1 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2 * new feature: reloading changed shared object files with
3 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
4 * new feature: references to foreign variables and functions
5 can now be compiled and loaded before the corresponding shared
6 object file is loaded, as long as the foreign definitions are
8 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
9 supported. (thanks to Dan Debertin)
10 * fixed bug #331: structure-class instances corresponding to
11 DEFSTRUCT forms are now created eagerly.
12 * fixed bug #345: backtraces from calls to undefined functions work
13 on x86 as well. Related bug #61 is now also partially fixed on x86
14 (backtraces from throws to unknown catch tags.)
15 * bug fix: lambda-list parsing is now stricter vrt. order and number
16 of lambda-list keywords.
17 * bug fix: as specified by AMOP, an error is signalled if a
18 class-option appears multiple times in a DEFCLASS form. (reported
20 * bug fix: RANDOM can be compiled when the compiler derives the type
21 of its numeric argument as a disjoint set of small integers.
22 (reported by Paul Dietz)
23 * bug fix: starting a core saved with shared objects loaded when
24 those objects are not available no longer causes threaded SBCL to
25 hang. (reported by Sean Ross)
26 * fixed some bugs related to Unicode integration:
27 ** RUN-PROGRAM can allow its child to take input from a Lisp
28 stream. (reported by Stefan Scholl)
29 ** ASDF-INSTALL successfully downloads .tar.gz files in all
30 locales. (reported by Ken Causey)
31 * fixed some bugs revealed by Paul Dietz' test suite:
32 ** INCF, DECF and REMF evaluate their place form as specified in
34 ** FORMATTER expands ~{ ~} iteration directives with V or #
36 ** FORMATTER deals with the ~@[ ~] conditional directive where the
37 consequent uses no arguments correctly.
38 ** the system has a partial understanding of the (COMPLEX RATIO)
40 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
43 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
44 * new feature: a build-time option (controlled by the :SB-UNICODE
45 keyword feature, enabled by default) for building the system with
46 support for the entire 21-bit character space defined by the
48 * new feature: the system now has rudimentary external-format
49 support; the primary user-visible change at this time is that
50 characters with the high bit set (such as lower-case-e-acute) will
51 print correctly to a terminal in a UTF-8 environment.
52 * minor incompatible change: BASE-CHAR no longer names a class;
53 however, CHARACTER continues to do so, as required by ANSI.
54 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
55 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
56 has been moved to the SB-EXT package (temporarily re-exported from
58 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
59 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
61 * fixed bug #351: better error-handlind and reporting for malformed
63 * fixed bug #350: bignum-printing is now more memory-efficient,
64 allowing printing of very large bignums, eg. (expt 2 10000000).
65 (reported by Bruno Haible)
66 * fixed bug #302: better primitive-type selection for intersection
68 * fixed bug #308: non-graphic characters now all have names, as
69 required. (reported by Bruno Haible)
70 * bug fix: redefining a class with different superclasses now correctly
71 removes it from the direct-subclasses of its previous superclasses.
72 (reported by David Morse)
73 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
74 new value now works. (reported by Bruno Haible)
75 * bug fix: correct canonicalization of multiple non-standard slot
76 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
77 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
78 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
79 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
80 now exists, an signals an error.
81 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
82 during macroexpansion for non-iteration variables as well. (reported
83 by Bruno Haible for CMUCL)
84 * bug fix: Cyclic structures and unprintable objects in compiler
85 messages no longer cause errors. (reported by Bruno Haible)
86 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
87 and READ-FROM-STRING all now return a primary value of NIL if
88 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
89 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
90 (reported by Bruno Haible for CMUCL)
91 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
92 do not designate a condition. (reported by Bruno Haible for
94 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
95 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
96 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
97 permitting resolution in favour of any of the conflicting symbols.
98 (reported by Bruno Haible for CMUCL)
99 * bug fix: EQUAL compiler optimizations is less aggressive on
100 strings which can potentially compare true despite having distinct
101 specialized array element types.
102 * bug fix: unit enumerations can be defined without dividing by
103 zero. (thanks to Vincent Arkesteijn)
104 * FORMAT compile-time argument count checking has been enhanced.
105 (report from Bruno Haible for CMUCL)
106 * a partial workaround for the bug 262: the compiler does not try to
107 inline-expand a local function doing RETURN-FROM from a deleted
108 BLOCK. (thanks to Peter Denno for the bug report and to David
109 Wragg for the simple test case)
110 * fixed some bugs revealed by Paul Dietz' test suite:
111 ** RENAME-PACKAGE allows all package designators as new package
113 ** constraint propagation and lambda variable substitution are
114 more cautious in dealing with partially deleted code.
115 ** compiler sometimes lost reoptimization passes.
116 ** CERROR, given a condition as condition designator, passes its
117 remaining arguments to the continue format control without
119 ** Case-altering FORMAT directives work correctly on non-ASCII
121 ** The REAL type specifier handles bounds outside the floating
122 point ranges without signalling FLOATING-POINT-OVERFLOW.
123 ** Functions with IR1-transformations can create intercomponent
124 references to global functions.
125 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
127 ** FORMAT ~R treats a nil value for its first parameter correctly.
129 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
130 * enhancement: saving cores with foreign code loaded is now
131 supported on x86/NetBSD and sparc/Linux in addition to the previously
133 * bug fix: on some platforms repeated installations caused multiple
134 copies of HTML documentation to be installed -- should not happen
135 any more. (reported by Stefan Scholl)
136 * bug fix: parsing self-recursive alien record types multiple times
137 no longer causes infinite recursion. (reported by Thomas F. Burdick,
138 original patch by Helmut Eller for CMUCL)
139 * bug fix: stack-exhaustion detection works now on NetBSD as well.
140 (thanks to Richard Kreuter)
141 * bug fix: defining classes whose accessors are methods on existing
142 generic functions in other (locked) packages no longer signals
143 bogus package lock violations. (reported by François-René Rideau)
144 * bug fix: special variables as DEFMETHOD parameters no longer have
145 associated bogus type declarations. (reported by David Wragg and
147 * bug fix: read-write consistency on streams of element-type
148 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
149 * bug fix: redefiniton of the only method of a generic function with
150 no DEFGENERIC no longer emits a full WARNING. In addition,
151 redefinition of generic functions with no DEFGENERIC to an
152 incompatible lambda list now signals an error. (thanks to Zach
154 * bug fix: DEFGENERIC now works even when there's a function of the
155 same name in an enclosing lexical environment. (thanks to Zach
157 * fixed compiler failure, caused by instrumenting code during
158 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
159 * optimization: added loop analysis and improved register allocation
160 * fixed some bugs revealed by Paul Dietz' test suite:
161 ** POSITION on displaced vectors with non-zero displacement
162 returns the right answer.
163 ** (SIMPLE-STRING) is a valid type specifier for sequence
165 ** *PRINT-LEVEL* handling for slotless structures is pedantically
167 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
168 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
170 ** The justification version of the FORMAT ~< directive treats
171 non-zero minpad parameter correctly.
173 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
174 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
175 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
176 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
177 the supported interface.
178 * new feature: Single-stepping of code compiled with DEBUG 2 or
179 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
180 * new feature: saving cores with foreign code loaded is now
181 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
182 Timothy Moore's work for CMUCL)
183 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
184 parameters to * instead of NIL if no initform is supplied.
185 (reported by Johan Bockgaard)
186 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
187 correctly when FUNCALL appears as the car of the form. Note:
188 despite this FUNCALL forms are not currently subject to
189 compiler-macro expansion. (port of Raymond Toy's fix for the
190 same from CMUCL, reported by Johan Bockgaard)
191 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
192 (thanks for Teemu Kalvas)
193 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
194 case instead of terminating the iteration (thanks for Julian
195 Squires, Sean Champ and Raymond Toy)
196 * bug fix: incorrect expansion of defgeneric that caused a style
197 warning. (thanks for Zach Beane)
198 * bug fix: slot accessor effective method computation works properly
199 for classes with multiple non-standard applicable methods on
200 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
201 * on x86 compiler supports stack allocation of results of LIST and
202 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
203 CMUCL implementation by Gerd Moellmann)
204 * fixed some bugs revealed by Paul Dietz' test suite:
205 ** FORMAT strings with both the ~<~:;~> form of the justification
206 directive and pretty-printing directives cause an error.
208 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
209 * incompatible change: the internal functions
210 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
211 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
212 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
213 instead of the old functions.
214 * new feature: on platforms where "dladdr" is available foreign
215 function names now appear in backtraces. (based on Helmut Eller's
217 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
218 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
220 * documentation: ASDF manual is now distributed with SBCL.
221 * bug fix: SBCL can now load its contributed modules with REQUIRE
222 even if the system-provided entries have been removed from
223 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
225 * bug fix: backtraces involving undefined functions or assembly
226 routines are more informative. (thanks to Brian Downing)
227 * bug fix: mutually referent alien structures now work correctly.
228 (reported by Rick Taube)
229 * bug fix: structures defined by WITH-ALIEN can be referred to
230 within other definitions in the same WITH-ALIEN.
231 * bug fix: division operators (MOD, TRUNCATE and the like) with
232 constant zero divisors and integer dividends no longer generate
234 * bug fix: provide default methods for INPUT-STREAM-P and
235 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
236 * bug fix: improve the detection and resolution of MOP metacycles
237 (where effective method computation depends on the generic function
238 having its effective method computed). (reported by Bruno Haible)
239 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
240 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
241 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
242 * bug fix: pathnames with relative directory components can be
243 represented relative to default pathnames.
244 * optimization: in taking the GCD of bignums, reduce the two bignums
245 to approximately the same size (using Euclid's algorithm) before
246 applying the more sophisticated binary GCD. (thanks to Juho
248 * optimization: COUNT on bitvectors now operates word-at-a-time.
249 * optimization: ASH with a positive, but not necessarily constant,
250 (leftwards) shift, when in a modular context, is compiled to a
252 * fixed some bugs revealed by Paul Dietz' test suite:
253 ** FORMAT variable parameters ("~V<char>") are defaulted properly
254 if the corresponding argument is NIL.
255 ** FORMAT directives accepting a minpad parameter treat negative
257 ** literal commas as character parameters to FORMAT directives are
259 ** literal spaces directly after ~<Newline> directives within a
260 format-logical-block (~:< ~@:>) do not induce :FILL-style
261 conditional newlines.
262 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
263 as specified: it no longer includes conditional newlines).
264 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
266 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
267 :PER-LINE-PREFIX argument does not evaluate to a string.
269 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
270 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
271 the manual for details; for now, package locks can be disabled by
272 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
273 if no major problems are found then it is likely that they will be
274 compiled in unconditionally.
275 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
276 unsupported operators on all platforms. To load a shared library
277 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
278 object file, link it into a shared library outside of SBCL and
279 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
281 * bug workaround: ROOM T and the GC were not working together
282 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
283 reporting facility aren't preserved. That reporting has been
284 disabled, so now until and unless someone figures out how to make it
285 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
286 * minor incompatible change: as threatened around sbcl-0.8.0, the
287 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
288 The symbols are also exported from SB-PCL for backwards
289 compatibility, but more so than before SB-PCL should be treated as
290 an implementation-internal package.
291 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
293 * the SB-SPROF contrib now works on (most) non-x86 architectures.
294 It is known as of this release not to work on the Alpha, however.
295 * fixed bug #167: errors signalled due to illegal syntax in method
296 bodies are now more legible.
297 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
298 designators and can hence be used with TYPEP.
299 * fixed bug #333: CHECK-TYPE now ensures that the type error
300 signalled, if any, has the right object to be accessed by
301 TYPE-ERROR-DATUM. (reported by Tony Martinez)
302 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
303 5.1.2.3 for multiple-value place subforms. (reported by Kalle
305 * fixed bug #334: programmatic addition of slots using specialized
306 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
307 and :ALLOCATION :CLASS slots.
308 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
309 float exponents. (rereported by Peter Seibel)
310 * fixed a bug: #\Space (and other whitespace characters) are no
311 longer considered to be macro characters in standard syntax by
313 * fixed bug: initialization of condition class metaobjects no longer
314 causes an instance of the condition to be created. (reported by
316 * fixed bug: it is now possible to have slots such that
317 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
318 description is neither :INSTANCE nor :CLASS.
319 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
320 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
321 SLOT-VALUE-USING-CLASS), and no longer causes errors with
322 non-standard SLOT-DEFINITION-ALLOCATION values.
323 * fixed bugs: various race conditions handling exiting threads.
324 CL-PPCRE's thread torture test now passes.
325 * fixed arguably-a-bug: GC time proportional to number of threads in
326 system even when most of them are idle
327 * optimization: improved performance of BIT and SBIT on bit-vectors.
328 * .fasl file incompatibility: The fasl file version number has
329 been incremented because of changes associated with package locks.
331 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
332 * minor incompatible change: the system no longer provides
333 optimization hints (conditions which are TYPEP
334 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
335 mode; these hints are still emitted when higher SPEED optimization
337 * new contrib module: a sampling profiler (profiling by statistical
338 sampling, rather than by instrumenting functions) is available as
339 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
340 * the behaviour of the standard function ED is now customizeable by
341 third parties through a hook variable: see ED's documentation
342 string for information on the protocol.
343 * the compiler no longer emits efficiency notes for (FUNCALL X)
344 when the type of X is uncertain under default optimization
346 * fixed bug 276: mutating a binding of a specialized parameter to a
347 method to something that is not TYPEP the specializer is now
349 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
350 exhibit the required behaviour on the x86 platform. (thanks to
351 Peter van Eynde, Eric Marsden and Bruno Haible)
352 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
353 even for difficult arguments. (reported by Peter Graves)
354 * fixed bug 141a: the backquote printer now descends quoted
356 * fixed another bug in backquote printing: no more destructive
357 modification of the form's list structure. (reported by Brian
359 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
360 move between its address being taken and the call to
361 interrupt_thread, fixing a crashing race condition.
362 * the SB-POSIX contrib implementation has been adjusted so that it
363 no longer exhibits ridiculously poor performance when constructing
364 instances corresponding to C structs.
366 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
367 * minor incompatible change: the sb-grovel contrib now treats C
368 structures as alien (in the sense of SB-ALIEN) objects rather than
369 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
370 has implications for memory management of client code
371 (sb-grovel-returned objects must now be manually managed) and for
372 type safety (alien objects now have full types).
