usable. When he's ready, I'd be happy
2168 to add them to the SourceForge "File Releases" section. (And if
2169 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2170 * new fasl file format version number (because of changes in
2171 internal representation of (OR ..) types to accommodate the new
2172 support for (AND ..) types, among other things)
2174 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2175 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2177 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2179 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2180 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2181 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2182 * DESCRIBE now works on structure objects again.
2183 * Most function call argument type mismatches are now handled as
2184 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2185 know whether the function will be redefined before the call is
2186 executed. (The compiler could flag local calls with full WARNINGs,
2187 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2188 it doesn't keep track of enough information to know whether calls
2189 are local in this sense.)
2190 * Compiler output is now more verbose, with messages truncated
2191 later than before. (There should be some supported way for users
2192 to override the default verbosity, but I haven't decided how to
2193 provide it yet, so this behavior is still controlled by the internal
2194 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2195 src/compiler/ir1util.lisp.)
2196 * Fasl file format version numbers have increased again, because
2197 support for the Gray streams extension changes the layout of the
2198 system's STREAM objects.
2199 * The Gray subclassable streams extension now works, thanks to a
2200 patch from Martin Atzmueller.
2201 * The full LOAD-FOREIGN extension (not just the primitive
2202 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2203 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2204 but like most other programs, it defaults to copying the Unix
2205 environment from the original process instead of starting the
2206 new process in an empty environment.
2207 * Extensions which manipulate the Unix environment now support
2208 an :ENVIRONMENT keyword option which doesn't smash case or
2209 do other bad things. The CMU-CL-style :ENV option is retained
2210 for porting convenience.
2211 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2212 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2214 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2216 * A patch from Martin Atzmueller seems to have solved the SIGINT
2217 problem, and as far as we know, signal-handling now works cleanly.
2218 (If you find any new bugs, please report them!)
2219 * The system no longer defaults Lisp source file names to types
2220 ".l", ".cl", or ".lsp", but only to ".lisp".
2221 * The compiler no longer uses special default file extensions for
2222 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2223 seems to expect a single default extension for all compiled code,
2224 and there's no compelling reason to try to stretch the standard
2225 to allow two different extensions.) Instead, byte-compiled files
2226 default to the same extension as native-compiled files.
2227 * Fasl file format version numbers have increased again, because
2228 a rearrangement of internal implementation packages made some
2229 dumped symbols in old fasl files unreadable in new cores.
2230 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2231 many fewer weird special cases.
2232 * Bug #17 (differing COMPILE-FILE behavior between logical and
2233 physical pathnames) has been fixed, and some related misbehavior too,
2234 thanks to a patch from Martin Atzmueller.
2235 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2236 by Tim Moore, ported to SBCL by Martin Atzmueller.
2237 * Martin Atzmueller fixed several filesystem-related problems,
2238 including bug #36, in part by porting CMU CL patches, which were
2239 written in part by Paul Werkowski.
2240 * More compiler warnings in src/runtime/ are gone, thanks to
2241 more patches from Martin Atzmueller.
2242 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2245 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2247 * DESCRIBE now works on CONDITION objects.
2248 * The debugger now handles errors which arise when trying to print
2249 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2251 * The build system now uses an additional file, customize-target-features.lisp,
2252 to allow local modifications to the target *FEATURES* list. (The point of
2253 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2254 debugging features enabled, without having to worry about propagating it
2255 into everyone's system when I do a "cvs update".) When no
2256 customize-target-features.lisp file exists, the target *FEATURES* list
2257 should be constructed the same way as before.
2258 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2259 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2260 warn about behavior which is undefined under the ANSI spec. Note: This
2261 is specified by ANSI, but it's not very popular with programmers.
2262 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2263 macro in the SBCL sources for an example of a workaround which you
2264 might use to make portable ANSI-standard code which does what you want.
2265 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2266 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2267 when it's used as a non-toplevel form. (This is required in order
2268 to implement the DEFCONSTANT-EQX macro.)
2269 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2270 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2271 funny things, due to bugs in the implementation of EVAL-WHEN
2272 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2273 non-toplevel DEFCONSTANTs are uncommon.)
