1 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2 * new feature: a build-time option (controlled by the :SB-UNICODE
3 keyword feature, enabled by default) for building the system with
4 support for the entire 21-bit character space defined by the
6 * new feature: the system now has rudimentary external-format
7 support; the primary user-visible change at this time is that
8 characters with the high bit set (such as lower-case-e-acute) will
9 print correctly to a terminal in a UTF-8 environment.
10 * minor incompatible change: BASE-CHAR no longer names a class;
11 however, CHARACTER continues to do so, as required by ANSI.
12 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
13 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
14 has been moved to the SB-EXT package (temporarily re-exported from
16 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
17 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
19 * fixed bug #351: better error-handlind and reporting for malformed
21 * fixed bug #350: bignum-printing is now more memory-efficient,
22 allowing printing of very large bignums, eg. (expt 2 10000000).
23 (reported by Bruno Haible)
24 * fixed bug #302: better primitive-type selection for intersection
26 * fixed bug #308: non-graphic characters now all have names, as
27 required. (reported by Bruno Haible)
28 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
29 during macroexpansion for non-iteration variables as well. (reported
30 by Bruno Haible for CMUCL)
31 * bug fix: Cyclic structures and unprintable objects in compiler
32 messages no longer cause errors. (reported by Bruno Haible)
33 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
34 and READ-FROM-STRING all now return a primary value of NIL if
35 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
36 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
37 (reported by Bruno Haible for CMUCL)
38 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
39 do not designate a condition. (reported by Bruno Haible for
41 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
42 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
43 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
44 permitting resolution in favour of any of the conflicting symbols.
45 (reported by Bruno Haible for CMUCL)
46 * bug fix: EQUAL compiler optimizations is less aggressive on
47 strings which can potentially compare true despite having distinct
48 specialized array element types.
49 * FORMAT compile-time argument count checking has been enhanced.
50 (report from Bruno Haible for CMUCL)
51 * fixed some bugs revealed by Paul Dietz' test suite:
52 ** RENAME-PACKAGE allows all package designators as new package
54 ** constraint propagation and lambda variable substitution are
55 more cautious in dealing with partially deleted code.
56 ** compiler sometimes lost reoptimization passes.
57 ** CERROR, given a condition as condition designator, passes its
58 remaining arguments to the continue format control without
60 ** Case-altering FORMAT directives work correctly on non-ASCII
62 ** The REAL type specifier handles bounds outside the floating
63 point ranges without signalling FLOATING-POINT-OVERFLOW.
64 ** Functions with IR1-transformations can create intercomponent
65 references to global functions.
67 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
68 * enhancement: saving cores with foreign code loaded is now
69 supported on x86/NetBSD and sparc/Linux in addition to the previously
71 * bug fix: on some platforms repeated installations caused multiple
72 copies of HTML documentation to be installed -- should not happen
73 any more. (reported by Stefan Scholl)
74 * bug fix: parsing self-recursive alien record types multiple times
75 no longer causes infinite recursion. (reported by Thomas F. Burdick,
76 original patch by Helmut Eller for CMUCL)
77 * bug fix: stack-exhaustion detection works now on NetBSD as well.
78 (thanks to Richard Kreuter)
79 * bug fix: defining classes whose accessors are methods on existing
80 generic functions in other (locked) packages no longer signals
81 bogus package lock violations. (reported by François-René Rideau)
82 * bug fix: special variables as DEFMETHOD parameters no longer have
83 associated bogus type declarations. (reported by David Wragg and
85 * bug fix: read-write consistency on streams of element-type
86 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
87 * bug fix: redefiniton of the only method of a generic function with
88 no DEFGENERIC no longer emits a full WARNING. In addition,
89 redefinition of generic functions with no DEFGENERIC to an
90 incompatible lambda list now signals an error. (thanks to Zach
92 * bug fix: DEFGENERIC now works even when there's a function of the
93 same name in an enclosing lexical environment. (thanks to Zach
95 * fixed compiler failure, caused by instrumenting code during
96 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
97 * optimization: added loop analysis and improved register allocation
98 * fixed some bugs revealed by Paul Dietz' test suite:
99 ** POSITION on displaced vectors with non-zero displacement
100 returns the right answer.
101 ** (SIMPLE-STRING) is a valid type specifier for sequence
103 ** *PRINT-LEVEL* handling for slotless structures is pedantically
105 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
106 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
108 ** The justification version of the FORMAT ~< directive treats
109 non-zero minpad parameter correctly.
111 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
112 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
113 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
114 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
115 the supported interface.
116 * new feature: Single-stepping of code compiled with DEBUG 2 or
117 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
118 * new feature: saving cores with foreign code loaded is now
119 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
120 Timothy Moore's work for CMUCL)
121 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
122 parameters to * instead of NIL if no initform is supplied.
123 (reported by Johan Bockgaard)
124 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
125 correctly when FUNCALL appears as the car of the form. Note:
126 despite this FUNCALL forms are not currently subject to
127 compiler-macro expansion. (port of Raymond Toy's fix for the
128 same from CMUCL, reported by Johan Bockgaard)
129 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
130 (thanks for Teemu Kalvas)
131 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
132 case instead of terminating the iteration (thanks for Julian
133 Squires, Sean Champ and Raymond Toy)
134 * bug fix: incorrect expansion of defgeneric that caused a style
135 warning. (thanks for Zach Beane)
136 * bug fix: slot accessor effective method computation works properly
137 for classes with multiple non-standard applicable methods on
138 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
139 * on x86 compiler supports stack allocation of results of LIST and
140 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
141 CMUCL implementation by Gerd Moellmann)
142 * fixed some bugs revealed by Paul Dietz' test suite:
143 ** FORMAT strings with both the ~<~:;~> form of the justification
144 directive and pretty-printing directives cause an error.
146 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
147 * incompatible change: the internal functions
148 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
149 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
150 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
151 instead of the old functions.
152 * new feature: on platforms where "dladdr" is available foreign
153 function names now appear in backtraces. (based on Helmut Eller's
155 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
156 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
158 * documentation: ASDF manual is now distributed with SBCL.
159 * bug fix: SBCL can now load its contributed modules with REQUIRE
160 even if the system-provided entries have been removed from
161 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
163 * bug fix: backtraces involving undefined functions or assembly
164 routines are more informative. (thanks to Brian Downing)
165 * bug fix: mutually referent alien structures now work correctly.
166 (reported by Rick Taube)
167 * bug fix: structures defined by WITH-ALIEN can be referred to
168 within other definitions in the same WITH-ALIEN.
169 * bug fix: division operators (MOD, TRUNCATE and the like) with
170 constant zero divisors and integer dividends no longer generate
172 * bug fix: provide default methods for INPUT-STREAM-P and
173 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
174 * bug fix: improve the detection and resolution of MOP metacycles
175 (where effective method computation depends on the generic function
176 having its effective method computed). (reported by Bruno Haible)
177 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
178 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
179 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
180 * bug fix: pathnames with relative directory components can be
181 represented relative to default pathnames.
182 * optimization: in taking the GCD of bignums, reduce the two bignums
183 to approximately the same size (using Euclid's algorithm) before
184 applying the more sophisticated binary GCD. (thanks to Juho
186 * optimization: COUNT on bitvectors now operates word-at-a-time.
187 * optimization: ASH with a positive, but not necessarily constant,
188 (leftwards) shift, when in a modular context, is compiled to a
190 * fixed some bugs revealed by Paul Dietz' test suite:
191 ** FORMAT variable parameters ("~V<char>") are defaulted properly
192 if the corresponding argument is NIL.
193 ** FORMAT directives accepting a minpad parameter treat negative
195 ** literal commas as character parameters to FORMAT directives are
197 ** literal spaces directly after ~<Newline> directives within a
198 format-logical-block (~:< ~@:>) do not induce :FILL-style
199 conditional newlines.
200 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
201 as specified: it no longer includes conditional newlines).
202 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
204 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
205 :PER-LINE-PREFIX argument does not evaluate to a string.
207 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
208 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
209 the manual for details; for now, package locks can be disabled by
210 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
211 if no major problems are found then it is likely that they will be
212 compiled in unconditionally.
213 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
214 unsupported operators on all platforms. To load a shared library
215 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
216 object file, link it into a shared library outside of SBCL and
217 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
219 * bug workaround: ROOM T and the GC were not working together
220 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
221 reporting facility aren't preserved. That reporting has been
222 disabled, so now until and unless someone figures out how to make it
223 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
224 * minor incompatible change: as threatened around sbcl-0.8.0, the
225 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
226 The symbols are also exported from SB-PCL for backwards
227 compatibility, but more so than before SB-PCL should be treated as
228 an implementation-internal package.
229 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
231 * the SB-SPROF contrib now works on (most) non-x86 architectures.
232 It is known as of this release not to work on the Alpha, however.
233 * fixed bug #167: errors signalled due to illegal syntax in method
234 bodies are now more legible.
235 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
236 designators and can hence be used with TYPEP.
237 * fixed bug #333: CHECK-TYPE now ensures that the type error
238 signalled, if any, has the right object to be accessed by
239 TYPE-ERROR-DATUM. (reported by Tony Martinez)
240 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
241 5.1.2.3 for multiple-value place subforms. (reported by Kalle
243 * fixed bug #334: programmatic addition of slots using specialized
244 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
245 and :ALLOCATION :CLASS slots.
246 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
247 float exponents. (rereported by Peter Seibel)
248 * fixed a bug: #\Space (and other whitespace characters) are no
249 longer considered to be macro characters in standard syntax by
251 * fixed bug: initialization of condition class metaobjects no longer
252 causes an instance of the condition to be created. (reported by
254 * fixed bug: it is now possible to have slots such that
255 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
256 description is neither :INSTANCE nor :CLASS.
257 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
258 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
259 SLOT-VALUE-USING-CLASS), and no longer causes errors with
260 non-standard SLOT-DEFINITION-ALLOCATION values.
261 * fixed bugs: various race conditions handling exiting threads.
262 CL-PPCRE's thread torture test now passes.
263 * fixed arguably-a-bug: GC time proportional to number of threads in
264 system even when most of them are idle
265 * optimization: improved performance of BIT and SBIT on bit-vectors.
266 * .fasl file incompatibility: The fasl file version number has
267 been incremented because of changes associated with package locks.
269 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
270 * minor incompatible change: the system no longer provides
271 optimization hints (conditions which are TYPEP
272 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
273 mode; these hints are still emitted when higher SPEED optimization
275 * new contrib module: a sampling profiler (profiling by statistical
276 sampling, rather than by instrumenting functions) is available as
277 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
278 * the behaviour of the standard function ED is now customizeable by
279 third parties through a hook variable: see ED's documentation
280 string for information on the protocol.
281 * the compiler no longer emits efficiency notes for (FUNCALL X)
282 when the type of X is uncertain under default optimization
284 * fixed bug 276: mutating a binding of a specialized parameter to a
285 method to something that is not TYPEP the specializer is now
287 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
288 exhibit the required behaviour on the x86 platform. (thanks to
289 Peter van Eynde, Eric Marsden and Bruno Haible)
290 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
291 even for difficult arguments. (reported by Peter Graves)
292 * fixed bug 141a: the backquote printer now descends quoted
294 * fixed another bug in backquote printing: no more destructive
295 modification of the form's list structure. (reported by Brian
297 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
298 move between its address being taken and the call to
299 interrupt_thread, fixing a crashing race condition.
300 * the SB-POSIX contrib implementation has been adjusted so that it
301 no longer exhibits ridiculously poor performance when constructing
302 instances corresponding to C structs.
304 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
305 * minor incompatible change: the sb-grovel contrib now treats C
306 structures as alien (in the sense of SB-ALIEN) objects rather than
307 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
308 has implications for memory management of client code
309 (sb-grovel-returned objects must now be manually managed) and for
310 type safety (alien objects now have full types).
311 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
312 used to control emission of compiler diagnostics, rather than the
313 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
314 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
315 quality should be considered deprecated.
316 * install.sh now installs the user manual as well
317 * (not quite a new documentable feature, but worth considering in
318 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
319 beginnings of a semantically meaningful condition hierarchy is
320 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
322 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
323 designator as the defaults argument.
324 * fixed bug: Displaced arrays whose displaced-to array has become
325 too small now cause ARRAY-DIMENSION to signal an error, providing
326 for safer bounds-checking. (reported by Bruno Haible)
327 * fixed bug: DEFCLASS slot definitions with identical :READER and
328 :WRITER names now signal a reasonable error. (reported by Thomas
330 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
332 * fixed bug: Invalid slot specification errors now print correctly.
333 (thanks to Zach Beane)
334 * fixed bug 320: Shared to local slot value transfers in class
335 redefinitions now happen corrently from superclasses as
336 well. (reported by Bruno Haible)
337 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
339 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
340 improper lists correctly. (reported by Bruno Haible)
341 * fixed bug 313: source-transform for <fun-name> was erroneously
342 applied to a call of a value of a variable with name <fun-name>.
343 (reported by Antonio Menezes Leitao)
344 * fixed bug 307: The obsolete instance protocol ensures that
345 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
346 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
347 remove dead unknown-values globs from the middle of the stack.
348 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
349 easier to resume long computations after using *BREAK-ON-SIGNALS*
350 to diagnose and fix failures. (thanks to Nikodemus Siivola)
351 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
352 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
353 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
354 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
355 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
357 * fixed a (fairly theoretical) bug in string printing: if
358 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
359 does not have array-element-type equal to the most general string
361 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
362 function name. (thanks to Nikodemus Siivola)
363 * fixed bug: compile-time format string checker failed on
364 non-closed ~{. (reported by Thomas F Burdick)
365 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
366 don't warn on legal loop code involving more than one aggregate
367 boolean. (thanks to Nikodemus Siivola)
368 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
369 in keyword position to MAKE-PACKAGE was not being reported as
371 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
372 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
374 * fixed bugs 280 and 312: the checking for multiple definitions in a
375 file is less likely to become confused by uses of inline
377 * fixed bug: the #S reader macro performs the keyword coercion
378 specified for slot names. (reported by Kalle Niemitalo)
379 * fixed bug: lambda lists may contain symbols whose names start with
380 & but are not lambda-list-keywords; their occurrence triggers a
382 * fixed bug 321: define-method-combination argument lambda lists do
383 not cause invalid code to be generated when &optional or &aux
384 variables have default values. (reported by Bruno Haible)
385 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
386 have CLOS classes; this appears also to have fixed reported
387 instability in stack exhaustion detection.