373 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
374 used to control emission of compiler diagnostics, rather than the
375 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
376 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
377 quality should be considered deprecated.
378 * install.sh now installs the user manual as well
379 * (not quite a new documentable feature, but worth considering in
380 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
381 beginnings of a semantically meaningful condition hierarchy is
382 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
384 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
385 designator as the defaults argument.
386 * fixed bug: Displaced arrays whose displaced-to array has become
387 too small now cause ARRAY-DIMENSION to signal an error, providing
388 for safer bounds-checking. (reported by Bruno Haible)
389 * fixed bug: DEFCLASS slot definitions with identical :READER and
390 :WRITER names now signal a reasonable error. (reported by Thomas
392 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
394 * fixed bug: Invalid slot specification errors now print correctly.
395 (thanks to Zach Beane)
396 * fixed bug 320: Shared to local slot value transfers in class
397 redefinitions now happen corrently from superclasses as
398 well. (reported by Bruno Haible)
399 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
401 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
402 improper lists correctly. (reported by Bruno Haible)
403 * fixed bug 313: source-transform for <fun-name> was erroneously
404 applied to a call of a value of a variable with name <fun-name>.
405 (reported by Antonio Menezes Leitao)
406 * fixed bug 307: The obsolete instance protocol ensures that
407 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
408 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
409 remove dead unknown-values globs from the middle of the stack.
410 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
411 easier to resume long computations after using *BREAK-ON-SIGNALS*
412 to diagnose and fix failures. (thanks to Nikodemus Siivola)
413 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
414 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
415 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
416 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
417 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
419 * fixed a (fairly theoretical) bug in string printing: if
420 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
421 does not have array-element-type equal to the most general string
423 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
424 function name. (thanks to Nikodemus Siivola)
425 * fixed bug: compile-time format string checker failed on
426 non-closed ~{. (reported by Thomas F Burdick)
427 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
428 don't warn on legal loop code involving more than one aggregate
429 boolean. (thanks to Nikodemus Siivola)
430 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
431 in keyword position to MAKE-PACKAGE was not being reported as
433 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
434 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
436 * fixed bugs 280 and 312: the checking for multiple definitions in a
437 file is less likely to become confused by uses of inline
439 * fixed bug: the #S reader macro performs the keyword coercion
440 specified for slot names. (reported by Kalle Niemitalo)
441 * fixed bug: lambda lists may contain symbols whose names start with
442 & but are not lambda-list-keywords; their occurrence triggers a
444 * fixed bug 321: define-method-combination argument lambda lists do
445 not cause invalid code to be generated when &optional or &aux
446 variables have default values. (reported by Bruno Haible)
447 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
448 have CLOS classes; this appears also to have fixed reported
449 instability in stack exhaustion detection.
450 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
451 MUFFLE-WARNING when no associated restart is present is now
453 * optimization: rearranged the expansion of various defining macros
454 so that each expands into only one top-level form in a
455 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
457 * optimization: used a previously little-used slot in symbols to
458 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
460 * fixed some bugs revealed by Paul Dietz' test suite:
461 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
463 ** Fixed an optimization bug related to inheritance of initforms
464 from local to shared slots.
465 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
466 ** CAST optimizer forgot to flush argument derived type.
467 ** print/read consistency on floats is now orders of magnitude
468 more likely. (thanks also to Bruno Haible for a similar report
470 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
471 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
472 for objects of type REAL. Make it so.
473 ** SXHASH is specified (infelicitously) to respect similarity,
474 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
475 it so. (thanks to Markus Ziegler)
476 ** on the Alpha, the compiler succeeds in compiling functions
477 returning a known number of arguments greater than 63.
478 ** fixed handling of invalid NIL arguments in keyword position
480 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
482 ** provided a readably-printable representation for RANDOM-STATE
484 ** ensured that pathnames, where they have a namestring, always
485 print using #P"..." syntax.
487 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
488 * Support for the forthcoming 2.0 version of the NetBSD kernel
489 running on x86 hardware has been added. (thanks to Perry
490 E. Metzger most immediately, and others for their past work)
491 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
492 current release version; SBCL's OpenBSD support had been broken
493 since about the time of OpenBSD's switch to ELF binary format.)
494 * [placeholder for DX summary]
495 ** user code with &REST lists declared dynamic-extent, under high
496 speed or space and low safety and debug optimization policy.
497 * The manual has been converted to Texinfo format and the debugger
498 chapter from the cmucl manual has been added.
499 * A facility has been added to extract documentation strings from
500 sbcl and store them as Texinfo-formatted snippets for inclusion in
501 the manual (via Texinfo's @include directive)
502 * bug fix: compiler emitted division in optimized DEREF. (thanks for
503 the test case to Dave Roberts)
504 * bug fix: multidimensional simple arrays loaded from FASLs had fill
505 pointers. (reported by Sean Ross)
506 * bug fix: PROFILE output is printed nicely even for large numerical
507 values. (thanks to Zach Beane)
508 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
509 greater than 32 handle EOF correctly.
510 * bug fix: on X86 an immediate argument of the IMUL instruction is
511 correctly printed in disassembly. (thanks to Lutz Euler)
512 * bug fix: class slots in redefined classes preserve their old
513 values. (thanks to Bruno Haible and Nikodemus Siivola)
514 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
515 succeeds. (reported by Marco Baringer)
516 * fixed some bugs revealed by Paul Dietz' test suite:
517 ** READ-SEQUENCE now works on ECHO-STREAMs.
518 ** RATIONALIZE works more according to its specification. (thanks
521 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
522 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
523 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
524 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
525 should matter to you only if you rebind the printer control
526 variables and then find you want different bindings in the
527 debugger than in the ordinary execution of your program.)
528 * The runtime build system has been tweaked to support building
529 (on SPARC/SunOS) using a C compiler which invokes Sun's own
530 assembler and linker. (thanks to Nikodemus Siivola)
531 * Unbound, undefined, undeclared variables now trigger full
532 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
533 more often programmer error than deliberate exploitation of undefined
535 * optimization: the hash algorithm for strings has changed to one
536 that is less vulnerable to spurious collisions. (thanks to Juho
538 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
539 needless bounds checking. (thanks to Juho Snellman)
540 * optimization: implemented multiplication as a modular
541 (UNSIGNED-BYTE 32) operation on the PPC backend.
542 * fixed some bugs revealed by Paul Dietz' test suite:
543 ** ADJUST-ARRAY now copies the datum in a zero rank array if
545 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
546 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
547 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
549 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
551 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
553 ** ECHO-STREAMs no longer attempt to echo the end of file value to
554 their output stream on EOF from read.
555 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
556 have been read to end-of-file.
557 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
559 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
560 description of determination of which consecutive characters
562 ** Printing the "Space" character with escaping on now yields "#\\ ",
563 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
564 ** Reading floating-point numbers with *READ-BASE* set to a number
565 less than 10 works correctly.
566 ** Reading floating-point numbers with *READ-BASE* set to a number
567 more than 10 works correctly.
568 ** Printing with *PRINT-READABLY* targets the standard readtable, not
569 the readtable currently in effect.
571 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
572 * minor incompatible change: parsing of namestrings on a physical
573 (Unix) host has changed; numbers after the final #\. in a
574 namestring are no longer interpreted as a version field. This is
575 intented to be largely invisible to the user, except that the
576 meaning of the namestring "*.*.*" has changed: it now refers to a
577 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
578 should usually be replaced by
579 (make-pathname :name :wild :type :wild :version :wild)
580 with the added benefit that this is more likely to be portable.
581 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
582 option to OPEN now signals an error if the file being opened
583 exists; this may have an impact on existing code.
584 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
585 signals received, on the PowerPC platforms (both Linux and
586 Darwin). (thanks to Pierre Mai for pointing out the location of
588 * several fixes on OS X: The system now builds and runs cleanly on
589 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
590 (thanks to Brian Mastenbrook)
591 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
592 second-resolution: integer multiples of 1/3600 between -24 and 24.
593 (thanks to Vincent Arkesteijn)
594 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
595 type when called with 1 argument; PEEK-CHAR checked type of
596 PEEK-TYPE only after having read first character from a
597 stream. (reported by Peter Graves)
598 * bug fix: the garbage collector now has much better locality
599 behaviour, and in particular no longer treats all memory as being
600 exhausted when many small objects point to each other in a deeply
602 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
603 recognized as being TYPEP their class.
604 * bug fix: the PUSHNEW documentation string has been corrected.
605 (thanks to Vincent Arkesteijn)
606 * bug fix: defaulting of the value for the last of an atypically
607 large number of multiple values being bound was not being
608 performed correctly on the Alpha or PPC platforms
609 * optimization: implemented multiplication as a modular
610 (UNSIGNED-BYTE 32) operation on the x86 backend.
611 * optimization: SEARCH on simple-base-strings can now be open-coded.
612 (see also contrib/compiler-extras.lisp for inspiration for
613 teaching the compiler about the Boyer-Moore algorithm).
614 * value, returned by MAX (and MIN) called with several EQUALP, but
615 not EQL, arguments now does not depend on compiler settings.
616 * fixed some bugs revealed by Paul Dietz' test suite:
617 ** in stack analysis liveness information is propagated from
618 non-local entry points.
619 ** pathwise CAST removing failed when the CAST node did not start
621 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
622 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
624 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
625 host is already defined.
626 ** RENAME-FILE works on streams instead of signalling an internal
628 ** PEEK-CHAR uses the current readtable when determining whether
629 or not a character is whitespace.
630 ** MERGE-PATHNAMES handles the case when the pathname does not
631 specify a name while the default-pathname specifies a version.
632 ** Pathnames now stand a better chance of respecting print/read
634 ** Attempting to use standardized file system operators with a
635 pathname with invalid :DIRECTORY components signals a
637 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
639 ** DIRECTORY on logical pathnames is more correct.
640 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
641 signal a TYPE-ERROR if their argument is not a stream.
642 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
643 designator argument does not designate a stream.
644 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
645 examining the synonym.
646 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
648 ** OPEN and WITH-OPEN-STREAM allow opening streams with
649 element-type larger than ([UN]SIGNED-BYTE 32).
651 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
652 * When built with the :SB-FUTEX feature, threaded builds now take
653 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
654 for faster/more reliable mutex and condition variable support.
655 * Incompatible change (but one you probably shouldn't have been using
656 anyway): the interface and code for arbitrating between multiple
657 threads in the same user session has been redesigned.
658 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
659 error when called without an explicit environment argument.
660 (thanks to Nikodemus Siivola)
661 * bug fix: buffered :DIRECTION :IO streams are less likely to become
662 confused about their position. (thanks to Adam Warner and Gerd
664 * bug fix: Pretty printing backquoted forms with unquotations in the
665 argument list position of various code constructs such as LAMBDA
666 now works correctly. (reported by Paul Dietz)
667 * bug fix: Pretty printing unquotations no longer loses all
668 stream position information.
669 * optimization: performance of string output streams is now less
670 poor for multiple small sequence writes.
671 * optimization: performance of CSUBTYPEP in the presence of complex
672 expressions involving CONS and NOT many times has been improved.
673 (reported by Paul Dietz)
674 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
676 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
677 streams for the REPL, for the debugger, and so on. (thanks to
679 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
680 * fixed some bugs revealed by Paul Dietz' test suite:
681 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
682 is ignored if it is too small, rather than propagating through
683 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
684 ** extremely complex negations of CONS types were not being
685 sufficiently canonicalized, leading to inconsistencies in
687 ** VALUES tranformer lost derived type.
689 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
690 * fixed a bootstrapping bug: the build process no longer assumes
691 that the various BOOLE-related constants have the same value in
692 host and target lisps. (noted by Paul Dietz' test suite on an
693 SBCL binary built from CLISP)
694 * The system can now be dynamically linked on the MIPS platform,
695 which enables dynamic loading of foreign code from Lisp. (thanks
696 to Ralf Baechle for discussions on the MIPS ABI)
697 * The system now records debugging information for its own source
698 files in a filesystem-position-independent manner, relative to
699 the "SYS" logical host.
700 * fixed a compiler bug: MV-LET convertion did not check references
701 to the "max args" entry point. (reported by Brian Downing)
702 * tweaked disassembly notes to be less confident about proclaiming
703 some instruction as an LRA. (thanks to Brian Downing)
704 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
705 now each have their own history, command character, and other
706 characteristics. (thanks to David Lichteblau)
707 * fixed some bugs revealed by Paul Dietz' test suite:
708 ** compiler failure in compiling LOGAND expressions including a
710 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
711 shift greater than 32.
712 ** FLUSH-DEST did not mark blocks for type check regeneration.
713 ** HANDLER-CASE failed to accept declarations in handler clauses
714 in some circumstances.
716 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
717 * New code in contrib/sb-introspect (still probably not entirely
718 stable yet) provides some support for smart Lisp development
719 environments like SLIME.
720 * The conditions signalled for errors occurring when loading .fasl
721 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
722 in a way which should help ASDF recover gracefully.
723 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
724 stuff has been cleaned up. If you code contrib/ stuff, this might
725 affect you, and you can look at contrib/README, contrib/STANDARDS,
726 and/or the 0.8.4.27 diff to check.
727 * In full calls the compiler now does not generate checks for declared
728 argument types for all arguments.
729 * various threading fixes
730 ** and some experimental patches which didn't make it into
731 the main tree for this release, but which are shipped in
732 contrib/experimental-thread.patch as a possible fix for some
733 failures (deadlock, spinning...) in GC-intensive multithreaded
735 * fixed PPC build problem (source code incompatibility of different
736 library versions): added offsetof() hackery which attempts to divine
737 where glibc maintainers put uc_mcontext today
738 * fixed bug 282: compiler does not trust type assertions while passing
739 arguments to a full call.
740 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
741 &OPTIONAL VALUES type parameter.
742 * fix bug 214: algorithm for noting rejected templates is now more
743 similar to that of template seletion. (also reported by rydis on
745 * fixed bug 141b: printing backquoted information readably and prettily
746 inserts a space where necessary.
747 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
748 (CHARACTER 10) are now reported as errors, rather than propagated
749 as unknown types. (reported by piso on #lisp)
750 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
751 with values NIL and :ERROR. (thanks to Milan Zamazal)
752 * fixed bug 191c: CLOS now does proper keyword argument checking as
753 described in CLHS 7.6.5 and 7.6.5.1.