2274 * The core file version number and fasl file version number have been
2275 incremented, because the old noncompliant DEFCONSTANT behavior involved
2276 calling functions which no longer exist, and because I also took the
2277 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2278 * fixed bug 1 (error handling before read-eval-print loop starts), and
2279 redid debugger restarts and related debugger commands somewhat while
2281 ** The QUIT debugger command is gone, since it did something
2282 rather different than the SB-EXT:QUIT command, and since it never
2283 worked properly outside the main toplevel read/eval/print loop.
2284 Invoking the new TOPLEVEL restart provides the same functionality.
2285 ** The GO debugger command is also gone, since you can just invoke
2286 the CONTINUE restart directly instead.
2287 ** The TOP debugger command is also gone, since it's redundant with the
2288 FRAME 0 command, and since it interfered with abbreviations for the
2290 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2292 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2293 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2294 some porting work by Martin Atzmueller)
2295 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2296 with a boatload of patches which he ported to SBCL. Now that those
2298 ** The system tries to make sure that its low-priority messages
2299 are prefixed by semicolons, to help people who like to use
2300 syntax highlighting in their ILISP buffer. (This patch
2301 was originally due to Raymond Toy.)
2302 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2303 patches originally written by Raymond Toy.
2304 ** The compiler understands coercion between single-value and
2305 multiple-VALUES type expressions better, getting rid of some very
2306 weird behavior, thanks to patches originally by Robert MacLachlan
2307 and Douglas Crosher.
2308 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2309 lambda lists, thanks to a patch originally by Pierre Mai.
2310 ** The system no longer bogusly warns about "abbreviated type
2312 ** The compiler gets less confused by inlining and RETURN-FROM,
2313 thanks to some patches originally by Tim Moore.
2314 ** The system no longer hangs when dumping circular lists to fasl
2315 files, thanks to a patch originally from Douglas Crosher.
2316 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2317 undefined function error.
2318 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2319 in structure slot initforms) for now, documented workaround instead:-|
2320 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2321 * fixed bug 5: added stubs for various Gray stream functions called
2322 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2323 installed, at least appropriate type errors are generated
2324 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2325 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2326 consistently in DEFMETHOD forms.
2327 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2328 it doesn't seem to affect SBCL after all
2329 * The C runtime system now builds with better optimization and many
2330 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2332 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2334 * The system is now under CVS at SourceForge (instead of the
2335 CVS repository on my home machine).
2336 * The new signal handling code has been tweaked to treat register
2337 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2338 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2339 downstream, e.g. in debug-int.lisp, has implicit dependencies
2340 on the unsignedness of integer representation of machine words,
2341 and that caused the system to bomb out with infinite regress
2342 when trying to recover from type errors involving signed values,
2343 e.g. (BUTLAST '(1 2 3) -1).)
2344 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2345 (This was one of the bugs Peter Van Eynde reported back in July.)
2346 * The system now uses code inspired by Colin Walters' O(N)
2347 implementation of MAP (from the cmucl-imp@cons.org mailing
2348 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2349 inline the MAP operation, and there is more than one
2350 sequence argument to the MAP call (so that it can't just
2351 do ETYPECASE once and for all based on the type of the
2352 single sequence argument). (The old non-inline implementation
2353 of the general M-argument sequence-of-length-N case required
2354 O(M*N*N) time when any of the sequence arguments were LISTs.)
2355 * The QUIT :UNIX-CODE keyword argument has been renamed to
2356 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2357 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2359 * Raymond Wiker's patches to port dynamic loading from Linux to
2360 FreeBSD have been added.
2361 * The BUGS file is now more nearly up to date, thanks in large part
2362 to Martin Atzmueller's review of it.
2363 * The debugger now flushes standard output streams before it begins
2364 its output ("debugger invoked" and so forth).
2365 * The core version number and fasl file version number have both
2366 been incremented, because of incompatible changes in the layout
2368 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2369 used to not be called for a saved Lisp image.)
2370 * Martin Atzmueller's version of a patch to fix a compiler crash,
2371 as posted on sbcl-devel 13 September 2000, has been installed.
2372 * Instead of installing Martin Atzmueller's patch for the
2373 compiler transform for SUBSEQ, I deleted the compiler transform,
2374 and transforms for some similar consing operations.