388 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
389 MUFFLE-WARNING when no associated restart is present is now
391 * optimization: rearranged the expansion of various defining macros
392 so that each expands into only one top-level form in a
393 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
395 * optimization: used a previously little-used slot in symbols to
396 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
398 * fixed some bugs revealed by Paul Dietz' test suite:
399 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
401 ** Fixed an optimization bug related to inheritance of initforms
402 from local to shared slots.
403 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
404 ** CAST optimizer forgot to flush argument derived type.
405 ** print/read consistency on floats is now orders of magnitude
406 more likely. (thanks also to Bruno Haible for a similar report
408 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
409 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
410 for objects of type REAL. Make it so.
411 ** SXHASH is specified (infelicitously) to respect similarity,
412 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
413 it so. (thanks to Markus Ziegler)
414 ** on the Alpha, the compiler succeeds in compiling functions
415 returning a known number of arguments greater than 63.
416 ** fixed handling of invalid NIL arguments in keyword position
418 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
420 ** provided a readably-printable representation for RANDOM-STATE
422 ** ensured that pathnames, where they have a namestring, always
423 print using #P"..." syntax.
425 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
426 * Support for the forthcoming 2.0 version of the NetBSD kernel
427 running on x86 hardware has been added. (thanks to Perry
428 E. Metzger most immediately, and others for their past work)
429 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
430 current release version; SBCL's OpenBSD support had been broken
431 since about the time of OpenBSD's switch to ELF binary format.)
432 * [placeholder for DX summary]
433 ** user code with &REST lists declared dynamic-extent, under high
434 speed or space and low safety and debug optimization policy.
435 * The manual has been converted to Texinfo format and the debugger
436 chapter from the cmucl manual has been added.
437 * A facility has been added to extract documentation strings from
438 sbcl and store them as Texinfo-formatted snippets for inclusion in
439 the manual (via Texinfo's @include directive)
440 * bug fix: compiler emitted division in optimized DEREF. (thanks for
441 the test case to Dave Roberts)
442 * bug fix: multidimensional simple arrays loaded from FASLs had fill
443 pointers. (reported by Sean Ross)
444 * bug fix: PROFILE output is printed nicely even for large numerical
445 values. (thanks to Zach Beane)
446 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
447 greater than 32 handle EOF correctly.
448 * bug fix: on X86 an immediate argument of the IMUL instruction is
449 correctly printed in disassembly. (thanks to Lutz Euler)
450 * bug fix: class slots in redefined classes preserve their old
451 values. (thanks to Bruno Haible and Nikodemus Siivola)
452 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
453 succeeds. (reported by Marco Baringer)
454 * fixed some bugs revealed by Paul Dietz' test suite:
455 ** READ-SEQUENCE now works on ECHO-STREAMs.
456 ** RATIONALIZE works more according to its specification. (thanks
459 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
460 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
461 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
462 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
463 should matter to you only if you rebind the printer control
464 variables and then find you want different bindings in the
465 debugger than in the ordinary execution of your program.)
466 * The runtime build system has been tweaked to support building
467 (on SPARC/SunOS) using a C compiler which invokes Sun's own
468 assembler and linker. (thanks to Nikodemus Siivola)
469 * Unbound, undefined, undeclared variables now trigger full
470 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
471 more often programmer error than deliberate exploitation of undefined
473 * optimization: the hash algorithm for strings has changed to one
474 that is less vulnerable to spurious collisions. (thanks to Juho
476 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
477 needless bounds checking. (thanks to Juho Snellman)
478 * optimization: implemented multiplication as a modular
479 (UNSIGNED-BYTE 32) operation on the PPC backend.
480 * fixed some bugs revealed by Paul Dietz' test suite:
481 ** ADJUST-ARRAY now copies the datum in a zero rank array if
483 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
484 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
485 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
487 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
489 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
491 ** ECHO-STREAMs no longer attempt to echo the end of file value to
492 their output stream on EOF from read.
493 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
494 have been read to end-of-file.
495 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
497 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
498 description of determination of which consecutive characters
500 ** Printing the "Space" character with escaping on now yields "#\\ ",
501 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
502 ** Reading floating-point numbers with *READ-BASE* set to a number
503 less than 10 works correctly.
504 ** Reading floating-point numbers with *READ-BASE* set to a number
505 more than 10 works correctly.
506 ** Printing with *PRINT-READABLY* targets the standard readtable, not
507 the readtable currently in effect.
509 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
510 * minor incompatible change: parsing of namestrings on a physical
511 (Unix) host has changed; numbers after the final #\. in a
512 namestring are no longer interpreted as a version field. This is
513 intented to be largely invisible to the user, except that the
514 meaning of the namestring "*.*.*" has changed: it now refers to a
515 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
516 should usually be replaced by
517 (make-pathname :name :wild :type :wild :version :wild)
518 with the added benefit that this is more likely to be portable.
519 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
520 option to OPEN now signals an error if the file being opened
521 exists; this may have an impact on existing code.
522 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
523 signals received, on the PowerPC platforms (both Linux and
524 Darwin). (thanks to Pierre Mai for pointing out the location of
526 * several fixes on OS X: The system now builds and runs cleanly on
527 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
528 (thanks to Brian Mastenbrook)
529 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
530 second-resolution: integer multiples of 1/3600 between -24 and 24.
531 (thanks to Vincent Arkesteijn)
532 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
533 type when called with 1 argument; PEEK-CHAR checked type of
534 PEEK-TYPE only after having read first character from a
535 stream. (reported by Peter Graves)
536 * bug fix: the garbage collector now has much better locality
537 behaviour, and in particular no longer treats all memory as being
538 exhausted when many small objects point to each other in a deeply
540 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
541 recognized as being TYPEP their class.
542 * bug fix: the PUSHNEW documentation string has been corrected.
543 (thanks to Vincent Arkesteijn)
544 * bug fix: defaulting of the value for the last of an atypically
545 large number of multiple values being bound was not being
546 performed correctly on the Alpha or PPC platforms
547 * optimization: implemented multiplication as a modular
548 (UNSIGNED-BYTE 32) operation on the x86 backend.
549 * optimization: SEARCH on simple-base-strings can now be open-coded.
550 (see also contrib/compiler-extras.lisp for inspiration for
551 teaching the compiler about the Boyer-Moore algorithm).
552 * value, returned by MAX (and MIN) called with several EQUALP, but
553 not EQL, arguments now does not depend on compiler settings.
554 * fixed some bugs revealed by Paul Dietz' test suite:
555 ** in stack analysis liveness information is propagated from
556 non-local entry points.
557 ** pathwise CAST removing failed when the CAST node did not start
559 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
560 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
562 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
563 host is already defined.
564 ** RENAME-FILE works on streams instead of signalling an internal
566 ** PEEK-CHAR uses the current readtable when determining whether
567 or not a character is whitespace.
568 ** MERGE-PATHNAMES handles the case when the pathname does not
569 specify a name while the default-pathname specifies a version.
570 ** Pathnames now stand a better chance of respecting print/read
572 ** Attempting to use standardized file system operators with a
573 pathname with invalid :DIRECTORY components signals a
575 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
577 ** DIRECTORY on logical pathnames is more correct.
578 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
579 signal a TYPE-ERROR if their argument is not a stream.
580 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
581 designator argument does not designate a stream.
582 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
583 examining the synonym.
584 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
586 ** OPEN and WITH-OPEN-STREAM allow opening streams with
587 element-type larger than ([UN]SIGNED-BYTE 32).
589 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
590 * When built with the :SB-FUTEX feature, threaded builds now take
591 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
592 for faster/more reliable mutex and condition variable support.
593 * Incompatible change (but one you probably shouldn't have been using
594 anyway): the interface and code for arbitrating between multiple
595 threads in the same user session has been redesigned.
596 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
597 error when called without an explicit environment argument.
598 (thanks to Nikodemus Siivola)
599 * bug fix: buffered :DIRECTION :IO streams are less likely to become
600 confused about their position. (thanks to Adam Warner and Gerd
602 * bug fix: Pretty printing backquoted forms with unquotations in the
603 argument list position of various code constructs such as LAMBDA
604 now works correctly. (reported by Paul Dietz)
605 * bug fix: Pretty printing unquotations no longer loses all
606 stream position information.
607 * optimization: performance of string output streams is now less
608 poor for multiple small sequence writes.
609 * optimization: performance of CSUBTYPEP in the presence of complex
610 expressions involving CONS and NOT many times has been improved.
611 (reported by Paul Dietz)
612 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
614 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
615 streams for the REPL, for the debugger, and so on. (thanks to
617 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
618 * fixed some bugs revealed by Paul Dietz' test suite:
619 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
620 is ignored if it is too small, rather than propagating through
621 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
622 ** extremely complex negations of CONS types were not being
623 sufficiently canonicalized, leading to inconsistencies in
625 ** VALUES tranformer lost derived type.
627 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
628 * fixed a bootstrapping bug: the build process no longer assumes
629 that the various BOOLE-related constants have the same value in
630 host and target lisps. (noted by Paul Dietz' test suite on an
631 SBCL binary built from CLISP)
632 * The system can now be dynamically linked on the MIPS platform,
633 which enables dynamic loading of foreign code from Lisp. (thanks
634 to Ralf Baechle for discussions on the MIPS ABI)
635 * The system now records debugging information for its own source
636 files in a filesystem-position-independent manner, relative to
637 the "SYS" logical host.
638 * fixed a compiler bug: MV-LET convertion did not check references
639 to the "max args" entry point. (reported by Brian Downing)
640 * tweaked disassembly notes to be less confident about proclaiming
641 some instruction as an LRA. (thanks to Brian Downing)
642 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
643 now each have their own history, command character, and other
644 characteristics. (thanks to David Lichteblau)
645 * fixed some bugs revealed by Paul Dietz' test suite:
646 ** compiler failure in compiling LOGAND expressions including a
648 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
649 shift greater than 32.
650 ** FLUSH-DEST did not mark blocks for type check regeneration.
651 ** HANDLER-CASE failed to accept declarations in handler clauses
652 in some circumstances.
654 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
655 * New code in contrib/sb-introspect (still probably not entirely
656 stable yet) provides some support for smart Lisp development
657 environments like SLIME.
658 * The conditions signalled for errors occurring when loading .fasl
659 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
660 in a way which should help ASDF recover gracefully.
661 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
662 stuff has been cleaned up. If you code contrib/ stuff, this might
663 affect you, and you can look at contrib/README, contrib/STANDARDS,
664 and/or the 0.8.4.27 diff to check.
665 * In full calls the compiler now does not generate checks for declared
666 argument types for all arguments.
667 * various threading fixes
668 ** and some experimental patches which didn't make it into
669 the main tree for this release, but which are shipped in
670 contrib/experimental-thread.patch as a possible fix for some
671 failures (deadlock, spinning...) in GC-intensive multithreaded
673 * fixed PPC build problem (source code incompatibility of different
674 library versions): added offsetof() hackery which attempts to divine
675 where glibc maintainers put uc_mcontext today
676 * fixed bug 282: compiler does not trust type assertions while passing
677 arguments to a full call.
678 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
679 &OPTIONAL VALUES type parameter.
680 * fix bug 214: algorithm for noting rejected templates is now more
681 similar to that of template seletion. (also reported by rydis on
683 * fixed bug 141b: printing backquoted information readably and prettily
684 inserts a space where necessary.
685 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
686 (CHARACTER 10) are now reported as errors, rather than propagated
687 as unknown types. (reported by piso on #lisp)
688 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
689 with values NIL and :ERROR. (thanks to Milan Zamazal)
690 * fixed bug 191c: CLOS now does proper keyword argument checking as
691 described in CLHS 7.6.5 and 7.6.5.1.
692 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
693 longer raise an error; further, using a list as a for-as-arithmetic
694 counter now raises a meaningful error.
695 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
696 now understood by sequence creation functions such as MAKE-SEQUENCE
698 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
699 a STRING-INPUT-STREAM.
700 * compiler enhancement: SIGNUM is now better able to derive the type
702 * type declarations inside WITH-SLOTS are checked. (reported by
704 * fixed some bugs revealed by Paul Dietz' test suite:
705 ** incorrect optimization of TRUNCATE for a positive first
706 argument and negative second.
707 ** compiler failure in let-convertion during flushing dead code.
708 ** compiler failure while deriving type of TRUNCATE on an
709 interval, containing 0.
710 ** ASH of a negative bignum by a negative bignum count now returns
712 ** intersection of CONS types now canonicalizes properly, fixing
713 inconsistencies in SUBTYPEP.
715 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
716 * incompatible change: The --disable-debugger command line
717 option now clobbers the debugger at a more fundamental
718 level, by redefining #'INVOKE-DEBUGGER instead of by
719 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
720 is specified by ANSI to ignore *DEBUGGER-HOOK* and
721 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
722 enter the debugger REPL and then suffer recursive errors
723 because *DEBUG-IO* is also messed up in --disable-debugger mode;
724 while under the new system, BREAK in --disable-debugger mode
725 terminates the system just as an unhandled error would.
726 * fixed compiler performance when processing loops with a step >1;
727 * bug fix: DOCUMENTATION now retrieves generic function
728 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
729 support has been systematized, and now supports the methods
730 specified by ANSI, along with a default method and a method for
731 slot documentation. (reported by Nathan Froyd)
732 * bug fix: effective methods associated with a generic function are
733 no longer cached over a change of that generic function's method
734 combination. (reported by Andreas Fuchs)
735 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
736 names a non-existent directory. (thanks to Andreas Fuchs)
737 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
738 the PPC platform, lead to stack corruption; this has been fixed.