754 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
755 longer raise an error; further, using a list as a for-as-arithmetic
756 counter now raises a meaningful error.
757 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
758 now understood by sequence creation functions such as MAKE-SEQUENCE
760 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
761 a STRING-INPUT-STREAM.
762 * compiler enhancement: SIGNUM is now better able to derive the type
764 * type declarations inside WITH-SLOTS are checked. (reported by
766 * fixed some bugs revealed by Paul Dietz' test suite:
767 ** incorrect optimization of TRUNCATE for a positive first
768 argument and negative second.
769 ** compiler failure in let-convertion during flushing dead code.
770 ** compiler failure while deriving type of TRUNCATE on an
771 interval, containing 0.
772 ** ASH of a negative bignum by a negative bignum count now returns
774 ** intersection of CONS types now canonicalizes properly, fixing
775 inconsistencies in SUBTYPEP.
777 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
778 * incompatible change: The --disable-debugger command line
779 option now clobbers the debugger at a more fundamental
780 level, by redefining #'INVOKE-DEBUGGER instead of by
781 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
782 is specified by ANSI to ignore *DEBUGGER-HOOK* and
783 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
784 enter the debugger REPL and then suffer recursive errors
785 because *DEBUG-IO* is also messed up in --disable-debugger mode;
786 while under the new system, BREAK in --disable-debugger mode
787 terminates the system just as an unhandled error would.
788 * fixed compiler performance when processing loops with a step >1;
789 * bug fix: DOCUMENTATION now retrieves generic function
790 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
791 support has been systematized, and now supports the methods
792 specified by ANSI, along with a default method and a method for
793 slot documentation. (reported by Nathan Froyd)
794 * bug fix: effective methods associated with a generic function are
795 no longer cached over a change of that generic function's method
796 combination. (reported by Andreas Fuchs)
797 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
798 names a non-existent directory. (thanks to Andreas Fuchs)
799 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
800 the PPC platform, lead to stack corruption; this has been fixed.
801 (reported by Rainer Joswig)
802 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
803 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
804 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
805 arithmetic in general, is now much more reliable on the PPC
807 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
808 platform now returns the right answer.
809 * optimization: restored some effective method precomputation in
810 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
811 precomputation is now tunable.
812 * optimization: compiler-internal data structure use has been
813 reviewed, and changes have been made that should improve the
814 performance of the compiler by about 20%.
815 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
816 simple-base-strings and simple-bit-vectors is improved.
817 * optimization: the optimization of 32-bit logical and arithmetic
818 functions introduced in version 0.8.3 on the x86 has been
819 implemented on the mips, ppc and sparc platforms; an
820 implementation of the same facility, but for 64-bit arithmetic,
821 has been added for the alpha.
822 * microoptimization: the compiler is better able to make use of the
823 x86 LEA instruction for multiplication by constants.
824 * bug fix: in some situations compiler did not report usage of
825 generic arithmetic in (SPEED 3) policy.
826 * bug 145b fix: compiler used wrong type specifier while converting
827 MEMBER-types to numeric.
828 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
830 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
831 index argument now works properly on the Alpha platform.
832 * bug fix: floating point exception treatment on the Alpha platform
834 * bug fix: FILE-POSITION works much better on string input and
835 output streams. (thanks to Nikodemus Siivola)
836 * bug fix: many threading/garbage collection symptoms sorted.
837 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
838 might be pseudo-atomic.
839 * internal change: Stopping for GC is now done with signals not
840 ptrace. GC is now done in whichever thread wanted it, instead of
842 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
844 * bug fix: result form in DO is not contained in the implicit
846 * incompatible change: ICR structure is changed; the value part of
847 CONTINUATION is now called LVAR; corresponding functions are
848 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
849 * added type deriver for ISQRT (thanks to Robert E. Brown).
850 * bug fix: better support for loading from the command line when an
851 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
853 * fixed some bugs revealed by Paul Dietz' test suite:
854 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
855 ** ROUND and FROUND now give the right answer when given very
856 small float arguments.
857 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
859 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
860 ** (GCD 0 <negative-integer>) returned <negative-integer>.
861 ** LCM should return a non-negative integer.
862 ** PARSE-INTEGER returned the index of a terminator instead of the
863 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
864 ** PARSE-INTEGER returned an incorrect index being applied to a
866 ** LCM with two arguments of 0 returns 0 rather than signalling
868 ** unsigned addition of a 32-bit constant with the high bit set no
869 longer causes an internal compiler error on the x86.
870 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
871 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
872 with negative last argument.
873 ** byte specifiers with zero size and position no longer cause
874 an error during type derivation.
875 ** bignum multiplication on the Alpha platform now returns the
877 * porting: The system now builds on SuSE AMD64, although it still
878 generates a 32-bit binary.
879 * .fasl file incompatibility: The fasl file version number has
880 been incremented (because of the changes to internal compiler
881 data structures referred to above).
883 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
884 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
885 more accurately, on the Darwin kernel running on PowerPC hardware.
886 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
887 * Compiler code deletion notes now signal a condition of type
888 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
889 an associated MUFFLE-WARNING restart.
890 * The compiler now performs limited argument count validation of
891 constant format strings in FORMAT, and where appropriate in ERROR,
892 CERROR and WARN. (thanks to Gerd Moellmann)
893 * New ASDF-INSTALL contrib can be used for automatic download and
894 installation of third-party Lisp code from CCLAN or other sites
896 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
897 which forces another thread to execute a function supplied by the
899 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
900 now accept and act upon their :ELEMENT-TYPE keyword argument.
901 (reported by Martin Atzmueller, Edi Weitz)
902 * bug fix: FILE-POSITION now accepts position designators up to
903 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
904 is the greater. (thanks to Patrik Nordebo)
905 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
906 Kalle Olavi Niemitalo)
907 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
908 printable. (reported by Eric Marsden)
909 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
910 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
911 * x86 bug fix in control stack exhaustion checking: now shows backtrace
912 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
913 (thanks to Stig Sandoe)
914 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
915 UNREAD-CHAR now correctly includes the unread character in the
916 target sequence. (thanks to Gerd Moellmann)
917 * bug fix in threaded builds: the system can now be suspended and
918 resumed by shell job control with minimal disruption.
919 * bug fixes in times and timezones >2038AD
920 * better handling of "where is GNU make?" problem in build scripts
921 (thanks to Nikodemus Siivola)
922 * new optimization: inside a named function any reference to a
923 function with the same name is considered to be a self-reference;
924 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
925 optimization quality.
926 * new optimization on x86: logical functions and + now have
927 optimized (UNSIGNED-BYTE 32) versions, which are automatically
928 used when the result is truncated to 32 bits.
929 * VALUES declaration is partially enabled.
930 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
931 * fixed some bugs revealed by Paul Dietz' test suite:
932 ** The system now obeys the constraint imposed by
933 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
934 types form a lattice under type intersection.
935 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
936 ** ASSOC now ignores NIL elements in an alist.
937 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
938 and (1+ MOST-POSITIVE-FIXNUM) answers.
939 ** The addition of a method with invalid qualifiers to a generic
940 function does not cause an error to be signalled immediately;
941 a warning is signalled, and the error is generated only on
942 calling the generic function.
943 * changed .fasl file version number, in part to add type codes for
944 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
945 obscure ANSI requirements
947 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
948 * fixed bug 148: failure to inline-expand a local function left
949 garbage, confusing the compiler.
950 * fixed bugs 3cd: structure slot readers perform type check if the
951 slot can have an invalid value (i.e. it is either not initialized
952 or can be written with a less specific slot writer).
953 * bug fix: the compiler now traps array references to elements off
954 the end of an array; previously, the bounds checking in some
955 circumstances could go off-by-one.
956 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
958 * type declarations for array element types now obey the description
959 on the CLHS page "Declaration TYPE", as per discussions on
960 sbcl-help around 2003-05-08. This means that a declaration
961 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
962 declaration, all references to BAR will be asserted or assumed
963 (with THE, so dependent on compiler policy) to involve objects of
964 type FOO. Note that no such declaration is implied in
965 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
966 * declared types of functions from the "Conditions"
967 chapter. (reported by Paul Dietz)
968 * bug fix: CERROR accepts a function as its first argument.
969 * bug fix: NTH an NTHCDR accept a bignum as index
970 arguments. (reported by Adam Warner)
971 * optimization: character compare routines now optimize comparing
972 against a constant character. (reported by Gilbert Baumann)
973 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
974 argument now works properly on the MIPS platform.
975 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
976 * repeated evaluation of the same DEFSTRUCT, a slot of which is
977 declared to have a functional type, does not cause an error
979 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
980 many :DEFAULT-INITARGS. (reported by Istvan Marko)
981 * fixed bug: if last continuation of a deleted block has a
982 destination, this destination should be deleted too. (reported by
984 * fixed a bug in the bootstrap process: the host compiler's values
985 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
986 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
987 test case from Patrik Nordebo)
988 * improved the ability of the disassembler on the PPC platform to
989 provide helpful disassembly notes.
990 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
991 the class in more cases than previously.
992 * bug fix: FILE-POSITION now understands :START and :END for
993 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
994 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
995 8) data. (thanks to David Lichteblau)
996 * bug fix: it is possible to add a method to a generic function
998 * bug fix: reader failed to signal END-OF-FILE inside an
999 object representation. (reported by Nikodemus Siivola)
1000 * fixed some bugs revealed by Paul Dietz' test suite:
1001 ** LAST and [N]BUTLAST should accept a bignum.
1002 ** condition slot accessors are methods.
1003 ** (VECTOR NIL) is a subtype of STRING.
1005 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1006 * minor incompatible change: some nonsensical specialized lambda
1007 lists (used in DEFMETHOD) which were previously ignored now signal
1009 * minor incompatible change: the system is now aware of the types of
1010 variables in the COMMON-LISP package, and will signal errors for
1011 most violations of these type constraints (where previously they
1012 were silently accepted).
1013 * minor incompatible change: COMPILE-FILE now uses the freedom
1014 afforded (ANSI 3.2.2.3) to use derived function types for
1015 functions defined in the same file. This also permits the system
1016 to warn on static type mismatches and function
1017 redefinition. (Currently it does not work with high DEBUG level.)
1018 * minor incompatible change: VALUES declaration is disabled.
1019 * When issuing notes, the compiler now signals a condition of type
1020 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1021 restart for use in user handlers. It is expected that the
1022 COMPILER-NOTE condition will eventually become a condition
1023 supertype to a hierarchy of note types, which will then be
1024 handleable in a similar fashion. However, at the moment, no such
1025 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1026 but it's an implementation detail, not a classification for the
1028 * Changes in type checking closed the following bugs:
1029 ** type checking of unused values (192b, 194d, 203);
1030 ** template selection based on unsafe type assertions (192c, 236);
1031 ** type checking in branches (194bc).
1032 * A short form of VALUES type specifier has ANSI meaning (it has
1033 increased the number of situations when SBCL cannot perform type
1035 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1036 keywords or constants is permissible.
1037 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1038 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1039 classes. (thanks to Antonio Martinez)
1040 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1041 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1042 argument to a pretty-print stream. Instead, it leaves any such
1043 operations to DESCRIBE-OBJECT methods.
1044 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1046 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1047 lambda lists. (thanks to David Lichteblau)
1048 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1049 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1050 (thanks to Gerd Moellmann)
1051 * fixed bug 246: increased compilation speed of long
1052 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1054 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1056 * DEFCONSTANT now throws a condition of type
1057 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1058 constant to a non-EQL value; CONTINUE and ABORT restarts
1059 respectively change and preserve the value.
1060 * fixed bug 63: The code walker, part of the implementation of CLOS,
1061 is now better at handling symbol macros.
1062 * bug fix: There is no longer an internal implementation type named
1063 CL:LENGTH. (reported by Raymond Toy)
1064 * bug fix: In macro-like defining macros/special operators the
1065 implicit block does not enclose the lambda list.
1066 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1067 enclosed in parentheses) are not suitable as type specifiers, and
1068 their use properly signals an error now.
1069 * bug fix: An argument count mismatch for a type specifier in code
1070 being compiled no longer causes an unhandled error at compile
1071 time, but signals a compile-time warning.
1072 * fixed simple vector readable printing
1073 * bug fix: DESCRIBE takes more care over whether the class
1074 precedence list slot of a class is bound before accessing it.
1075 (reported by Markus Krummenacker)
1076 * bug fix: FORMATTER can successfully compile pretty-printer format
1077 strings which use variants of the ~* directive inside.
1078 * bug fix: SEARCH now applies its TEST predicate to the elements of
1079 the arguments in the correct order. (thanks to Wolfhard Buss)
1080 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1081 in type inference. (thanks to Robert E. Brown)
1082 * bug fix: Reading in symbols with an explicit package name of ""
1083 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1084 * fixed some bugs revealed by Paul Dietz' test suite:
1085 ** NIL is now allowed as a structure slot name.
1086 ** Arbitrary numbers, not just REALs, are allowed in certain
1087 circumstances in LOOP for-as-arithmetic clauses.
1088 ** Multiple class redefinitions before slot access no longer
1089 causes a type error.
1090 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1091 association between the name and a class.
1092 ** Generic functions with non-standard method-combination and over
1093 five methods all of which return constants no longer return NIL
1094 after the first few invocations. (thanks to Gerd Moellmann)
1095 ** CALL-NEXT-METHOD with no arguments now passes the original
1096 values of the arguments, even in the presence of assignment.
1097 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1099 ** DISASSEMBLE works with closures and funcallable instances.
1100 ** ADD-METHOD now returns the generic function, not the new method.
1101 ** FIND-METHOD signals an error if the lengths of the specializers
1102 is incompatible with the generic function, even if the ERRORP
1104 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1105 which its argument is a member.
1106 ** DEFCLASS only redefines the class named by its class-name
1107 argument if that name is the proper name of the class;
1108 otherwise, it creates a new class.
1109 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1110 of the UNBOUND-SLOT condition to the name of the slot.
1111 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1112 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1113 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1114 SLOT-MAKUNBOUND in the specified fashion.