2375 * A bug in signal handling which kept TRACE from working on OpenBSD
2377 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2378 away by constant folding
2379 * The system now defines its address space constants in one place
2380 (in the Lisp sources), and propagates them automatically elsewhere
2381 (through GENESIS and the sbcl.h file). Therefore, patching the
2382 address map is less unnecessarily tedious and error-prone. The
2383 Lisp names of address space constants have also been systematized.
2384 * CVS tags like dollar-Header-dollar have been removed from
2385 the sources, because they have never saved me trouble and
2386 they've been source of trouble working with patches and other
2387 diff-related operations.
2388 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2389 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2391 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2393 * The system has been ported to OpenBSD.
2394 * The system now compiles with a simple "sh make.sh" on the systems
2395 that it's supported on. I.e., now you no longer need to tweak
2396 text in the target-features.lisp-expr and symlinks in src/runtime/
2397 by hand, the make.sh takes care of it for you.
2398 * The system is no longer so grossly inefficient when compiling code
2399 involving vectors implemented as general (not simple) vectors (VECTOR T),
2400 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2401 which dares to use the various sequence functions on non-simple
2402 vectors, takes less of a performance hit.
2403 * There is now a primitive type predicate VECTOR-T-P
2404 to test for the (VECTOR T) type, so that e.g.
2405 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2406 can now be compiled with some semblance of efficiency. (The old code
2407 turned the type declaration into a full call to %TYPEP at runtime!)
2408 * AREF on (VECTOR T) is still not fast, since it's still compiled
2409 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2410 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2411 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2412 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2413 without making full calls to SUBTYPEP at runtime.
2414 (Some analogous efficiency issues for non-simple vectors specialized to
2415 element types other than T, or for non-simple multidimensional arrays,
2416 have not been addressed. They could almost certainly be handled the
2417 same way if anyone is motivated to do so.)
2418 * The changes in array handling break binary compatibility, so
2419 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2420 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2421 instead of (VALUES T T).
2422 * By following the instructions that Dan Barlow posted to sbcl-devel
2423 on 2 July 2000, I was able to enable primitive dynamic object
2424 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2425 functionality is not implemented (since it calls ld to resolve
2426 library references automatically, requiring RUN-PROGRAM for its
2427 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2428 try to resolve library references) is now supported.
2429 * The system now flushes the standard output streams when it terminates,
2430 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2431 them at several other probably-convenient times, e.g. in each pass of
2432 the toplevel read-eval-print loop, and after evaluating a form given
2433 as an "--eval" command-line option. (These changes were motivated by a
2434 discussion of stream flushing issues on cmucl-imp in August 2000.)
2435 * The source transform for TYPEP of array types no longer assumes
2436 that an array whose element type is a not-yet-defined type
2437 is implemented as an array of T, but instead punts, so that the
2438 type will be interpreted at runtime.
2439 * There is now some support for cross-compiling in make.sh: each of
2440 the phases of make.sh has its own script. (This should be transparent
2441 to people doing ordinary, non-cross-compile builds.)
2442 * Since my laptop doesn't have hundreds of megabytes of memory like
2443 my desktop machine, I became more motivated to do some items on
2444 my to-do list in order to reduce the size of the system a little:
2445 ** Arrange for various needed-only-at-cold-init things to be
2446 uninterned after cold init. To support this, those things have
2447 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2448 symbols with such names are now uninterned after cold init).
2449 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2450 fasl files for cold load.
2451 ** Remove the old compiler structure pooling code (which used to
2452 be conditional on the target feature :SB-ALLOC) completely.
2453 ** Redo the representation of some data in cold init to be more compact.
2454 (I also looked into supporting byte compiled code at bootstrap time,
2455 which would probably reduce the size of the system a lot, but that
2456 looked too complicated, so I punted for now.)
2457 * The maximum signal nesting depth in the src/runtime/ support code has
2458 been reduced from 4096 to 256. (I don't know any reason for the very
2459 large old value. If the new smaller value turns out to break something,
2460 I'll probably just bump it back up.)
2461 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2463 * Many, many bugs reported by Peter Van Eynde have been added to
2464 the BUGS list; some have even been fixed.