739 (reported by Rainer Joswig)
740 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
741 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
742 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
743 arithmetic in general, is now much more reliable on the PPC
745 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
746 platform now returns the right answer.
747 * optimization: restored some effective method precomputation in
748 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
749 precomputation is now tunable.
750 * optimization: compiler-internal data structure use has been
751 reviewed, and changes have been made that should improve the
752 performance of the compiler by about 20%.
753 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
754 simple-base-strings and simple-bit-vectors is improved.
755 * optimization: the optimization of 32-bit logical and arithmetic
756 functions introduced in version 0.8.3 on the x86 has been
757 implemented on the mips, ppc and sparc platforms; an
758 implementation of the same facility, but for 64-bit arithmetic,
759 has been added for the alpha.
760 * microoptimization: the compiler is better able to make use of the
761 x86 LEA instruction for multiplication by constants.
762 * bug fix: in some situations compiler did not report usage of
763 generic arithmetic in (SPEED 3) policy.
764 * bug 145b fix: compiler used wrong type specifier while converting
765 MEMBER-types to numeric.
766 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
768 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
769 index argument now works properly on the Alpha platform.
770 * bug fix: floating point exception treatment on the Alpha platform
772 * bug fix: FILE-POSITION works much better on string input and
773 output streams. (thanks to Nikodemus Siivola)
774 * bug fix: many threading/garbage collection symptoms sorted.
775 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
776 might be pseudo-atomic.
777 * internal change: Stopping for GC is now done with signals not
778 ptrace. GC is now done in whichever thread wanted it, instead of
780 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
782 * bug fix: result form in DO is not contained in the implicit
784 * incompatible change: ICR structure is changed; the value part of
785 CONTINUATION is now called LVAR; corresponding functions are
786 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
787 * added type deriver for ISQRT (thanks to Robert E. Brown).
788 * bug fix: better support for loading from the command line when an
789 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
791 * fixed some bugs revealed by Paul Dietz' test suite:
792 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
793 ** ROUND and FROUND now give the right answer when given very
794 small float arguments.
795 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
797 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
798 ** (GCD 0 <negative-integer>) returned <negative-integer>.
799 ** LCM should return a non-negative integer.
800 ** PARSE-INTEGER returned the index of a terminator instead of the
801 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
802 ** PARSE-INTEGER returned an incorrect index being applied to a
804 ** LCM with two arguments of 0 returns 0 rather than signalling
806 ** unsigned addition of a 32-bit constant with the high bit set no
807 longer causes an internal compiler error on the x86.
808 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
809 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
810 with negative last argument.
811 ** byte specifiers with zero size and position no longer cause
812 an error during type derivation.
813 ** bignum multiplication on the Alpha platform now returns the
815 * porting: The system now builds on SuSE AMD64, although it still
816 generates a 32-bit binary.
817 * .fasl file incompatibility: The fasl file version number has
818 been incremented (because of the changes to internal compiler
819 data structures referred to above).
821 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
822 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
823 more accurately, on the Darwin kernel running on PowerPC hardware.
824 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
825 * Compiler code deletion notes now signal a condition of type
826 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
827 an associated MUFFLE-WARNING restart.
828 * The compiler now performs limited argument count validation of
829 constant format strings in FORMAT, and where appropriate in ERROR,
830 CERROR and WARN. (thanks to Gerd Moellmann)
831 * New ASDF-INSTALL contrib can be used for automatic download and
832 installation of third-party Lisp code from CCLAN or other sites
834 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
835 which forces another thread to execute a function supplied by the
837 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
838 now accept and act upon their :ELEMENT-TYPE keyword argument.
839 (reported by Martin Atzmueller, Edi Weitz)
840 * bug fix: FILE-POSITION now accepts position designators up to
841 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
842 is the greater. (thanks to Patrik Nordebo)
843 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
844 Kalle Olavi Niemitalo)
845 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
846 printable. (reported by Eric Marsden)
847 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
848 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
849 * x86 bug fix in control stack exhaustion checking: now shows backtrace
850 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
851 (thanks to Stig Sandoe)
852 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
853 UNREAD-CHAR now correctly includes the unread character in the
854 target sequence. (thanks to Gerd Moellmann)
855 * bug fix in threaded builds: the system can now be suspended and
856 resumed by shell job control with minimal disruption.
857 * bug fixes in times and timezones >2038AD
858 * better handling of "where is GNU make?" problem in build scripts
859 (thanks to Nikodemus Siivola)
860 * new optimization: inside a named function any reference to a
861 function with the same name is considered to be a self-reference;
862 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
863 optimization quality.
864 * new optimization on x86: logical functions and + now have
865 optimized (UNSIGNED-BYTE 32) versions, which are automatically
866 used when the result is truncated to 32 bits.
867 * VALUES declaration is partially enabled.
868 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
869 * fixed some bugs revealed by Paul Dietz' test suite:
870 ** The system now obeys the constraint imposed by
871 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
872 types form a lattice under type intersection.
873 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
874 ** ASSOC now ignores NIL elements in an alist.
875 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
876 and (1+ MOST-POSITIVE-FIXNUM) answers.
877 ** The addition of a method with invalid qualifiers to a generic
878 function does not cause an error to be signalled immediately;
879 a warning is signalled, and the error is generated only on
880 calling the generic function.
881 * changed .fasl file version number, in part to add type codes for
882 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
883 obscure ANSI requirements
885 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
886 * fixed bug 148: failure to inline-expand a local function left
887 garbage, confusing the compiler.
888 * fixed bugs 3cd: structure slot readers perform type check if the
889 slot can have an invalid value (i.e. it is either not initialized
890 or can be written with a less specific slot writer).
891 * bug fix: the compiler now traps array references to elements off
892 the end of an array; previously, the bounds checking in some
893 circumstances could go off-by-one.
894 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
896 * type declarations for array element types now obey the description
897 on the CLHS page "Declaration TYPE", as per discussions on
898 sbcl-help around 2003-05-08. This means that a declaration
899 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
900 declaration, all references to BAR will be asserted or assumed
901 (with THE, so dependent on compiler policy) to involve objects of
902 type FOO. Note that no such declaration is implied in
903 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
904 * declared types of functions from the "Conditions"
905 chapter. (reported by Paul Dietz)
906 * bug fix: CERROR accepts a function as its first argument.
907 * bug fix: NTH an NTHCDR accept a bignum as index
908 arguments. (reported by Adam Warner)
909 * optimization: character compare routines now optimize comparing
910 against a constant character. (reported by Gilbert Baumann)
911 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
912 argument now works properly on the MIPS platform.
913 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
914 * repeated evaluation of the same DEFSTRUCT, a slot of which is
915 declared to have a functional type, does not cause an error
917 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
918 many :DEFAULT-INITARGS. (reported by Istvan Marko)
919 * fixed bug: if last continuation of a deleted block has a
920 destination, this destination should be deleted too. (reported by
922 * fixed a bug in the bootstrap process: the host compiler's values
923 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
924 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
925 test case from Patrik Nordebo)
926 * improved the ability of the disassembler on the PPC platform to
927 provide helpful disassembly notes.
928 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
929 the class in more cases than previously.
930 * bug fix: FILE-POSITION now understands :START and :END for
931 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
932 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
933 8) data. (thanks to David Lichteblau)
934 * bug fix: it is possible to add a method to a generic function
936 * bug fix: reader failed to signal END-OF-FILE inside an
937 object representation. (reported by Nikodemus Siivola)
938 * fixed some bugs revealed by Paul Dietz' test suite:
939 ** LAST and [N]BUTLAST should accept a bignum.
940 ** condition slot accessors are methods.
941 ** (VECTOR NIL) is a subtype of STRING.
943 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
944 * minor incompatible change: some nonsensical specialized lambda
945 lists (used in DEFMETHOD) which were previously ignored now signal
947 * minor incompatible change: the system is now aware of the types of
948 variables in the COMMON-LISP package, and will signal errors for
949 most violations of these type constraints (where previously they
950 were silently accepted).
951 * minor incompatible change: COMPILE-FILE now uses the freedom
952 afforded (ANSI 3.2.2.3) to use derived function types for
953 functions defined in the same file. This also permits the system
954 to warn on static type mismatches and function
955 redefinition. (Currently it does not work with high DEBUG level.)
956 * minor incompatible change: VALUES declaration is disabled.
957 * When issuing notes, the compiler now signals a condition of type
958 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
959 restart for use in user handlers. It is expected that the
960 COMPILER-NOTE condition will eventually become a condition
961 supertype to a hierarchy of note types, which will then be
962 handleable in a similar fashion. However, at the moment, no such
963 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
964 but it's an implementation detail, not a classification for the
966 * Changes in type checking closed the following bugs:
967 ** type checking of unused values (192b, 194d, 203);
968 ** template selection based on unsafe type assertions (192c, 236);
969 ** type checking in branches (194bc).
970 * A short form of VALUES type specifier has ANSI meaning (it has
971 increased the number of situations when SBCL cannot perform type
973 * fixed bug in DEFSTRUCT: once again, naming structure slots with
974 keywords or constants is permissible.
975 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
976 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
977 classes. (thanks to Antonio Martinez)
978 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
979 outputs FRESH-LINE or TERPRI, and no longer converts its stream
980 argument to a pretty-print stream. Instead, it leaves any such
981 operations to DESCRIBE-OBJECT methods.
982 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
984 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
985 lambda lists. (thanks to David Lichteblau)
986 * bug fix: Defining a generic function with a :METHOD-CLASS being a
987 subclass of STANDARD-METHOD no longer causes stack exhaustion.
988 (thanks to Gerd Moellmann)
989 * fixed bug 246: increased compilation speed of long
990 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
992 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
994 * DEFCONSTANT now throws a condition of type
995 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
996 constant to a non-EQL value; CONTINUE and ABORT restarts
997 respectively change and preserve the value.
998 * fixed bug 63: The code walker, part of the implementation of CLOS,
999 is now better at handling symbol macros.
1000 * bug fix: There is no longer an internal implementation type named
1001 CL:LENGTH. (reported by Raymond Toy)
1002 * bug fix: In macro-like defining macros/special operators the
1003 implicit block does not enclose the lambda list.
1004 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1005 enclosed in parentheses) are not suitable as type specifiers, and
1006 their use properly signals an error now.
1007 * bug fix: An argument count mismatch for a type specifier in code
1008 being compiled no longer causes an unhandled error at compile
1009 time, but signals a compile-time warning.
1010 * fixed simple vector readable printing
1011 * bug fix: DESCRIBE takes more care over whether the class
1012 precedence list slot of a class is bound before accessing it.
1013 (reported by Markus Krummenacker)
1014 * bug fix: FORMATTER can successfully compile pretty-printer format
1015 strings which use variants of the ~* directive inside.
1016 * bug fix: SEARCH now applies its TEST predicate to the elements of
1017 the arguments in the correct order. (thanks to Wolfhard Buss)
1018 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1019 in type inference. (thanks to Robert E. Brown)
1020 * bug fix: Reading in symbols with an explicit package name of ""
1021 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1022 * fixed some bugs revealed by Paul Dietz' test suite:
1023 ** NIL is now allowed as a structure slot name.
1024 ** Arbitrary numbers, not just REALs, are allowed in certain
1025 circumstances in LOOP for-as-arithmetic clauses.
1026 ** Multiple class redefinitions before slot access no longer
1027 causes a type error.
1028 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1029 association between the name and a class.
1030 ** Generic functions with non-standard method-combination and over
1031 five methods all of which return constants no longer return NIL
1032 after the first few invocations. (thanks to Gerd Moellmann)
1033 ** CALL-NEXT-METHOD with no arguments now passes the original
1034 values of the arguments, even in the presence of assignment.
1035 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1037 ** DISASSEMBLE works with closures and funcallable instances.
1038 ** ADD-METHOD now returns the generic function, not the new method.
1039 ** FIND-METHOD signals an error if the lengths of the specializers
1040 is incompatible with the generic function, even if the ERRORP
1042 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1043 which its argument is a member.
1044 ** DEFCLASS only redefines the class named by its class-name
1045 argument if that name is the proper name of the class;
1046 otherwise, it creates a new class.
1047 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1048 of the UNBOUND-SLOT condition to the name of the slot.
1049 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1050 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1051 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1052 SLOT-MAKUNBOUND in the specified fashion.
1054 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1055 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1056 cross-compilation host. As a consequence, we can now bootstrap our
1057 way up to SBCL starting with a bare gcc toolchain and human-readable
1058 source code (first the source to CLISP, then the source to SBCL).
1059 * A contributed module containing a partial implementation of the
1060 simple-streams interface has been included. (thanks to Rudi
1062 * A contributed module implementing the RFC1321 Message Digest
1063 Algorithm, known as MD5, has been included.
1064 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1065 no longer has any effect, as the code controlled by this feature
1066 has been deleted. (As far as we know, no-one has ever built using
1067 this feature, and its semantics were confused in any case).
1068 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1069 work on conditions (as required by the ANSI specification),
1070 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1071 expected behaviour on conditions. Users should note, however,
1072 that such behaviour is not required by the ANSI specification,
1073 and so use of this behaviour may render their code unportable.
1074 * fixed some bugs revealed by Paul Dietz' test suite:
1075 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1077 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1078 ** SXHASH on condition objects no longer returns NIL.
1079 ** :ALLOCATION :CLASS slots are better treated; their values are
1080 updated on class redefinition, and initforms inherited from
1081 superclasses are applied.
1082 ** REMOVE-METHOD returns its generic function argument even when
1083 no method was removed.
1084 ** SHARED-INITIALIZE now initializes the values of the requested
1085 slots, including those with :ALLOCATION :CLASS.
1086 ** ALLOCATE-INSTANCE now works on structure classes defined via
1087 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1089 ** SLOT-EXISTS-P now works on conditions, as well as structures
1091 ** MAKE-LOAD-FORM now has the required methods on
1092 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1093 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1094 keyword, and now implements the SLOT-NAMES argument.