1116 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1117 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1118 cross-compilation host. As a consequence, we can now bootstrap our
1119 way up to SBCL starting with a bare gcc toolchain and human-readable
1120 source code (first the source to CLISP, then the source to SBCL).
1121 * A contributed module containing a partial implementation of the
1122 simple-streams interface has been included. (thanks to Rudi
1124 * A contributed module implementing the RFC1321 Message Digest
1125 Algorithm, known as MD5, has been included.
1126 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1127 no longer has any effect, as the code controlled by this feature
1128 has been deleted. (As far as we know, no-one has ever built using
1129 this feature, and its semantics were confused in any case).
1130 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1131 work on conditions (as required by the ANSI specification),
1132 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1133 expected behaviour on conditions. Users should note, however,
1134 that such behaviour is not required by the ANSI specification,
1135 and so use of this behaviour may render their code unportable.
1136 * fixed some bugs revealed by Paul Dietz' test suite:
1137 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1139 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1140 ** SXHASH on condition objects no longer returns NIL.
1141 ** :ALLOCATION :CLASS slots are better treated; their values are
1142 updated on class redefinition, and initforms inherited from
1143 superclasses are applied.
1144 ** REMOVE-METHOD returns its generic function argument even when
1145 no method was removed.
1146 ** SHARED-INITIALIZE now initializes the values of the requested
1147 slots, including those with :ALLOCATION :CLASS.
1148 ** ALLOCATE-INSTANCE now works on structure classes defined via
1149 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1151 ** SLOT-EXISTS-P now works on conditions, as well as structures
1153 ** MAKE-LOAD-FORM now has the required methods on
1154 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1155 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1156 keyword, and now implements the SLOT-NAMES argument.
1157 ** methods with &OPTIONAL arguments no longer allow too many
1158 arguments to be passed in the call without error.
1159 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1160 option is consistent with the required arguments of the generic
1161 function lambda list.
1162 * bug fix: REQUIRE accepts a string designator. (Thanks to
1164 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1165 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1166 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1167 * bug fix: The compiler now checks for duplicated variables in macro
1169 * bug fix: SETQ on globals returns the correct value.
1170 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1171 not a COMPILER-ERROR (followed by some other strange error on
1172 choosing the CONTINUE restart).
1173 * bug fix: make.sh and friends are now more consistent in the way that
1174 they look for GNU "make".
1176 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1177 * experimental native threads support (on x86 Linux >=2.4 only).
1178 This is not compiled in by default: you need to add :SB-THREAD to
1179 the target features. See the "Beyond ANSI" chapter of the manual
1181 * fix for longstanding nonANSIism: The old distinction between
1182 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1183 The return value from CL:FIND-CLASS is now a CLOS class, and
1184 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1185 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1186 * An interface to the MetaObject Protocol, as described in Kiczales,
1187 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1188 MIT Press, 1991, is available from the SB-MOP package.
1189 * incompatible change: the SB-PCL package should now be considered
1190 a private implementation detail, and no longer a semi-private MOP
1192 * minor incompatible change: due to rearrangement for threads, the
1193 control stack and binding stack are now allocated at arbitrary
1194 addresses instead of being hardcoded per-port. Users affected by
1195 this probably have to be doing advanced things with shared
1196 libraries, and will know who they are.
1197 * minor incompatible change: Previously, all --eval forms used were
1198 processed with READ before any of them were processed with EVAL.
1199 Now each --eval form is processed with both READ and EVAL before
1200 the next --eval form is processed. (Thus package operations like
1201 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1202 work as the user might reasonably expect.)
1203 * minor incompatible change: *STANDARD-INPUT* is now only an
1204 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1206 * minor incompatible change: Y-OR-N-P is now character-oriented,
1207 not line oriented. Also, YES-OR-NO-P now works without errors.
1208 (thanks to Antonio Martinez)
1209 * sb-aclrepl module improvements: an integrated inspector, added
1210 repl features, and a bug fix to :trace command.
1211 * Known functions, which cannot be open coded by the backend, are
1212 considered to be able to check types of their arguments. (fixing
1213 a bug report by Nathan J. Froyd)
1214 * fixed a bug in computing method discriminating functions: It is
1215 now possible to define methods specialized on classes which have
1216 forward-referenced superclasses. (thanks to Gerd Moellmann)
1217 * fixed evaluation order in optional entries (reported by Gilbert
1219 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1220 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1221 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1222 required-by-AMOP NAME argument, as well as CLASS and
1223 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1224 * fixed bug 20: DEFMETHOD can define methods using names that are
1225 not the proper names of classes to designate class specializers.
1226 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1227 Unix file descriptors, instead of blowing up. (thanks to Antonio
1229 * Garbage collection refactoring: user-visible change is that a
1230 call to the GC function during WITHOUT-GCING will not do garbage
1231 collection until the end of the WITHOUT-GCING. If you were doing
1232 this you were probably losing anyway.
1233 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1234 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1235 TYPEP the latter but not the former.
1236 * The compiler issues a full WARNING for calls to undefined functions
1237 with names from the CL package.
1238 * MAP-INTO for a vector destination is open coded. (reported by
1239 Brian Downing on c.l.l)
1240 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1241 documentation string.
1242 * fixed some bugs revealed by Paul Dietz' test suite:
1243 ** COPY-ALIST now signals an error if its argument is a dotted
1245 ** Condition slots are now accessed more correctly in the presence
1246 of multiple initargs for a given slot.
1247 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1248 exclude restarts of the same name associated with a different
1250 ** DEFCLASS of forward-referenced classes with another
1251 forward-referenced class in the superclasses list no longer
1253 ** Condition slots are now initialized once each, not multiple
1254 times. (thanks to Gerd Moellmann)
1255 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1257 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1258 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1259 arguments contain duplicated elements.
1260 ** RESTART-CASE understands local macros.
1261 ** RESTART-CASE associates exactly its own restarts with a condition.
1262 ** ENDP in safe mode checks its argument to be of type LIST.
1263 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1264 in question is unbound.
1265 ** Optimized MAKE-INSTANCE functions no longer cause internal
1266 assertion failures in the presence of duplicate initargs.
1267 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1268 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1269 propagated even in the ctor optimized implementation of
1271 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1273 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1274 * a better implementation of SXHASH on (simple) bit vectors,
1275 measured both in execution speed and in distribution of results
1276 over the positive fixnums, has been installed. Likewise, a better
1277 implementation of EQUAL for simple bit vectors is now available.
1278 * fixed CEILING optimization for a divisor of form 2^k.
1279 * fixed bug 240 (emitting extra style warnings "using the lexical
1280 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1281 by Antonio Martinez)
1282 * fixed SXHASH, giving different results for NIL depending on type
1283 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1284 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1285 variable. (found by Rolf Wester)
1286 * SBCL does not ignore type declarations for special
1287 variables. (reported by rif on c.l.l 2003-03-05)
1288 * some bug fixes in contrib/sb-aclrepl/
1289 * fixed some bugs revealed by Paul Dietz' test suite:
1290 ** a bug in the CONS type specifier, whereby the CAR and CDR
1291 types got intertwined, has been fixed;
1292 ** the type system is now able to reason about the interaction
1293 between INTEGER and RATIO types more completely;
1294 ** APPEND, [N]REVERSE and NRECONC check that those their
1295 arguments, which must be proper lists, are really so;
1296 ** An array specialized to be unable to hold elements has been
1297 implemented, as required -- yes, really -- by ANSI;
1298 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1299 on malformed property lists;
1301 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1302 * incompatible packaging change: in line with Unix convention,
1303 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1304 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1305 * REQUIRE and PROVIDE are now optionally capable of doing something
1306 useful. See the documentation string for REQUIRE.
1307 * infrastructure for a managed SBCL contrib system: contributed
1308 modules in this release include:
1309 ** the ASDF system definition facility;
1310 ** an interface to the BSD Sockets API;
1311 ** an ACL-like convenience interface to the repl;
1312 (thanks to Kevin Rosenberg)
1313 ** an implementation of ROTATE-BYTE, with efficient implementation
1315 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1316 gives a closer approximation to the right answer than previously.
1317 (thanks to Raymond Toy)
1318 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1319 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1320 optional environment arguments, as required by ANSI.
1321 * fixed bugs in other functions taking environment objects, allowing
1322 calls with an explicit NIL environment argument to be compiled
1324 * fixed bug 228: primary return values from
1325 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1326 COMPILE or FUNCTION.
1327 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1328 :TYPE will no longer signal errors on innocuous objects.
1329 * fixed bug 231b: SETQ is better at respecting type declarations in
1330 the lexical environment.
1331 * fixed a bug in DEFCLASS: classes named by symbols with no or
1332 unprintable packages can now be defined.
1333 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1334 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1335 * fixed compiler failure related to checking types of functions.
1336 (reported by Robert E. Brown)
1337 * the compiler is now much more consistent in its error-checking
1338 treatment of bounding index arguments to sequence functions: in
1339 (SAFETY 3) code, errors will be signalled in almost all cases if
1340 invalid sequence bounding indices are passed to functions defined
1341 by ANSI to operate on sequences.
1342 * fixed a bug in the build procedure: documentation of SBCL-specific
1343 packages is now preserved and available in the final Lisp image.
1344 * lifted FDEFINITION lookup out of loops in the implementation of
1345 many list operations. (thanks to Robert E. Brown)
1346 * fixed a bug in the reader: the #n# reader macro now works for
1347 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1348 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1349 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1350 E. Brown and Tony Martinez respectively)
1351 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1352 SLOT-BOUNDP: in particular, it is now able to optimize them much
1353 better, and is now not vulnerable to having packages renamed.
1354 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1355 always called when a slot is not present in an instance. (thanks
1357 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1358 longer signal an error on unreading a character following EOF on
1359 the previous constituent stream. (thanks to Tony Martinez)
1360 * fixed some bugs revealed by Paul Dietz' test suite:
1361 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1362 not just nonnegative fixnums;
1363 ** the logical bit-array operators such as BIT-AND now accept an
1364 explicit NIL for their "opt-arg" argument (to indicate a
1365 freshly-consed result bit-array);
1366 ** ELT now signals an error on an invalid sequence index in safe
1368 ** the type system is now cleverer about negations of numeric
1369 types, and consequently understands the BIGNUM and RATIO types
1371 ** the type system is now cleverer about the interaction between
1372 INTEGER and RATIO types: while bugs still remain, many more
1373 cases are accurately computed;
1374 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1375 if it is in the last clause;
1376 ** CONSTANTLY now correctly returns a side-effect-free function in
1378 ** DECLARE is no longer treated as a special-operator; in
1379 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1380 * incremented fasl file version number due to the change in the
1381 DEFSTRUCT-SLOT-DESCRIPTION structure.
1383 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1384 * minor incompatible change: code processed by the "interpreter" or
1385 EVAL now has a compilation optimization policy of (DEBUG 2)
1386 (changed from (DEBUG 1)) to improve debuggability of interactive
1387 development, and to allow the use of the debug RETURN command in
1389 * an experimental implementation of the RETURN command for the
1390 debugger has been included. (thanks to Frederik Kuivinen)
1391 * fixed bug 62: constraints were not propagated into a loop.
1392 * fixed bug in embedded calls of SORT (reported and investigated by
1394 * fixed some bugs revealed by Paul Dietz' test suite:
1395 ** printing and reading of arrays with some dimensions having
1396 length 0 (thanks to Gerd Moellmann);
1397 ** BOA constructor with &AUX argument without a default value does
1398 not cause a type error;
1399 ** CONSTANTP now returns true for all self-evaluating objects.
1401 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1402 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1403 accessors that are related by inheritance, as specified in the
1404 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1405 Valtteri Vuorikoski)
1406 * The compiler is now able to inline functions that were defined in
1407 a complex lexical environment (e.g. inside a MACROLET).
1408 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1410 * Support for the upcoming FreeBSD-5.0 release has been included.
1411 (thanks to Dag-Erling Smorgrav)
1412 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1413 effect when it is not in a toplevel context.
1414 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1415 stand a better chance of being correct. (thanks to Gerd
1417 * fixed bug in COERCE, which now signals an error on coercing a
1418 rational to a bounded real type which excludes the expected
1420 * The compiler is now able to derive types more accurately from the
1421 COERCE and COMPILE functions.
1422 * fixed bug 223: functional binding is considered to be constant
1423 only for symbols in the CL package.
1424 * fixed bug 231: SETQ did not check the type of a variable being set
1425 (reported by Robert E. Brown)
1426 * A new optimization for MAKE-INSTANCE has been included, fixing
1427 various bugs (including relating to :ALLOCATION :CLASS slots and
1428 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1430 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1431 ** As required by ANSI, LOOP now disallows anonymous collection
1432 clauses such as COLLECT I in conjunction with aggregate boolean
1433 clauses such as THEREIS (= I 1);
1434 ** LOOP now signals an error when any variable is reused in the
1435 same loop (including the potentially useful construct analogous
1436 to WITH A = 1 WITH A = (1+ A);
1437 ** IT is only a special loop symbol within the first clause of a
1438 conditional loop clause;
1439 ** LOOP with a typed iteration variable over a hashtable now
1440 signals a type error iff it should.
1441 * fixed some other bugs revealed by Paul Dietz' test suite:
1442 ** FILE-STREAM now names the class previously known as FD-STREAM;
1443 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1444 argument) no longer signals an error;
1445 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1446 of the slot symbol, rather than using the current package
1447 ((:CONC-NAME "") continues to intern the slot's name in the
1449 * incremented fasl file version number, because of the incompatible
1450 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1451 of the new implementation of DEFINE-COMPILER-MACRO.
1453 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1454 * Support for building SBCL for MIPS platforms running in
1455 little-endian mode has now been checked in, and basic
1456 functionality on said platforms verified.
1457 * minor incompatible change: PCL now records the pathname of a file
1458 in which methods and the like are defined, rather than its
1460 * minor incompatible change: TRUENAME now considers the truename of
1461 a file naming a directory to be the pathname with :DIRECTORY
1462 component indicating that directory.
1463 * minor incompatible change: a NAMED clause in the extended form of
1464 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1465 reason for the previous behaviour is unclear.