2465 * While enabling dynamic object file loading, I tried to make the
2466 code easier to understand, renaming various functions and variables
2467 with less ambiguous names, and changing some function calling
2468 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2469 * While trying to figure out how to do the OpenBSD port, I tried to
2470 clean up some of the code in src/runtime/. In particular, I dropped
2471 support for non-POSIX signal handling, added various comments,
2472 tweaked the code to reduce the number of compilation warnings, and
2473 renamed some files to increase consistency.
2474 * To support the new automatic configuration functionality in make.sh,
2475 the source file target-features.lisp-expr has been replaced with the
2476 source file base-target-features.lisp-expr and the machine-generated
2477 file local-target-features.lisp-expr.
2478 * fixed a stupid quoting error in make.sh so that using CMU CL
2479 "lisp -batch" as cross-compilation host works again
2481 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2483 * DESCRIBE no longer tries to call itself recursively to describe
2484 bound/fbound values, so that it no longer fails on symbols which are
2485 bound to themselves (like keywords, T, and NIL).
2486 * DESCRIBE now works on generic functions.
2487 * The printer now prints less-screwed-up representations of closures
2488 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2489 * A private symbol is used instead of the :EMPTY keyword previously
2490 used to mark empty slots in hash tables. Thus
2491 (DEFVAR *HT* (MAKE-HASH-TABLE))
2492 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2493 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2494 now does what ANSI says that it should. (You can still get
2495 similar noncompliant behavior if bang on the hash table
2496 implementation with all the symbols you get back from
2497 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2498 This breaks binary compatibility, since tests for equality to
2499 :EMPTY are wired into things like the macroexpansion of
2500 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2502 * There's now a minimal placeholder implementation for CL:STEP,
2503 as required by ANSI.
2504 * An obscure bug in the interaction of the normal compiler, the byte
2505 compiler, inlining, and structure predicates has been patched
2506 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2507 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2509 * Missing ordinary arguments in a macro call are now detected even
2510 when the macro lambda list contains &KEY or &REST.
2511 * The debugger no longer complains about encountering the top of the
2512 stack when you type "FRAME 0" to explicitly instruct it to go to
2513 the top of the stack. And it now prints the frame you request even
2514 if it's the current frame (instead of saying "You are here.").
2515 * As specified by ANSI, the system now always prints keywords
2516 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2518 * The default initial SIZE of HASH-TABLEs is now smaller.
2519 * Type information from CLOS class dispatch is now propagated
2520 into DEFMETHOD bodies, so that e.g.
2521 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2523 is now basically equivalent to
2524 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2525 (DECLARE (TYPE SINGLE-FLOAT X))
2527 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2528 operation, without having to do run-time type dispatch.
2529 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2530 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2531 * Since I don't seem to be making big file reorganizations very often
2532 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2533 is overflowing my ability to conveniently back them up), I've finally
2534 checked the system into CVS. (The CVS repository is on my home system,
2535 not at SourceForge -- putting it on SourceForge might come later.)
2536 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2537 high-level GC-NOTIFY-FOO functions send their output. (There's
2538 still very little control of where low-level verbose GC functions
2539 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2540 less than it used to -- the GC-NOTIFY-FOO functions are now under
2541 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2542 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2543 in only one place in the source code, and propagates it automatically
2544 everywhere that it's needed. Thus e.g. when I bump the version from
2545 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2546 * The C source files now include boilerplate legalese and documentation
2547 at the head of each file (just as the Lisp source files already did).
2548 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2549 to his page will be replaced with a link to his new CLiki service.
2551 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2553 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2554 * The build process now looks for GNU make under the default name "gmake",
2555 instead of "make" as it used to. If GNU make is not available as "gmake"
2556 on your system, you can change this default behavior by setting the
2557 GNUMAKE environment variable.
2558 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2559 can build without error under CMU CL.
2561 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2563 * There is now a partial SBCL user manual (with some new text and some
2564 text cribbed from the CMU CL manual).
2565 * The beginnings of a profiler have been added (starting with the
2566 CMU CL profiler and simplifying and cleaning up). Eventually the
2567 main interface should be through the TRACE macro, but for now,
2568 it's still accessed through vaguely CMU-CL-style functions and macros
2569 exported from the package SB-PROFILE.
2570 * Some problems left over from porting CMU CL to the new
2571 cross-compilation bootstrap process have been cleaned up:
2572 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2573 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2574 part when getting rid of the compiler *BACKEND* stuff.)