1095 ** methods with &OPTIONAL arguments no longer allow too many
1096 arguments to be passed in the call without error.
1097 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1098 option is consistent with the required arguments of the generic
1099 function lambda list.
1100 * bug fix: REQUIRE accepts a string designator. (Thanks to
1102 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1103 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1104 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1105 * bug fix: The compiler now checks for duplicated variables in macro
1107 * bug fix: SETQ on globals returns the correct value.
1108 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1109 not a COMPILER-ERROR (followed by some other strange error on
1110 choosing the CONTINUE restart).
1111 * bug fix: make.sh and friends are now more consistent in the way that
1112 they look for GNU "make".
1114 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1115 * experimental native threads support (on x86 Linux >=2.4 only).
1116 This is not compiled in by default: you need to add :SB-THREAD to
1117 the target features. See the "Beyond ANSI" chapter of the manual
1119 * fix for longstanding nonANSIism: The old distinction between
1120 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1121 The return value from CL:FIND-CLASS is now a CLOS class, and
1122 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1123 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1124 * An interface to the MetaObject Protocol, as described in Kiczales,
1125 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1126 MIT Press, 1991, is available from the SB-MOP package.
1127 * incompatible change: the SB-PCL package should now be considered
1128 a private implementation detail, and no longer a semi-private MOP
1130 * minor incompatible change: due to rearrangement for threads, the
1131 control stack and binding stack are now allocated at arbitrary
1132 addresses instead of being hardcoded per-port. Users affected by
1133 this probably have to be doing advanced things with shared
1134 libraries, and will know who they are.
1135 * minor incompatible change: Previously, all --eval forms used were
1136 processed with READ before any of them were processed with EVAL.
1137 Now each --eval form is processed with both READ and EVAL before
1138 the next --eval form is processed. (Thus package operations like
1139 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1140 work as the user might reasonably expect.)
1141 * minor incompatible change: *STANDARD-INPUT* is now only an
1142 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1144 * minor incompatible change: Y-OR-N-P is now character-oriented,
1145 not line oriented. Also, YES-OR-NO-P now works without errors.
1146 (thanks to Antonio Martinez)
1147 * sb-aclrepl module improvements: an integrated inspector, added
1148 repl features, and a bug fix to :trace command.
1149 * Known functions, which cannot be open coded by the backend, are
1150 considered to be able to check types of their arguments. (fixing
1151 a bug report by Nathan J. Froyd)
1152 * fixed a bug in computing method discriminating functions: It is
1153 now possible to define methods specialized on classes which have
1154 forward-referenced superclasses. (thanks to Gerd Moellmann)
1155 * fixed evaluation order in optional entries (reported by Gilbert
1157 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1158 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1159 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1160 required-by-AMOP NAME argument, as well as CLASS and
1161 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1162 * fixed bug 20: DEFMETHOD can define methods using names that are
1163 not the proper names of classes to designate class specializers.
1164 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1165 Unix file descriptors, instead of blowing up. (thanks to Antonio
1167 * Garbage collection refactoring: user-visible change is that a
1168 call to the GC function during WITHOUT-GCING will not do garbage
1169 collection until the end of the WITHOUT-GCING. If you were doing
1170 this you were probably losing anyway.
1171 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1172 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1173 TYPEP the latter but not the former.
1174 * The compiler issues a full WARNING for calls to undefined functions
1175 with names from the CL package.
1176 * MAP-INTO for a vector destination is open coded. (reported by
1177 Brian Downing on c.l.l)
1178 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1179 documentation string.
1180 * fixed some bugs revealed by Paul Dietz' test suite:
1181 ** COPY-ALIST now signals an error if its argument is a dotted
1183 ** Condition slots are now accessed more correctly in the presence
1184 of multiple initargs for a given slot.
1185 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1186 exclude restarts of the same name associated with a different
1188 ** DEFCLASS of forward-referenced classes with another
1189 forward-referenced class in the superclasses list no longer
1191 ** Condition slots are now initialized once each, not multiple
1192 times. (thanks to Gerd Moellmann)
1193 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1195 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1196 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1197 arguments contain duplicated elements.
1198 ** RESTART-CASE understands local macros.
1199 ** RESTART-CASE associates exactly its own restarts with a condition.
1200 ** ENDP in safe mode checks its argument to be of type LIST.
1201 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1202 in question is unbound.
1203 ** Optimized MAKE-INSTANCE functions no longer cause internal
1204 assertion failures in the presence of duplicate initargs.
1205 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1206 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1207 propagated even in the ctor optimized implementation of
1209 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1211 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1212 * a better implementation of SXHASH on (simple) bit vectors,
1213 measured both in execution speed and in distribution of results
1214 over the positive fixnums, has been installed. Likewise, a better
1215 implementation of EQUAL for simple bit vectors is now available.
1216 * fixed CEILING optimization for a divisor of form 2^k.
1217 * fixed bug 240 (emitting extra style warnings "using the lexical
1218 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1219 by Antonio Martinez)
1220 * fixed SXHASH, giving different results for NIL depending on type
1221 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1222 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1223 variable. (found by Rolf Wester)
1224 * SBCL does not ignore type declarations for special
1225 variables. (reported by rif on c.l.l 2003-03-05)
1226 * some bug fixes in contrib/sb-aclrepl/
1227 * fixed some bugs revealed by Paul Dietz' test suite:
1228 ** a bug in the CONS type specifier, whereby the CAR and CDR
1229 types got intertwined, has been fixed;
1230 ** the type system is now able to reason about the interaction
1231 between INTEGER and RATIO types more completely;
1232 ** APPEND, [N]REVERSE and NRECONC check that those their
1233 arguments, which must be proper lists, are really so;
1234 ** An array specialized to be unable to hold elements has been
1235 implemented, as required -- yes, really -- by ANSI;
1236 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1237 on malformed property lists;
1239 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1240 * incompatible packaging change: in line with Unix convention,
1241 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1242 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1243 * REQUIRE and PROVIDE are now optionally capable of doing something
1244 useful. See the documentation string for REQUIRE.
1245 * infrastructure for a managed SBCL contrib system: contributed
1246 modules in this release include:
1247 ** the ASDF system definition facility;
1248 ** an interface to the BSD Sockets API;
1249 ** an ACL-like convenience interface to the repl;
1250 (thanks to Kevin Rosenberg)
1251 ** an implementation of ROTATE-BYTE, with efficient implementation
1253 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1254 gives a closer approximation to the right answer than previously.
1255 (thanks to Raymond Toy)
1256 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1257 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1258 optional environment arguments, as required by ANSI.
1259 * fixed bugs in other functions taking environment objects, allowing
1260 calls with an explicit NIL environment argument to be compiled
1262 * fixed bug 228: primary return values from
1263 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1264 COMPILE or FUNCTION.
1265 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1266 :TYPE will no longer signal errors on innocuous objects.
1267 * fixed bug 231b: SETQ is better at respecting type declarations in
1268 the lexical environment.
1269 * fixed a bug in DEFCLASS: classes named by symbols with no or
1270 unprintable packages can now be defined.
1271 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1272 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1273 * fixed compiler failure related to checking types of functions.
1274 (reported by Robert E. Brown)
1275 * the compiler is now much more consistent in its error-checking
1276 treatment of bounding index arguments to sequence functions: in
1277 (SAFETY 3) code, errors will be signalled in almost all cases if
1278 invalid sequence bounding indices are passed to functions defined
1279 by ANSI to operate on sequences.
1280 * fixed a bug in the build procedure: documentation of SBCL-specific
1281 packages is now preserved and available in the final Lisp image.
1282 * lifted FDEFINITION lookup out of loops in the implementation of
1283 many list operations. (thanks to Robert E. Brown)
1284 * fixed a bug in the reader: the #n# reader macro now works for
1285 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1286 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1287 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1288 E. Brown and Tony Martinez respectively)
1289 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1290 SLOT-BOUNDP: in particular, it is now able to optimize them much
1291 better, and is now not vulnerable to having packages renamed.
1292 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1293 always called when a slot is not present in an instance. (thanks
1295 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1296 longer signal an error on unreading a character following EOF on
1297 the previous constituent stream. (thanks to Tony Martinez)
1298 * fixed some bugs revealed by Paul Dietz' test suite:
1299 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1300 not just nonnegative fixnums;
1301 ** the logical bit-array operators such as BIT-AND now accept an
1302 explicit NIL for their "opt-arg" argument (to indicate a
1303 freshly-consed result bit-array);
1304 ** ELT now signals an error on an invalid sequence index in safe
1306 ** the type system is now cleverer about negations of numeric
1307 types, and consequently understands the BIGNUM and RATIO types
1309 ** the type system is now cleverer about the interaction between
1310 INTEGER and RATIO types: while bugs still remain, many more
1311 cases are accurately computed;
1312 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1313 if it is in the last clause;
1314 ** CONSTANTLY now correctly returns a side-effect-free function in
1316 ** DECLARE is no longer treated as a special-operator; in
1317 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1318 * incremented fasl file version number due to the change in the
1319 DEFSTRUCT-SLOT-DESCRIPTION structure.
1321 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1322 * minor incompatible change: code processed by the "interpreter" or
1323 EVAL now has a compilation optimization policy of (DEBUG 2)
1324 (changed from (DEBUG 1)) to improve debuggability of interactive
1325 development, and to allow the use of the debug RETURN command in
1327 * an experimental implementation of the RETURN command for the
1328 debugger has been included. (thanks to Frederik Kuivinen)
1329 * fixed bug 62: constraints were not propagated into a loop.
1330 * fixed bug in embedded calls of SORT (reported and investigated by
1332 * fixed some bugs revealed by Paul Dietz' test suite:
1333 ** printing and reading of arrays with some dimensions having
1334 length 0 (thanks to Gerd Moellmann);
1335 ** BOA constructor with &AUX argument without a default value does
1336 not cause a type error;
1337 ** CONSTANTP now returns true for all self-evaluating objects.
1339 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1340 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1341 accessors that are related by inheritance, as specified in the
1342 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1343 Valtteri Vuorikoski)
1344 * The compiler is now able to inline functions that were defined in
1345 a complex lexical environment (e.g. inside a MACROLET).
1346 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1348 * Support for the upcoming FreeBSD-5.0 release has been included.
1349 (thanks to Dag-Erling Smorgrav)
1350 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1351 effect when it is not in a toplevel context.
1352 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1353 stand a better chance of being correct. (thanks to Gerd
1355 * fixed bug in COERCE, which now signals an error on coercing a
1356 rational to a bounded real type which excludes the expected
1358 * The compiler is now able to derive types more accurately from the
1359 COERCE and COMPILE functions.
1360 * fixed bug 223: functional binding is considered to be constant
1361 only for symbols in the CL package.
1362 * fixed bug 231: SETQ did not check the type of a variable being set
1363 (reported by Robert E. Brown)
1364 * A new optimization for MAKE-INSTANCE has been included, fixing
1365 various bugs (including relating to :ALLOCATION :CLASS slots and
1366 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1368 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1369 ** As required by ANSI, LOOP now disallows anonymous collection
1370 clauses such as COLLECT I in conjunction with aggregate boolean
1371 clauses such as THEREIS (= I 1);
1372 ** LOOP now signals an error when any variable is reused in the
1373 same loop (including the potentially useful construct analogous
1374 to WITH A = 1 WITH A = (1+ A);
1375 ** IT is only a special loop symbol within the first clause of a
1376 conditional loop clause;
1377 ** LOOP with a typed iteration variable over a hashtable now
1378 signals a type error iff it should.
1379 * fixed some other bugs revealed by Paul Dietz' test suite:
1380 ** FILE-STREAM now names the class previously known as FD-STREAM;
1381 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1382 argument) no longer signals an error;
1383 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1384 of the slot symbol, rather than using the current package
1385 ((:CONC-NAME "") continues to intern the slot's name in the
1387 * incremented fasl file version number, because of the incompatible
1388 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1389 of the new implementation of DEFINE-COMPILER-MACRO.
1391 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1392 * Support for building SBCL for MIPS platforms running in
1393 little-endian mode has now been checked in, and basic
1394 functionality on said platforms verified.
1395 * minor incompatible change: PCL now records the pathname of a file
1396 in which methods and the like are defined, rather than its
1398 * minor incompatible change: TRUENAME now considers the truename of
1399 a file naming a directory to be the pathname with :DIRECTORY
1400 component indicating that directory.
1401 * minor incompatible change: a NAMED clause in the extended form of
1402 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1403 reason for the previous behaviour is unclear.
1404 * more systematization and improvement of CLOS and MOP conformance
1405 in PCL (thanks to Gerd Moellman and Pierre Mai):
1406 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1408 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1409 primary methods with no specializers;
1410 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1412 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1413 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1414 FORWARD-REFERENCED-CLASSes; error reporting on
1415 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1417 ** SXHASH on CLOS instances now uses a slot internal to the
1418 instance to return different numbers on distinct instances,
1419 while preserving the same return value through invocations of
1421 ** DEFMETHOD signals errors when methods with longer incongruent
1422 lambda lists are added to generic functions;
1423 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1424 CLASS, as specified in AMOP;
1425 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1426 on the order returned by the primary method for classes of
1427 class STANDARD-CLASS;
1428 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1429 * fixed some bugs shown by Paul Dietz' test suite:
1430 ** DOLIST puts its body in TAGBODY;
1431 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1433 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1434 value producing form;
1435 ** if more variables are given to PROGV than values, extra
1436 variables are bound and made to have no value;
1437 ** NSUBSTITUTE on list arguments gets the right answer with
1439 ** ELT signals an error of type TYPE-ERROR when the index argument
1440 is not a valid sequence index;
1441 ** LOOP signals (at macroexpansion time) an error of type
1442 PROGRAM-ERROR when duplicate variable names are found;
1443 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1444 ** FUNCALL of special-operators now cause an error of type
1446 ** PSETQ now works as required in the presence of side-effecting
1447 symbol-macro places;
1448 ** NCONC accepts any object as its last argument;
1449 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1451 ** loop-for-as-package does not require a package to be explicitely
1453 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1455 * fixed bug 166: compiler preserves "there is a way to go"
1456 invariant when deleting code.