1466 * more systematization and improvement of CLOS and MOP conformance
1467 in PCL (thanks to Gerd Moellman and Pierre Mai):
1468 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1470 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1471 primary methods with no specializers;
1472 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1474 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1475 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1476 FORWARD-REFERENCED-CLASSes; error reporting on
1477 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1479 ** SXHASH on CLOS instances now uses a slot internal to the
1480 instance to return different numbers on distinct instances,
1481 while preserving the same return value through invocations of
1483 ** DEFMETHOD signals errors when methods with longer incongruent
1484 lambda lists are added to generic functions;
1485 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1486 CLASS, as specified in AMOP;
1487 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1488 on the order returned by the primary method for classes of
1489 class STANDARD-CLASS;
1490 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1491 * fixed some bugs shown by Paul Dietz' test suite:
1492 ** DOLIST puts its body in TAGBODY;
1493 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1495 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1496 value producing form;
1497 ** if more variables are given to PROGV than values, extra
1498 variables are bound and made to have no value;
1499 ** NSUBSTITUTE on list arguments gets the right answer with
1501 ** ELT signals an error of type TYPE-ERROR when the index argument
1502 is not a valid sequence index;
1503 ** LOOP signals (at macroexpansion time) an error of type
1504 PROGRAM-ERROR when duplicate variable names are found;
1505 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1506 ** FUNCALL of special-operators now cause an error of type
1508 ** PSETQ now works as required in the presence of side-effecting
1509 symbol-macro places;
1510 ** NCONC accepts any object as its last argument;
1511 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1513 ** loop-for-as-package does not require a package to be explicitely
1515 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1517 * fixed bug 166: compiler preserves "there is a way to go"
1518 invariant when deleting code.
1519 * fixed bug 172: macro lambda lists with required arguments after
1520 &REST arguments now cause an error to be signalled. (thanks to
1522 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1523 bug. (thanks to Matthew Danish)
1524 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1526 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1527 arguments are lexically rebound. (thanks to Gerd Moellmann and
1529 * fixed bug 194: error messages are now more informative when there
1530 is no primary method applicable in a call to a generic
1531 function. (thanks to Gerd Moellmann)
1532 * fixed bug in command line argument checking (thanks to Julian
1534 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1535 (thanks to Matthew Danish)
1536 * incremented fasl file version number, because of the
1537 SXHASH-related changes in the layout of CLOS data structures
1539 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1540 * minor incompatible change: The runtime (the Unix executable named
1541 "sbcl") is now much pickier about the .core files it will load.
1542 Essentially it now requires .core files to descend from the same
1543 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1544 as the runtime does. (The intent is to prevent the crashes which
1545 can occur, and which can even be reported as mysterious failures,
1546 when people patch the sources or change the build parameters
1547 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1548 sbcl and .core files.)
1549 * fixed bug: VALUES-LIST is no longer optimized away.
1550 * fixed bug 142: The FFI conversion of C string values to Lisp
1551 string values no longer conses excessively. (thanks to Nathan
1552 Froyd porting Raymond Toy's fix to CMU CL)
1553 * began to systematize and improve MOP conformance in PCL (thanks to
1554 Nathan Froyd, Gerd Moellman and Pierre Mai):
1555 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1557 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1558 ** FINALIZE-INHERITANCE is now called on class finalization;
1559 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1560 argument precedence order.
1561 * fixed bug 202: The compiler no longer fails on functions whose
1562 derived types contradict their declared type.
1563 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1564 so it can be non-toplevel.
1565 * The fasl file version number has changed (because of the new
1566 implementation of DEFMACRO).
1567 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1568 safe code, that any length requirement by their type-specifier
1569 argument is valid. The exceptions to this are described in bug
1571 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1572 functions now check the types of their inputs as required by ANSI.
1573 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1574 introduced symbol is DECLAREd to be SPECIAL.
1575 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1576 * fixed bug: :COUNT argument to sequence functions may be negative
1577 * fixed bug: body of DO-SYMBOLS may contain declarations
1578 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1579 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1580 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1581 symbol macro only once
1582 * fixed printing of call frame when argument list is unavailable
1583 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1584 * compiler no longer signals WARNING on unknown keyword
1587 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1588 * A beta-quality port to the mips architecture running Linux,
1589 based on the old CMUCL backend, has been made. It has been tested
1590 on a big-endian kernel, and works sufficiently well to be able to
1591 rebuild itself; it has not been tested in little-endian mode.
1592 * fixed an inconsistency between gencgc.c and purify.c which made
1593 dumping/loading .core files unreliable
1594 * fixed bug 120a: The compiler now deals correctly with IFs where
1595 the consequent is the same as the alternative, instead of
1596 misderiving the return type. (thanks to Alexey Dejneka)
1597 * fixed bug 113: Logical pathnames are now dumpable (the logical
1598 host is resolved at load-time, throwing an error if it is not
1600 * fixed bug 174: FORMAT's error message is slightly clearer when a
1601 non-printing character is used in a format directive.
1602 * fixed several bugs in compiler checking of type declarations, i.e.
1603 violations of the Python "declarations are assertions" principle
1604 (thanks to Alexey Dejneka)
1605 * fixed several bugs in PCL's error checking (thanks to Gerd
1607 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1609 * fixed bug in compilation of functions as first class values
1610 (thanks to Antonio Martinez-Shotton)
1611 * The compiler's handling TYPE-ERRORs which it can prove will
1612 inevitably happen at runtime has been cleaned up and corrected
1613 in several ways. (thanks to Alexey Dejneka)
1614 * improved argument type checking for various basic arithmetic
1615 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1616 lavished on them in the past that they can be compiled in many
1617 ways in different special cases
1618 * fixed bug 181: compiler checks validity of user supplied type
1620 * cleaned up code flushing in optimization: Function calls which
1621 should signal errors for safety purposes (e.g. which ANSI says
1622 should signal errors when their arguments are of incorrect type)
1623 are no longer optimized away.
1624 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1625 * incremented fasl file version number, because changes in the
1626 implementation of sequence functions like COERCE caused
1627 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1628 in old inline expansions) to become undefined. (Actually these
1629 changes were later undone, so we might very well be binary
1630 compatible with 0.7.7 after all, but leaving the version number
1631 incremented seemed like the simplest and most conservative
1634 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1635 * An alpha-quality port to the parisc architecture running Linux,
1636 based on the old CMUCL backend, has been made. This, even more so
1637 than the other backends, should be considered still a work in
1638 progress; known problems include that the Linux kernel in 64-bit
1639 mode does not propagate the correct sigcontext structure to
1640 userspace, and consequently SBCL on a parisc64 kernel will not
1642 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1643 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1644 suppressed.) Also now that the compiler is looking at declarations
1645 in the environment, it checks optimization declarations as well,
1646 and suppresses inlining when (> DEBUG SPEED).
1647 * More fixes have been made to treatment of floating point exception
1648 treatment and other Unix signals. In particular, floating point
1649 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1650 * The detection and handling of control stack exhaustion (infinite
1651 or very deeply nested recursion) has changed. Stack exhaustion
1652 detection is now done by write-protecting pages at the OS level
1653 and applies at all optimization settings; when found, a
1654 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1655 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1656 be caught using IGNORE-ERRORS.
1657 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1658 correctly when transforming recursive LABELS functions to LETs.
1659 (thanks to Alexey Dejneka)
1660 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1661 that are names of constants or global variables.
1662 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1663 alien routines with docstrings.
1664 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1665 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1667 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1668 (thanks to Eric Marsden)
1669 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1670 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1671 * Bug fix: The SPARC backend can now compile functions involving
1672 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1673 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1674 to a functions expecting a PATHNAME-DESIGNATOR.
1675 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1676 lambda lists. (thanks to Alexey Dejneka)
1677 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1678 functions so that EXPORTing the name of the class doesn't cause
1679 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1680 to Antonio Martinez for reporting this)
1681 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1682 constant-folded without failing an assertion. (thanks to Einar
1683 Floystad Dorum for reporting this)
1684 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1685 types are now dealt with more correctly by the compiler.
1686 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1687 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1688 behaviour analogous to RENAME-FILE. This puts its behaviour more
1689 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1691 * The fasl file version number has changed again. (because of the
1692 bug fix involving the names of PCL MAKE-INSTANCE functions)
1694 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1695 * bug fix: Floating point exceptions are treated much more
1696 consistently on the x86/Linux and PPC/Linux platforms.
1697 * Array initialization with :INITIAL-ELEMENT is now much faster for
1698 cases when the compiler cannot open code the array creation, but
1699 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1700 array accesses have also seen a speed increase.
1701 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1702 specified. (This was at the root of some bad interactions between
1703 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1705 * bug fix: Internal error arguments for undefined functions are now
1706 computed correctly on the PPC/Linux platform.
1707 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1708 Raymond Toy's patch for CMU CL)
1709 * Support for the Solaris 9 operating environment has been included
1710 (thanks to Daniel Merritt)
1711 * A very ugly but hopefully complete draft of the missing FFI chapter
1712 of the manual has been created by reformatting the corresponding
1713 CMU CL manual chapter into (currently very ugly and incoherent)
1714 DocBook and bringing it up to date for SBCL behavior. Thus, the
1715 manual is now essentially complete, at least by my extreme
1716 once-and-only-once standards, whereby it's acceptable to refer to
1717 the doc strings of SB-EXT functions as the primary documentation.
1718 * The fasl file version number has changed again, due to cleanup of
1719 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1720 is no longer a static symbol.)
1722 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1723 * SBCL now builds with OpenMCL (version 0.12) as the
1724 cross-compilation host; also, more progress has been made toward
1725 bootstrapping under CLISP.
1726 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1728 * bug 158 fixed: The compiler can now deal with integer loop
1729 increments different from 1; fixing this turned out also to fix
1731 * bug 169 fixed: no more bogus warnings about using lexical bindings
1732 despite the presence of perfectly good SPECIAL declarations (thanks
1733 to David Lichteblau)
1734 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1735 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1736 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1738 * bug 180 fixed: Method combination specifications no longer ignore
1739 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1740 * bug fix: Structure type predicate functions now check their argument
1741 count as they should.
1742 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1743 correctly. (thanks to Pierre Mai)
1744 * minor incompatible change: The --noprogrammer option is deprecated
1745 in favor of the new --disable-debugger option, which is very similar.
1746 (The major difference is that it takes effect at a slightly different
1747 time at startup, causing handling of errors in --sysinit and
1748 --userinit files will be affected differently.) The
1749 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1750 been added to allow this functionality to be controlled from ordinary
1751 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1752 maintainers, who might want to run non-interactive scripts to
1753 build SBCL cores which will later be used interactively.)
1754 * minor incompatible change: The LOAD function no longer, when given
1755 a wild pathname to load, loads all files matching that pathname.
1756 Instead, an error of type FILE-ERROR is signalled.
1758 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1759 * bug 147 fixed: The compiler preserves its block link/count
1760 invariants more correctly now so that it doesn't crash. (thanks
1762 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1764 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1765 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1766 no-value-for-this-character as NIL (as specified by ANSI).
1767 * HOST-NAMESTRING on physical pathnames now returns a string that is
1768 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1769 (thanks to Christophe Rhodes)
1770 * The Alpha port handles icache flushing more correctly. (thanks to
1772 * More progress has been made toward bootstrapping under CLISP. (thanks
1773 to Christophe Rhodes)
1774 * The fasl file format has changed again, because dynamic loading
1775 on OpenBSD (which has non-ELF object files) motivated some cleanups
1776 in the way that foreign symbols are transformed and passed around.
1777 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1778 is no longer treated as whitespace by the reader, but instead as
1779 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1780 |A
\7fB|, instead of A as it used to.
1782 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1783 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1784 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1785 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1786 this as of 0.7.1.45, but was left out of the previous news section
1787 (thanks to Dan Barlow)
1788 * SBCL now runs on the Solaris operating system on SPARC architectures
1789 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1790 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1791 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1792 * SPARC backend cleanups, allowing builds of cores optimized for V8
1793 and V9 SPARCS, and also emission of code targeted to a particular
1794 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1796 * SBCL is closer to bootstrapping under CLISP, thanks to various
1797 fixes by Christophe Rhodes.
1798 * The fasl file format has changed again, to allow the compiler's
1799 INFO database to support symbol macros.
1800 * The user manual (in doc/) is formatted into HTML more nicely.
1801 (thanks to coreythomas)
1802 * The system is smarter about SUBTYPEP relationships, especially
1803 those involving NOT types (including types such as ATOM which are
1804 represented internally using NOT types). Thus SUBTYPEP is less
1805 likely to return (VALUES NIL NIL) in general, and in particular
1806 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1808 * The fasl file format has changed again, because the internal
1809 representation of types now includes a new slot to support the new
1810 SUBTYPEP-of-NOT-types logic.
1811 * (not a change in the main branch of SBCL, but a related prototype
1812 which can hopefully be merged into the main branch of SBCL in the
1813 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1814 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1815 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1816 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1817 dispatch macro character combinations. (thanks to Alexey Dejneka)
1818 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1819 that superficially look like logical namestrings correctly.
1820 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1822 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1823 default supplied-p) for &optional and &key arguments. (thanks to
1826 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1827 * incompatible change: The compiler is now less aggressive about
1828 tail call optimization, doing it only when (> SPACE DEBUG) or
1829 (> SPEED DEBUG). (This is an incompatible change because there are
1830 programs which relied on the old CMU-CL-style behavior to optimize
1831 away their unbounded recursion which will now die of stack overflow.)
1832 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1833 for non-GENCGC systems has been increased to 20M (since that
1834 seems much closer to the likely performance optimum for modern
1835 systems than the old 4M value was)
1836 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1837 and *DEBUG-PRINT-LEVEL*
1838 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1839 of CMU CL's support for SPARC, and various endianness and other
1840 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1841 * new syntactic sugar for the Unix command line: --load foo.bar is now
1842 an alternate notation for --eval '(load "foo.bar")'.
1844 ** The system now detects stack overflow and handles it gracefully,
1845 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1846 optimization settings. (This is a good thing in general, and
1847 its introduction in this version should be particularly timely
1848 for anyone whose code fails because of suppression of tail
1850 ** The system now hunts for the C variable "environ" in a more
1851 devious way, to avoid segfaults when the C library version differs
1852 between compile time and run time. (thanks to Christophe Rhodes)
1853 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1854 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1855 ** The compiler no longer issues bogus style warnings for undefined
1856 classes in the same source file as the DEFCLASSes which defined
1857 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1859 ** fixes in CONDITION class precedence list for undefined function
1860 errors (thanks to Alexei Dejneka)
1861 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1862 correctly. (thanks to Dan Barlow)
1863 ** portability fixes aiming at bootstrapping under CLISP (thanks
1864 to Dave McDonald and Christophe Rhodes)
1865 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1866 ** fixes in type translation and and type inference (thanks to
1868 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1869 ** various fixes in the new ports (thanks to Dan Barlow)
1870 * several changes related to debugging:
1871 ** suppression of tail recursion, as noted above
1872 ** stack overflow detection, as noted above
1873 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1874 is now the default. (For some time encapsulation has been more
1875 reliable than the breakpoint-based :ENCAPSULATE NIL
1876 implementation, at least on X86 systems; and I just noticed that
1877 encapsulation also seems closer to the spirit of the ANSI
1880 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1882 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1883 up properly again. (There was a packaging bug in 0.7.0 which
1884 left their definitions in SB-SYS::LOAD-FOREIGN and
1885 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1886 vital for most things which interface to C-level interfaces,
1887 like extensions working with sockets or databases or
1888 Perl-compatible regexes or whatever, and the need to fix
1889 this bug is the main reason that 0.7.1 was released so
1891 ** DEFGENERIC is now choosier about the methods it redefines, so that
1892 reLOADing a previously-LOADed file containing DEFGENERICs does
1893 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1894 cycle now works as it should. (thanks to Alexey Dejneka)
1895 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1896 fixed by Christophe Rhodes. (He actually submitted this patch
1897 months ago, and I delayed until after 0.7.0.)
1898 ** Bug 111 (internal compiler confusion about runtime checks on
1899 FUNCTION types) was fixed by Alexey Dejneka.
1900 * Some internal cleanups (getting rid of variables which aren't
1901 needed now that the byte interpreter is gone) caused the fasl
1902 file format number to change again.
1904 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1905 * major incompatible change: The default fasl file extension, i.e. the
1906 default extension for files produced by COMPILE-FILE, has changed
1907 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1909 ** There are many changes in the implementation of the compiler.
1910 SBCL is now essentially a compiler-only implementation of ANSI
1911 Common Lisp. EVAL still "interprets" a few special cases, but
1912 almost all the interesting cases are handled by creating
1913 a LAMBDA expression, calling COMPILE on it, then calling
1914 FUNCALL on the result.
1915 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1916 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1917 Since the code is newer, there might still be some new bugs
1918 (though not as many as before Martin Atzmueller's fixes:-). But
1919 the new code is substantially simpler and clearer, and hopefully
1920 any remaining bugs will be simpler, less fundamental, and more
1921 fixable then the bugs in the old code.
1922 ** The revised compiler is still a little unsteady on its feet.
1924 *** The debugging information it produces (particularly the names
1925 of FUNCTION objects) is sometimes much less useful than what
1926 the old compiler produced.
1927 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1928 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1929 within a MACROLET) has been temporarily weakened.
1930 ** There are new compiler optimizations for various functions:
1931 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1932 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1933 *** the math functions TRUNCATE, FLOOR, and CEILING
1934 *** the function-of-all-trades COERCE
1935 Mostly these should be transparent, but there's one
1936 potentially-annoying problem (bug 117): when the compiler
1937 inline-expands a function and does type analysis on the result,
1938 it can create control paths which have type mismatches, and
1939 when it can't prove that those control paths aren't taken,
1940 it will issue WARNINGs about the type mismatches. This is
1941 a particular problem in practice for the new sequence functions.
1942 It's not clear how this should be fixed, and for now, a
1943 workaround is given in the entry for 117 in the BUGS file.
1944 ** (Because of the interaction between the two previous items --
1945 occasional inlining problems and new inline expansions -- some
1946 of the new sequence function optimizations won't really kick in
1947 completely until debugging information, and then inlining, are
1948 straightened out in some future version.)
1949 * minor incompatible changes:
1950 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1951 DIRECTORY no longer implicitly promotes NIL slots of its
1952 pathname argument to :WILD. In particular, when you ask for the
1953 contents of a directory (which you used to be able to do without
1954 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1955 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1956 ** changes in behavior that ANSI explicitly defines to be
1957 implementation dependent:
1958 *** The new compiler-only implementation still conforms with ANSI,
1959 but acts a little different than before. Besides the obvious
1960 changes in performance tradeoffs (that the cost per form passed
1961 to EVAL has gone up, and the cost per form executed by EVAL
1962 has gone down), the behavior of the system changes a little
1963 because there are no longer any interpreted function objects.
1964 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1965 e.g. doing COMPILE on the output of interactive DEFUN is
1967 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
1969 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
1970 has changed from (:CL) to NIL.
1971 *** The CHAR-NAME of unprintable ASCII characters which, unlike
1972 e.g. #\Newline and #\Tab, don't have names specified in the
1973 ANSI Common Lisp standard, is now based on their ASCII symbolic
1974 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
1975 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
1976 are no longer used for output.
1977 ** changes in internal implementation constants:
1978 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
1979 4 million. (If your application spends a lot of time GCing and
1980 you have a lot of RAM, you might want to experiment with
1981 increasing it even more.)
1982 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
1983 However, almost all old code should still continue to work without
1984 immediate update, as SB-C-CALL is now a (deprecated) nickname
1986 ** Old operator names in the style DEF-FOO are now deprecated in
1987 favor of new corresponding names DEFINE-FOO, for consistency with
1988 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
1989 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
1990 internal symbols, but a few supported extensions like
1991 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
1992 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
1993 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
1994 etc. as you get deeper into recursive calls to the debugger
1995 command loop, instead of the old "5]", "5]]", "5]]]"
1996 sequence. (I was motivated to do this when squabbles between
1997 ILISP and SBCL left me very deeply nested in the debugger. In the
1998 short term, this change will probably provoke more ILISP/SBCL
1999 squabbles, but hopefully it will be an improvement in the long run.)
2000 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2001 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2002 function, since declaiming return types involving aliens is
2003 (1) annoyingly messy to do by hand and (2) vital to efficient
2004 compilation of code which calls such functions.
2005 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2006 longer reexported by the SB-EXT package. They're solely useful
2007 for alien code, so it seems more logical that you should get
2008 them from the SB-ALIEN package, not in SB-EXT.
2009 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2010 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2011 features. Instead, the code that they used to control is always
2012 built into the system.
2013 * many other bug fixes
2014 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2015 advantage of the new EVAL-WHEN stuff and to clean them up in
2016 general, and they are now more ANSI-compliant in a number of
2017 ways. Martin Atzmueller is responsible for a lot of this.
2018 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2020 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2021 *** correct ERROR type for various file operations
2022 *** some fixes for Lisp streams
2023 *** DEFMETHOD syntax checking
2024 *** changing old weird representation of debug information as
2025 strings (which, among their other deficiencies, don't transform
2026 correctly when you rename packages, and don't change their
2027 print representation when you change things like *PACKAGE*
2028 and *PRINT-LENGTH*) to symbols and lists of symbols
2029 He also made several improvements and fixed several bugs in DESCRIBE.
2030 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2032 *** misbehavior of WRITE-STRING/WRITE-LINE
2033 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2034 and several other LOOP problems as well
2035 *** DIRECTORY when similar filenames are present
2036 *** DEFGENERIC with :METHOD options
2037 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2038 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2039 *** argument ordering in FIND with :TEST option
2040 *** mishandled package designator argument in APROPOS-LIST
2041 *** various problems in the backquote readmacro
2042 *** a bug in APROPOS
2043 *** probably some others that I'm not describing very well here,
2044 since the CVS log documents them by reference to sbcl-devel
2045 messages, and the SourceForge archives aren't working well.:-(
2046 ** Dan Barlow improved the Alpha port (and is making progress on the
2047 PPC port, for those of you who think different).
2048 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2049 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2050 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2051 continues to work on the SPARC port (for those of you in a position
2052 to look down upon our little PC-compatible boxes from a great height).
2053 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2054 and uses that copy, rather than the current dynamic value, when
2055 it's trying to decide whether to truncate output. Thus e.g.
2056 (let ((*print-lines* 50))
2057 (pprint-logical-block (stream nil)
2059 (let ((*print-lines* 8))
2060 (print (aref possiblybigthings i) stream)))))
2061 should now truncate the logical block only at 50 lines, instead of
2062 often truncating it at 8 lines, as it did before.
2063 * The doc/cmucl/ directory, containing old CMU CL documentation
2064 from the time of the fork, is no longer part of the base system.
2065 SourceForge has shut down its anonymous FTP service, and with it
2066 my original plan for distributing the old CMU CL documentation
2067 there. For now, if you need these files you can download an old
2068 SBCL source release and extract them from it.
2069 * The fasl file version number changed again, for dozens of reasons,
2070 some of which are apparent above.
2072 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2073 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2074 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2075 patch, so that DISASSEMBLE output is much nicer.
2076 * The code in the SB-PROFILE package now seems reasonably stable.
2077 I still haven't decided what the final interface should look like
2078 (I'd like PROFILE to interact cleanly with TRACE, since both
2079 facilities use function encapsulation) but if you have a need
2080 for profiling now, you can probably use it successfully with
2081 the current CMU-CL-style interface.
2082 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2083 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2084 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2085 link now returns the dangling link itself, and for similar
2086 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2087 link itself. (In these cases the old code signalled an error and
2088 looped endlessly, respectively.) Thus, DIRECTORY now works even
2089 in the presence of dangling and cyclic symbolic links.
2090 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2091 is now a supported extension again, since the consensus on
2092 sbcl-devel was that it can be useful for ordinary development
2093 work, not just for debugging SBCL itself.
2094 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2095 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2096 that currently-defined functions might be redefined later with
2097 different return types.
2098 * Hash tables can be printed readably, as inspired by CMU CL code
2099 of Eric Marsden and SBCL code of Martin Atzmueller.
2100 * better error handling in CLOS method combination, thanks to
2101 Martin Atzmueller porting Pierre Mai's CMU CL patches
2102 * more overflow fixes for >16Mbyte I/O buffers
2103 * A bug in READ has been fixed, so that now a single Ctrl-D
2104 character suffices to cause end-of-file on character streams.
2105 In particular, now you only need one Ctrl-D at the command
2106 line (not two) to exit SBCL.
2107 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2109 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2110 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2111 does the right thing.
2112 * The implementation of some type tests, especially for CONDITION
2113 types, is now tidier and maybe faster, due to CMU CL code
2114 originally by Douglas Crosher, ported by Martin Atzmueller.
2115 * Some math functions have been fixed, and there are new
2116 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2117 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2118 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2119 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2120 likely to be built into sbcl-0.7.0.)
2121 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2122 CL, ported to SBCL by Martin Atzmueller)
2123 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2124 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2125 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2126 * There's a new slam.sh hack to shorten the edit/compile/debug
2127 cycle for low-level changes to SBCL itself, and a new
2128 :SB-AFTER-XC-CORE target feature to control the generation of
2129 the after-xc.core file needed by slam.sh.
2130 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2131 COMPILE-FILE is no longer supported, so that now every function
2132 gets an entry point, so that block compilation looks a little
2133 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2134 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2135 returns the number of bytes consed since the system started,
2136 rather than the number consed since the first time the function
2137 was called. (The new definition parallels ANSI functions like
2138 CL:GET-INTERNAL-RUN-TIME.)
2139 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2140 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2141 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2142 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2143 * Fasl file version numbers are now independent of the target CPU,
2144 since historically most system changes which required version
2145 number changes have affected all CPUs equally. Similarly,
2146 the byte fasl file version is now equal to the ordinary
2149 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2150 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2151 is no longer recognized. I apologize for this, because it was
2152 listed in SB-EXT as a supported extension, but I found that
2153 its existing behavior was poorly specified, as well as incorrectly
2154 specified, and it looked like too much of a mess to straighten it
2155 out. I have enough on my hands trying to get ANSI stuff to work..
2156 * many patches ported from CMU CL by Martin Atzmueller, with
2157 half a dozen bug fixes in pretty-printing and the debugger, and
2158 half a dozen others elsewhere
2159 * fixed bug 13: Floating point infinities are now supported again.
2160 They might still be a little bit flaky, but thanks to bug reports
2161 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2162 as flaky as they were.
2163 * The --noprogrammer command line option is now supported. (Its
2164 behavior is slightly different in detail from what the old man
2165 page claimed it would do, but it's still appropriate under the
2166 same circumstances that the man page talks about.)
2167 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2168 are now supported, and enabled by default. Thus, the compiler can
2169 handle many floating point and complex operations much less
2170 inefficiently. (Thus e.g. you can implement a complex FFT
2172 * The compiler now detects type mismatches between DECLAIM FTYPE
2173 and DEFUN better, and implements CHECK-TYPE more correctly, and
2174 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2175 thanks to patches from Martin Atzmueller.
2176 * various fixes to make the cross-compiler more portable to
2177 ANSI-conforming-but-different cross-compilation hosts (notably
2178 Lispworks for Windows, following bug reports from Arthur Lemmens)
2179 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2180 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2181 fixed thanks to Pierre Mai's CMU CL patches.
2182 * fixes to keep the system from overflowing internal counters when
2183 it tries to use i/o buffers larger than 16M bytes
2184 * fixed bug 45a: Various internal functions required to support
2185 complex special functions have been merged from CMU CL sources.
2186 (When I was first setting up SBCL, I misunderstood a compile-time
2187 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2188 * improved support for type intersection and union, fixing bug 12
2189 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2190 more obscure bugs as well
2191 * some steps toward byte-compiling non-performance-critical
2192 parts of the system, courtesy of patches from Martin Atzmueller
2193 * Christophe Rhodes has made some debian packages of sbcl at
2194 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2195 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2196 stable, but are nonetheless usable. When he's ready, I'd be happy
2197 to add them to the SourceForge "File Releases" section. (And if
2198 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2199 * new fasl file format version number (because of changes in
2200 internal representation of (OR ..) types to accommodate the new
2201 support for (AND ..) types, among other things)
2203 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2204 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2206 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2208 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2209 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2210 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2211 * DESCRIBE now works on structure objects again.
2212 * Most function call argument type mismatches are now handled as
2213 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2214 know whether the function will be redefined before the call is
2215 executed. (The compiler could flag local calls with full WARNINGs,
2216 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2217 it doesn't keep track of enough information to know whether calls
2218 are local in this sense.)
2219 * Compiler output is now more verbose, with messages truncated
2220 later than before. (There should be some supported way for users
2221 to override the default verbosity, but I haven't decided how to
2222 provide it yet, so this behavior is still controlled by the internal
2223 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2224 src/compiler/ir1util.lisp.)
2225 * Fasl file format version numbers have increased again, because
2226 support for the Gray streams extension changes the layout of the
2227 system's STREAM objects.
2228 * The Gray subclassable streams extension now works, thanks to a
2229 patch from Martin Atzmueller.
2230 * The full LOAD-FOREIGN extension (not just the primitive
2231 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2232 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2233 but like most other programs, it defaults to copying the Unix
2234 environment from the original process instead of starting the
2235 new process in an empty environment.
2236 * Extensions which manipulate the Unix environment now support
2237 an :ENVIRONMENT keyword option which doesn't smash case or
2238 do other bad things. The CMU-CL-style :ENV option is retained
2239 for porting convenience.
2240 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2241 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2243 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2245 * A patch from Martin Atzmueller seems to have solved the SIGINT
2246 problem, and as far as we know, signal-handling now works cleanly.
2247 (If you find any new bugs, please report them!)
2248 * The system no longer defaults Lisp source file names to types
2249 ".l", ".cl", or ".lsp", but only to ".lisp".
2250 * The compiler no longer uses special default file extensions for
2251 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2252 seems to expect a single default extension for all compiled code,
2253 and there's no compelling reason to try to stretch the standard
2254 to allow two different extensions.) Instead, byte-compiled files
2255 default to the same extension as native-compiled files.
2256 * Fasl file format version numbers have increased again, because
2257 a rearrangement of internal implementation packages made some
2258 dumped symbols in old fasl files unreadable in new cores.
2259 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2260 many fewer weird special cases.
2261 * Bug #17 (differing COMPILE-FILE behavior between logical and
2262 physical pathnames) has been fixed, and some related misbehavior too,
2263 thanks to a patch from Martin Atzmueller.
2264 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2265 by Tim Moore, ported to SBCL by Martin Atzmueller.
2266 * Martin Atzmueller fixed several filesystem-related problems,
2267 including bug #36, in part by porting CMU CL patches, which were
2268 written in part by Paul Werkowski.
2269 * More compiler warnings in src/runtime/ are gone, thanks to
2270 more patches from Martin Atzmueller.
2271 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2274 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2276 * DESCRIBE now works on CONDITION objects.
2277 * The debugger now handles errors which arise when trying to print
2278 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2280 * The build system now uses an additional file, customize-target-features.lisp,
2281 to allow local modifications to the target *FEATURES* list. (The point of
2282 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2283 debugging features enabled, without having to worry about propagating it
2284 into everyone's system when I do a "cvs update".) When no
2285 customize-target-features.lisp file exists, the target *FEATURES* list
2286 should be constructed the same way as before.
2287 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2288 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2289 warn about behavior which is undefined under the ANSI spec. Note: This
2290 is specified by ANSI, but it's not very popular with programmers.
2291 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2292 macro in the SBCL sources for an example of a workaround which you
2293 might use to make portable ANSI-standard code which does what you want.
2294 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2295 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2296 when it's used as a non-toplevel form. (This is required in order
2297 to implement the DEFCONSTANT-EQX macro.)
2298 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2299 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2300 funny things, due to bugs in the implementation of EVAL-WHEN
2301 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2302 non-toplevel DEFCONSTANTs are uncommon.)
2303 * The core file version number and fasl file version number have been
2304 incremented, because the old noncompliant DEFCONSTANT behavior involved
2305 calling functions which no longer exist, and because I also took the
2306 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2307 * fixed bug 1 (error handling before read-eval-print loop starts), and
2308 redid debugger restarts and related debugger commands somewhat while
2310 ** The QUIT debugger command is gone, since it did something
2311 rather different than the SB-EXT:QUIT command, and since it never
2312 worked properly outside the main toplevel read/eval/print loop.
2313 Invoking the new TOPLEVEL restart provides the same functionality.
2314 ** The GO debugger command is also gone, since you can just invoke
2315 the CONTINUE restart directly instead.
2316 ** The TOP debugger command is also gone, since it's redundant with the
2317 FRAME 0 command, and since it interfered with abbreviations for the
2319 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2321 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2322 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2323 some porting work by Martin Atzmueller)
2324 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2325 with a boatload of patches which he ported to SBCL. Now that those
2327 ** The system tries to make sure that its low-priority messages
2328 are prefixed by semicolons, to help people who like to use
2329 syntax highlighting in their ILISP buffer. (This patch
2330 was originally due to Raymond Toy.)
2331 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2332 patches originally written by Raymond Toy.
2333 ** The compiler understands coercion between single-value and
2334 multiple-VALUES type expressions better, getting rid of some very
2335 weird behavior, thanks to patches originally by Robert MacLachlan
2336 and Douglas Crosher.
2337 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2338 lambda lists, thanks to a patch originally by Pierre Mai.
2339 ** The system no longer bogusly warns about "abbreviated type
2341 ** The compiler gets less confused by inlining and RETURN-FROM,
2342 thanks to some patches originally by Tim Moore.
2343 ** The system no longer hangs when dumping circular lists to fasl
2344 files, thanks to a patch originally from Douglas Crosher.
2345 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2346 undefined function error.
2347 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2348 in structure slot initforms) for now, documented workaround instead:-|
2349 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2350 * fixed bug 5: added stubs for various Gray stream functions called
2351 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2352 installed, at least appropriate type errors are generated
2353 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2354 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2355 consistently in DEFMETHOD forms.
2356 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2357 it doesn't seem to affect SBCL after all
2358 * The C runtime system now builds with better optimization and many
2359 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2361 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2363 * The system is now under CVS at SourceForge (instead of the
2364 CVS repository on my home machine).
2365 * The new signal handling code has been tweaked to treat register
2366 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2367 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2368 downstream, e.g. in debug-int.lisp, has implicit dependencies
2369 on the unsignedness of integer representation of machine words,
2370 and that caused the system to bomb out with infinite regress
2371 when trying to recover from type errors involving signed values,
2372 e.g. (BUTLAST '(1 2 3) -1).)
2373 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2374 (This was one of the bugs Peter Van Eynde reported back in July.)
2375 * The system now uses code inspired by Colin Walters' O(N)
2376 implementation of MAP (from the cmucl-imp@cons.org mailing
2377 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2378 inline the MAP operation, and there is more than one
2379 sequence argument to the MAP call (so that it can't just
2380 do ETYPECASE once and for all based on the type of the
2381 single sequence argument). (The old non-inline implementation
2382 of the general M-argument sequence-of-length-N case required
2383 O(M*N*N) time when any of the sequence arguments were LISTs.)
2384 * The QUIT :UNIX-CODE keyword argument has been renamed to
2385 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2386 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2388 * Raymond Wiker's patches to port dynamic loading from Linux to
2389 FreeBSD have been added.
2390 * The BUGS file is now more nearly up to date, thanks in large part
2391 to Martin Atzmueller's review of it.
2392 * The debugger now flushes standard output streams before it begins
2393 its output ("debugger invoked" and so forth).
2394 * The core version number and fasl file version number have both
2395 been incremented, because of incompatible changes in the layout
2397 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2398 used to not be called for a saved Lisp image.)
2399 * Martin Atzmueller's version of a patch to fix a compiler crash,
2400 as posted on sbcl-devel 13 September 2000, has been installed.
2401 * Instead of installing Martin Atzmueller's patch for the
2402 compiler transform for SUBSEQ, I deleted the compiler transform,
2403 and transforms for some similar consing operations.
2404 * A bug in signal handling which kept TRACE from working on OpenBSD
2406 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2407 away by constant folding
2408 * The system now defines its address space constants in one place
2409 (in the Lisp sources), and propagates them automatically elsewhere
2410 (through GENESIS and the sbcl.h file). Therefore, patching the
2411 address map is less unnecessarily tedious and error-prone. The
2412 Lisp names of address space constants have also been systematized.
2413 * CVS tags like dollar-Header-dollar have been removed from
2414 the sources, because they have never saved me trouble and
2415 they've been source of trouble working with patches and other
2416 diff-related operations.
2417 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2418 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2420 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2422 * The system has been ported to OpenBSD.
2423 * The system now compiles with a simple "sh make.sh" on the systems
2424 that it's supported on. I.e., now you no longer need to tweak
2425 text in the target-features.lisp-expr and symlinks in src/runtime/
2426 by hand, the make.sh takes care of it for you.
2427 * The system is no longer so grossly inefficient when compiling code
2428 involving vectors implemented as general (not simple) vectors (VECTOR T),
2429 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2430 which dares to use the various sequence functions on non-simple
2431 vectors, takes less of a performance hit.
2432 * There is now a primitive type predicate VECTOR-T-P
2433 to test for the (VECTOR T) type, so that e.g.
2434 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2435 can now be compiled with some semblance of efficiency. (The old code
2436 turned the type declaration into a full call to %TYPEP at runtime!)
2437 * AREF on (VECTOR T) is still not fast, since it's still compiled
2438 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2439 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2440 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2441 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2442 without making full calls to SUBTYPEP at runtime.
2443 (Some analogous efficiency issues for non-simple vectors specialized to
2444 element types other than T, or for non-simple multidimensional arrays,
2445 have not been addressed. They could almost certainly be handled the
2446 same way if anyone is motivated to do so.)
2447 * The changes in array handling break binary compatibility, so
2448 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2449 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2450 instead of (VALUES T T).
2451 * By following the instructions that Dan Barlow posted to sbcl-devel
2452 on 2 July 2000, I was able to enable primitive dynamic object
2453 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2454 functionality is not implemented (since it calls ld to resolve
2455 library references automatically, requiring RUN-PROGRAM for its
2456 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2457 try to resolve library references) is now supported.
2458 * The system now flushes the standard output streams when it terminates,
2459 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2460 them at several other probably-convenient times, e.g. in each pass of
2461 the toplevel read-eval-print loop, and after evaluating a form given
2462 as an "--eval" command-line option. (These changes were motivated by a
2463 discussion of stream flushing issues on cmucl-imp in August 2000.)
2464 * The source transform for TYPEP of array types no longer assumes
2465 that an array whose element type is a not-yet-defined type
2466 is implemented as an array of T, but instead punts, so that the
2467 type will be interpreted at runtime.
2468 * There is now some support for cross-compiling in make.sh: each of
2469 the phases of make.sh has its own script. (This should be transparent
2470 to people doing ordinary, non-cross-compile builds.)
2471 * Since my laptop doesn't have hundreds of megabytes of memory like
2472 my desktop machine, I became more motivated to do some items on
2473 my to-do list in order to reduce the size of the system a little:
2474 ** Arrange for various needed-only-at-cold-init things to be
2475 uninterned after cold init. To support this, those things have
2476 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2477 symbols with such names are now uninterned after cold init).
2478 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2479 fasl files for cold load.
2480 ** Remove the old compiler structure pooling code (which used to
2481 be conditional on the target feature :SB-ALLOC) completely.
2482 ** Redo the representation of some data in cold init to be more compact.
2483 (I also looked into supporting byte compiled code at bootstrap time,
2484 which would probably reduce the size of the system a lot, but that
2485 looked too complicated, so I punted for now.)
2486 * The maximum signal nesting depth in the src/runtime/ support code has
2487 been reduced from 4096 to 256. (I don't know any reason for the very
2488 large old value. If the new smaller value turns out to break something,
2489 I'll probably just bump it back up.)
2490 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2492 * Many, many bugs reported by Peter Van Eynde have been added to
2493 the BUGS list; some have even been fixed.
2494 * While enabling dynamic object file loading, I tried to make the
2495 code easier to understand, renaming various functions and variables
2496 with less ambiguous names, and changing some function calling
2497 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2498 * While trying to figure out how to do the OpenBSD port, I tried to
2499 clean up some of the code in src/runtime/. In particular, I dropped
2500 support for non-POSIX signal handling, added various comments,
2501 tweaked the code to reduce the number of compilation warnings, and
2502 renamed some files to increase consistency.
2503 * To support the new automatic configuration functionality in make.sh,
2504 the source file target-features.lisp-expr has been replaced with the
2505 source file base-target-features.lisp-expr and the machine-generated
2506 file local-target-features.lisp-expr.
2507 * fixed a stupid quoting error in make.sh so that using CMU CL
2508 "lisp -batch" as cross-compilation host works again
2510 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2512 * DESCRIBE no longer tries to call itself recursively to describe
2513 bound/fbound values, so that it no longer fails on symbols which are
2514 bound to themselves (like keywords, T, and NIL).
2515 * DESCRIBE now works on generic functions.
2516 * The printer now prints less-screwed-up representations of closures
2517 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2518 * A private symbol is used instead of the :EMPTY keyword previously
2519 used to mark empty slots in hash tables. Thus
2520 (DEFVAR *HT* (MAKE-HASH-TABLE))
2521 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2522 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2523 now does what ANSI says that it should. (You can still get
2524 similar noncompliant behavior if bang on the hash table
2525 implementation with all the symbols you get back from
2526 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2527 This breaks binary compatibility, since tests for equality to
2528 :EMPTY are wired into things like the macroexpansion of
2529 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2531 * There's now a minimal placeholder implementation for CL:STEP,
2532 as required by ANSI.
2533 * An obscure bug in the interaction of the normal compiler, the byte
2534 compiler, inlining, and structure predicates has been patched
2535 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2536 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2538 * Missing ordinary arguments in a macro call are now detected even
2539 when the macro lambda list contains &KEY or &REST.
2540 * The debugger no longer complains about encountering the top of the
2541 stack when you type "FRAME 0" to explicitly instruct it to go to
2542 the top of the stack. And it now prints the frame you request even
2543 if it's the current frame (instead of saying "You are here.").
2544 * As specified by ANSI, the system now always prints keywords
2545 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2547 * The default initial SIZE of HASH-TABLEs is now smaller.
2548 * Type information from CLOS class dispatch is now propagated
2549 into DEFMETHOD bodies, so that e.g.
2550 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2552 is now basically equivalent to
2553 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2554 (DECLARE (TYPE SINGLE-FLOAT X))
2556 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2557 operation, without having to do run-time type dispatch.
2558 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2559 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2560 * Since I don't seem to be making big file reorganizations very often
2561 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2562 is overflowing my ability to conveniently back them up), I've finally
2563 checked the system into CVS. (The CVS repository is on my home system,
2564 not at SourceForge -- putting it on SourceForge might come later.)
2565 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2566 high-level GC-NOTIFY-FOO functions send their output. (There's
2567 still very little control of where low-level verbose GC functions
2568 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2569 less than it used to -- the GC-NOTIFY-FOO functions are now under
2570 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2571 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2572 in only one place in the source code, and propagates it automatically
2573 everywhere that it's needed. Thus e.g. when I bump the version from
2574 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2575 * The C source files now include boilerplate legalese and documentation
2576 at the head of each file (just as the Lisp source files already did).
2577 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2578 to his page will be replaced with a link to his new CLiki service.
2580 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2582 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2583 * The build process now looks for GNU make under the default name "gmake",
2584 instead of "make" as it used to. If GNU make is not available as "gmake"
2585 on your system, you can change this default behavior by setting the
2586 GNUMAKE environment variable.
2587 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2588 can build without error under CMU CL.
2590 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2592 * There is now a partial SBCL user manual (with some new text and some
2593 text cribbed from the CMU CL manual).
2594 * The beginnings of a profiler have been added (starting with the
2595 CMU CL profiler and simplifying and cleaning up). Eventually the
2596 main interface should be through the TRACE macro, but for now,
2597 it's still accessed through vaguely CMU-CL-style functions and macros
2598 exported from the package SB-PROFILE.
2599 * Some problems left over from porting CMU CL to the new
2600 cross-compilation bootstrap process have been cleaned up:
2601 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2602 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2603 part when getting rid of the compiler *BACKEND* stuff.)
2604 ** The value of *NULL-TYPE* was screwed up, because it was
2605 being initialized before the type system knew the final
2606 definition of the 'NULL type. This screwed up several key
2607 optimizations in the compiler, causing inefficiency in all sorts
2608 of places. (I found it because I wanted to understand why
2609 GET-INTERNAL-RUN-TIME was consing.)
2610 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2611 PROCLAIM FTYPE information. Unfortunately this broke binary
2612 compatibility again, since now the forms output by DEFGENERIC
2613 to refer to functions which didn't exist in 0.6.3.
2614 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2615 can use the new (as of 0.6.3) transform for SOME into MAP into
2617 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2618 values to (INTEGER 0 1000000), so that the time code will no longer
2619 occasionally get blown up by Linux returning 1000000 microseconds
2620 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2621 its output conform to the ANSI spec. (Alas, this makes its output
2622 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2623 * A full call to MAP NIL with a single sequence argument no longer conses.
2624 * fixes to problems pointed out by Martin Atzmueller:
2625 * The manual page no longer talks about multiprocessing as though
2626 it were currently supported.
2627 * The ILISP support patches have been removed from the distribution,
2628 because as of version 5.10.1, ILISP now supports SBCL without us
2629 having to maintain patches.
2630 * added a modified version of Raymond Toy's recent CMU CL patch for
2631 EQUALP comparison of HASH-TABLE
2633 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2635 * The system still can't cross-compile itself with
2636 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2637 entails), but at least it can compile more of itself that way
2638 than it used to be able to, and various buglets which were uncovered
2639 by trying to cross-compile itself that way have now been fixed.
2640 * This release breaks binary compatibility again. This time
2641 at least I've incremented the FASL file format version to 2, so that the
2642 problem can be detected reliably instead of just causing weird errors.
2643 * various new style warnings:
2644 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2645 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2646 ** using the lexical binding of a variable named in the *FOO* style
2647 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2648 as specified by ANSI.
2649 * *RANDOM-STATE* is no longer automatically initialized from
2650 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2651 default behavior of the system is to repeat its behavior every time
2652 it's run. If you'd like to change this behavior, you can always
2653 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2654 old convention there was no comparably easy way to get the system to
2655 repeat its behavior every time it was run.
2656 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2657 FTYPE declarations has been removed, in favor of their ANSI
2658 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2659 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2660 the types of their sequence arguments can be determined at compile time.
2661 This is done through a new open code expansion for MAP which eliminates
2662 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2663 argument types can be determined at compile time.
2664 * The optimizer now transforms COERCE into an identity operation when it
2665 can prove that the coerced object is already of the correct type. (This
2666 can be a win for machine generated code, including the output of other
2667 optimization transforms, such as the MAP transform above.)
2668 * Credit information has been moved from source file headers into CREDITS.
2669 * Source file headers have been made more standard.
2670 * The CASE macro now compiles without complaining even when it has
2673 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2675 * (Note that the way that the PCL macroexpansions were rewritten
2676 to accommodate the change in DEFGENERIC below breaks binary
2677 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2678 not run under sbcl-0.6.2. Once we get out of alpha releases,
2679 i.e. hit release 1.0.0, we'll probably try to maintain binary
2680 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2681 and sbcl-1.4.4. Until then, however, it might be fairly common
2682 for maintenance releases to break binary compatibility.)
2683 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2685 * The reporting of unhandled signals has been changed to print some
2686 explanatory text as well as the report form. (Previously only
2687 the report form was printed.)
2688 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2689 it defines, so that the compiler no longer issues undefined function
2690 warnings for compiled-but-not-yet-loaded generic functions.
2691 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2692 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2693 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2694 section of the ANSI spec are supported.
2695 * The "" nickname for the "KEYWORD" package has been removed.
2696 The reader still handles symbol tokens which begin with a package marker
2697 as keywords, but it doesn't expose its mechanism for doing so in the
2698 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2699 * The system now issues STYLE-WARNINGs for contradictory TYPE
2700 proclamations. (Warnings for contradictory FTYPE proclamations would
2701 be nice too, but those can't be done usefully unless the type system
2702 is made smarter about FUNCTION types.)
2703 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2704 systematized, so that "*target-*.lisp is supposed to exist only on the
2705 target and imply that there's a related file which exists on the
2706 host, and *host-*.lisp is supposed to exist only on the host and imply
2707 that there's a related file which exists on the target. This involves a
2708 lot of renaming. Hopefully the acute confusion caused by the renaming
2709 will be justified by the reduction in chronic confusion..
2710 ** runtime-type.lisp -> early-target-type.lisp
2711 ** target-type.lisp -> late-target-type.lisp
2712 ** early-host-format.lisp -> early-format.lisp
2713 ** late-host-format.lisp -> late-format.lisp
2714 ** host-error.lisp -> misc-error.lisp
2715 ** early-error.lisp -> early-target-error.lisp
2716 ** late-error.lisp -> late-target-error.lisp
2717 ** host-defboot.lisp -> early-defboot.lisp
2718 ** code/misc.lisp -> code/target-misc.lisp
2719 ** code/host-misc.lisp -> code/misc.lisp
2720 ** code/numbers.lisp -> code/target-numbers.lisp
2721 ** code/early-numbers.lisp -> numbers.lisp
2722 ** early-host-type.lisp -> early-type.lisp
2723 ** late-host-type.lisp -> late-type.lisp
2724 ** host-typep.lisp -> typep.lisp
2725 ** load.lisp -> target-load.lisp
2726 ** host-load.lisp -> load.lisp
2727 ** host-disassem.lisp -> disassem.lisp
2728 ** host-insts.lisp -> insts.lisp
2729 ** byte-comp.lisp -> target-byte-comp.lisp
2730 ** host-byte-comp.lisp -> byte-comp.lisp
2731 ** host-signal.lisp -> signal.lisp
2732 ** host-defstruct.lisp -> defstruct.lisp
2733 ** late-target-type.lisp -> deftypes-for-target.lisp
2734 Furthermore, several other previously target-only files foo.lisp (e.g.
2735 hash-table.lisp and random.lisp) have been split into a target-and-host
2736 foo.lisp file and a target-only target-foo.lisp file, with their key type
2737 definitions in the target-and-host part, so that the cross-compiler will
2738 know more about target types.
2739 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2740 gone away. In their place are various *BACKEND-FOO* variables
2741 corresponding to the slots of the old structure.
2742 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2743 into the target SBCL has been fixed.
2744 * The chill.lisp system for loading cold code into a running SBCL
2746 * Support for the CMU CL "scavenger hook" extension has been removed.
2747 (It was undocumented and unused in the CMU CL sources that SBCL was
2748 derived from, and stale in sbcl-0.6.1.)
2749 * Various errors in the cross-compiler type system were detected
2750 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2751 (enabling various consistency checks). Many of them were fixed,
2752 but some hard problems remain, so the compiler is back to
2753 running without *TYPE-SYSTEM-INITIALIZED* for now.
2754 * As part of the cross-compiler type system cleanup, I implemented
2755 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2756 * I have started adding UNCROSS calls throughout the type system
2757 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2758 on cross-compiler input files will be able to go away, and various
2760 * CONSTANTP now returns true for quoted forms (as explicitly required
2763 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2765 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2766 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2767 (MAKE-LIST -1). In the longer term, it ought to become true
2768 that declarations are assertions even at SAFETY 1. For now, it's not
2769 quite true even at SAFETY 3, but it's at least more nearly true..
2770 (Note that this change seems to increases the size of the system by
2771 O(5%) and to decrease the speed of the compiler by 20% or more.)
2772 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2773 for the problem of printing dozens of lines of distracting information
2774 about low-level system machinery as part of the top stack frame
2775 on entry to the debugger when an undefined function was called.
2776 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2777 is not reset to COMMON-LISP-USER.
2778 * Compilation of stuff related to dyncount.lisp has been made conditional
2779 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2780 smaller. The various dyncount-related symbols have been moved into
2781 a new "SB-DYNCOUNT" package.
2782 * tty-inspect.lisp has been renamed to inspect.lisp.
2783 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2784 feature has gone away. (When we eventually port to other flavors of
2785 libc and/or Unix, we'll try to make the differences between flavors
2786 invisible at the user level.)
2787 * Various other *FEATURES* tags, and/or their associated conditionals,
2788 have been removed if obsolescent, or given better documentation, or
2789 sometimes given more-mnemonic names.
2791 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2793 * tidied up "make.sh" script
2794 * tidied up system directory structure
2795 * better "clean.sh" behavior
2796 * added doc/FOR-CMUCL-DEVELOPERS
2797 * many many small tweaks to output format, e.g. removing possibly-confusing
2798 trailing #\. character in DESCRIBE-INSTANCE
2799 * (EQUALP #\A 'A) no longer signals an error.
2800 * new hashing code, including EQUALP hashing
2801 * tidied up Lisp initialization and toplevel
2802 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2803 * command line argument processing
2804 * added POSIX-GETENV function to deal with Unix-ish environment variables
2805 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2806 terminating SBCL on EOF
2807 * non-verbose GC by default
2808 * There is no more "sbcl" shell script; the sbcl file is now the C
2809 runtime executable (just like CMU CL).
2810 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2812 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2813 symbol and package information as Lisp native symbol and package objects
2814 instead of strings naming symbols and strings naming packages. This way,
2815 whenever packages are renamed (as in warm init), debug information is
2816 transformed along with everything else.
2817 * tweaked the optimization policy declarations which control the building
2818 of SBCL itself. Now, among other things, the system no longer saves
2819 source location debugging information. (This helps two problems at once
2820 by reducing SBCL size and by keeping SBCL from trying to look for its
2821 sources -- which may not exist -- when reporting errors.)
2822 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2823 debugging and testing purposes
2824 * cleaned up printing, making the printer call PRINT-OBJECT for
2825 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2826 methods, giving nearly-ANSI behavior
2827 * converted almost all special variables to use *FOO* naming convention
2828 * deleted PARSE-TIME functionality, since it can be done portably
2829 * moved some files out of cold init into warm init
2830 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2832 * regularized formatting of source files
2833 * added an install.sh script
2834 * fixed ridiculous memory usage of cross-compiler by making
2835 compiler/alloc.lisp not try to do pooling unless it can hook
2836 itself into the GC of the cross-compilation host. Now the system
2837 builds nicely on my old laptop.
2838 * added :SB-ALLOC in target-features.lisp-expr
2839 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2840 was not implemented)
2841 * re-did condition handling and note reporting in the compiler. Notes
2842 are no longer handled by signalling conditions. Style warnings
2843 and warnings are handled more correctly and reported in such a way
2844 that it's easy to find one or the other in your output (so that you
2845 can e.g. figure out which of many problems caused COMPILE-FILE to
2847 * changed the severity of several compiler warnings from full WARNING
2848 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2849 compiler note reporting so that it doesn't use the condition system
2850 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2851 COMPILE-FILE command)
2852 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2853 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2854 no longer supported, since their implementation was deeply intertwingled
2855 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2856 * removed bogus "support" for compiler macros named (SETF FOO), and
2857 removed the compiler macro for SETF INFO (but only after making a fool
2858 of myself on the cmucl-imp mailing list by posting a bogus patch for
2859 DEFINE-COMPILER-MACRO..)
2860 * Compiled files containing forms which have side effects on the Lisp
2861 reader (such as DEFPACKAGE forms) are now handled more correctly.
2862 (Compiler queuing of top level lambdas has been suppressed by setting
2863 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2864 * deleted various currently-unused source files, e.g. gengc.lisp. They
2865 may be added back at some point e.g. when porting to other architectures,
2866 but until they are it's distracting to distribute them and to try to
2868 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2869 there were so many of them they're just distractions, and UNCROSS is
2870 known to be able to handle the current sources
2871 * moved PROFILE functionality into TRACE, so that it will be clear
2872 how the wrapping and unwrapping of functions when you profile them
2873 interacts with the wrapping and unwrapping of functions when you
2874 trace them. (Actually, the functionality isn't there yet, but at least
2875 the interface specification is there. Hopefully, the functionality will
2876 arrive with some future maintenance release.)
2877 * removed host-oops.lisp
2878 * changed signature of QUIT function to allow UNIX-CODE argument
2879 * fixed READ-SEQUENCE bug
2880 * tweaked verbose GC output so that it looks more like the progress
2881 output that ANSI specifies for functions like LOAD
2882 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2883 * added <http://sbcl.sourceforge.com> to the banner information printed by