2575 ** The value of *NULL-TYPE* was screwed up, because it was
2576 being initialized before the type system knew the final
2577 definition of the 'NULL type. This screwed up several key
2578 optimizations in the compiler, causing inefficiency in all sorts
2579 of places. (I found it because I wanted to understand why
2580 GET-INTERNAL-RUN-TIME was consing.)
2581 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2582 PROCLAIM FTYPE information. Unfortunately this broke binary
2583 compatibility again, since now the forms output by DEFGENERIC
2584 to refer to functions which didn't exist in 0.6.3.
2585 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2586 can use the new (as of 0.6.3) transform for SOME into MAP into
2588 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2589 values to (INTEGER 0 1000000), so that the time code will no longer
2590 occasionally get blown up by Linux returning 1000000 microseconds
2591 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2592 its output conform to the ANSI spec. (Alas, this makes its output
2593 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2594 * A full call to MAP NIL with a single sequence argument no longer conses.
2595 * fixes to problems pointed out by Martin Atzmueller:
2596 * The manual page no longer talks about multiprocessing as though
2597 it were currently supported.
2598 * The ILISP support patches have been removed from the distribution,
2599 because as of version 5.10.1, ILISP now supports SBCL without us
2600 having to maintain patches.
2601 * added a modified version of Raymond Toy's recent CMU CL patch for
2602 EQUALP comparison of HASH-TABLE
2604 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2606 * The system still can't cross-compile itself with
2607 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2608 entails), but at least it can compile more of itself that way
2609 than it used to be able to, and various buglets which were uncovered
2610 by trying to cross-compile itself that way have now been fixed.
2611 * This release breaks binary compatibility again. This time
2612 at least I've incremented the FASL file format version to 2, so that the
2613 problem can be detected reliably instead of just causing weird errors.
2614 * various new style warnings:
2615 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2616 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2617 ** using the lexical binding of a variable named in the *FOO* style
2618 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2619 as specified by ANSI.
2620 * *RANDOM-STATE* is no longer automatically initialized from
2621 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2622 default behavior of the system is to repeat its behavior every time
2623 it's run. If you'd like to change this behavior, you can always
2624 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2625 old convention there was no comparably easy way to get the system to
2626 repeat its behavior every time it was run.
2627 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2628 FTYPE declarations has been removed, in favor of their ANSI
2629 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2630 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2631 the types of their sequence arguments can be determined at compile time.
2632 This is done through a new open code expansion for MAP which eliminates
2633 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2634 argument types can be determined at compile time.
2635 * The optimizer now transforms COERCE into an identity operation when it
2636 can prove that the coerced object is already of the correct type. (This
2637 can be a win for machine generated code, including the output of other
2638 optimization transforms, such as the MAP transform above.)
2639 * Credit information has been moved from source file headers into CREDITS.
2640 * Source file headers have been made more standard.
2641 * The CASE macro now compiles without complaining even when it has
2644 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2646 * (Note that the way that the PCL macroexpansions were rewritten
2647 to accommodate the change in DEFGENERIC below breaks binary
2648 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2649 not run under sbcl-0.6.2. Once we get out of alpha releases,
2650 i.e. hit release 1.0.0, we'll probably try to maintain binary
2651 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2652 and sbcl-1.4.4. Until then, however, it might be fairly common
2653 for maintenance releases to break binary compatibility.)
2654 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2656 * The reporting of unhandled signals has been changed to print some
2657 explanatory text as well as the report form. (Previously only
2658 the report form was printed.)
2659 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2660 it defines, so that the compiler no longer issues undefined function
2661 warnings for compiled-but-not-yet-loaded generic functions.
2662 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2663 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2664 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2665 section of the ANSI spec are supported.
2666 * The "" nickname for the "KEYWORD" package has been removed.
2667 The reader still handles symbol tokens which begin with a package marker
2668 as keywords, but it doesn't expose its mechanism for doing so in the
2669 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2670 * The system now issues STYLE-WARNINGs for contradictory TYPE
2671 proclamations. (Warnings for contradictory FTYPE proclamations would
2672 be nice too, but those can't be done usefully unless the type system
2673 is made smarter about FUNCTION types.)