1457 * fixed bug 172: macro lambda lists with required arguments after
1458 &REST arguments now cause an error to be signalled. (thanks to
1460 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1461 bug. (thanks to Matthew Danish)
1462 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1464 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1465 arguments are lexically rebound. (thanks to Gerd Moellmann and
1467 * fixed bug 194: error messages are now more informative when there
1468 is no primary method applicable in a call to a generic
1469 function. (thanks to Gerd Moellmann)
1470 * fixed bug in command line argument checking (thanks to Julian
1472 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1473 (thanks to Matthew Danish)
1474 * incremented fasl file version number, because of the
1475 SXHASH-related changes in the layout of CLOS data structures
1477 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1478 * minor incompatible change: The runtime (the Unix executable named
1479 "sbcl") is now much pickier about the .core files it will load.
1480 Essentially it now requires .core files to descend from the same
1481 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1482 as the runtime does. (The intent is to prevent the crashes which
1483 can occur, and which can even be reported as mysterious failures,
1484 when people patch the sources or change the build parameters
1485 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1486 sbcl and .core files.)
1487 * fixed bug: VALUES-LIST is no longer optimized away.
1488 * fixed bug 142: The FFI conversion of C string values to Lisp
1489 string values no longer conses excessively. (thanks to Nathan
1490 Froyd porting Raymond Toy's fix to CMU CL)
1491 * began to systematize and improve MOP conformance in PCL (thanks to
1492 Nathan Froyd, Gerd Moellman and Pierre Mai):
1493 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1495 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1496 ** FINALIZE-INHERITANCE is now called on class finalization;
1497 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1498 argument precedence order.
1499 * fixed bug 202: The compiler no longer fails on functions whose
1500 derived types contradict their declared type.
1501 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1502 so it can be non-toplevel.
1503 * The fasl file version number has changed (because of the new
1504 implementation of DEFMACRO).
1505 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1506 safe code, that any length requirement by their type-specifier
1507 argument is valid. The exceptions to this are described in bug
1509 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1510 functions now check the types of their inputs as required by ANSI.
1511 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1512 introduced symbol is DECLAREd to be SPECIAL.
1513 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1514 * fixed bug: :COUNT argument to sequence functions may be negative
1515 * fixed bug: body of DO-SYMBOLS may contain declarations
1516 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1517 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1518 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1519 symbol macro only once
1520 * fixed printing of call frame when argument list is unavailable
1521 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1522 * compiler no longer signals WARNING on unknown keyword
1525 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1526 * A beta-quality port to the mips architecture running Linux,
1527 based on the old CMUCL backend, has been made. It has been tested
1528 on a big-endian kernel, and works sufficiently well to be able to
1529 rebuild itself; it has not been tested in little-endian mode.
1530 * fixed an inconsistency between gencgc.c and purify.c which made
1531 dumping/loading .core files unreliable
1532 * fixed bug 120a: The compiler now deals correctly with IFs where
1533 the consequent is the same as the alternative, instead of
1534 misderiving the return type. (thanks to Alexey Dejneka)
1535 * fixed bug 113: Logical pathnames are now dumpable (the logical
1536 host is resolved at load-time, throwing an error if it is not
1538 * fixed bug 174: FORMAT's error message is slightly clearer when a
1539 non-printing character is used in a format directive.
1540 * fixed several bugs in compiler checking of type declarations, i.e.
1541 violations of the Python "declarations are assertions" principle
1542 (thanks to Alexey Dejneka)
1543 * fixed several bugs in PCL's error checking (thanks to Gerd
1545 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1547 * fixed bug in compilation of functions as first class values
1548 (thanks to Antonio Martinez-Shotton)
1549 * The compiler's handling TYPE-ERRORs which it can prove will
1550 inevitably happen at runtime has been cleaned up and corrected
1551 in several ways. (thanks to Alexey Dejneka)
1552 * improved argument type checking for various basic arithmetic
1553 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1554 lavished on them in the past that they can be compiled in many
1555 ways in different special cases
1556 * fixed bug 181: compiler checks validity of user supplied type
1558 * cleaned up code flushing in optimization: Function calls which
1559 should signal errors for safety purposes (e.g. which ANSI says
1560 should signal errors when their arguments are of incorrect type)
1561 are no longer optimized away.
1562 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1563 * incremented fasl file version number, because changes in the
1564 implementation of sequence functions like COERCE caused
1565 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1566 in old inline expansions) to become undefined. (Actually these
1567 changes were later undone, so we might very well be binary
1568 compatible with 0.7.7 after all, but leaving the version number
1569 incremented seemed like the simplest and most conservative
1572 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1573 * An alpha-quality port to the parisc architecture running Linux,
1574 based on the old CMUCL backend, has been made. This, even more so
1575 than the other backends, should be considered still a work in
1576 progress; known problems include that the Linux kernel in 64-bit
1577 mode does not propagate the correct sigcontext structure to
1578 userspace, and consequently SBCL on a parisc64 kernel will not
1580 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1581 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1582 suppressed.) Also now that the compiler is looking at declarations
1583 in the environment, it checks optimization declarations as well,
1584 and suppresses inlining when (> DEBUG SPEED).
1585 * More fixes have been made to treatment of floating point exception
1586 treatment and other Unix signals. In particular, floating point
1587 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1588 * The detection and handling of control stack exhaustion (infinite
1589 or very deeply nested recursion) has changed. Stack exhaustion
1590 detection is now done by write-protecting pages at the OS level
1591 and applies at all optimization settings; when found, a
1592 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1593 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1594 be caught using IGNORE-ERRORS.
1595 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1596 correctly when transforming recursive LABELS functions to LETs.
1597 (thanks to Alexey Dejneka)
1598 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1599 that are names of constants or global variables.
1600 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1601 alien routines with docstrings.
1602 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1603 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1605 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1606 (thanks to Eric Marsden)
1607 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1608 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1609 * Bug fix: The SPARC backend can now compile functions involving
1610 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1611 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1612 to a functions expecting a PATHNAME-DESIGNATOR.
1613 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1614 lambda lists. (thanks to Alexey Dejneka)
1615 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1616 functions so that EXPORTing the name of the class doesn't cause
1617 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1618 to Antonio Martinez for reporting this)
1619 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1620 constant-folded without failing an assertion. (thanks to Einar
1621 Floystad Dorum for reporting this)
1622 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1623 types are now dealt with more correctly by the compiler.
1624 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1625 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1626 behaviour analogous to RENAME-FILE. This puts its behaviour more
1627 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1629 * The fasl file version number has changed again. (because of the
1630 bug fix involving the names of PCL MAKE-INSTANCE functions)
1632 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1633 * bug fix: Floating point exceptions are treated much more
1634 consistently on the x86/Linux and PPC/Linux platforms.
1635 * Array initialization with :INITIAL-ELEMENT is now much faster for
1636 cases when the compiler cannot open code the array creation, but
1637 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1638 array accesses have also seen a speed increase.
1639 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1640 specified. (This was at the root of some bad interactions between
1641 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1643 * bug fix: Internal error arguments for undefined functions are now
1644 computed correctly on the PPC/Linux platform.
1645 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1646 Raymond Toy's patch for CMU CL)
1647 * Support for the Solaris 9 operating environment has been included
1648 (thanks to Daniel Merritt)
1649 * A very ugly but hopefully complete draft of the missing FFI chapter
1650 of the manual has been created by reformatting the corresponding
1651 CMU CL manual chapter into (currently very ugly and incoherent)
1652 DocBook and bringing it up to date for SBCL behavior. Thus, the
1653 manual is now essentially complete, at least by my extreme
1654 once-and-only-once standards, whereby it's acceptable to refer to
1655 the doc strings of SB-EXT functions as the primary documentation.
1656 * The fasl file version number has changed again, due to cleanup of
1657 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1658 is no longer a static symbol.)
1660 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1661 * SBCL now builds with OpenMCL (version 0.12) as the
1662 cross-compilation host; also, more progress has been made toward
1663 bootstrapping under CLISP.
1664 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1666 * bug 158 fixed: The compiler can now deal with integer loop
1667 increments different from 1; fixing this turned out also to fix
1669 * bug 169 fixed: no more bogus warnings about using lexical bindings
1670 despite the presence of perfectly good SPECIAL declarations (thanks
1671 to David Lichteblau)
1672 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1673 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1674 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1676 * bug 180 fixed: Method combination specifications no longer ignore
1677 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1678 * bug fix: Structure type predicate functions now check their argument
1679 count as they should.
1680 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1681 correctly. (thanks to Pierre Mai)
1682 * minor incompatible change: The --noprogrammer option is deprecated
1683 in favor of the new --disable-debugger option, which is very similar.
1684 (The major difference is that it takes effect at a slightly different
1685 time at startup, causing handling of errors in --sysinit and
1686 --userinit files will be affected differently.) The
1687 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1688 been added to allow this functionality to be controlled from ordinary
1689 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1690 maintainers, who might want to run non-interactive scripts to
1691 build SBCL cores which will later be used interactively.)
1692 * minor incompatible change: The LOAD function no longer, when given
1693 a wild pathname to load, loads all files matching that pathname.
1694 Instead, an error of type FILE-ERROR is signalled.
1696 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1697 * bug 147 fixed: The compiler preserves its block link/count
1698 invariants more correctly now so that it doesn't crash. (thanks
1700 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1702 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1703 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1704 no-value-for-this-character as NIL (as specified by ANSI).
1705 * HOST-NAMESTRING on physical pathnames now returns a string that is
1706 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1707 (thanks to Christophe Rhodes)
1708 * The Alpha port handles icache flushing more correctly. (thanks to
1710 * More progress has been made toward bootstrapping under CLISP. (thanks
1711 to Christophe Rhodes)
1712 * The fasl file format has changed again, because dynamic loading
1713 on OpenBSD (which has non-ELF object files) motivated some cleanups
1714 in the way that foreign symbols are transformed and passed around.
1715 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1716 is no longer treated as whitespace by the reader, but instead as
1717 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
1718 |A
\x7fB|, instead of A as it used to.
1720 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1721 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1722 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1723 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1724 this as of 0.7.1.45, but was left out of the previous news section
1725 (thanks to Dan Barlow)
1726 * SBCL now runs on the Solaris operating system on SPARC architectures
1727 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1728 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1729 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1730 * SPARC backend cleanups, allowing builds of cores optimized for V8
1731 and V9 SPARCS, and also emission of code targeted to a particular
1732 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1734 * SBCL is closer to bootstrapping under CLISP, thanks to various
1735 fixes by Christophe Rhodes.
1736 * The fasl file format has changed again, to allow the compiler's
1737 INFO database to support symbol macros.
1738 * The user manual (in doc/) is formatted into HTML more nicely.
1739 (thanks to coreythomas)
1740 * The system is smarter about SUBTYPEP relationships, especially
1741 those involving NOT types (including types such as ATOM which are
1742 represented internally using NOT types). Thus SUBTYPEP is less
1743 likely to return (VALUES NIL NIL) in general, and in particular
1744 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1746 * The fasl file format has changed again, because the internal
1747 representation of types now includes a new slot to support the new
1748 SUBTYPEP-of-NOT-types logic.
1749 * (not a change in the main branch of SBCL, but a related prototype
1750 which can hopefully be merged into the main branch of SBCL in the
1751 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1752 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1753 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1754 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1755 dispatch macro character combinations. (thanks to Alexey Dejneka)
1756 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1757 that superficially look like logical namestrings correctly.
1758 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1760 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1761 default supplied-p) for &optional and &key arguments. (thanks to
1764 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1765 * incompatible change: The compiler is now less aggressive about
1766 tail call optimization, doing it only when (> SPACE DEBUG) or
1767 (> SPEED DEBUG). (This is an incompatible change because there are
1768 programs which relied on the old CMU-CL-style behavior to optimize
1769 away their unbounded recursion which will now die of stack overflow.)
1770 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1771 for non-GENCGC systems has been increased to 20M (since that
1772 seems much closer to the likely performance optimum for modern
1773 systems than the old 4M value was)
1774 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1775 and *DEBUG-PRINT-LEVEL*
1776 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1777 of CMU CL's support for SPARC, and various endianness and other
1778 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1779 * new syntactic sugar for the Unix command line: --load foo.bar is now
1780 an alternate notation for --eval '(load "foo.bar")'.
1782 ** The system now detects stack overflow and handles it gracefully,
1783 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1784 optimization settings. (This is a good thing in general, and
1785 its introduction in this version should be particularly timely
1786 for anyone whose code fails because of suppression of tail
1788 ** The system now hunts for the C variable "environ" in a more
1789 devious way, to avoid segfaults when the C library version differs
1790 between compile time and run time. (thanks to Christophe Rhodes)
1791 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1792 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1793 ** The compiler no longer issues bogus style warnings for undefined
1794 classes in the same source file as the DEFCLASSes which defined
1795 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1797 ** fixes in CONDITION class precedence list for undefined function
1798 errors (thanks to Alexei Dejneka)
1799 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1800 correctly. (thanks to Dan Barlow)
1801 ** portability fixes aiming at bootstrapping under CLISP (thanks
1802 to Dave McDonald and Christophe Rhodes)
1803 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1804 ** fixes in type translation and and type inference (thanks to
1806 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1807 ** various fixes in the new ports (thanks to Dan Barlow)
1808 * several changes related to debugging:
1809 ** suppression of tail recursion, as noted above
1810 ** stack overflow detection, as noted above
1811 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1812 is now the default. (For some time encapsulation has been more
1813 reliable than the breakpoint-based :ENCAPSULATE NIL
1814 implementation, at least on X86 systems; and I just noticed that
1815 encapsulation also seems closer to the spirit of the ANSI
1818 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1820 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1821 up properly again. (There was a packaging bug in 0.7.0 which
1822 left their definitions in SB-SYS::LOAD-FOREIGN and
1823 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1824 vital for most things which interface to C-level interfaces,
1825 like extensions working with sockets or databases or
1826 Perl-compatible regexes or whatever, and the need to fix
1827 this bug is the main reason that 0.7.1 was released so
1829 ** DEFGENERIC is now choosier about the methods it redefines, so that
1830 reLOADing a previously-LOADed file containing DEFGENERICs does
1831 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1832 cycle now works as it should. (thanks to Alexey Dejneka)
1833 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1834 fixed by Christophe Rhodes. (He actually submitted this patch
1835 months ago, and I delayed until after 0.7.0.)
1836 ** Bug 111 (internal compiler confusion about runtime checks on
1837 FUNCTION types) was fixed by Alexey Dejneka.
1838 * Some internal cleanups (getting rid of variables which aren't
1839 needed now that the byte interpreter is gone) caused the fasl
1840 file format number to change again.
1842 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1843 * major incompatible change: The default fasl file extension, i.e. the
1844 default extension for files produced by COMPILE-FILE, has changed
1845 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1847 ** There are many changes in the implementation of the compiler.
1848 SBCL is now essentially a compiler-only implementation of ANSI
1849 Common Lisp. EVAL still "interprets" a few special cases, but
1850 almost all the interesting cases are handled by creating
1851 a LAMBDA expression, calling COMPILE on it, then calling
1852 FUNCALL on the result.
1853 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1854 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1855 Since the code is newer, there might still be some new bugs
1856 (though not as many as before Martin Atzmueller's fixes:-). But
1857 the new code is substantially simpler and clearer, and hopefully
1858 any remaining bugs will be simpler, less fundamental, and more
1859 fixable then the bugs in the old code.
1860 ** The revised compiler is still a little unsteady on its feet.
1862 *** The debugging information it produces (particularly the names
1863 of FUNCTION objects) is sometimes much less useful than what
1864 the old compiler produced.
1865 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1866 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1867 within a MACROLET) has been temporarily weakened.
1868 ** There are new compiler optimizations for various functions:
1869 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1870 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1871 *** the math functions TRUNCATE, FLOOR, and CEILING
1872 *** the function-of-all-trades COERCE
1873 Mostly these should be transparent, but there's one
1874 potentially-annoying problem (bug 117): when the compiler
1875 inline-expands a function and does type analysis on the result,
1876 it can create control paths which have type mismatches, and
1877 when it can't prove that those control paths aren't taken,
1878 it will issue WARNINGs about the type mismatches. This is
1879 a particular problem in practice for the new sequence functions.
1880 It's not clear how this should be fixed, and for now, a
1881 workaround is given in the entry for 117 in the BUGS file.
1882 ** (Because of the interaction between the two previous items --
1883 occasional inlining problems and new inline expansions -- some
1884 of the new sequence function optimizations won't really kick in
1885 completely until debugging information, and then inlining, are
1886 straightened out in some future version.)
1887 * minor incompatible changes:
1888 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1889 DIRECTORY no longer implicitly promotes NIL slots of its
1890 pathname argument to :WILD. In particular, when you ask for the
1891 contents of a directory (which you used to be able to do without
1892 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1893 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1894 ** changes in behavior that ANSI explicitly defines to be
1895 implementation dependent:
1896 *** The new compiler-only implementation still conforms with ANSI,
1897 but acts a little different than before. Besides the obvious
1898 changes in performance tradeoffs (that the cost per form passed
1899 to EVAL has gone up, and the cost per form executed by EVAL
1900 has gone down), the behavior of the system changes a little
1901 because there are no longer any interpreted function objects.
1902 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1903 e.g. doing COMPILE on the output of interactive DEFUN is
1905 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1907 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1908 has changed from (:CL) to NIL.
1909 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1910 e.g. #\Newline and #\Tab, don't have names specified in the
1911 ANSI Common Lisp standard, is now based on their ASCII symbolic
1912 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1913 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1914 are no longer used for output.
1915 ** changes in internal implementation constants:
1916 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1917 4 million. (If your application spends a lot of time GCing and
1918 you have a lot of RAM, you might want to experiment with
1919 increasing it even more.)
1920 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1921 However, almost all old code should still continue to work without
1922 immediate update, as SB-C-CALL is now a (deprecated) nickname
1924 ** Old operator names in the style DEF-FOO are now deprecated in
1925 favor of new corresponding names DEFINE-FOO, for consistency with
1926 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1927 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1928 internal symbols, but a few supported extensions like
1929 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1930 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1931 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1932 etc. as you get deeper into recursive calls to the debugger
1933 command loop, instead of the old "5]", "5]]", "5]]]"
1934 sequence. (I was motivated to do this when squabbles between
1935 ILISP and SBCL left me very deeply nested in the debugger. In the
1936 short term, this change will probably provoke more ILISP/SBCL
1937 squabbles, but hopefully it will be an improvement in the long run.)
1938 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
1939 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
1940 function, since declaiming return types involving aliens is
1941 (1) annoyingly messy to do by hand and (2) vital to efficient
1942 compilation of code which calls such functions.
1943 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
1944 longer reexported by the SB-EXT package. They're solely useful
1945 for alien code, so it seems more logical that you should get
1946 them from the SB-ALIEN package, not in SB-EXT.
1947 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
1948 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
1949 features. Instead, the code that they used to control is always
1950 built into the system.
1951 * many other bug fixes
1952 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
1953 advantage of the new EVAL-WHEN stuff and to clean them up in
1954 general, and they are now more ANSI-compliant in a number of
1955 ways. Martin Atzmueller is responsible for a lot of this.
1956 ** Besides the cleanups discussed above, Martin Atzmueller fixed
1958 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
1959 *** correct ERROR type for various file operations
1960 *** some fixes for Lisp streams
1961 *** DEFMETHOD syntax checking
1962 *** changing old weird representation of debug information as
1963 strings (which, among their other deficiencies, don't transform
1964 correctly when you rename packages, and don't change their
1965 print representation when you change things like *PACKAGE*
1966 and *PRINT-LENGTH*) to symbols and lists of symbols
1967 He also made several improvements and fixed several bugs in DESCRIBE.
1968 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
1970 *** misbehavior of WRITE-STRING/WRITE-LINE
1971 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
1972 and several other LOOP problems as well
1973 *** DIRECTORY when similar filenames are present
1974 *** DEFGENERIC with :METHOD options
1975 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
1976 *** bug in the optimization of ARRAY-ELEMENT-TYPE
1977 *** argument ordering in FIND with :TEST option
1978 *** mishandled package designator argument in APROPOS-LIST
1979 *** various problems in the backquote readmacro
1980 *** a bug in APROPOS
1981 *** probably some others that I'm not describing very well here,
1982 since the CVS log documents them by reference to sbcl-devel
1983 messages, and the SourceForge archives aren't working well.:-(
1984 ** Dan Barlow improved the Alpha port (and is making progress on the
1985 PPC port, for those of you who think different).
1986 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
1987 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
1988 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
1989 continues to work on the SPARC port (for those of you in a position
1990 to look down upon our little PC-compatible boxes from a great height).
1991 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
1992 and uses that copy, rather than the current dynamic value, when
1993 it's trying to decide whether to truncate output. Thus e.g.
1994 (let ((*print-lines* 50))
1995 (pprint-logical-block (stream nil)
1997 (let ((*print-lines* 8))
1998 (print (aref possiblybigthings i) stream)))))
1999 should now truncate the logical block only at 50 lines, instead of
2000 often truncating it at 8 lines, as it did before.
2001 * The doc/cmucl/ directory, containing old CMU CL documentation
2002 from the time of the fork, is no longer part of the base system.
2003 SourceForge has shut down its anonymous FTP service, and with it
2004 my original plan for distributing the old CMU CL documentation
2005 there. For now, if you need these files you can download an old
2006 SBCL source release and extract them from it.
2007 * The fasl file version number changed again, for dozens of reasons,
2008 some of which are apparent above.
2010 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2011 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2012 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2013 patch, so that DISASSEMBLE output is much nicer.
2014 * The code in the SB-PROFILE package now seems reasonably stable.
2015 I still haven't decided what the final interface should look like
2016 (I'd like PROFILE to interact cleanly with TRACE, since both
2017 facilities use function encapsulation) but if you have a need
2018 for profiling now, you can probably use it successfully with
2019 the current CMU-CL-style interface.
2020 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2021 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2022 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2023 link now returns the dangling link itself, and for similar
2024 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2025 link itself. (In these cases the old code signalled an error and
2026 looped endlessly, respectively.) Thus, DIRECTORY now works even
2027 in the presence of dangling and cyclic symbolic links.
2028 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2029 is now a supported extension again, since the consensus on
2030 sbcl-devel was that it can be useful for ordinary development
2031 work, not just for debugging SBCL itself.
2032 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2033 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2034 that currently-defined functions might be redefined later with
2035 different return types.
2036 * Hash tables can be printed readably, as inspired by CMU CL code
2037 of Eric Marsden and SBCL code of Martin Atzmueller.
2038 * better error handling in CLOS method combination, thanks to
2039 Martin Atzmueller porting Pierre Mai's CMU CL patches
2040 * more overflow fixes for >16Mbyte I/O buffers
2041 * A bug in READ has been fixed, so that now a single Ctrl-D
2042 character suffices to cause end-of-file on character streams.
2043 In particular, now you only need one Ctrl-D at the command
2044 line (not two) to exit SBCL.
2045 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2047 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2048 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2049 does the right thing.
2050 * The implementation of some type tests, especially for CONDITION
2051 types, is now tidier and maybe faster, due to CMU CL code
2052 originally by Douglas Crosher, ported by Martin Atzmueller.
2053 * Some math functions have been fixed, and there are new
2054 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2055 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2056 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2057 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2058 likely to be built into sbcl-0.7.0.)
2059 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2060 CL, ported to SBCL by Martin Atzmueller)
2061 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2062 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2063 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2064 * There's a new slam.sh hack to shorten the edit/compile/debug
2065 cycle for low-level changes to SBCL itself, and a new
2066 :SB-AFTER-XC-CORE target feature to control the generation of
2067 the after-xc.core file needed by slam.sh.
2068 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2069 COMPILE-FILE is no longer supported, so that now every function
2070 gets an entry point, so that block compilation looks a little
2071 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2072 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2073 returns the number of bytes consed since the system started,
2074 rather than the number consed since the first time the function
2075 was called. (The new definition parallels ANSI functions like
2076 CL:GET-INTERNAL-RUN-TIME.)
2077 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2078 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2079 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2080 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2081 * Fasl file version numbers are now independent of the target CPU,
2082 since historically most system changes which required version
2083 number changes have affected all CPUs equally. Similarly,
2084 the byte fasl file version is now equal to the ordinary
2087 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2088 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2089 is no longer recognized. I apologize for this, because it was
2090 listed in SB-EXT as a supported extension, but I found that
2091 its existing behavior was poorly specified, as well as incorrectly
2092 specified, and it looked like too much of a mess to straighten it
2093 out. I have enough on my hands trying to get ANSI stuff to work..
2094 * many patches ported from CMU CL by Martin Atzmueller, with
2095 half a dozen bug fixes in pretty-printing and the debugger, and
2096 half a dozen others elsewhere
2097 * fixed bug 13: Floating point infinities are now supported again.
2098 They might still be a little bit flaky, but thanks to bug reports
2099 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2100 as flaky as they were.
2101 * The --noprogrammer command line option is now supported. (Its
2102 behavior is slightly different in detail from what the old man
2103 page claimed it would do, but it's still appropriate under the
2104 same circumstances that the man page talks about.)
2105 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2106 are now supported, and enabled by default. Thus, the compiler can
2107 handle many floating point and complex operations much less
2108 inefficiently. (Thus e.g. you can implement a complex FFT
2110 * The compiler now detects type mismatches between DECLAIM FTYPE
2111 and DEFUN better, and implements CHECK-TYPE more correctly, and
2112 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2113 thanks to patches from Martin Atzmueller.
2114 * various fixes to make the cross-compiler more portable to
2115 ANSI-conforming-but-different cross-compilation hosts (notably
2116 Lispworks for Windows, following bug reports from Arthur Lemmens)
2117 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2118 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2119 fixed thanks to Pierre Mai's CMU CL patches.
2120 * fixes to keep the system from overflowing internal counters when
2121 it tries to use i/o buffers larger than 16M bytes
2122 * fixed bug 45a: Various internal functions required to support
2123 complex special functions have been merged from CMU CL sources.
2124 (When I was first setting up SBCL, I misunderstood a compile-time
2125 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2126 * improved support for type intersection and union, fixing bug 12
2127 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2128 more obscure bugs as well
2129 * some steps toward byte-compiling non-performance-critical
2130 parts of the system, courtesy of patches from Martin Atzmueller
2131 * Christophe Rhodes has made some debian packages of sbcl at
2132 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2133 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2134 stable, but are nonetheless usable. When he's ready, I'd be happy
2135 to add them to the SourceForge "File Releases" section. (And if
2136 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2137 * new fasl file format version number (because of changes in
2138 internal representation of (OR ..) types to accommodate the new
2139 support for (AND ..) types, among other things)
2141 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2142 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2144 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2146 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2147 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2148 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2149 * DESCRIBE now works on structure objects again.
2150 * Most function call argument type mismatches are now handled as
2151 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2152 know whether the function will be redefined before the call is
2153 executed. (The compiler could flag local calls with full WARNINGs,
2154 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2155 it doesn't keep track of enough information to know whether calls
2156 are local in this sense.)
2157 * Compiler output is now more verbose, with messages truncated
2158 later than before. (There should be some supported way for users
2159 to override the default verbosity, but I haven't decided how to
2160 provide it yet, so this behavior is still controlled by the internal
2161 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2162 src/compiler/ir1util.lisp.)
2163 * Fasl file format version numbers have increased again, because
2164 support for the Gray streams extension changes the layout of the
2165 system's STREAM objects.
2166 * The Gray subclassable streams extension now works, thanks to a
2167 patch from Martin Atzmueller.
2168 * The full LOAD-FOREIGN extension (not just the primitive
2169 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2170 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2171 but like most other programs, it defaults to copying the Unix
2172 environment from the original process instead of starting the
2173 new process in an empty environment.
2174 * Extensions which manipulate the Unix environment now support
2175 an :ENVIRONMENT keyword option which doesn't smash case or
2176 do other bad things. The CMU-CL-style :ENV option is retained
2177 for porting convenience.
2178 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2179 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2181 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2183 * A patch from Martin Atzmueller seems to have solved the SIGINT
2184 problem, and as far as we know, signal-handling now works cleanly.
2185 (If you find any new bugs, please report them!)
2186 * The system no longer defaults Lisp source file names to types
2187 ".l", ".cl", or ".lsp", but only to ".lisp".
2188 * The compiler no longer uses special default file extensions for
2189 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2190 seems to expect a single default extension for all compiled code,
2191 and there's no compelling reason to try to stretch the standard
2192 to allow two different extensions.) Instead, byte-compiled files
2193 default to the same extension as native-compiled files.
2194 * Fasl file format version numbers have increased again, because
2195 a rearrangement of internal implementation packages made some
2196 dumped symbols in old fasl files unreadable in new cores.
2197 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2198 many fewer weird special cases.
2199 * Bug #17 (differing COMPILE-FILE behavior between logical and
2200 physical pathnames) has been fixed, and some related misbehavior too,
2201 thanks to a patch from Martin Atzmueller.
2202 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2203 by Tim Moore, ported to SBCL by Martin Atzmueller.
2204 * Martin Atzmueller fixed several filesystem-related problems,
2205 including bug #36, in part by porting CMU CL patches, which were
2206 written in part by Paul Werkowski.
2207 * More compiler warnings in src/runtime/ are gone, thanks to
2208 more patches from Martin Atzmueller.
2209 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2212 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2214 * DESCRIBE now works on CONDITION objects.
2215 * The debugger now handles errors which arise when trying to print
2216 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2218 * The build system now uses an additional file, customize-target-features.lisp,
2219 to allow local modifications to the target *FEATURES* list. (The point of
2220 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2221 debugging features enabled, without having to worry about propagating it
2222 into everyone's system when I do a "cvs update".) When no
2223 customize-target-features.lisp file exists, the target *FEATURES* list
2224 should be constructed the same way as before.
2225 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2226 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2227 warn about behavior which is undefined under the ANSI spec. Note: This
2228 is specified by ANSI, but it's not very popular with programmers.
2229 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2230 macro in the SBCL sources for an example of a workaround which you
2231 might use to make portable ANSI-standard code which does what you want.
2232 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2233 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2234 when it's used as a non-toplevel form. (This is required in order
2235 to implement the DEFCONSTANT-EQX macro.)
2236 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2237 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2238 funny things, due to bugs in the implementation of EVAL-WHEN
2239 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2240 non-toplevel DEFCONSTANTs are uncommon.)
2241 * The core file version number and fasl file version number have been
2242 incremented, because the old noncompliant DEFCONSTANT behavior involved
2243 calling functions which no longer exist, and because I also took the
2244 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2245 * fixed bug 1 (error handling before read-eval-print loop starts), and
2246 redid debugger restarts and related debugger commands somewhat while
2248 ** The QUIT debugger command is gone, since it did something
2249 rather different than the SB-EXT:QUIT command, and since it never
2250 worked properly outside the main toplevel read/eval/print loop.
2251 Invoking the new TOPLEVEL restart provides the same functionality.
2252 ** The GO debugger command is also gone, since you can just invoke
2253 the CONTINUE restart directly instead.
2254 ** The TOP debugger command is also gone, since it's redundant with the
2255 FRAME 0 command, and since it interfered with abbreviations for the
2257 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2259 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2260 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2261 some porting work by Martin Atzmueller)
2262 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2263 with a boatload of patches which he ported to SBCL. Now that those
2265 ** The system tries to make sure that its low-priority messages
2266 are prefixed by semicolons, to help people who like to use
2267 syntax highlighting in their ILISP buffer. (This patch
2268 was originally due to Raymond Toy.)
2269 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2270 patches originally written by Raymond Toy.
2271 ** The compiler understands coercion between single-value and
2272 multiple-VALUES type expressions better, getting rid of some very
2273 weird behavior, thanks to patches originally by Robert MacLachlan
2274 and Douglas Crosher.
2275 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2276 lambda lists, thanks to a patch originally by Pierre Mai.
2277 ** The system no longer bogusly warns about "abbreviated type
2279 ** The compiler gets less confused by inlining and RETURN-FROM,
2280 thanks to some patches originally by Tim Moore.
2281 ** The system no longer hangs when dumping circular lists to fasl
2282 files, thanks to a patch originally from Douglas Crosher.
2283 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2284 undefined function error.
2285 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2286 in structure slot initforms) for now, documented workaround instead:-|
2287 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2288 * fixed bug 5: added stubs for various Gray stream functions called
2289 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2290 installed, at least appropriate type errors are generated
2291 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2292 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2293 consistently in DEFMETHOD forms.
2294 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2295 it doesn't seem to affect SBCL after all
2296 * The C runtime system now builds with better optimization and many
2297 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2299 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2301 * The system is now under CVS at SourceForge (instead of the
2302 CVS repository on my home machine).
2303 * The new signal handling code has been tweaked to treat register
2304 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2305 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2306 downstream, e.g. in debug-int.lisp, has implicit dependencies
2307 on the unsignedness of integer representation of machine words,
2308 and that caused the system to bomb out with infinite regress
2309 when trying to recover from type errors involving signed values,
2310 e.g. (BUTLAST '(1 2 3) -1).)
2311 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2312 (This was one of the bugs Peter Van Eynde reported back in July.)
2313 * The system now uses code inspired by Colin Walters' O(N)
2314 implementation of MAP (from the cmucl-imp@cons.org mailing
2315 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2316 inline the MAP operation, and there is more than one
2317 sequence argument to the MAP call (so that it can't just
2318 do ETYPECASE once and for all based on the type of the
2319 single sequence argument). (The old non-inline implementation
2320 of the general M-argument sequence-of-length-N case required
2321 O(M*N*N) time when any of the sequence arguments were LISTs.)
2322 * The QUIT :UNIX-CODE keyword argument has been renamed to
2323 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2324 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2326 * Raymond Wiker's patches to port dynamic loading from Linux to
2327 FreeBSD have been added.
2328 * The BUGS file is now more nearly up to date, thanks in large part
2329 to Martin Atzmueller's review of it.
2330 * The debugger now flushes standard output streams before it begins
2331 its output ("debugger invoked" and so forth).
2332 * The core version number and fasl file version number have both
2333 been incremented, because of incompatible changes in the layout
2335 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2336 used to not be called for a saved Lisp image.)
2337 * Martin Atzmueller's version of a patch to fix a compiler crash,
2338 as posted on sbcl-devel 13 September 2000, has been installed.
2339 * Instead of installing Martin Atzmueller's patch for the
2340 compiler transform for SUBSEQ, I deleted the compiler transform,
2341 and transforms for some similar consing operations.
2342 * A bug in signal handling which kept TRACE from working on OpenBSD
2344 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2345 away by constant folding
2346 * The system now defines its address space constants in one place
2347 (in the Lisp sources), and propagates them automatically elsewhere
2348 (through GENESIS and the sbcl.h file). Therefore, patching the
2349 address map is less unnecessarily tedious and error-prone. The
2350 Lisp names of address space constants have also been systematized.
2351 * CVS tags like dollar-Header-dollar have been removed from
2352 the sources, because they have never saved me trouble and
2353 they've been source of trouble working with patches and other
2354 diff-related operations.
2355 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2356 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2358 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2360 * The system has been ported to OpenBSD.
2361 * The system now compiles with a simple "sh make.sh" on the systems
2362 that it's supported on. I.e., now you no longer need to tweak
2363 text in the target-features.lisp-expr and symlinks in src/runtime/
2364 by hand, the make.sh takes care of it for you.
2365 * The system is no longer so grossly inefficient when compiling code
2366 involving vectors implemented as general (not simple) vectors (VECTOR T),
2367 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2368 which dares to use the various sequence functions on non-simple
2369 vectors, takes less of a performance hit.
2370 * There is now a primitive type predicate VECTOR-T-P
2371 to test for the (VECTOR T) type, so that e.g.
2372 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2373 can now be compiled with some semblance of efficiency. (The old code
2374 turned the type declaration into a full call to %TYPEP at runtime!)
2375 * AREF on (VECTOR T) is still not fast, since it's still compiled
2376 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2377 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2378 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2379 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2380 without making full calls to SUBTYPEP at runtime.
2381 (Some analogous efficiency issues for non-simple vectors specialized to
2382 element types other than T, or for non-simple multidimensional arrays,
2383 have not been addressed. They could almost certainly be handled the
2384 same way if anyone is motivated to do so.)
2385 * The changes in array handling break binary compatibility, so
2386 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2387 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2388 instead of (VALUES T T).
2389 * By following the instructions that Dan Barlow posted to sbcl-devel
2390 on 2 July 2000, I was able to enable primitive dynamic object
2391 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2392 functionality is not implemented (since it calls ld to resolve
2393 library references automatically, requiring RUN-PROGRAM for its
2394 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2395 try to resolve library references) is now supported.
2396 * The system now flushes the standard output streams when it terminates,
2397 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2398 them at several other probably-convenient times, e.g. in each pass of
2399 the toplevel read-eval-print loop, and after evaluating a form given
2400 as an "--eval" command-line option. (These changes were motivated by a
2401 discussion of stream flushing issues on cmucl-imp in August 2000.)
2402 * The source transform for TYPEP of array types no longer assumes
2403 that an array whose element type is a not-yet-defined type
2404 is implemented as an array of T, but instead punts, so that the
2405 type will be interpreted at runtime.
2406 * There is now some support for cross-compiling in make.sh: each of
2407 the phases of make.sh has its own script. (This should be transparent
2408 to people doing ordinary, non-cross-compile builds.)
2409 * Since my laptop doesn't have hundreds of megabytes of memory like
2410 my desktop machine, I became more motivated to do some items on
2411 my to-do list in order to reduce the size of the system a little:
2412 ** Arrange for various needed-only-at-cold-init things to be
2413 uninterned after cold init. To support this, those things have
2414 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2415 symbols with such names are now uninterned after cold init).
2416 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2417 fasl files for cold load.
2418 ** Remove the old compiler structure pooling code (which used to
2419 be conditional on the target feature :SB-ALLOC) completely.
2420 ** Redo the representation of some data in cold init to be more compact.
2421 (I also looked into supporting byte compiled code at bootstrap time,
2422 which would probably reduce the size of the system a lot, but that
2423 looked too complicated, so I punted for now.)
2424 * The maximum signal nesting depth in the src/runtime/ support code has
2425 been reduced from 4096 to 256. (I don't know any reason for the very
2426 large old value. If the new smaller value turns out to break something,
2427 I'll probably just bump it back up.)
2428 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2430 * Many, many bugs reported by Peter Van Eynde have been added to
2431 the BUGS list; some have even been fixed.
2432 * While enabling dynamic object file loading, I tried to make the
2433 code easier to understand, renaming various functions and variables
2434 with less ambiguous names, and changing some function calling
2435 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2436 * While trying to figure out how to do the OpenBSD port, I tried to
2437 clean up some of the code in src/runtime/. In particular, I dropped
2438 support for non-POSIX signal handling, added various comments,
2439 tweaked the code to reduce the number of compilation warnings, and
2440 renamed some files to increase consistency.
2441 * To support the new automatic configuration functionality in make.sh,
2442 the source file target-features.lisp-expr has been replaced with the
2443 source file base-target-features.lisp-expr and the machine-generated
2444 file local-target-features.lisp-expr.
2445 * fixed a stupid quoting error in make.sh so that using CMU CL
2446 "lisp -batch" as cross-compilation host works again
2448 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2450 * DESCRIBE no longer tries to call itself recursively to describe
2451 bound/fbound values, so that it no longer fails on symbols which are
2452 bound to themselves (like keywords, T, and NIL).
2453 * DESCRIBE now works on generic functions.
2454 * The printer now prints less-screwed-up representations of closures
2455 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2456 * A private symbol is used instead of the :EMPTY keyword previously
2457 used to mark empty slots in hash tables. Thus
2458 (DEFVAR *HT* (MAKE-HASH-TABLE))
2459 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2460 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2461 now does what ANSI says that it should. (You can still get
2462 similar noncompliant behavior if bang on the hash table
2463 implementation with all the symbols you get back from
2464 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2465 This breaks binary compatibility, since tests for equality to
2466 :EMPTY are wired into things like the macroexpansion of
2467 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2469 * There's now a minimal placeholder implementation for CL:STEP,
2470 as required by ANSI.
2471 * An obscure bug in the interaction of the normal compiler, the byte
2472 compiler, inlining, and structure predicates has been patched
2473 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2474 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2476 * Missing ordinary arguments in a macro call are now detected even
2477 when the macro lambda list contains &KEY or &REST.
2478 * The debugger no longer complains about encountering the top of the
2479 stack when you type "FRAME 0" to explicitly instruct it to go to
2480 the top of the stack. And it now prints the frame you request even
2481 if it's the current frame (instead of saying "You are here.").
2482 * As specified by ANSI, the system now always prints keywords
2483 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2485 * The default initial SIZE of HASH-TABLEs is now smaller.
2486 * Type information from CLOS class dispatch is now propagated
2487 into DEFMETHOD bodies, so that e.g.
2488 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2490 is now basically equivalent to
2491 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2492 (DECLARE (TYPE SINGLE-FLOAT X))
2494 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2495 operation, without having to do run-time type dispatch.
2496 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2497 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2498 * Since I don't seem to be making big file reorganizations very often
2499 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2500 is overflowing my ability to conveniently back them up), I've finally
2501 checked the system into CVS. (The CVS repository is on my home system,
2502 not at SourceForge -- putting it on SourceForge might come later.)
2503 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2504 high-level GC-NOTIFY-FOO functions send their output. (There's
2505 still very little control of where low-level verbose GC functions
2506 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2507 less than it used to -- the GC-NOTIFY-FOO functions are now under
2508 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2509 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2510 in only one place in the source code, and propagates it automatically
2511 everywhere that it's needed. Thus e.g. when I bump the version from
2512 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2513 * The C source files now include boilerplate legalese and documentation
2514 at the head of each file (just as the Lisp source files already did).
2515 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2516 to his page will be replaced with a link to his new CLiki service.
2518 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2520 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2521 * The build process now looks for GNU make under the default name "gmake",
2522 instead of "make" as it used to. If GNU make is not available as "gmake"
2523 on your system, you can change this default behavior by setting the
2524 GNUMAKE environment variable.
2525 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2526 can build without error under CMU CL.
2528 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2530 * There is now a partial SBCL user manual (with some new text and some
2531 text cribbed from the CMU CL manual).
2532 * The beginnings of a profiler have been added (starting with the
2533 CMU CL profiler and simplifying and cleaning up). Eventually the
2534 main interface should be through the TRACE macro, but for now,
2535 it's still accessed through vaguely CMU-CL-style functions and macros
2536 exported from the package SB-PROFILE.
2537 * Some problems left over from porting CMU CL to the new
2538 cross-compilation bootstrap process have been cleaned up:
2539 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2540 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2541 part when getting rid of the compiler *BACKEND* stuff.)
2542 ** The value of *NULL-TYPE* was screwed up, because it was
2543 being initialized before the type system knew the final
2544 definition of the 'NULL type. This screwed up several key
2545 optimizations in the compiler, causing inefficiency in all sorts
2546 of places. (I found it because I wanted to understand why
2547 GET-INTERNAL-RUN-TIME was consing.)
2548 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2549 PROCLAIM FTYPE information. Unfortunately this broke binary
2550 compatibility again, since now the forms output by DEFGENERIC
2551 to refer to functions which didn't exist in 0.6.3.
2552 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2553 can use the new (as of 0.6.3) transform for SOME into MAP into
2555 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2556 values to (INTEGER 0 1000000), so that the time code will no longer
2557 occasionally get blown up by Linux returning 1000000 microseconds
2558 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2559 its output conform to the ANSI spec. (Alas, this makes its output
2560 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2561 * A full call to MAP NIL with a single sequence argument no longer conses.
2562 * fixes to problems pointed out by Martin Atzmueller:
2563 * The manual page no longer talks about multiprocessing as though
2564 it were currently supported.
2565 * The ILISP support patches have been removed from the distribution,
2566 because as of version 5.10.1, ILISP now supports SBCL without us
2567 having to maintain patches.
2568 * added a modified version of Raymond Toy's recent CMU CL patch for
2569 EQUALP comparison of HASH-TABLE
2571 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2573 * The system still can't cross-compile itself with
2574 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2575 entails), but at least it can compile more of itself that way
2576 than it used to be able to, and various buglets which were uncovered
2577 by trying to cross-compile itself that way have now been fixed.
2578 * This release breaks binary compatibility again. This time
2579 at least I've incremented the FASL file format version to 2, so that the
2580 problem can be detected reliably instead of just causing weird errors.
2581 * various new style warnings:
2582 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2583 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2584 ** using the lexical binding of a variable named in the *FOO* style
2585 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2586 as specified by ANSI.
2587 * *RANDOM-STATE* is no longer automatically initialized from
2588 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2589 default behavior of the system is to repeat its behavior every time
2590 it's run. If you'd like to change this behavior, you can always
2591 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2592 old convention there was no comparably easy way to get the system to
2593 repeat its behavior every time it was run.
2594 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2595 FTYPE declarations has been removed, in favor of their ANSI
2596 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2597 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2598 the types of their sequence arguments can be determined at compile time.
2599 This is done through a new open code expansion for MAP which eliminates
2600 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2601 argument types can be determined at compile time.
2602 * The optimizer now transforms COERCE into an identity operation when it
2603 can prove that the coerced object is already of the correct type. (This
2604 can be a win for machine generated code, including the output of other
2605 optimization transforms, such as the MAP transform above.)
2606 * Credit information has been moved from source file headers into CREDITS.
2607 * Source file headers have been made more standard.
2608 * The CASE macro now compiles without complaining even when it has
2611 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2613 * (Note that the way that the PCL macroexpansions were rewritten
2614 to accommodate the change in DEFGENERIC below breaks binary
2615 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2616 not run under sbcl-0.6.2. Once we get out of alpha releases,
2617 i.e. hit release 1.0.0, we'll probably try to maintain binary
2618 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2619 and sbcl-1.4.4. Until then, however, it might be fairly common
2620 for maintenance releases to break binary compatibility.)
2621 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2623 * The reporting of unhandled signals has been changed to print some
2624 explanatory text as well as the report form. (Previously only
2625 the report form was printed.)
2626 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2627 it defines, so that the compiler no longer issues undefined function
2628 warnings for compiled-but-not-yet-loaded generic functions.
2629 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2630 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2631 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2632 section of the ANSI spec are supported.
2633 * The "" nickname for the "KEYWORD" package has been removed.
2634 The reader still handles symbol tokens which begin with a package marker
2635 as keywords, but it doesn't expose its mechanism for doing so in the
2636 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2637 * The system now issues STYLE-WARNINGs for contradictory TYPE
2638 proclamations. (Warnings for contradictory FTYPE proclamations would
2639 be nice too, but those can't be done usefully unless the type system
2640 is made smarter about FUNCTION types.)
2641 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2642 systematized, so that "*target-*.lisp is supposed to exist only on the
2643 target and imply that there's a related file which exists on the
2644 host, and *host-*.lisp is supposed to exist only on the host and imply
2645 that there's a related file which exists on the target. This involves a
2646 lot of renaming. Hopefully the acute confusion caused by the renaming
2647 will be justified by the reduction in chronic confusion..
2648 ** runtime-type.lisp -> early-target-type.lisp
2649 ** target-type.lisp -> late-target-type.lisp
2650 ** early-host-format.lisp -> early-format.lisp
2651 ** late-host-format.lisp -> late-format.lisp
2652 ** host-error.lisp -> misc-error.lisp
2653 ** early-error.lisp -> early-target-error.lisp
2654 ** late-error.lisp -> late-target-error.lisp
2655 ** host-defboot.lisp -> early-defboot.lisp
2656 ** code/misc.lisp -> code/target-misc.lisp
2657 ** code/host-misc.lisp -> code/misc.lisp
2658 ** code/numbers.lisp -> code/target-numbers.lisp
2659 ** code/early-numbers.lisp -> numbers.lisp
2660 ** early-host-type.lisp -> early-type.lisp
2661 ** late-host-type.lisp -> late-type.lisp
2662 ** host-typep.lisp -> typep.lisp
2663 ** load.lisp -> target-load.lisp
2664 ** host-load.lisp -> load.lisp
2665 ** host-disassem.lisp -> disassem.lisp
2666 ** host-insts.lisp -> insts.lisp
2667 ** byte-comp.lisp -> target-byte-comp.lisp
2668 ** host-byte-comp.lisp -> byte-comp.lisp
2669 ** host-signal.lisp -> signal.lisp
2670 ** host-defstruct.lisp -> defstruct.lisp
2671 ** late-target-type.lisp -> deftypes-for-target.lisp
2672 Furthermore, several other previously target-only files foo.lisp (e.g.
2673 hash-table.lisp and random.lisp) have been split into a target-and-host
2674 foo.lisp file and a target-only target-foo.lisp file, with their key type
2675 definitions in the target-and-host part, so that the cross-compiler will
2676 know more about target types.
2677 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2678 gone away. In their place are various *BACKEND-FOO* variables
2679 corresponding to the slots of the old structure.
2680 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2681 into the target SBCL has been fixed.
2682 * The chill.lisp system for loading cold code into a running SBCL
2684 * Support for the CMU CL "scavenger hook" extension has been removed.
2685 (It was undocumented and unused in the CMU CL sources that SBCL was
2686 derived from, and stale in sbcl-0.6.1.)
2687 * Various errors in the cross-compiler type system were detected
2688 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2689 (enabling various consistency checks). Many of them were fixed,
2690 but some hard problems remain, so the compiler is back to
2691 running without *TYPE-SYSTEM-INITIALIZED* for now.
2692 * As part of the cross-compiler type system cleanup, I implemented
2693 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2694 * I have started adding UNCROSS calls throughout the type system
2695 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2696 on cross-compiler input files will be able to go away, and various
2698 * CONSTANTP now returns true for quoted forms (as explicitly required
2701 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2703 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2704 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2705 (MAKE-LIST -1). In the longer term, it ought to become true
2706 that declarations are assertions even at SAFETY 1. For now, it's not
2707 quite true even at SAFETY 3, but it's at least more nearly true..
2708 (Note that this change seems to increases the size of the system by
2709 O(5%) and to decrease the speed of the compiler by 20% or more.)
2710 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2711 for the problem of printing dozens of lines of distracting information
2712 about low-level system machinery as part of the top stack frame
2713 on entry to the debugger when an undefined function was called.
2714 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2715 is not reset to COMMON-LISP-USER.
2716 * Compilation of stuff related to dyncount.lisp has been made conditional
2717 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2718 smaller. The various dyncount-related symbols have been moved into
2719 a new "SB-DYNCOUNT" package.
2720 * tty-inspect.lisp has been renamed to inspect.lisp.
2721 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2722 feature has gone away. (When we eventually port to other flavors of
2723 libc and/or Unix, we'll try to make the differences between flavors
2724 invisible at the user level.)
2725 * Various other *FEATURES* tags, and/or their associated conditionals,
2726 have been removed if obsolescent, or given better documentation, or
2727 sometimes given more-mnemonic names.
2729 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2731 * tidied up "make.sh" script
2732 * tidied up system directory structure
2733 * better "clean.sh" behavior
2734 * added doc/FOR-CMUCL-DEVELOPERS
2735 * many many small tweaks to output format, e.g. removing possibly-confusing
2736 trailing #\. character in DESCRIBE-INSTANCE
2737 * (EQUALP #\A 'A) no longer signals an error.
2738 * new hashing code, including EQUALP hashing
2739 * tidied up Lisp initialization and toplevel
2740 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2741 * command line argument processing
2742 * added POSIX-GETENV function to deal with Unix-ish environment variables
2743 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2744 terminating SBCL on EOF
2745 * non-verbose GC by default
2746 * There is no more "sbcl" shell script; the sbcl file is now the C
2747 runtime executable (just like CMU CL).
2748 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2750 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2751 symbol and package information as Lisp native symbol and package objects
2752 instead of strings naming symbols and strings naming packages. This way,
2753 whenever packages are renamed (as in warm init), debug information is
2754 transformed along with everything else.
2755 * tweaked the optimization policy declarations which control the building
2756 of SBCL itself. Now, among other things, the system no longer saves
2757 source location debugging information. (This helps two problems at once
2758 by reducing SBCL size and by keeping SBCL from trying to look for its
2759 sources -- which may not exist -- when reporting errors.)
2760 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2761 debugging and testing purposes
2762 * cleaned up printing, making the printer call PRINT-OBJECT for
2763 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2764 methods, giving nearly-ANSI behavior
2765 * converted almost all special variables to use *FOO* naming convention
2766 * deleted PARSE-TIME functionality, since it can be done portably
2767 * moved some files out of cold init into warm init
2768 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2770 * regularized formatting of source files
2771 * added an install.sh script
2772 * fixed ridiculous memory usage of cross-compiler by making
2773 compiler/alloc.lisp not try to do pooling unless it can hook
2774 itself into the GC of the cross-compilation host. Now the system
2775 builds nicely on my old laptop.
2776 * added :SB-ALLOC in target-features.lisp-expr
2777 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2778 was not implemented)
2779 * re-did condition handling and note reporting in the compiler. Notes
2780 are no longer handled by signalling conditions. Style warnings
2781 and warnings are handled more correctly and reported in such a way
2782 that it's easy to find one or the other in your output (so that you
2783 can e.g. figure out which of many problems caused COMPILE-FILE to
2785 * changed the severity of several compiler warnings from full WARNING
2786 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2787 compiler note reporting so that it doesn't use the condition system
2788 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2789 COMPILE-FILE command)
2790 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2791 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2792 no longer supported, since their implementation was deeply intertwingled
2793 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2794 * removed bogus "support" for compiler macros named (SETF FOO), and
2795 removed the compiler macro for SETF INFO (but only after making a fool
2796 of myself on the cmucl-imp mailing list by posting a bogus patch for
2797 DEFINE-COMPILER-MACRO..)
2798 * Compiled files containing forms which have side effects on the Lisp
2799 reader (such as DEFPACKAGE forms) are now handled more correctly.
2800 (Compiler queuing of top level lambdas has been suppressed by setting
2801 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2802 * deleted various currently-unused source files, e.g. gengc.lisp. They
2803 may be added back at some point e.g. when porting to other architectures,
2804 but until they are it's distracting to distribute them and to try to
2806 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2807 there were so many of them they're just distractions, and UNCROSS is
2808 known to be able to handle the current sources
2809 * moved PROFILE functionality into TRACE, so that it will be clear
2810 how the wrapping and unwrapping of functions when you profile them
2811 interacts with the wrapping and unwrapping of functions when you
2812 trace them. (Actually, the functionality isn't there yet, but at least
2813 the interface specification is there. Hopefully, the functionality will
2814 arrive with some future maintenance release.)
2815 * removed host-oops.lisp
2816 * changed signature of QUIT function to allow UNIX-CODE argument
2817 * fixed READ-SEQUENCE bug
2818 * tweaked verbose GC output so that it looks more like the progress
2819 output that ANSI specifies for functions like LOAD
2820 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2821 * added <http://sbcl.sourceforge.com> to the banner information printed by