2674 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2675 systematized, so that "*target-*.lisp is supposed to exist only on the
2676 target and imply that there's a related file which exists on the
2677 host, and *host-*.lisp is supposed to exist only on the host and imply
2678 that there's a related file which exists on the target. This involves a
2679 lot of renaming. Hopefully the acute confusion caused by the renaming
2680 will be justified by the reduction in chronic confusion..
2681 ** runtime-type.lisp -> early-target-type.lisp
2682 ** target-type.lisp -> late-target-type.lisp
2683 ** early-host-format.lisp -> early-format.lisp
2684 ** late-host-format.lisp -> late-format.lisp
2685 ** host-error.lisp -> misc-error.lisp
2686 ** early-error.lisp -> early-target-error.lisp
2687 ** late-error.lisp -> late-target-error.lisp
2688 ** host-defboot.lisp -> early-defboot.lisp
2689 ** code/misc.lisp -> code/target-misc.lisp
2690 ** code/host-misc.lisp -> code/misc.lisp
2691 ** code/numbers.lisp -> code/target-numbers.lisp
2692 ** code/early-numbers.lisp -> numbers.lisp
2693 ** early-host-type.lisp -> early-type.lisp
2694 ** late-host-type.lisp -> late-type.lisp
2695 ** host-typep.lisp -> typep.lisp
2696 ** load.lisp -> target-load.lisp
2697 ** host-load.lisp -> load.lisp
2698 ** host-disassem.lisp -> disassem.lisp
2699 ** host-insts.lisp -> insts.lisp
2700 ** byte-comp.lisp -> target-byte-comp.lisp
2701 ** host-byte-comp.lisp -> byte-comp.lisp
2702 ** host-signal.lisp -> signal.lisp
2703 ** host-defstruct.lisp -> defstruct.lisp
2704 ** late-target-type.lisp -> deftypes-for-target.lisp
2705 Furthermore, several other previously target-only files foo.lisp (e.g.
2706 hash-table.lisp and random.lisp) have been split into a target-and-host
2707 foo.lisp file and a target-only target-foo.lisp file, with their key type
2708 definitions in the target-and-host part, so that the cross-compiler will
2709 know more about target types.
2710 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2711 gone away. In their place are various *BACKEND-FOO* variables
2712 corresponding to the slots of the old structure.
2713 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2714 into the target SBCL has been fixed.
2715 * The chill.lisp system for loading cold code into a running SBCL
2717 * Support for the CMU CL "scavenger hook" extension has been removed.
2718 (It was undocumented and unused in the CMU CL sources that SBCL was
2719 derived from, and stale in sbcl-0.6.1.)
2720 * Various errors in the cross-compiler type system were detected
2721 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2722 (enabling various consistency checks). Many of them were fixed,
2723 but some hard problems remain, so the compiler is back to
2724 running without *TYPE-SYSTEM-INITIALIZED* for now.
2725 * As part of the cross-compiler type system cleanup, I implemented
2726 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2727 * I have started adding UNCROSS calls throughout the type system
2728 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2729 on cross-compiler input files will be able to go away, and various
2731 * CONSTANTP now returns true for quoted forms (as explicitly required
2734 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2736 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2737 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2738 (MAKE-LIST -1). In the longer term, it ought to become true
2739 that declarations are assertions even at SAFETY 1. For now, it's not
2740 quite true even at SAFETY 3, but it's at least more nearly true..
2741 (Note that this change seems to increases the size of the system by
2742 O(5%) and to decrease the speed of the compiler by 20% or more.)
2743 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2744 for the problem of printing dozens of lines of distracting information
2745 about low-level system machinery as part of the top stack frame
2746 on entry to the debugger when an undefined function was called.
2747 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2748 is not reset to COMMON-LISP-USER.
2749 * Compilation of stuff related to dyncount.lisp has been made conditional
2750 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2751 smaller. The various dyncount-related symbols have been moved into
2752 a new "SB-DYNCOUNT" package.
2753 * tty-inspect.lisp has been renamed to inspect.lisp.
2754 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2755 feature has gone away. (When we eventually port to other flavors of
2756 libc and/or Unix, we'll try to make the differences between flavors
2757 invisible at the user level.)
2758 * Various other *FEATURES* tags, and/or their associated conditionals,
2759 have been removed if obsolescent, or given better documentation, or
2760 sometimes given more-mnemonic names.
2762 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2764 * tidied up "make.sh" script
2765 * tidied up system directory structure
2766 * better "clean.sh" behavior
2767 * added doc/FOR-CMUCL-DEVELOPERS
2768 * many many small tweaks to output format, e.g. removing possibly-confusing
2769 trailing #\. character in DESCRIBE-INSTANCE
2770 * (EQUALP #\A 'A) no longer signals an error.
2771 * new hashing code, including EQUALP hashing
2772 * tidied up Lisp initialization and toplevel
2773 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2774 * command line argument processing
2775 * added POSIX-GETENV function to deal with Unix-ish environment variables
2776 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2777 terminating SBCL on EOF
2778 * non-verbose GC by default
2779 * There is no more "sbcl" shell script; the sbcl file is now the C
2780 runtime executable (just like CMU CL).
2781 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2783 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2784 symbol and package information as Lisp native symbol and package objects
2785 instead of strings naming symbols and strings naming packages. This way,
2786 whenever packages are renamed (as in warm init), debug information is
2787 transformed along with everything else.
2788 * tweaked the optimization policy declarations which control the building
2789 of SBCL itself. Now, among other things, the system no longer saves
2790 source location debugging information. (This helps two problems at once
2791 by reducing SBCL size and by keeping SBCL from trying to look for its
2792 sources -- which may not exist -- when reporting errors.)
2793 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2794 debugging and testing purposes
2795 * cleaned up printing, making the printer call PRINT-OBJECT for
2796 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2797 methods, giving nearly-ANSI behavior
2798 * converted almost all special variables to use *FOO* naming convention
2799 * deleted PARSE-TIME functionality, since it can be done portably
2800 * moved some files out of cold init into warm init
2801 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2803 * regularized formatting of source files
2804 * added an install.sh script
2805 * fixed ridiculous memory usage of cross-compiler by making
2806 compiler/alloc.lisp not try to do pooling unless it can hook
2807 itself into the GC of the cross-compilation host. Now the system
2808 builds nicely on my old laptop.
2809 * added :SB-ALLOC in target-features.lisp-expr
2810 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2811 was not implemented)
2812 * re-did condition handling and note reporting in the compiler. Notes
2813 are no longer handled by signalling conditions. Style warnings
2814 and warnings are handled more correctly and reported in such a way
2815 that it's easy to find one or the other in your output (so that you
2816 can e.g. figure out which of many problems caused COMPILE-FILE to
2818 * changed the severity of several compiler warnings from full WARNING
2819 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2820 compiler note reporting so that it doesn't use the condition system
2821 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2822 COMPILE-FILE command)
2823 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2824 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2825 no longer supported, since their implementation was deeply intertwingled
2826 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2827 * removed bogus "support" for compiler macros named (SETF FOO), and
2828 removed the compiler macro for SETF INFO (but only after making a fool
2829 of myself on the cmucl-imp mailing list by posting a bogus patch for
2830 DEFINE-COMPILER-MACRO..)
2831 * Compiled files containing forms which have side effects on the Lisp
2832 reader (such as DEFPACKAGE forms) are now handled more correctly.
2833 (Compiler queuing of top level lambdas has been suppressed by setting
2834 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2835 * deleted various currently-unused source files, e.g. gengc.lisp. They
2836 may be added back at some point e.g. when porting to other architectures,
2837 but until they are it's distracting to distribute them and to try to
2839 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2840 there were so many of them they're just distractions, and UNCROSS is
2841 known to be able to handle the current sources
2842 * moved PROFILE functionality into TRACE, so that it will be clear
2843 how the wrapping and unwrapping of functions when you profile them
2844 interacts with the wrapping and unwrapping of functions when you
2845 trace them. (Actually, the functionality isn't there yet, but at least
2846 the interface specification is there. Hopefully, the functionality will
2847 arrive with some future maintenance release.)
2848 * removed host-oops.lisp
2849 * changed signature of QUIT function to allow UNIX-CODE argument
2850 * fixed READ-SEQUENCE bug
2851 * tweaked verbose GC output so that it looks more like the progress
2852 output that ANSI specifies for functions like LOAD
2853 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2854 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob