1 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
3 platform. The port supports 61-bit fixnums, large memory spaces
4 and reloading shared object files.
5 * enhancement: saving cores with foreign code loaded is now
6 supported on ppc/Darwin in addition to the previously supported
8 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
10 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
11 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
12 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
13 produces an error. (thanks to Vincent Arkesteijn)
14 * bug fix: NAMESTRING on pathnames with :WILD components in their
15 directories works correctly. (thanks to Artem V. Andreev)
16 * fixed bug 125: compiler preserves identity of closures. (reported
18 * build fix: fixed the dependence on *LOAD-PATHNAME* and
19 *COMPILE-FILE-PATHNAME* being absolute pathnames.
20 * on x86 compiler partially supports stack allocation of dynamic-extent
22 * fixed some bugs related to Unicode integration:
23 ** encoding and decoding errors are now much more robustly
24 handled; it should now be possible to recover even from invalid
25 input or output to the terminal. (thanks to Teemu Kalvas)
26 * fixed some bugs revealed by Paul Dietz' test suite:
27 ** the FORMATTER-generated functions for ~V[ conditionals require
28 the correct number of arguments.
30 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
31 * new feature: reloading changed shared object files with
32 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
33 * new feature: references to foreign variables and functions
34 can now be compiled and loaded before the corresponding shared
35 object file is loaded, as long as the foreign definitions are
37 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
38 supported. (thanks to Dan Debertin)
39 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
40 just keywords. (thanks to Vincent Arkesteijn)
41 * fixed bug #331: structure-class instances corresponding to
42 DEFSTRUCT forms are now created eagerly.
43 * fixed bug #345: backtraces from calls to undefined functions work
44 on x86 as well. Related bug #61 is now also partially fixed on x86
45 (backtraces from throws to unknown catch tags.)
46 * bug fix: lambda-list parsing is now stricter vrt. order and number
47 of lambda-list keywords.
48 * bug fix: as specified by AMOP, an error is signalled if a
49 class-option appears multiple times in a DEFCLASS form. (reported
51 * bug fix: RANDOM can be compiled when the compiler derives the type
52 of its numeric argument as a disjoint set of small integers.
53 (reported by Paul Dietz)
54 * bug fix: starting a core saved with shared objects loaded when
55 those objects are not available no longer causes threaded SBCL to
56 hang. (reported by Sean Ross)
57 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
59 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
60 argument, which must always be NIL. (reported by Kalle Niemitalo)
61 * bug fix: printing 1.0d+23 no longer results in an error.
62 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
63 * bug fix: structure slot setters preserve evaluation order. (thanks
65 * fixed some bugs related to Unicode integration:
66 ** RUN-PROGRAM can allow its child to take input from a Lisp
67 stream. (reported by Stefan Scholl)
68 ** ASDF-INSTALL successfully downloads .tar.gz files in all
69 locales. (reported by Ken Causey)
70 * fixed some bugs revealed by Paul Dietz' test suite:
71 ** INCF, DECF and REMF evaluate their place form as specified in
73 ** FORMATTER expands ~{ ~} iteration directives with V or #
75 ** FORMATTER deals with the ~@[ ~] conditional directive where the
76 consequent uses no arguments correctly.
77 ** the system has a partial understanding of the (COMPLEX RATIO)
79 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
82 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
83 * new feature: a build-time option (controlled by the :SB-UNICODE
84 keyword feature, enabled by default) for building the system with
85 support for the entire 21-bit character space defined by the
87 * new feature: the system now has rudimentary external-format
88 support; the primary user-visible change at this time is that
89 characters with the high bit set (such as lower-case-e-acute) will
90 print correctly to a terminal in a UTF-8 environment.
91 * minor incompatible change: BASE-CHAR no longer names a class;
92 however, CHARACTER continues to do so, as required by ANSI.
93 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
94 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
95 has been moved to the SB-EXT package (temporarily re-exported from
97 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
98 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
100 * fixed bug #351: better error-handlind and reporting for malformed
102 * fixed bug #350: bignum-printing is now more memory-efficient,
103 allowing printing of very large bignums, eg. (expt 2 10000000).
104 (reported by Bruno Haible)
105 * fixed bug #302: better primitive-type selection for intersection
107 * fixed bug #308: non-graphic characters now all have names, as
108 required. (reported by Bruno Haible)
109 * bug fix: redefining a class with different superclasses now correctly
110 removes it from the direct-subclasses of its previous superclasses.
111 (reported by David Morse)
112 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
113 new value now works. (reported by Bruno Haible)
114 * bug fix: correct canonicalization of multiple non-standard slot
115 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
116 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
117 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
118 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
119 now exists, an signals an error.
120 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
121 during macroexpansion for non-iteration variables as well. (reported
122 by Bruno Haible for CMUCL)
123 * bug fix: Cyclic structures and unprintable objects in compiler
124 messages no longer cause errors. (reported by Bruno Haible)
125 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
126 and READ-FROM-STRING all now return a primary value of NIL if
127 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
128 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
129 (reported by Bruno Haible for CMUCL)
130 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
131 do not designate a condition. (reported by Bruno Haible for
133 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
134 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
135 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
136 permitting resolution in favour of any of the conflicting symbols.
137 (reported by Bruno Haible for CMUCL)
138 * bug fix: EQUAL compiler optimizations is less aggressive on
139 strings which can potentially compare true despite having distinct
140 specialized array element types.
141 * bug fix: unit enumerations can be defined without dividing by
142 zero. (thanks to Vincent Arkesteijn)
143 * FORMAT compile-time argument count checking has been enhanced.
144 (report from Bruno Haible for CMUCL)
145 * a partial workaround for the bug 262: the compiler does not try to
146 inline-expand a local function doing RETURN-FROM from a deleted
147 BLOCK. (thanks to Peter Denno for the bug report and to David
148 Wragg for the simple test case)
149 * fixed some bugs revealed by Paul Dietz' test suite:
150 ** RENAME-PACKAGE allows all package designators as new package
152 ** constraint propagation and lambda variable substitution are
153 more cautious in dealing with partially deleted code.
154 ** compiler sometimes lost reoptimization passes.
155 ** CERROR, given a condition as condition designator, passes its
156 remaining arguments to the continue format control without
158 ** Case-altering FORMAT directives work correctly on non-ASCII
160 ** The REAL type specifier handles bounds outside the floating
161 point ranges without signalling FLOATING-POINT-OVERFLOW.
162 ** Functions with IR1-transformations can create intercomponent
163 references to global functions.
164 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
166 ** FORMAT ~R treats a nil value for its first parameter correctly.
168 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
169 * enhancement: saving cores with foreign code loaded is now
170 supported on x86/NetBSD and sparc/Linux in addition to the previously
172 * bug fix: on some platforms repeated installations caused multiple
173 copies of HTML documentation to be installed -- should not happen
174 any more. (reported by Stefan Scholl)
175 * bug fix: parsing self-recursive alien record types multiple times
176 no longer causes infinite recursion. (reported by Thomas F. Burdick,
177 original patch by Helmut Eller for CMUCL)
178 * bug fix: stack-exhaustion detection works now on NetBSD as well.
179 (thanks to Richard Kreuter)
180 * bug fix: defining classes whose accessors are methods on existing
181 generic functions in other (locked) packages no longer signals
182 bogus package lock violations. (reported by François-René Rideau)
183 * bug fix: special variables as DEFMETHOD parameters no longer have
184 associated bogus type declarations. (reported by David Wragg and
186 * bug fix: read-write consistency on streams of element-type
187 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
188 * bug fix: redefiniton of the only method of a generic function with
189 no DEFGENERIC no longer emits a full WARNING. In addition,
190 redefinition of generic functions with no DEFGENERIC to an
191 incompatible lambda list now signals an error. (thanks to Zach
193 * bug fix: DEFGENERIC now works even when there's a function of the
194 same name in an enclosing lexical environment. (thanks to Zach
196 * fixed compiler failure, caused by instrumenting code during
197 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
198 * optimization: added loop analysis and improved register allocation
199 * fixed some bugs revealed by Paul Dietz' test suite:
200 ** POSITION on displaced vectors with non-zero displacement
201 returns the right answer.
202 ** (SIMPLE-STRING) is a valid type specifier for sequence
204 ** *PRINT-LEVEL* handling for slotless structures is pedantically
206 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
207 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
209 ** The justification version of the FORMAT ~< directive treats
210 non-zero minpad parameter correctly.
212 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
213 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
214 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
215 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
216 the supported interface.
217 * new feature: Single-stepping of code compiled with DEBUG 2 or
218 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
219 * new feature: saving cores with foreign code loaded is now
220 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
221 Timothy Moore's work for CMUCL)
222 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
223 parameters to * instead of NIL if no initform is supplied.
224 (reported by Johan Bockgaard)
225 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
226 correctly when FUNCALL appears as the car of the form. Note:
227 despite this FUNCALL forms are not currently subject to
228 compiler-macro expansion. (port of Raymond Toy's fix for the
229 same from CMUCL, reported by Johan Bockgaard)
230 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
231 (thanks for Teemu Kalvas)
232 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
233 case instead of terminating the iteration (thanks for Julian
234 Squires, Sean Champ and Raymond Toy)
235 * bug fix: incorrect expansion of defgeneric that caused a style
236 warning. (thanks for Zach Beane)
237 * bug fix: slot accessor effective method computation works properly
238 for classes with multiple non-standard applicable methods on
239 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
240 * on x86 compiler supports stack allocation of results of LIST and
241 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
242 CMUCL implementation by Gerd Moellmann)
243 * fixed some bugs revealed by Paul Dietz' test suite:
244 ** FORMAT strings with both the ~<~:;~> form of the justification
245 directive and pretty-printing directives cause an error.
247 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
248 * incompatible change: the internal functions
249 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
250 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
251 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
252 instead of the old functions.
253 * new feature: on platforms where "dladdr" is available foreign
254 function names now appear in backtraces. (based on Helmut Eller's
256 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
257 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
259 * documentation: ASDF manual is now distributed with SBCL.
260 * bug fix: SBCL can now load its contributed modules with REQUIRE
261 even if the system-provided entries have been removed from
262 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
264 * bug fix: backtraces involving undefined functions or assembly
265 routines are more informative. (thanks to Brian Downing)
266 * bug fix: mutually referent alien structures now work correctly.
267 (reported by Rick Taube)
268 * bug fix: structures defined by WITH-ALIEN can be referred to
269 within other definitions in the same WITH-ALIEN.
270 * bug fix: division operators (MOD, TRUNCATE and the like) with
271 constant zero divisors and integer dividends no longer generate
273 * bug fix: provide default methods for INPUT-STREAM-P and
274 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
275 * bug fix: improve the detection and resolution of MOP metacycles
276 (where effective method computation depends on the generic function
277 having its effective method computed). (reported by Bruno Haible)
278 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
279 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
280 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
281 * bug fix: pathnames with relative directory components can be
282 represented relative to default pathnames.
283 * optimization: in taking the GCD of bignums, reduce the two bignums
284 to approximately the same size (using Euclid's algorithm) before
285 applying the more sophisticated binary GCD. (thanks to Juho
287 * optimization: COUNT on bitvectors now operates word-at-a-time.
288 * optimization: ASH with a positive, but not necessarily constant,
289 (leftwards) shift, when in a modular context, is compiled to a
291 * fixed some bugs revealed by Paul Dietz' test suite:
292 ** FORMAT variable parameters ("~V<char>") are defaulted properly
293 if the corresponding argument is NIL.
294 ** FORMAT directives accepting a minpad parameter treat negative
296 ** literal commas as character parameters to FORMAT directives are
298 ** literal spaces directly after ~<Newline> directives within a
299 format-logical-block (~:< ~@:>) do not induce :FILL-style
300 conditional newlines.
301 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
302 as specified: it no longer includes conditional newlines).
303 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
305 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
306 :PER-LINE-PREFIX argument does not evaluate to a string.
308 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
309 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
310 the manual for details; for now, package locks can be disabled by
311 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
312 if no major problems are found then it is likely that they will be
313 compiled in unconditionally.
314 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
315 unsupported operators on all platforms. To load a shared library
316 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
317 object file, link it into a shared library outside of SBCL and
318 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
320 * bug workaround: ROOM T and the GC were not working together
321 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
322 reporting facility aren't preserved. That reporting has been
323 disabled, so now until and unless someone figures out how to make it
324 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
325 * minor incompatible change: as threatened around sbcl-0.8.0, the
326 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
327 The symbols are also exported from SB-PCL for backwards
328 compatibility, but more so than before SB-PCL should be treated as
329 an implementation-internal package.
330 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
332 * the SB-SPROF contrib now works on (most) non-x86 architectures.
333 It is known as of this release not to work on the Alpha, however.
334 * fixed bug #167: errors signalled due to illegal syntax in method
335 bodies are now more legible.
336 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
337 designators and can hence be used with TYPEP.
338 * fixed bug #333: CHECK-TYPE now ensures that the type error
339 signalled, if any, has the right object to be accessed by
340 TYPE-ERROR-DATUM. (reported by Tony Martinez)
341 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
342 5.1.2.3 for multiple-value place subforms. (reported by Kalle
344 * fixed bug #334: programmatic addition of slots using specialized
345 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
346 and :ALLOCATION :CLASS slots.
347 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
348 float exponents. (rereported by Peter Seibel)
349 * fixed a bug: #\Space (and other whitespace characters) are no
350 longer considered to be macro characters in standard syntax by
352 * fixed bug: initialization of condition class metaobjects no longer
353 causes an instance of the condition to be created. (reported by
355 * fixed bug: it is now possible to have slots such that
356 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
357 description is neither :INSTANCE nor :CLASS.
358 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
359 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
360 SLOT-VALUE-USING-CLASS), and no longer causes errors with
361 non-standard SLOT-DEFINITION-ALLOCATION values.
362 * fixed bugs: various race conditions handling exiting threads.
363 CL-PPCRE's thread torture test now passes.
364 * fixed arguably-a-bug: GC time proportional to number of threads in
365 system even when most of them are idle
366 * optimization: improved performance of BIT and SBIT on bit-vectors.
367 * .fasl file incompatibility: The fasl file version number has
368 been incremented because of changes associated with package locks.
370 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
371 * minor incompatible change: the system no longer provides
372 optimization hints (conditions which are TYPEP
373 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
374 mode; these hints are still emitted when higher SPEED optimization
376 * new contrib module: a sampling profiler (profiling by statistical
377 sampling, rather than by instrumenting functions) is available as
378 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
379 * the behaviour of the standard function ED is now customizeable by
380 third parties through a hook variable: see ED's documentation
381 string for information on the protocol.
382 * the compiler no longer emits efficiency notes for (FUNCALL X)
383 when the type of X is uncertain under default optimization
385 * fixed bug 276: mutating a binding of a specialized parameter to a
386 method to something that is not TYPEP the specializer is now
388 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
389 exhibit the required behaviour on the x86 platform. (thanks to
390 Peter van Eynde, Eric Marsden and Bruno Haible)
391 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
392 even for difficult arguments. (reported by Peter Graves)
393 * fixed bug 141a: the backquote printer now descends quoted
395 * fixed another bug in backquote printing: no more destructive
396 modification of the form's list structure. (reported by Brian
398 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
399 move between its address being taken and the call to
400 interrupt_thread, fixing a crashing race condition.
401 * the SB-POSIX contrib implementation has been adjusted so that it
402 no longer exhibits ridiculously poor performance when constructing
403 instances corresponding to C structs.
405 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
406 * minor incompatible change: the sb-grovel contrib now treats C
407 structures as alien (in the sense of SB-ALIEN) objects rather than
408 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
409 has implications for memory management of client code
410 (sb-grovel-returned objects must now be manually managed) and for
411 type safety (alien objects now have full types).
412 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
413 used to control emission of compiler diagnostics, rather than the
414 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
415 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
416 quality should be considered deprecated.
417 * install.sh now installs the user manual as well
418 * (not quite a new documentable feature, but worth considering in
419 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
420 beginnings of a semantically meaningful condition hierarchy is
421 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
423 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
424 designator as the defaults argument.
425 * fixed bug: Displaced arrays whose displaced-to array has become
426 too small now cause ARRAY-DIMENSION to signal an error, providing
427 for safer bounds-checking. (reported by Bruno Haible)
428 * fixed bug: DEFCLASS slot definitions with identical :READER and
429 :WRITER names now signal a reasonable error. (reported by Thomas
431 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
433 * fixed bug: Invalid slot specification errors now print correctly.
434 (thanks to Zach Beane)
435 * fixed bug 320: Shared to local slot value transfers in class
436 redefinitions now happen corrently from superclasses as
437 well. (reported by Bruno Haible)
438 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
440 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
441 improper lists correctly. (reported by Bruno Haible)
442 * fixed bug 313: source-transform for <fun-name> was erroneously
443 applied to a call of a value of a variable with name <fun-name>.
444 (reported by Antonio Menezes Leitao)
445 * fixed bug 307: The obsolete instance protocol ensures that
446 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
447 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
448 remove dead unknown-values globs from the middle of the stack.
449 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
450 easier to resume long computations after using *BREAK-ON-SIGNALS*
451 to diagnose and fix failures. (thanks to Nikodemus Siivola)
452 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
453 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
454 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
455 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
456 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
458 * fixed a (fairly theoretical) bug in string printing: if
459 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
460 does not have array-element-type equal to the most general string
462 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
463 function name. (thanks to Nikodemus Siivola)
464 * fixed bug: compile-time format string checker failed on
465 non-closed ~{. (reported by Thomas F Burdick)
466 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
467 don't warn on legal loop code involving more than one aggregate
468 boolean. (thanks to Nikodemus Siivola)
469 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
470 in keyword position to MAKE-PACKAGE was not being reported as
472 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
473 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
475 * fixed bugs 280 and 312: the checking for multiple definitions in a
476 file is less likely to become confused by uses of inline
478 * fixed bug: the #S reader macro performs the keyword coercion
479 specified for slot names. (reported by Kalle Niemitalo)
480 * fixed bug: lambda lists may contain symbols whose names start with
481 & but are not lambda-list-keywords; their occurrence triggers a
483 * fixed bug 321: define-method-combination argument lambda lists do
484 not cause invalid code to be generated when &optional or &aux
485 variables have default values. (reported by Bruno Haible)
486 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
487 have CLOS classes; this appears also to have fixed reported
488 instability in stack exhaustion detection.
489 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
490 MUFFLE-WARNING when no associated restart is present is now
492 * optimization: rearranged the expansion of various defining macros
493 so that each expands into only one top-level form in a
494 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
496 * optimization: used a previously little-used slot in symbols to
497 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
499 * fixed some bugs revealed by Paul Dietz' test suite:
500 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
502 ** Fixed an optimization bug related to inheritance of initforms
503 from local to shared slots.
504 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
505 ** CAST optimizer forgot to flush argument derived type.
506 ** print/read consistency on floats is now orders of magnitude
507 more likely. (thanks also to Bruno Haible for a similar report
509 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
510 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
511 for objects of type REAL. Make it so.
512 ** SXHASH is specified (infelicitously) to respect similarity,
513 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
514 it so. (thanks to Markus Ziegler)
515 ** on the Alpha, the compiler succeeds in compiling functions
516 returning a known number of arguments greater than 63.
517 ** fixed handling of invalid NIL arguments in keyword position
519 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
521 ** provided a readably-printable representation for RANDOM-STATE
523 ** ensured that pathnames, where they have a namestring, always
524 print using #P"..." syntax.
526 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
527 * Support for the forthcoming 2.0 version of the NetBSD kernel
528 running on x86 hardware has been added. (thanks to Perry
529 E. Metzger most immediately, and others for their past work)
530 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
531 current release version; SBCL's OpenBSD support had been broken
532 since about the time of OpenBSD's switch to ELF binary format.)
533 * [placeholder for DX summary]
534 ** user code with &REST lists declared dynamic-extent, under high
535 speed or space and low safety and debug optimization policy.
536 * The manual has been converted to Texinfo format and the debugger
537 chapter from the cmucl manual has been added.
538 * A facility has been added to extract documentation strings from
539 sbcl and store them as Texinfo-formatted snippets for inclusion in
540 the manual (via Texinfo's @include directive)
541 * bug fix: compiler emitted division in optimized DEREF. (thanks for
542 the test case to Dave Roberts)
543 * bug fix: multidimensional simple arrays loaded from FASLs had fill
544 pointers. (reported by Sean Ross)
545 * bug fix: PROFILE output is printed nicely even for large numerical
546 values. (thanks to Zach Beane)
547 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
548 greater than 32 handle EOF correctly.
549 * bug fix: on X86 an immediate argument of the IMUL instruction is
550 correctly printed in disassembly. (thanks to Lutz Euler)
551 * bug fix: class slots in redefined classes preserve their old
552 values. (thanks to Bruno Haible and Nikodemus Siivola)
553 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
554 succeeds. (reported by Marco Baringer)
555 * fixed some bugs revealed by Paul Dietz' test suite:
556 ** READ-SEQUENCE now works on ECHO-STREAMs.
557 ** RATIONALIZE works more according to its specification. (thanks
560 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
561 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
562 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
563 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
564 should matter to you only if you rebind the printer control
565 variables and then find you want different bindings in the
566 debugger than in the ordinary execution of your program.)
567 * The runtime build system has been tweaked to support building
568 (on SPARC/SunOS) using a C compiler which invokes Sun's own
569 assembler and linker. (thanks to Nikodemus Siivola)
570 * Unbound, undefined, undeclared variables now trigger full
571 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
572 more often programmer error than deliberate exploitation of undefined
574 * optimization: the hash algorithm for strings has changed to one
575 that is less vulnerable to spurious collisions. (thanks to Juho
577 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
578 needless bounds checking. (thanks to Juho Snellman)
579 * optimization: implemented multiplication as a modular
580 (UNSIGNED-BYTE 32) operation on the PPC backend.
581 * fixed some bugs revealed by Paul Dietz' test suite:
582 ** ADJUST-ARRAY now copies the datum in a zero rank array if
584 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
585 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
586 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
588 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
590 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
592 ** ECHO-STREAMs no longer attempt to echo the end of file value to
593 their output stream on EOF from read.
594 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
595 have been read to end-of-file.
596 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
598 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
599 description of determination of which consecutive characters
601 ** Printing the "Space" character with escaping on now yields "#\\ ",
602 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
603 ** Reading floating-point numbers with *READ-BASE* set to a number
604 less than 10 works correctly.
605 ** Reading floating-point numbers with *READ-BASE* set to a number
606 more than 10 works correctly.
607 ** Printing with *PRINT-READABLY* targets the standard readtable, not
608 the readtable currently in effect.
610 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
611 * minor incompatible change: parsing of namestrings on a physical
612 (Unix) host has changed; numbers after the final #\. in a
613 namestring are no longer interpreted as a version field. This is
614 intented to be largely invisible to the user, except that the
615 meaning of the namestring "*.*.*" has changed: it now refers to a
616 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
617 should usually be replaced by
618 (make-pathname :name :wild :type :wild :version :wild)
619 with the added benefit that this is more likely to be portable.
620 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
621 option to OPEN now signals an error if the file being opened
622 exists; this may have an impact on existing code.
623 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
624 signals received, on the PowerPC platforms (both Linux and
625 Darwin). (thanks to Pierre Mai for pointing out the location of
627 * several fixes on OS X: The system now builds and runs cleanly on
628 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
629 (thanks to Brian Mastenbrook)
630 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
631 second-resolution: integer multiples of 1/3600 between -24 and 24.
632 (thanks to Vincent Arkesteijn)
633 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
634 type when called with 1 argument; PEEK-CHAR checked type of
635 PEEK-TYPE only after having read first character from a
636 stream. (reported by Peter Graves)
637 * bug fix: the garbage collector now has much better locality
638 behaviour, and in particular no longer treats all memory as being
639 exhausted when many small objects point to each other in a deeply
641 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
642 recognized as being TYPEP their class.
643 * bug fix: the PUSHNEW documentation string has been corrected.
644 (thanks to Vincent Arkesteijn)
645 * bug fix: defaulting of the value for the last of an atypically
646 large number of multiple values being bound was not being
647 performed correctly on the Alpha or PPC platforms
648 * optimization: implemented multiplication as a modular
649 (UNSIGNED-BYTE 32) operation on the x86 backend.
650 * optimization: SEARCH on simple-base-strings can now be open-coded.
651 (see also contrib/compiler-extras.lisp for inspiration for
652 teaching the compiler about the Boyer-Moore algorithm).
653 * value, returned by MAX (and MIN) called with several EQUALP, but
654 not EQL, arguments now does not depend on compiler settings.
655 * fixed some bugs revealed by Paul Dietz' test suite:
656 ** in stack analysis liveness information is propagated from
657 non-local entry points.
658 ** pathwise CAST removing failed when the CAST node did not start
660 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
661 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
663 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
664 host is already defined.
665 ** RENAME-FILE works on streams instead of signalling an internal
667 ** PEEK-CHAR uses the current readtable when determining whether
668 or not a character is whitespace.
669 ** MERGE-PATHNAMES handles the case when the pathname does not
670 specify a name while the default-pathname specifies a version.
671 ** Pathnames now stand a better chance of respecting print/read
673 ** Attempting to use standardized file system operators with a
674 pathname with invalid :DIRECTORY components signals a
676 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
678 ** DIRECTORY on logical pathnames is more correct.
679 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
680 signal a TYPE-ERROR if their argument is not a stream.
681 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
682 designator argument does not designate a stream.
683 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
684 examining the synonym.
685 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
687 ** OPEN and WITH-OPEN-STREAM allow opening streams with
688 element-type larger than ([UN]SIGNED-BYTE 32).
690 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
691 * When built with the :SB-FUTEX feature, threaded builds now take
692 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
693 for faster/more reliable mutex and condition variable support.
694 * Incompatible change (but one you probably shouldn't have been using
695 anyway): the interface and code for arbitrating between multiple
696 threads in the same user session has been redesigned.
697 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
698 error when called without an explicit environment argument.
699 (thanks to Nikodemus Siivola)
700 * bug fix: buffered :DIRECTION :IO streams are less likely to become
701 confused about their position. (thanks to Adam Warner and Gerd
703 * bug fix: Pretty printing backquoted forms with unquotations in the
704 argument list position of various code constructs such as LAMBDA
705 now works correctly. (reported by Paul Dietz)
706 * bug fix: Pretty printing unquotations no longer loses all
707 stream position information.
708 * optimization: performance of string output streams is now less
709 poor for multiple small sequence writes.
710 * optimization: performance of CSUBTYPEP in the presence of complex
711 expressions involving CONS and NOT many times has been improved.
712 (reported by Paul Dietz)
713 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
715 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
716 streams for the REPL, for the debugger, and so on. (thanks to
718 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
719 * fixed some bugs revealed by Paul Dietz' test suite:
720 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
721 is ignored if it is too small, rather than propagating through
722 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
723 ** extremely complex negations of CONS types were not being
724 sufficiently canonicalized, leading to inconsistencies in
726 ** VALUES tranformer lost derived type.
728 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
729 * fixed a bootstrapping bug: the build process no longer assumes
730 that the various BOOLE-related constants have the same value in
731 host and target lisps. (noted by Paul Dietz' test suite on an
732 SBCL binary built from CLISP)
733 * The system can now be dynamically linked on the MIPS platform,
734 which enables dynamic loading of foreign code from Lisp. (thanks
735 to Ralf Baechle for discussions on the MIPS ABI)
736 * The system now records debugging information for its own source
737 files in a filesystem-position-independent manner, relative to
738 the "SYS" logical host.
739 * fixed a compiler bug: MV-LET convertion did not check references
740 to the "max args" entry point. (reported by Brian Downing)
741 * tweaked disassembly notes to be less confident about proclaiming
742 some instruction as an LRA. (thanks to Brian Downing)
743 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
744 now each have their own history, command character, and other
745 characteristics. (thanks to David Lichteblau)
746 * fixed some bugs revealed by Paul Dietz' test suite:
747 ** compiler failure in compiling LOGAND expressions including a
749 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
750 shift greater than 32.
751 ** FLUSH-DEST did not mark blocks for type check regeneration.
752 ** HANDLER-CASE failed to accept declarations in handler clauses
753 in some circumstances.
755 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
756 * New code in contrib/sb-introspect (still probably not entirely
757 stable yet) provides some support for smart Lisp development
758 environments like SLIME.
759 * The conditions signalled for errors occurring when loading .fasl
760 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
761 in a way which should help ASDF recover gracefully.
762 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
763 stuff has been cleaned up. If you code contrib/ stuff, this might
764 affect you, and you can look at contrib/README, contrib/STANDARDS,
765 and/or the 0.8.4.27 diff to check.
766 * In full calls the compiler now does not generate checks for declared
767 argument types for all arguments.
768 * various threading fixes
769 ** and some experimental patches which didn't make it into
770 the main tree for this release, but which are shipped in
771 contrib/experimental-thread.patch as a possible fix for some
772 failures (deadlock, spinning...) in GC-intensive multithreaded
774 * fixed PPC build problem (source code incompatibility of different
775 library versions): added offsetof() hackery which attempts to divine
776 where glibc maintainers put uc_mcontext today
777 * fixed bug 282: compiler does not trust type assertions while passing
778 arguments to a full call.
779 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
780 &OPTIONAL VALUES type parameter.
781 * fix bug 214: algorithm for noting rejected templates is now more
782 similar to that of template seletion. (also reported by rydis on
784 * fixed bug 141b: printing backquoted information readably and prettily
785 inserts a space where necessary.
786 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
787 (CHARACTER 10) are now reported as errors, rather than propagated
788 as unknown types. (reported by piso on #lisp)
789 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
790 with values NIL and :ERROR. (thanks to Milan Zamazal)
791 * fixed bug 191c: CLOS now does proper keyword argument checking as
792 described in CLHS 7.6.5 and 7.6.5.1.
793 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
794 longer raise an error; further, using a list as a for-as-arithmetic
795 counter now raises a meaningful error.
796 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
797 now understood by sequence creation functions such as MAKE-SEQUENCE
799 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
800 a STRING-INPUT-STREAM.
801 * compiler enhancement: SIGNUM is now better able to derive the type
803 * type declarations inside WITH-SLOTS are checked. (reported by
805 * fixed some bugs revealed by Paul Dietz' test suite:
806 ** incorrect optimization of TRUNCATE for a positive first
807 argument and negative second.
808 ** compiler failure in let-convertion during flushing dead code.
809 ** compiler failure while deriving type of TRUNCATE on an
810 interval, containing 0.
811 ** ASH of a negative bignum by a negative bignum count now returns
813 ** intersection of CONS types now canonicalizes properly, fixing
814 inconsistencies in SUBTYPEP.
816 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
817 * incompatible change: The --disable-debugger command line
818 option now clobbers the debugger at a more fundamental
819 level, by redefining #'INVOKE-DEBUGGER instead of by
820 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
821 is specified by ANSI to ignore *DEBUGGER-HOOK* and
822 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
823 enter the debugger REPL and then suffer recursive errors
824 because *DEBUG-IO* is also messed up in --disable-debugger mode;
825 while under the new system, BREAK in --disable-debugger mode
826 terminates the system just as an unhandled error would.
827 * fixed compiler performance when processing loops with a step >1;
828 * bug fix: DOCUMENTATION now retrieves generic function
829 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
830 support has been systematized, and now supports the methods
831 specified by ANSI, along with a default method and a method for
832 slot documentation. (reported by Nathan Froyd)
833 * bug fix: effective methods associated with a generic function are
834 no longer cached over a change of that generic function's method
835 combination. (reported by Andreas Fuchs)
836 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
837 names a non-existent directory. (thanks to Andreas Fuchs)
838 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
839 the PPC platform, lead to stack corruption; this has been fixed.
840 (reported by Rainer Joswig)
841 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
842 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
843 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
844 arithmetic in general, is now much more reliable on the PPC
846 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
847 platform now returns the right answer.
848 * optimization: restored some effective method precomputation in
849 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
850 precomputation is now tunable.
851 * optimization: compiler-internal data structure use has been
852 reviewed, and changes have been made that should improve the
853 performance of the compiler by about 20%.
854 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
855 simple-base-strings and simple-bit-vectors is improved.
856 * optimization: the optimization of 32-bit logical and arithmetic
857 functions introduced in version 0.8.3 on the x86 has been
858 implemented on the mips, ppc and sparc platforms; an
859 implementation of the same facility, but for 64-bit arithmetic,
860 has been added for the alpha.
861 * microoptimization: the compiler is better able to make use of the
862 x86 LEA instruction for multiplication by constants.
863 * bug fix: in some situations compiler did not report usage of
864 generic arithmetic in (SPEED 3) policy.
865 * bug 145b fix: compiler used wrong type specifier while converting
866 MEMBER-types to numeric.
867 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
869 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
870 index argument now works properly on the Alpha platform.
871 * bug fix: floating point exception treatment on the Alpha platform
873 * bug fix: FILE-POSITION works much better on string input and
874 output streams. (thanks to Nikodemus Siivola)
875 * bug fix: many threading/garbage collection symptoms sorted.
876 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
877 might be pseudo-atomic.
878 * internal change: Stopping for GC is now done with signals not
879 ptrace. GC is now done in whichever thread wanted it, instead of
881 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
883 * bug fix: result form in DO is not contained in the implicit
885 * incompatible change: ICR structure is changed; the value part of
886 CONTINUATION is now called LVAR; corresponding functions are
887 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
888 * added type deriver for ISQRT (thanks to Robert E. Brown).
889 * bug fix: better support for loading from the command line when an
890 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
892 * fixed some bugs revealed by Paul Dietz' test suite:
893 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
894 ** ROUND and FROUND now give the right answer when given very
895 small float arguments.
896 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
898 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
899 ** (GCD 0 <negative-integer>) returned <negative-integer>.
900 ** LCM should return a non-negative integer.
901 ** PARSE-INTEGER returned the index of a terminator instead of the
902 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
903 ** PARSE-INTEGER returned an incorrect index being applied to a
905 ** LCM with two arguments of 0 returns 0 rather than signalling
907 ** unsigned addition of a 32-bit constant with the high bit set no
908 longer causes an internal compiler error on the x86.
909 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
910 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
911 with negative last argument.
912 ** byte specifiers with zero size and position no longer cause
913 an error during type derivation.
914 ** bignum multiplication on the Alpha platform now returns the
916 * porting: The system now builds on SuSE AMD64, although it still
917 generates a 32-bit binary.
918 * .fasl file incompatibility: The fasl file version number has
919 been incremented (because of the changes to internal compiler
920 data structures referred to above).
922 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
923 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
924 more accurately, on the Darwin kernel running on PowerPC hardware.
925 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
926 * Compiler code deletion notes now signal a condition of type
927 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
928 an associated MUFFLE-WARNING restart.
929 * The compiler now performs limited argument count validation of
930 constant format strings in FORMAT, and where appropriate in ERROR,
931 CERROR and WARN. (thanks to Gerd Moellmann)
932 * New ASDF-INSTALL contrib can be used for automatic download and
933 installation of third-party Lisp code from CCLAN or other sites
935 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
936 which forces another thread to execute a function supplied by the
938 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
939 now accept and act upon their :ELEMENT-TYPE keyword argument.
940 (reported by Martin Atzmueller, Edi Weitz)
941 * bug fix: FILE-POSITION now accepts position designators up to
942 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
943 is the greater. (thanks to Patrik Nordebo)
944 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
945 Kalle Olavi Niemitalo)
946 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
947 printable. (reported by Eric Marsden)
948 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
949 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
950 * x86 bug fix in control stack exhaustion checking: now shows backtrace
951 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
952 (thanks to Stig Sandoe)
953 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
954 UNREAD-CHAR now correctly includes the unread character in the
955 target sequence. (thanks to Gerd Moellmann)
956 * bug fix in threaded builds: the system can now be suspended and
957 resumed by shell job control with minimal disruption.
958 * bug fixes in times and timezones >2038AD
959 * better handling of "where is GNU make?" problem in build scripts
960 (thanks to Nikodemus Siivola)
961 * new optimization: inside a named function any reference to a
962 function with the same name is considered to be a self-reference;
963 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
964 optimization quality.
965 * new optimization on x86: logical functions and + now have
966 optimized (UNSIGNED-BYTE 32) versions, which are automatically
967 used when the result is truncated to 32 bits.
968 * VALUES declaration is partially enabled.
969 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
970 * fixed some bugs revealed by Paul Dietz' test suite:
971 ** The system now obeys the constraint imposed by
972 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
973 types form a lattice under type intersection.
974 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
975 ** ASSOC now ignores NIL elements in an alist.
976 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
977 and (1+ MOST-POSITIVE-FIXNUM) answers.
978 ** The addition of a method with invalid qualifiers to a generic
979 function does not cause an error to be signalled immediately;
980 a warning is signalled, and the error is generated only on
981 calling the generic function.
982 * changed .fasl file version number, in part to add type codes for
983 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
984 obscure ANSI requirements
986 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
987 * fixed bug 148: failure to inline-expand a local function left
988 garbage, confusing the compiler.
989 * fixed bugs 3cd: structure slot readers perform type check if the
990 slot can have an invalid value (i.e. it is either not initialized
991 or can be written with a less specific slot writer).
992 * bug fix: the compiler now traps array references to elements off
993 the end of an array; previously, the bounds checking in some
994 circumstances could go off-by-one.
995 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
997 * type declarations for array element types now obey the description
998 on the CLHS page "Declaration TYPE", as per discussions on
999 sbcl-help around 2003-05-08. This means that a declaration
1000 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1001 declaration, all references to BAR will be asserted or assumed
1002 (with THE, so dependent on compiler policy) to involve objects of
1003 type FOO. Note that no such declaration is implied in
1004 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1005 * declared types of functions from the "Conditions"
1006 chapter. (reported by Paul Dietz)
1007 * bug fix: CERROR accepts a function as its first argument.
1008 * bug fix: NTH an NTHCDR accept a bignum as index
1009 arguments. (reported by Adam Warner)
1010 * optimization: character compare routines now optimize comparing
1011 against a constant character. (reported by Gilbert Baumann)
1012 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1013 argument now works properly on the MIPS platform.
1014 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1015 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1016 declared to have a functional type, does not cause an error
1018 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1019 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1020 * fixed bug: if last continuation of a deleted block has a
1021 destination, this destination should be deleted too. (reported by
1023 * fixed a bug in the bootstrap process: the host compiler's values
1024 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1025 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1026 test case from Patrik Nordebo)
1027 * improved the ability of the disassembler on the PPC platform to
1028 provide helpful disassembly notes.
1029 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1030 the class in more cases than previously.
1031 * bug fix: FILE-POSITION now understands :START and :END for
1032 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1033 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1034 8) data. (thanks to David Lichteblau)
1035 * bug fix: it is possible to add a method to a generic function
1036 without lambda list.
1037 * bug fix: reader failed to signal END-OF-FILE inside an
1038 object representation. (reported by Nikodemus Siivola)
1039 * fixed some bugs revealed by Paul Dietz' test suite:
1040 ** LAST and [N]BUTLAST should accept a bignum.
1041 ** condition slot accessors are methods.
1042 ** (VECTOR NIL) is a subtype of STRING.
1044 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1045 * minor incompatible change: some nonsensical specialized lambda
1046 lists (used in DEFMETHOD) which were previously ignored now signal
1048 * minor incompatible change: the system is now aware of the types of
1049 variables in the COMMON-LISP package, and will signal errors for
1050 most violations of these type constraints (where previously they
1051 were silently accepted).
1052 * minor incompatible change: COMPILE-FILE now uses the freedom
1053 afforded (ANSI 3.2.2.3) to use derived function types for
1054 functions defined in the same file. This also permits the system
1055 to warn on static type mismatches and function
1056 redefinition. (Currently it does not work with high DEBUG level.)
1057 * minor incompatible change: VALUES declaration is disabled.
1058 * When issuing notes, the compiler now signals a condition of type
1059 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1060 restart for use in user handlers. It is expected that the
1061 COMPILER-NOTE condition will eventually become a condition
1062 supertype to a hierarchy of note types, which will then be
1063 handleable in a similar fashion. However, at the moment, no such
1064 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1065 but it's an implementation detail, not a classification for the
1067 * Changes in type checking closed the following bugs:
1068 ** type checking of unused values (192b, 194d, 203);
1069 ** template selection based on unsafe type assertions (192c, 236);
1070 ** type checking in branches (194bc).
1071 * A short form of VALUES type specifier has ANSI meaning (it has
1072 increased the number of situations when SBCL cannot perform type
1074 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1075 keywords or constants is permissible.
1076 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1077 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1078 classes. (thanks to Antonio Martinez)
1079 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1080 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1081 argument to a pretty-print stream. Instead, it leaves any such
1082 operations to DESCRIBE-OBJECT methods.
1083 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1085 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1086 lambda lists. (thanks to David Lichteblau)
1087 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1088 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1089 (thanks to Gerd Moellmann)
1090 * fixed bug 246: increased compilation speed of long
1091 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1093 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1095 * DEFCONSTANT now throws a condition of type
1096 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1097 constant to a non-EQL value; CONTINUE and ABORT restarts
1098 respectively change and preserve the value.
1099 * fixed bug 63: The code walker, part of the implementation of CLOS,
1100 is now better at handling symbol macros.
1101 * bug fix: There is no longer an internal implementation type named
1102 CL:LENGTH. (reported by Raymond Toy)
1103 * bug fix: In macro-like defining macros/special operators the
1104 implicit block does not enclose the lambda list.
1105 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1106 enclosed in parentheses) are not suitable as type specifiers, and
1107 their use properly signals an error now.
1108 * bug fix: An argument count mismatch for a type specifier in code
1109 being compiled no longer causes an unhandled error at compile
1110 time, but signals a compile-time warning.
1111 * fixed simple vector readable printing
1112 * bug fix: DESCRIBE takes more care over whether the class
1113 precedence list slot of a class is bound before accessing it.
1114 (reported by Markus Krummenacker)
1115 * bug fix: FORMATTER can successfully compile pretty-printer format
1116 strings which use variants of the ~* directive inside.
1117 * bug fix: SEARCH now applies its TEST predicate to the elements of
1118 the arguments in the correct order. (thanks to Wolfhard Buss)
1119 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1120 in type inference. (thanks to Robert E. Brown)
1121 * bug fix: Reading in symbols with an explicit package name of ""
1122 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1123 * fixed some bugs revealed by Paul Dietz' test suite:
1124 ** NIL is now allowed as a structure slot name.
1125 ** Arbitrary numbers, not just REALs, are allowed in certain
1126 circumstances in LOOP for-as-arithmetic clauses.
1127 ** Multiple class redefinitions before slot access no longer
1128 causes a type error.
1129 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1130 association between the name and a class.
1131 ** Generic functions with non-standard method-combination and over
1132 five methods all of which return constants no longer return NIL
1133 after the first few invocations. (thanks to Gerd Moellmann)
1134 ** CALL-NEXT-METHOD with no arguments now passes the original
1135 values of the arguments, even in the presence of assignment.
1136 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1138 ** DISASSEMBLE works with closures and funcallable instances.
1139 ** ADD-METHOD now returns the generic function, not the new method.
1140 ** FIND-METHOD signals an error if the lengths of the specializers
1141 is incompatible with the generic function, even if the ERRORP
1143 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1144 which its argument is a member.
1145 ** DEFCLASS only redefines the class named by its class-name
1146 argument if that name is the proper name of the class;
1147 otherwise, it creates a new class.
1148 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1149 of the UNBOUND-SLOT condition to the name of the slot.
1150 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1151 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1152 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1153 SLOT-MAKUNBOUND in the specified fashion.
1155 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1156 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1157 cross-compilation host. As a consequence, we can now bootstrap our
1158 way up to SBCL starting with a bare gcc toolchain and human-readable
1159 source code (first the source to CLISP, then the source to SBCL).
1160 * A contributed module containing a partial implementation of the
1161 simple-streams interface has been included. (thanks to Rudi
1163 * A contributed module implementing the RFC1321 Message Digest
1164 Algorithm, known as MD5, has been included.
1165 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1166 no longer has any effect, as the code controlled by this feature
1167 has been deleted. (As far as we know, no-one has ever built using
1168 this feature, and its semantics were confused in any case).
1169 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1170 work on conditions (as required by the ANSI specification),
1171 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1172 expected behaviour on conditions. Users should note, however,
1173 that such behaviour is not required by the ANSI specification,
1174 and so use of this behaviour may render their code unportable.
1175 * fixed some bugs revealed by Paul Dietz' test suite:
1176 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1178 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1179 ** SXHASH on condition objects no longer returns NIL.
1180 ** :ALLOCATION :CLASS slots are better treated; their values are
1181 updated on class redefinition, and initforms inherited from
1182 superclasses are applied.
1183 ** REMOVE-METHOD returns its generic function argument even when
1184 no method was removed.
1185 ** SHARED-INITIALIZE now initializes the values of the requested
1186 slots, including those with :ALLOCATION :CLASS.
1187 ** ALLOCATE-INSTANCE now works on structure classes defined via
1188 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1190 ** SLOT-EXISTS-P now works on conditions, as well as structures
1192 ** MAKE-LOAD-FORM now has the required methods on
1193 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1194 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1195 keyword, and now implements the SLOT-NAMES argument.
1196 ** methods with &OPTIONAL arguments no longer allow too many
1197 arguments to be passed in the call without error.
1198 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1199 option is consistent with the required arguments of the generic
1200 function lambda list.
1201 * bug fix: REQUIRE accepts a string designator. (Thanks to
1203 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1204 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1205 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1206 * bug fix: The compiler now checks for duplicated variables in macro
1208 * bug fix: SETQ on globals returns the correct value.
1209 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1210 not a COMPILER-ERROR (followed by some other strange error on
1211 choosing the CONTINUE restart).
1212 * bug fix: make.sh and friends are now more consistent in the way that
1213 they look for GNU "make".
1215 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1216 * experimental native threads support (on x86 Linux >=2.4 only).
1217 This is not compiled in by default: you need to add :SB-THREAD to
1218 the target features. See the "Beyond ANSI" chapter of the manual
1220 * fix for longstanding nonANSIism: The old distinction between
1221 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1222 The return value from CL:FIND-CLASS is now a CLOS class, and
1223 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1224 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1225 * An interface to the MetaObject Protocol, as described in Kiczales,
1226 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1227 MIT Press, 1991, is available from the SB-MOP package.
1228 * incompatible change: the SB-PCL package should now be considered
1229 a private implementation detail, and no longer a semi-private MOP
1231 * minor incompatible change: due to rearrangement for threads, the
1232 control stack and binding stack are now allocated at arbitrary
1233 addresses instead of being hardcoded per-port. Users affected by
1234 this probably have to be doing advanced things with shared
1235 libraries, and will know who they are.
1236 * minor incompatible change: Previously, all --eval forms used were
1237 processed with READ before any of them were processed with EVAL.
1238 Now each --eval form is processed with both READ and EVAL before
1239 the next --eval form is processed. (Thus package operations like
1240 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1241 work as the user might reasonably expect.)
1242 * minor incompatible change: *STANDARD-INPUT* is now only an
1243 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1245 * minor incompatible change: Y-OR-N-P is now character-oriented,
1246 not line oriented. Also, YES-OR-NO-P now works without errors.
1247 (thanks to Antonio Martinez)
1248 * sb-aclrepl module improvements: an integrated inspector, added
1249 repl features, and a bug fix to :trace command.
1250 * Known functions, which cannot be open coded by the backend, are
1251 considered to be able to check types of their arguments. (fixing
1252 a bug report by Nathan J. Froyd)
1253 * fixed a bug in computing method discriminating functions: It is
1254 now possible to define methods specialized on classes which have
1255 forward-referenced superclasses. (thanks to Gerd Moellmann)
1256 * fixed evaluation order in optional entries (reported by Gilbert
1258 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1259 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1260 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1261 required-by-AMOP NAME argument, as well as CLASS and
1262 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1263 * fixed bug 20: DEFMETHOD can define methods using names that are
1264 not the proper names of classes to designate class specializers.
1265 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1266 Unix file descriptors, instead of blowing up. (thanks to Antonio
1268 * Garbage collection refactoring: user-visible change is that a
1269 call to the GC function during WITHOUT-GCING will not do garbage
1270 collection until the end of the WITHOUT-GCING. If you were doing
1271 this you were probably losing anyway.
1272 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1273 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1274 TYPEP the latter but not the former.
1275 * The compiler issues a full WARNING for calls to undefined functions
1276 with names from the CL package.
1277 * MAP-INTO for a vector destination is open coded. (reported by
1278 Brian Downing on c.l.l)
1279 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1280 documentation string.
1281 * fixed some bugs revealed by Paul Dietz' test suite:
1282 ** COPY-ALIST now signals an error if its argument is a dotted
1284 ** Condition slots are now accessed more correctly in the presence
1285 of multiple initargs for a given slot.
1286 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1287 exclude restarts of the same name associated with a different
1289 ** DEFCLASS of forward-referenced classes with another
1290 forward-referenced class in the superclasses list no longer
1292 ** Condition slots are now initialized once each, not multiple
1293 times. (thanks to Gerd Moellmann)
1294 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1296 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1297 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1298 arguments contain duplicated elements.
1299 ** RESTART-CASE understands local macros.
1300 ** RESTART-CASE associates exactly its own restarts with a condition.
1301 ** ENDP in safe mode checks its argument to be of type LIST.
1302 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1303 in question is unbound.
1304 ** Optimized MAKE-INSTANCE functions no longer cause internal
1305 assertion failures in the presence of duplicate initargs.
1306 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1307 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1308 propagated even in the ctor optimized implementation of
1310 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1312 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1313 * a better implementation of SXHASH on (simple) bit vectors,
1314 measured both in execution speed and in distribution of results
1315 over the positive fixnums, has been installed. Likewise, a better
1316 implementation of EQUAL for simple bit vectors is now available.
1317 * fixed CEILING optimization for a divisor of form 2^k.
1318 * fixed bug 240 (emitting extra style warnings "using the lexical
1319 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1320 by Antonio Martinez)
1321 * fixed SXHASH, giving different results for NIL depending on type
1322 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1323 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1324 variable. (found by Rolf Wester)
1325 * SBCL does not ignore type declarations for special
1326 variables. (reported by rif on c.l.l 2003-03-05)
1327 * some bug fixes in contrib/sb-aclrepl/
1328 * fixed some bugs revealed by Paul Dietz' test suite:
1329 ** a bug in the CONS type specifier, whereby the CAR and CDR
1330 types got intertwined, has been fixed;
1331 ** the type system is now able to reason about the interaction
1332 between INTEGER and RATIO types more completely;
1333 ** APPEND, [N]REVERSE and NRECONC check that those their
1334 arguments, which must be proper lists, are really so;
1335 ** An array specialized to be unable to hold elements has been
1336 implemented, as required -- yes, really -- by ANSI;
1337 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1338 on malformed property lists;
1340 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1341 * incompatible packaging change: in line with Unix convention,
1342 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1343 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1344 * REQUIRE and PROVIDE are now optionally capable of doing something
1345 useful. See the documentation string for REQUIRE.
1346 * infrastructure for a managed SBCL contrib system: contributed
1347 modules in this release include:
1348 ** the ASDF system definition facility;
1349 ** an interface to the BSD Sockets API;
1350 ** an ACL-like convenience interface to the repl;
1351 (thanks to Kevin Rosenberg)
1352 ** an implementation of ROTATE-BYTE, with efficient implementation
1354 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1355 gives a closer approximation to the right answer than previously.
1356 (thanks to Raymond Toy)
1357 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1358 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1359 optional environment arguments, as required by ANSI.
1360 * fixed bugs in other functions taking environment objects, allowing
1361 calls with an explicit NIL environment argument to be compiled
1363 * fixed bug 228: primary return values from
1364 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1365 COMPILE or FUNCTION.
1366 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1367 :TYPE will no longer signal errors on innocuous objects.
1368 * fixed bug 231b: SETQ is better at respecting type declarations in
1369 the lexical environment.
1370 * fixed a bug in DEFCLASS: classes named by symbols with no or
1371 unprintable packages can now be defined.
1372 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1373 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1374 * fixed compiler failure related to checking types of functions.
1375 (reported by Robert E. Brown)
1376 * the compiler is now much more consistent in its error-checking
1377 treatment of bounding index arguments to sequence functions: in
1378 (SAFETY 3) code, errors will be signalled in almost all cases if
1379 invalid sequence bounding indices are passed to functions defined
1380 by ANSI to operate on sequences.
1381 * fixed a bug in the build procedure: documentation of SBCL-specific
1382 packages is now preserved and available in the final Lisp image.
1383 * lifted FDEFINITION lookup out of loops in the implementation of
1384 many list operations. (thanks to Robert E. Brown)
1385 * fixed a bug in the reader: the #n# reader macro now works for
1386 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1387 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1388 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1389 E. Brown and Tony Martinez respectively)
1390 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1391 SLOT-BOUNDP: in particular, it is now able to optimize them much
1392 better, and is now not vulnerable to having packages renamed.
1393 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1394 always called when a slot is not present in an instance. (thanks
1396 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1397 longer signal an error on unreading a character following EOF on
1398 the previous constituent stream. (thanks to Tony Martinez)
1399 * fixed some bugs revealed by Paul Dietz' test suite:
1400 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1401 not just nonnegative fixnums;
1402 ** the logical bit-array operators such as BIT-AND now accept an
1403 explicit NIL for their "opt-arg" argument (to indicate a
1404 freshly-consed result bit-array);
1405 ** ELT now signals an error on an invalid sequence index in safe
1407 ** the type system is now cleverer about negations of numeric
1408 types, and consequently understands the BIGNUM and RATIO types
1410 ** the type system is now cleverer about the interaction between
1411 INTEGER and RATIO types: while bugs still remain, many more
1412 cases are accurately computed;
1413 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1414 if it is in the last clause;
1415 ** CONSTANTLY now correctly returns a side-effect-free function in
1417 ** DECLARE is no longer treated as a special-operator; in
1418 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1419 * incremented fasl file version number due to the change in the
1420 DEFSTRUCT-SLOT-DESCRIPTION structure.
1422 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1423 * minor incompatible change: code processed by the "interpreter" or
1424 EVAL now has a compilation optimization policy of (DEBUG 2)
1425 (changed from (DEBUG 1)) to improve debuggability of interactive
1426 development, and to allow the use of the debug RETURN command in
1428 * an experimental implementation of the RETURN command for the
1429 debugger has been included. (thanks to Frederik Kuivinen)
1430 * fixed bug 62: constraints were not propagated into a loop.
1431 * fixed bug in embedded calls of SORT (reported and investigated by
1433 * fixed some bugs revealed by Paul Dietz' test suite:
1434 ** printing and reading of arrays with some dimensions having
1435 length 0 (thanks to Gerd Moellmann);
1436 ** BOA constructor with &AUX argument without a default value does
1437 not cause a type error;
1438 ** CONSTANTP now returns true for all self-evaluating objects.
1440 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1441 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1442 accessors that are related by inheritance, as specified in the
1443 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1444 Valtteri Vuorikoski)
1445 * The compiler is now able to inline functions that were defined in
1446 a complex lexical environment (e.g. inside a MACROLET).
1447 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1449 * Support for the upcoming FreeBSD-5.0 release has been included.
1450 (thanks to Dag-Erling Smorgrav)
1451 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1452 effect when it is not in a toplevel context.
1453 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1454 stand a better chance of being correct. (thanks to Gerd
1456 * fixed bug in COERCE, which now signals an error on coercing a
1457 rational to a bounded real type which excludes the expected
1459 * The compiler is now able to derive types more accurately from the
1460 COERCE and COMPILE functions.
1461 * fixed bug 223: functional binding is considered to be constant
1462 only for symbols in the CL package.
1463 * fixed bug 231: SETQ did not check the type of a variable being set
1464 (reported by Robert E. Brown)
1465 * A new optimization for MAKE-INSTANCE has been included, fixing
1466 various bugs (including relating to :ALLOCATION :CLASS slots and
1467 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1469 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1470 ** As required by ANSI, LOOP now disallows anonymous collection
1471 clauses such as COLLECT I in conjunction with aggregate boolean
1472 clauses such as THEREIS (= I 1);
1473 ** LOOP now signals an error when any variable is reused in the
1474 same loop (including the potentially useful construct analogous
1475 to WITH A = 1 WITH A = (1+ A);
1476 ** IT is only a special loop symbol within the first clause of a
1477 conditional loop clause;
1478 ** LOOP with a typed iteration variable over a hashtable now
1479 signals a type error iff it should.
1480 * fixed some other bugs revealed by Paul Dietz' test suite:
1481 ** FILE-STREAM now names the class previously known as FD-STREAM;
1482 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1483 argument) no longer signals an error;
1484 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1485 of the slot symbol, rather than using the current package
1486 ((:CONC-NAME "") continues to intern the slot's name in the
1488 * incremented fasl file version number, because of the incompatible
1489 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1490 of the new implementation of DEFINE-COMPILER-MACRO.
1492 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1493 * Support for building SBCL for MIPS platforms running in
1494 little-endian mode has now been checked in, and basic
1495 functionality on said platforms verified.
1496 * minor incompatible change: PCL now records the pathname of a file
1497 in which methods and the like are defined, rather than its
1499 * minor incompatible change: TRUENAME now considers the truename of
1500 a file naming a directory to be the pathname with :DIRECTORY
1501 component indicating that directory.
1502 * minor incompatible change: a NAMED clause in the extended form of
1503 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1504 reason for the previous behaviour is unclear.
1505 * more systematization and improvement of CLOS and MOP conformance
1506 in PCL (thanks to Gerd Moellman and Pierre Mai):
1507 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1509 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1510 primary methods with no specializers;
1511 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1513 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1514 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1515 FORWARD-REFERENCED-CLASSes; error reporting on
1516 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1518 ** SXHASH on CLOS instances now uses a slot internal to the
1519 instance to return different numbers on distinct instances,
1520 while preserving the same return value through invocations of
1522 ** DEFMETHOD signals errors when methods with longer incongruent
1523 lambda lists are added to generic functions;
1524 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1525 CLASS, as specified in AMOP;
1526 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1527 on the order returned by the primary method for classes of
1528 class STANDARD-CLASS;
1529 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1530 * fixed some bugs shown by Paul Dietz' test suite:
1531 ** DOLIST puts its body in TAGBODY;
1532 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1534 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1535 value producing form;
1536 ** if more variables are given to PROGV than values, extra
1537 variables are bound and made to have no value;
1538 ** NSUBSTITUTE on list arguments gets the right answer with
1540 ** ELT signals an error of type TYPE-ERROR when the index argument
1541 is not a valid sequence index;
1542 ** LOOP signals (at macroexpansion time) an error of type
1543 PROGRAM-ERROR when duplicate variable names are found;
1544 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1545 ** FUNCALL of special-operators now cause an error of type
1547 ** PSETQ now works as required in the presence of side-effecting
1548 symbol-macro places;
1549 ** NCONC accepts any object as its last argument;
1550 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1552 ** loop-for-as-package does not require a package to be explicitely
1554 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1556 * fixed bug 166: compiler preserves "there is a way to go"
1557 invariant when deleting code.
1558 * fixed bug 172: macro lambda lists with required arguments after
1559 &REST arguments now cause an error to be signalled. (thanks to
1561 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1562 bug. (thanks to Matthew Danish)
1563 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1565 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1566 arguments are lexically rebound. (thanks to Gerd Moellmann and
1568 * fixed bug 194: error messages are now more informative when there
1569 is no primary method applicable in a call to a generic
1570 function. (thanks to Gerd Moellmann)
1571 * fixed bug in command line argument checking (thanks to Julian
1573 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1574 (thanks to Matthew Danish)
1575 * incremented fasl file version number, because of the
1576 SXHASH-related changes in the layout of CLOS data structures
1578 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1579 * minor incompatible change: The runtime (the Unix executable named
1580 "sbcl") is now much pickier about the .core files it will load.
1581 Essentially it now requires .core files to descend from the same
1582 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1583 as the runtime does. (The intent is to prevent the crashes which
1584 can occur, and which can even be reported as mysterious failures,
1585 when people patch the sources or change the build parameters
1586 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1587 sbcl and .core files.)
1588 * fixed bug: VALUES-LIST is no longer optimized away.
1589 * fixed bug 142: The FFI conversion of C string values to Lisp
1590 string values no longer conses excessively. (thanks to Nathan
1591 Froyd porting Raymond Toy's fix to CMU CL)
1592 * began to systematize and improve MOP conformance in PCL (thanks to
1593 Nathan Froyd, Gerd Moellman and Pierre Mai):
1594 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1596 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1597 ** FINALIZE-INHERITANCE is now called on class finalization;
1598 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1599 argument precedence order.
1600 * fixed bug 202: The compiler no longer fails on functions whose
1601 derived types contradict their declared type.
1602 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1603 so it can be non-toplevel.
1604 * The fasl file version number has changed (because of the new
1605 implementation of DEFMACRO).
1606 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1607 safe code, that any length requirement by their type-specifier
1608 argument is valid. The exceptions to this are described in bug
1610 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1611 functions now check the types of their inputs as required by ANSI.
1612 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1613 introduced symbol is DECLAREd to be SPECIAL.
1614 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1615 * fixed bug: :COUNT argument to sequence functions may be negative
1616 * fixed bug: body of DO-SYMBOLS may contain declarations
1617 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1618 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1619 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1620 symbol macro only once
1621 * fixed printing of call frame when argument list is unavailable
1622 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1623 * compiler no longer signals WARNING on unknown keyword
1626 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1627 * A beta-quality port to the mips architecture running Linux,
1628 based on the old CMUCL backend, has been made. It has been tested
1629 on a big-endian kernel, and works sufficiently well to be able to
1630 rebuild itself; it has not been tested in little-endian mode.
1631 * fixed an inconsistency between gencgc.c and purify.c which made
1632 dumping/loading .core files unreliable
1633 * fixed bug 120a: The compiler now deals correctly with IFs where
1634 the consequent is the same as the alternative, instead of
1635 misderiving the return type. (thanks to Alexey Dejneka)
1636 * fixed bug 113: Logical pathnames are now dumpable (the logical
1637 host is resolved at load-time, throwing an error if it is not
1639 * fixed bug 174: FORMAT's error message is slightly clearer when a
1640 non-printing character is used in a format directive.
1641 * fixed several bugs in compiler checking of type declarations, i.e.
1642 violations of the Python "declarations are assertions" principle
1643 (thanks to Alexey Dejneka)
1644 * fixed several bugs in PCL's error checking (thanks to Gerd
1646 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1648 * fixed bug in compilation of functions as first class values
1649 (thanks to Antonio Martinez-Shotton)
1650 * The compiler's handling TYPE-ERRORs which it can prove will
1651 inevitably happen at runtime has been cleaned up and corrected
1652 in several ways. (thanks to Alexey Dejneka)
1653 * improved argument type checking for various basic arithmetic
1654 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1655 lavished on them in the past that they can be compiled in many
1656 ways in different special cases
1657 * fixed bug 181: compiler checks validity of user supplied type
1659 * cleaned up code flushing in optimization: Function calls which
1660 should signal errors for safety purposes (e.g. which ANSI says
1661 should signal errors when their arguments are of incorrect type)
1662 are no longer optimized away.
1663 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1664 * incremented fasl file version number, because changes in the
1665 implementation of sequence functions like COERCE caused
1666 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1667 in old inline expansions) to become undefined. (Actually these
1668 changes were later undone, so we might very well be binary
1669 compatible with 0.7.7 after all, but leaving the version number
1670 incremented seemed like the simplest and most conservative
1673 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1674 * An alpha-quality port to the parisc architecture running Linux,
1675 based on the old CMUCL backend, has been made. This, even more so
1676 than the other backends, should be considered still a work in
1677 progress; known problems include that the Linux kernel in 64-bit
1678 mode does not propagate the correct sigcontext structure to
1679 userspace, and consequently SBCL on a parisc64 kernel will not
1681 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1682 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1683 suppressed.) Also now that the compiler is looking at declarations
1684 in the environment, it checks optimization declarations as well,
1685 and suppresses inlining when (> DEBUG SPEED).
1686 * More fixes have been made to treatment of floating point exception
1687 treatment and other Unix signals. In particular, floating point
1688 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1689 * The detection and handling of control stack exhaustion (infinite
1690 or very deeply nested recursion) has changed. Stack exhaustion
1691 detection is now done by write-protecting pages at the OS level
1692 and applies at all optimization settings; when found, a
1693 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1694 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1695 be caught using IGNORE-ERRORS.
1696 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1697 correctly when transforming recursive LABELS functions to LETs.
1698 (thanks to Alexey Dejneka)
1699 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1700 that are names of constants or global variables.
1701 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1702 alien routines with docstrings.
1703 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1704 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1706 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1707 (thanks to Eric Marsden)
1708 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1709 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1710 * Bug fix: The SPARC backend can now compile functions involving
1711 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1712 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1713 to a functions expecting a PATHNAME-DESIGNATOR.
1714 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1715 lambda lists. (thanks to Alexey Dejneka)
1716 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1717 functions so that EXPORTing the name of the class doesn't cause
1718 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1719 to Antonio Martinez for reporting this)
1720 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1721 constant-folded without failing an assertion. (thanks to Einar
1722 Floystad Dorum for reporting this)
1723 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1724 types are now dealt with more correctly by the compiler.
1725 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1726 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1727 behaviour analogous to RENAME-FILE. This puts its behaviour more
1728 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1730 * The fasl file version number has changed again. (because of the
1731 bug fix involving the names of PCL MAKE-INSTANCE functions)
1733 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1734 * bug fix: Floating point exceptions are treated much more
1735 consistently on the x86/Linux and PPC/Linux platforms.
1736 * Array initialization with :INITIAL-ELEMENT is now much faster for
1737 cases when the compiler cannot open code the array creation, but
1738 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1739 array accesses have also seen a speed increase.
1740 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1741 specified. (This was at the root of some bad interactions between
1742 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1744 * bug fix: Internal error arguments for undefined functions are now
1745 computed correctly on the PPC/Linux platform.
1746 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1747 Raymond Toy's patch for CMU CL)
1748 * Support for the Solaris 9 operating environment has been included
1749 (thanks to Daniel Merritt)
1750 * A very ugly but hopefully complete draft of the missing FFI chapter
1751 of the manual has been created by reformatting the corresponding
1752 CMU CL manual chapter into (currently very ugly and incoherent)
1753 DocBook and bringing it up to date for SBCL behavior. Thus, the
1754 manual is now essentially complete, at least by my extreme
1755 once-and-only-once standards, whereby it's acceptable to refer to
1756 the doc strings of SB-EXT functions as the primary documentation.
1757 * The fasl file version number has changed again, due to cleanup of
1758 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1759 is no longer a static symbol.)
1761 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1762 * SBCL now builds with OpenMCL (version 0.12) as the
1763 cross-compilation host; also, more progress has been made toward
1764 bootstrapping under CLISP.
1765 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1767 * bug 158 fixed: The compiler can now deal with integer loop
1768 increments different from 1; fixing this turned out also to fix
1770 * bug 169 fixed: no more bogus warnings about using lexical bindings
1771 despite the presence of perfectly good SPECIAL declarations (thanks
1772 to David Lichteblau)
1773 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1774 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1775 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1777 * bug 180 fixed: Method combination specifications no longer ignore
1778 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1779 * bug fix: Structure type predicate functions now check their argument
1780 count as they should.
1781 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1782 correctly. (thanks to Pierre Mai)
1783 * minor incompatible change: The --noprogrammer option is deprecated
1784 in favor of the new --disable-debugger option, which is very similar.
1785 (The major difference is that it takes effect at a slightly different
1786 time at startup, causing handling of errors in --sysinit and
1787 --userinit files will be affected differently.) The
1788 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1789 been added to allow this functionality to be controlled from ordinary
1790 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1791 maintainers, who might want to run non-interactive scripts to
1792 build SBCL cores which will later be used interactively.)
1793 * minor incompatible change: The LOAD function no longer, when given
1794 a wild pathname to load, loads all files matching that pathname.
1795 Instead, an error of type FILE-ERROR is signalled.
1797 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1798 * bug 147 fixed: The compiler preserves its block link/count
1799 invariants more correctly now so that it doesn't crash. (thanks
1801 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1803 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1804 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1805 no-value-for-this-character as NIL (as specified by ANSI).
1806 * HOST-NAMESTRING on physical pathnames now returns a string that is
1807 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1808 (thanks to Christophe Rhodes)
1809 * The Alpha port handles icache flushing more correctly. (thanks to
1811 * More progress has been made toward bootstrapping under CLISP. (thanks
1812 to Christophe Rhodes)
1813 * The fasl file format has changed again, because dynamic loading
1814 on OpenBSD (which has non-ELF object files) motivated some cleanups
1815 in the way that foreign symbols are transformed and passed around.
1816 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1817 is no longer treated as whitespace by the reader, but instead as
1818 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1819 |A
\7fB|, instead of A as it used to.
1821 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1822 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1823 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1824 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1825 this as of 0.7.1.45, but was left out of the previous news section
1826 (thanks to Dan Barlow)
1827 * SBCL now runs on the Solaris operating system on SPARC architectures
1828 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1829 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1830 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1831 * SPARC backend cleanups, allowing builds of cores optimized for V8
1832 and V9 SPARCS, and also emission of code targeted to a particular
1833 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1835 * SBCL is closer to bootstrapping under CLISP, thanks to various
1836 fixes by Christophe Rhodes.
1837 * The fasl file format has changed again, to allow the compiler's
1838 INFO database to support symbol macros.
1839 * The user manual (in doc/) is formatted into HTML more nicely.
1840 (thanks to coreythomas)
1841 * The system is smarter about SUBTYPEP relationships, especially
1842 those involving NOT types (including types such as ATOM which are
1843 represented internally using NOT types). Thus SUBTYPEP is less
1844 likely to return (VALUES NIL NIL) in general, and in particular
1845 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1847 * The fasl file format has changed again, because the internal
1848 representation of types now includes a new slot to support the new
1849 SUBTYPEP-of-NOT-types logic.
1850 * (not a change in the main branch of SBCL, but a related prototype
1851 which can hopefully be merged into the main branch of SBCL in the
1852 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1853 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1854 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1855 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1856 dispatch macro character combinations. (thanks to Alexey Dejneka)
1857 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1858 that superficially look like logical namestrings correctly.
1859 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1861 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1862 default supplied-p) for &optional and &key arguments. (thanks to
1865 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1866 * incompatible change: The compiler is now less aggressive about
1867 tail call optimization, doing it only when (> SPACE DEBUG) or
1868 (> SPEED DEBUG). (This is an incompatible change because there are
1869 programs which relied on the old CMU-CL-style behavior to optimize
1870 away their unbounded recursion which will now die of stack overflow.)
1871 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1872 for non-GENCGC systems has been increased to 20M (since that
1873 seems much closer to the likely performance optimum for modern
1874 systems than the old 4M value was)
1875 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1876 and *DEBUG-PRINT-LEVEL*
1877 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1878 of CMU CL's support for SPARC, and various endianness and other
1879 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1880 * new syntactic sugar for the Unix command line: --load foo.bar is now
1881 an alternate notation for --eval '(load "foo.bar")'.
1883 ** The system now detects stack overflow and handles it gracefully,
1884 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1885 optimization settings. (This is a good thing in general, and
1886 its introduction in this version should be particularly timely
1887 for anyone whose code fails because of suppression of tail
1889 ** The system now hunts for the C variable "environ" in a more
1890 devious way, to avoid segfaults when the C library version differs
1891 between compile time and run time. (thanks to Christophe Rhodes)
1892 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1893 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1894 ** The compiler no longer issues bogus style warnings for undefined
1895 classes in the same source file as the DEFCLASSes which defined
1896 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1898 ** fixes in CONDITION class precedence list for undefined function
1899 errors (thanks to Alexei Dejneka)
1900 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1901 correctly. (thanks to Dan Barlow)
1902 ** portability fixes aiming at bootstrapping under CLISP (thanks
1903 to Dave McDonald and Christophe Rhodes)
1904 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1905 ** fixes in type translation and and type inference (thanks to
1907 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1908 ** various fixes in the new ports (thanks to Dan Barlow)
1909 * several changes related to debugging:
1910 ** suppression of tail recursion, as noted above
1911 ** stack overflow detection, as noted above
1912 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1913 is now the default. (For some time encapsulation has been more
1914 reliable than the breakpoint-based :ENCAPSULATE NIL
1915 implementation, at least on X86 systems; and I just noticed that
1916 encapsulation also seems closer to the spirit of the ANSI
1919 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1921 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1922 up properly again. (There was a packaging bug in 0.7.0 which
1923 left their definitions in SB-SYS::LOAD-FOREIGN and
1924 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1925 vital for most things which interface to C-level interfaces,
1926 like extensions working with sockets or databases or
1927 Perl-compatible regexes or whatever, and the need to fix
1928 this bug is the main reason that 0.7.1 was released so
1930 ** DEFGENERIC is now choosier about the methods it redefines, so that
1931 reLOADing a previously-LOADed file containing DEFGENERICs does
1932 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1933 cycle now works as it should. (thanks to Alexey Dejneka)
1934 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1935 fixed by Christophe Rhodes. (He actually submitted this patch
1936 months ago, and I delayed until after 0.7.0.)
1937 ** Bug 111 (internal compiler confusion about runtime checks on
1938 FUNCTION types) was fixed by Alexey Dejneka.
1939 * Some internal cleanups (getting rid of variables which aren't
1940 needed now that the byte interpreter is gone) caused the fasl
1941 file format number to change again.
1943 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1944 * major incompatible change: The default fasl file extension, i.e. the
1945 default extension for files produced by COMPILE-FILE, has changed
1946 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1948 ** There are many changes in the implementation of the compiler.
1949 SBCL is now essentially a compiler-only implementation of ANSI
1950 Common Lisp. EVAL still "interprets" a few special cases, but
1951 almost all the interesting cases are handled by creating
1952 a LAMBDA expression, calling COMPILE on it, then calling
1953 FUNCALL on the result.
1954 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1955 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1956 Since the code is newer, there might still be some new bugs
1957 (though not as many as before Martin Atzmueller's fixes:-). But
1958 the new code is substantially simpler and clearer, and hopefully
1959 any remaining bugs will be simpler, less fundamental, and more
1960 fixable then the bugs in the old code.
1961 ** The revised compiler is still a little unsteady on its feet.
1963 *** The debugging information it produces (particularly the names
1964 of FUNCTION objects) is sometimes much less useful than what
1965 the old compiler produced.
1966 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1967 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1968 within a MACROLET) has been temporarily weakened.
1969 ** There are new compiler optimizations for various functions:
1970 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1971 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1972 *** the math functions TRUNCATE, FLOOR, and CEILING
1973 *** the function-of-all-trades COERCE
1974 Mostly these should be transparent, but there's one
1975 potentially-annoying problem (bug 117): when the compiler
1976 inline-expands a function and does type analysis on the result,
1977 it can create control paths which have type mismatches, and
1978 when it can't prove that those control paths aren't taken,
1979 it will issue WARNINGs about the type mismatches. This is
1980 a particular problem in practice for the new sequence functions.
1981 It's not clear how this should be fixed, and for now, a
1982 workaround is given in the entry for 117 in the BUGS file.
1983 ** (Because of the interaction between the two previous items --
1984 occasional inlining problems and new inline expansions -- some
1985 of the new sequence function optimizations won't really kick in
1986 completely until debugging information, and then inlining, are
1987 straightened out in some future version.)
1988 * minor incompatible changes:
1989 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1990 DIRECTORY no longer implicitly promotes NIL slots of its
1991 pathname argument to :WILD. In particular, when you ask for the
1992 contents of a directory (which you used to be able to do without
1993 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1994 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1995 ** changes in behavior that ANSI explicitly defines to be
1996 implementation dependent:
1997 *** The new compiler-only implementation still conforms with ANSI,
1998 but acts a little different than before. Besides the obvious
1999 changes in performance tradeoffs (that the cost per form passed
2000 to EVAL has gone up, and the cost per form executed by EVAL
2001 has gone down), the behavior of the system changes a little
2002 because there are no longer any interpreted function objects.
2003 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2004 e.g. doing COMPILE on the output of interactive DEFUN is
2006 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2008 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2009 has changed from (:CL) to NIL.
2010 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2011 e.g. #\Newline and #\Tab, don't have names specified in the
2012 ANSI Common Lisp standard, is now based on their ASCII symbolic
2013 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2014 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2015 are no longer used for output.
2016 ** changes in internal implementation constants:
2017 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2018 4 million. (If your application spends a lot of time GCing and
2019 you have a lot of RAM, you might want to experiment with
2020 increasing it even more.)
2021 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2022 However, almost all old code should still continue to work without
2023 immediate update, as SB-C-CALL is now a (deprecated) nickname
2025 ** Old operator names in the style DEF-FOO are now deprecated in
2026 favor of new corresponding names DEFINE-FOO, for consistency with
2027 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2028 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2029 internal symbols, but a few supported extensions like
2030 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2031 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2032 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2033 etc. as you get deeper into recursive calls to the debugger
2034 command loop, instead of the old "5]", "5]]", "5]]]"
2035 sequence. (I was motivated to do this when squabbles between
2036 ILISP and SBCL left me very deeply nested in the debugger. In the
2037 short term, this change will probably provoke more ILISP/SBCL
2038 squabbles, but hopefully it will be an improvement in the long run.)
2039 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2040 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2041 function, since declaiming return types involving aliens is
2042 (1) annoyingly messy to do by hand and (2) vital to efficient
2043 compilation of code which calls such functions.
2044 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2045 longer reexported by the SB-EXT package. They're solely useful
2046 for alien code, so it seems more logical that you should get
2047 them from the SB-ALIEN package, not in SB-EXT.
2048 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2049 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2050 features. Instead, the code that they used to control is always
2051 built into the system.
2052 * many other bug fixes
2053 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2054 advantage of the new EVAL-WHEN stuff and to clean them up in
2055 general, and they are now more ANSI-compliant in a number of
2056 ways. Martin Atzmueller is responsible for a lot of this.
2057 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2059 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2060 *** correct ERROR type for various file operations
2061 *** some fixes for Lisp streams
2062 *** DEFMETHOD syntax checking
2063 *** changing old weird representation of debug information as
2064 strings (which, among their other deficiencies, don't transform
2065 correctly when you rename packages, and don't change their
2066 print representation when you change things like *PACKAGE*
2067 and *PRINT-LENGTH*) to symbols and lists of symbols
2068 He also made several improvements and fixed several bugs in DESCRIBE.
2069 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2071 *** misbehavior of WRITE-STRING/WRITE-LINE
2072 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2073 and several other LOOP problems as well
2074 *** DIRECTORY when similar filenames are present
2075 *** DEFGENERIC with :METHOD options
2076 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2077 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2078 *** argument ordering in FIND with :TEST option
2079 *** mishandled package designator argument in APROPOS-LIST
2080 *** various problems in the backquote readmacro
2081 *** a bug in APROPOS
2082 *** probably some others that I'm not describing very well here,
2083 since the CVS log documents them by reference to sbcl-devel
2084 messages, and the SourceForge archives aren't working well.:-(
2085 ** Dan Barlow improved the Alpha port (and is making progress on the
2086 PPC port, for those of you who think different).
2087 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2088 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2089 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2090 continues to work on the SPARC port (for those of you in a position
2091 to look down upon our little PC-compatible boxes from a great height).
2092 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2093 and uses that copy, rather than the current dynamic value, when
2094 it's trying to decide whether to truncate output. Thus e.g.
2095 (let ((*print-lines* 50))
2096 (pprint-logical-block (stream nil)
2098 (let ((*print-lines* 8))
2099 (print (aref possiblybigthings i) stream)))))
2100 should now truncate the logical block only at 50 lines, instead of
2101 often truncating it at 8 lines, as it did before.
2102 * The doc/cmucl/ directory, containing old CMU CL documentation
2103 from the time of the fork, is no longer part of the base system.
2104 SourceForge has shut down its anonymous FTP service, and with it
2105 my original plan for distributing the old CMU CL documentation
2106 there. For now, if you need these files you can download an old
2107 SBCL source release and extract them from it.
2108 * The fasl file version number changed again, for dozens of reasons,
2109 some of which are apparent above.
2111 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2112 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2113 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2114 patch, so that DISASSEMBLE output is much nicer.
2115 * The code in the SB-PROFILE package now seems reasonably stable.
2116 I still haven't decided what the final interface should look like
2117 (I'd like PROFILE to interact cleanly with TRACE, since both
2118 facilities use function encapsulation) but if you have a need
2119 for profiling now, you can probably use it successfully with
2120 the current CMU-CL-style interface.
2121 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2122 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2123 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2124 link now returns the dangling link itself, and for similar
2125 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2126 link itself. (In these cases the old code signalled an error and
2127 looped endlessly, respectively.) Thus, DIRECTORY now works even
2128 in the presence of dangling and cyclic symbolic links.
2129 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2130 is now a supported extension again, since the consensus on
2131 sbcl-devel was that it can be useful for ordinary development
2132 work, not just for debugging SBCL itself.
2133 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2134 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2135 that currently-defined functions might be redefined later with
2136 different return types.
2137 * Hash tables can be printed readably, as inspired by CMU CL code
2138 of Eric Marsden and SBCL code of Martin Atzmueller.
2139 * better error handling in CLOS method combination, thanks to
2140 Martin Atzmueller porting Pierre Mai's CMU CL patches
2141 * more overflow fixes for >16Mbyte I/O buffers
2142 * A bug in READ has been fixed, so that now a single Ctrl-D
2143 character suffices to cause end-of-file on character streams.
2144 In particular, now you only need one Ctrl-D at the command
2145 line (not two) to exit SBCL.
2146 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2148 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2149 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2150 does the right thing.
2151 * The implementation of some type tests, especially for CONDITION
2152 types, is now tidier and maybe faster, due to CMU CL code
2153 originally by Douglas Crosher, ported by Martin Atzmueller.
2154 * Some math functions have been fixed, and there are new
2155 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2156 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2157 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2158 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2159 likely to be built into sbcl-0.7.0.)
2160 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2161 CL, ported to SBCL by Martin Atzmueller)
2162 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2163 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2164 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2165 * There's a new slam.sh hack to shorten the edit/compile/debug
2166 cycle for low-level changes to SBCL itself, and a new
2167 :SB-AFTER-XC-CORE target feature to control the generation of
2168 the after-xc.core file needed by slam.sh.
2169 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2170 COMPILE-FILE is no longer supported, so that now every function
2171 gets an entry point, so that block compilation looks a little
2172 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2173 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2174 returns the number of bytes consed since the system started,
2175 rather than the number consed since the first time the function
2176 was called. (The new definition parallels ANSI functions like
2177 CL:GET-INTERNAL-RUN-TIME.)
2178 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2179 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2180 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2181 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2182 * Fasl file version numbers are now independent of the target CPU,
2183 since historically most system changes which required version
2184 number changes have affected all CPUs equally. Similarly,
2185 the byte fasl file version is now equal to the ordinary
2188 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2189 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2190 is no longer recognized. I apologize for this, because it was
2191 listed in SB-EXT as a supported extension, but I found that
2192 its existing behavior was poorly specified, as well as incorrectly
2193 specified, and it looked like too much of a mess to straighten it
2194 out. I have enough on my hands trying to get ANSI stuff to work..
2195 * many patches ported from CMU CL by Martin Atzmueller, with
2196 half a dozen bug fixes in pretty-printing and the debugger, and
2197 half a dozen others elsewhere
2198 * fixed bug 13: Floating point infinities are now supported again.
2199 They might still be a little bit flaky, but thanks to bug reports
2200 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2201 as flaky as they were.
2202 * The --noprogrammer command line option is now supported. (Its
2203 behavior is slightly different in detail from what the old man
2204 page claimed it would do, but it's still appropriate under the
2205 same circumstances that the man page talks about.)
2206 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2207 are now supported, and enabled by default. Thus, the compiler can
2208 handle many floating point and complex operations much less
2209 inefficiently. (Thus e.g. you can implement a complex FFT
2211 * The compiler now detects type mismatches between DECLAIM FTYPE
2212 and DEFUN better, and implements CHECK-TYPE more correctly, and
2213 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2214 thanks to patches from Martin Atzmueller.
2215 * various fixes to make the cross-compiler more portable to
2216 ANSI-conforming-but-different cross-compilation hosts (notably
2217 Lispworks for Windows, following bug reports from Arthur Lemmens)
2218 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2219 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2220 fixed thanks to Pierre Mai's CMU CL patches.
2221 * fixes to keep the system from overflowing internal counters when
2222 it tries to use i/o buffers larger than 16M bytes
2223 * fixed bug 45a: Various internal functions required to support
2224 complex special functions have been merged from CMU CL sources.
2225 (When I was first setting up SBCL, I misunderstood a compile-time
2226 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2227 * improved support for type intersection and union, fixing bug 12
2228 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2229 more obscure bugs as well
2230 * some steps toward byte-compiling non-performance-critical
2231 parts of the system, courtesy of patches from Martin Atzmueller
2232 * Christophe Rhodes has made some debian packages of sbcl at
2233 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2234 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2235 stable, but are nonetheless usable. When he's ready, I'd be happy
2236 to add them to the SourceForge "File Releases" section. (And if
2237 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2238 * new fasl file format version number (because of changes in
2239 internal representation of (OR ..) types to accommodate the new
2240 support for (AND ..) types, among other things)
2242 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2243 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2245 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2247 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2248 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2249 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2250 * DESCRIBE now works on structure objects again.
2251 * Most function call argument type mismatches are now handled as
2252 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2253 know whether the function will be redefined before the call is
2254 executed. (The compiler could flag local calls with full WARNINGs,
2255 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2256 it doesn't keep track of enough information to know whether calls
2257 are local in this sense.)
2258 * Compiler output is now more verbose, with messages truncated
2259 later than before. (There should be some supported way for users
2260 to override the default verbosity, but I haven't decided how to
2261 provide it yet, so this behavior is still controlled by the internal
2262 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2263 src/compiler/ir1util.lisp.)
2264 * Fasl file format version numbers have increased again, because
2265 support for the Gray streams extension changes the layout of the
2266 system's STREAM objects.
2267 * The Gray subclassable streams extension now works, thanks to a
2268 patch from Martin Atzmueller.
2269 * The full LOAD-FOREIGN extension (not just the primitive
2270 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2271 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2272 but like most other programs, it defaults to copying the Unix
2273 environment from the original process instead of starting the
2274 new process in an empty environment.
2275 * Extensions which manipulate the Unix environment now support
2276 an :ENVIRONMENT keyword option which doesn't smash case or
2277 do other bad things. The CMU-CL-style :ENV option is retained
2278 for porting convenience.
2279 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2280 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2282 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2284 * A patch from Martin Atzmueller seems to have solved the SIGINT
2285 problem, and as far as we know, signal-handling now works cleanly.
2286 (If you find any new bugs, please report them!)
2287 * The system no longer defaults Lisp source file names to types
2288 ".l", ".cl", or ".lsp", but only to ".lisp".
2289 * The compiler no longer uses special default file extensions for
2290 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2291 seems to expect a single default extension for all compiled code,
2292 and there's no compelling reason to try to stretch the standard
2293 to allow two different extensions.) Instead, byte-compiled files
2294 default to the same extension as native-compiled files.
2295 * Fasl file format version numbers have increased again, because
2296 a rearrangement of internal implementation packages made some
2297 dumped symbols in old fasl files unreadable in new cores.
2298 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2299 many fewer weird special cases.
2300 * Bug #17 (differing COMPILE-FILE behavior between logical and
2301 physical pathnames) has been fixed, and some related misbehavior too,
2302 thanks to a patch from Martin Atzmueller.
2303 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2304 by Tim Moore, ported to SBCL by Martin Atzmueller.
2305 * Martin Atzmueller fixed several filesystem-related problems,
2306 including bug #36, in part by porting CMU CL patches, which were
2307 written in part by Paul Werkowski.
2308 * More compiler warnings in src/runtime/ are gone, thanks to
2309 more patches from Martin Atzmueller.
2310 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2313 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2315 * DESCRIBE now works on CONDITION objects.
2316 * The debugger now handles errors which arise when trying to print
2317 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2319 * The build system now uses an additional file, customize-target-features.lisp,
2320 to allow local modifications to the target *FEATURES* list. (The point of
2321 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2322 debugging features enabled, without having to worry about propagating it
2323 into everyone's system when I do a "cvs update".) When no
2324 customize-target-features.lisp file exists, the target *FEATURES* list
2325 should be constructed the same way as before.
2326 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2327 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2328 warn about behavior which is undefined under the ANSI spec. Note: This
2329 is specified by ANSI, but it's not very popular with programmers.
2330 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2331 macro in the SBCL sources for an example of a workaround which you
2332 might use to make portable ANSI-standard code which does what you want.
2333 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2334 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2335 when it's used as a non-toplevel form. (This is required in order
2336 to implement the DEFCONSTANT-EQX macro.)
2337 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2338 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2339 funny things, due to bugs in the implementation of EVAL-WHEN
2340 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2341 non-toplevel DEFCONSTANTs are uncommon.)
2342 * The core file version number and fasl file version number have been
2343 incremented, because the old noncompliant DEFCONSTANT behavior involved
2344 calling functions which no longer exist, and because I also took the
2345 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2346 * fixed bug 1 (error handling before read-eval-print loop starts), and
2347 redid debugger restarts and related debugger commands somewhat while
2349 ** The QUIT debugger command is gone, since it did something
2350 rather different than the SB-EXT:QUIT command, and since it never
2351 worked properly outside the main toplevel read/eval/print loop.
2352 Invoking the new TOPLEVEL restart provides the same functionality.
2353 ** The GO debugger command is also gone, since you can just invoke
2354 the CONTINUE restart directly instead.
2355 ** The TOP debugger command is also gone, since it's redundant with the
2356 FRAME 0 command, and since it interfered with abbreviations for the
2358 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2360 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2361 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2362 some porting work by Martin Atzmueller)
2363 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2364 with a boatload of patches which he ported to SBCL. Now that those
2366 ** The system tries to make sure that its low-priority messages
2367 are prefixed by semicolons, to help people who like to use
2368 syntax highlighting in their ILISP buffer. (This patch
2369 was originally due to Raymond Toy.)
2370 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2371 patches originally written by Raymond Toy.
2372 ** The compiler understands coercion between single-value and
2373 multiple-VALUES type expressions better, getting rid of some very
2374 weird behavior, thanks to patches originally by Robert MacLachlan
2375 and Douglas Crosher.
2376 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2377 lambda lists, thanks to a patch originally by Pierre Mai.
2378 ** The system no longer bogusly warns about "abbreviated type
2380 ** The compiler gets less confused by inlining and RETURN-FROM,
2381 thanks to some patches originally by Tim Moore.
2382 ** The system no longer hangs when dumping circular lists to fasl
2383 files, thanks to a patch originally from Douglas Crosher.
2384 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2385 undefined function error.
2386 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2387 in structure slot initforms) for now, documented workaround instead:-|
2388 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2389 * fixed bug 5: added stubs for various Gray stream functions called
2390 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2391 installed, at least appropriate type errors are generated
2392 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2393 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2394 consistently in DEFMETHOD forms.
2395 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2396 it doesn't seem to affect SBCL after all
2397 * The C runtime system now builds with better optimization and many
2398 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2400 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2402 * The system is now under CVS at SourceForge (instead of the
2403 CVS repository on my home machine).
2404 * The new signal handling code has been tweaked to treat register
2405 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2406 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2407 downstream, e.g. in debug-int.lisp, has implicit dependencies
2408 on the unsignedness of integer representation of machine words,
2409 and that caused the system to bomb out with infinite regress
2410 when trying to recover from type errors involving signed values,
2411 e.g. (BUTLAST '(1 2 3) -1).)
2412 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2413 (This was one of the bugs Peter Van Eynde reported back in July.)
2414 * The system now uses code inspired by Colin Walters' O(N)
2415 implementation of MAP (from the cmucl-imp@cons.org mailing
2416 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2417 inline the MAP operation, and there is more than one
2418 sequence argument to the MAP call (so that it can't just
2419 do ETYPECASE once and for all based on the type of the
2420 single sequence argument). (The old non-inline implementation
2421 of the general M-argument sequence-of-length-N case required
2422 O(M*N*N) time when any of the sequence arguments were LISTs.)
2423 * The QUIT :UNIX-CODE keyword argument has been renamed to
2424 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2425 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2427 * Raymond Wiker's patches to port dynamic loading from Linux to
2428 FreeBSD have been added.
2429 * The BUGS file is now more nearly up to date, thanks in large part
2430 to Martin Atzmueller's review of it.
2431 * The debugger now flushes standard output streams before it begins
2432 its output ("debugger invoked" and so forth).
2433 * The core version number and fasl file version number have both
2434 been incremented, because of incompatible changes in the layout
2436 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2437 used to not be called for a saved Lisp image.)
2438 * Martin Atzmueller's version of a patch to fix a compiler crash,
2439 as posted on sbcl-devel 13 September 2000, has been installed.
2440 * Instead of installing Martin Atzmueller's patch for the
2441 compiler transform for SUBSEQ, I deleted the compiler transform,
2442 and transforms for some similar consing operations.
2443 * A bug in signal handling which kept TRACE from working on OpenBSD
2445 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2446 away by constant folding
2447 * The system now defines its address space constants in one place
2448 (in the Lisp sources), and propagates them automatically elsewhere
2449 (through GENESIS and the sbcl.h file). Therefore, patching the
2450 address map is less unnecessarily tedious and error-prone. The
2451 Lisp names of address space constants have also been systematized.
2452 * CVS tags like dollar-Header-dollar have been removed from
2453 the sources, because they have never saved me trouble and
2454 they've been source of trouble working with patches and other
2455 diff-related operations.
2456 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2457 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2459 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2461 * The system has been ported to OpenBSD.
2462 * The system now compiles with a simple "sh make.sh" on the systems
2463 that it's supported on. I.e., now you no longer need to tweak
2464 text in the target-features.lisp-expr and symlinks in src/runtime/
2465 by hand, the make.sh takes care of it for you.
2466 * The system is no longer so grossly inefficient when compiling code
2467 involving vectors implemented as general (not simple) vectors (VECTOR T),
2468 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2469 which dares to use the various sequence functions on non-simple
2470 vectors, takes less of a performance hit.
2471 * There is now a primitive type predicate VECTOR-T-P
2472 to test for the (VECTOR T) type, so that e.g.
2473 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2474 can now be compiled with some semblance of efficiency. (The old code
2475 turned the type declaration into a full call to %TYPEP at runtime!)
2476 * AREF on (VECTOR T) is still not fast, since it's still compiled
2477 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2478 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2479 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2480 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2481 without making full calls to SUBTYPEP at runtime.
2482 (Some analogous efficiency issues for non-simple vectors specialized to
2483 element types other than T, or for non-simple multidimensional arrays,
2484 have not been addressed. They could almost certainly be handled the
2485 same way if anyone is motivated to do so.)
2486 * The changes in array handling break binary compatibility, so
2487 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2488 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2489 instead of (VALUES T T).
2490 * By following the instructions that Dan Barlow posted to sbcl-devel
2491 on 2 July 2000, I was able to enable primitive dynamic object
2492 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2493 functionality is not implemented (since it calls ld to resolve
2494 library references automatically, requiring RUN-PROGRAM for its
2495 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2496 try to resolve library references) is now supported.
2497 * The system now flushes the standard output streams when it terminates,
2498 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2499 them at several other probably-convenient times, e.g. in each pass of
2500 the toplevel read-eval-print loop, and after evaluating a form given
2501 as an "--eval" command-line option. (These changes were motivated by a
2502 discussion of stream flushing issues on cmucl-imp in August 2000.)
2503 * The source transform for TYPEP of array types no longer assumes
2504 that an array whose element type is a not-yet-defined type
2505 is implemented as an array of T, but instead punts, so that the
2506 type will be interpreted at runtime.
2507 * There is now some support for cross-compiling in make.sh: each of
2508 the phases of make.sh has its own script. (This should be transparent
2509 to people doing ordinary, non-cross-compile builds.)
2510 * Since my laptop doesn't have hundreds of megabytes of memory like
2511 my desktop machine, I became more motivated to do some items on
2512 my to-do list in order to reduce the size of the system a little:
2513 ** Arrange for various needed-only-at-cold-init things to be
2514 uninterned after cold init. To support this, those things have
2515 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2516 symbols with such names are now uninterned after cold init).
2517 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2518 fasl files for cold load.
2519 ** Remove the old compiler structure pooling code (which used to
2520 be conditional on the target feature :SB-ALLOC) completely.
2521 ** Redo the representation of some data in cold init to be more compact.
2522 (I also looked into supporting byte compiled code at bootstrap time,
2523 which would probably reduce the size of the system a lot, but that
2524 looked too complicated, so I punted for now.)
2525 * The maximum signal nesting depth in the src/runtime/ support code has
2526 been reduced from 4096 to 256. (I don't know any reason for the very
2527 large old value. If the new smaller value turns out to break something,
2528 I'll probably just bump it back up.)
2529 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2531 * Many, many bugs reported by Peter Van Eynde have been added to
2532 the BUGS list; some have even been fixed.
2533 * While enabling dynamic object file loading, I tried to make the
2534 code easier to understand, renaming various functions and variables
2535 with less ambiguous names, and changing some function calling
2536 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2537 * While trying to figure out how to do the OpenBSD port, I tried to
2538 clean up some of the code in src/runtime/. In particular, I dropped
2539 support for non-POSIX signal handling, added various comments,
2540 tweaked the code to reduce the number of compilation warnings, and
2541 renamed some files to increase consistency.
2542 * To support the new automatic configuration functionality in make.sh,
2543 the source file target-features.lisp-expr has been replaced with the
2544 source file base-target-features.lisp-expr and the machine-generated
2545 file local-target-features.lisp-expr.
2546 * fixed a stupid quoting error in make.sh so that using CMU CL
2547 "lisp -batch" as cross-compilation host works again
2549 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2551 * DESCRIBE no longer tries to call itself recursively to describe
2552 bound/fbound values, so that it no longer fails on symbols which are
2553 bound to themselves (like keywords, T, and NIL).
2554 * DESCRIBE now works on generic functions.
2555 * The printer now prints less-screwed-up representations of closures
2556 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2557 * A private symbol is used instead of the :EMPTY keyword previously
2558 used to mark empty slots in hash tables. Thus
2559 (DEFVAR *HT* (MAKE-HASH-TABLE))
2560 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2561 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2562 now does what ANSI says that it should. (You can still get
2563 similar noncompliant behavior if bang on the hash table
2564 implementation with all the symbols you get back from
2565 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2566 This breaks binary compatibility, since tests for equality to
2567 :EMPTY are wired into things like the macroexpansion of
2568 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2570 * There's now a minimal placeholder implementation for CL:STEP,
2571 as required by ANSI.
2572 * An obscure bug in the interaction of the normal compiler, the byte
2573 compiler, inlining, and structure predicates has been patched
2574 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2575 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2577 * Missing ordinary arguments in a macro call are now detected even
2578 when the macro lambda list contains &KEY or &REST.
2579 * The debugger no longer complains about encountering the top of the
2580 stack when you type "FRAME 0" to explicitly instruct it to go to
2581 the top of the stack. And it now prints the frame you request even
2582 if it's the current frame (instead of saying "You are here.").
2583 * As specified by ANSI, the system now always prints keywords
2584 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2586 * The default initial SIZE of HASH-TABLEs is now smaller.
2587 * Type information from CLOS class dispatch is now propagated
2588 into DEFMETHOD bodies, so that e.g.
2589 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2591 is now basically equivalent to
2592 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2593 (DECLARE (TYPE SINGLE-FLOAT X))
2595 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2596 operation, without having to do run-time type dispatch.
2597 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2598 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2599 * Since I don't seem to be making big file reorganizations very often
2600 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2601 is overflowing my ability to conveniently back them up), I've finally
2602 checked the system into CVS. (The CVS repository is on my home system,
2603 not at SourceForge -- putting it on SourceForge might come later.)
2604 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2605 high-level GC-NOTIFY-FOO functions send their output. (There's
2606 still very little control of where low-level verbose GC functions
2607 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2608 less than it used to -- the GC-NOTIFY-FOO functions are now under
2609 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2610 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2611 in only one place in the source code, and propagates it automatically
2612 everywhere that it's needed. Thus e.g. when I bump the version from
2613 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2614 * The C source files now include boilerplate legalese and documentation
2615 at the head of each file (just as the Lisp source files already did).
2616 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2617 to his page will be replaced with a link to his new CLiki service.
2619 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2621 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2622 * The build process now looks for GNU make under the default name "gmake",
2623 instead of "make" as it used to. If GNU make is not available as "gmake"
2624 on your system, you can change this default behavior by setting the
2625 GNUMAKE environment variable.
2626 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2627 can build without error under CMU CL.
2629 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2631 * There is now a partial SBCL user manual (with some new text and some
2632 text cribbed from the CMU CL manual).
2633 * The beginnings of a profiler have been added (starting with the
2634 CMU CL profiler and simplifying and cleaning up). Eventually the
2635 main interface should be through the TRACE macro, but for now,
2636 it's still accessed through vaguely CMU-CL-style functions and macros
2637 exported from the package SB-PROFILE.
2638 * Some problems left over from porting CMU CL to the new
2639 cross-compilation bootstrap process have been cleaned up:
2640 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2641 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2642 part when getting rid of the compiler *BACKEND* stuff.)
2643 ** The value of *NULL-TYPE* was screwed up, because it was
2644 being initialized before the type system knew the final
2645 definition of the 'NULL type. This screwed up several key
2646 optimizations in the compiler, causing inefficiency in all sorts
2647 of places. (I found it because I wanted to understand why
2648 GET-INTERNAL-RUN-TIME was consing.)
2649 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2650 PROCLAIM FTYPE information. Unfortunately this broke binary
2651 compatibility again, since now the forms output by DEFGENERIC
2652 to refer to functions which didn't exist in 0.6.3.
2653 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2654 can use the new (as of 0.6.3) transform for SOME into MAP into
2656 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2657 values to (INTEGER 0 1000000), so that the time code will no longer
2658 occasionally get blown up by Linux returning 1000000 microseconds
2659 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2660 its output conform to the ANSI spec. (Alas, this makes its output
2661 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2662 * A full call to MAP NIL with a single sequence argument no longer conses.
2663 * fixes to problems pointed out by Martin Atzmueller:
2664 * The manual page no longer talks about multiprocessing as though
2665 it were currently supported.
2666 * The ILISP support patches have been removed from the distribution,
2667 because as of version 5.10.1, ILISP now supports SBCL without us
2668 having to maintain patches.
2669 * added a modified version of Raymond Toy's recent CMU CL patch for
2670 EQUALP comparison of HASH-TABLE
2672 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2674 * The system still can't cross-compile itself with
2675 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2676 entails), but at least it can compile more of itself that way
2677 than it used to be able to, and various buglets which were uncovered
2678 by trying to cross-compile itself that way have now been fixed.
2679 * This release breaks binary compatibility again. This time
2680 at least I've incremented the FASL file format version to 2, so that the
2681 problem can be detected reliably instead of just causing weird errors.
2682 * various new style warnings:
2683 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2684 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2685 ** using the lexical binding of a variable named in the *FOO* style
2686 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2687 as specified by ANSI.
2688 * *RANDOM-STATE* is no longer automatically initialized from
2689 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2690 default behavior of the system is to repeat its behavior every time
2691 it's run. If you'd like to change this behavior, you can always
2692 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2693 old convention there was no comparably easy way to get the system to
2694 repeat its behavior every time it was run.
2695 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2696 FTYPE declarations has been removed, in favor of their ANSI
2697 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2698 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2699 the types of their sequence arguments can be determined at compile time.
2700 This is done through a new open code expansion for MAP which eliminates
2701 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2702 argument types can be determined at compile time.
2703 * The optimizer now transforms COERCE into an identity operation when it
2704 can prove that the coerced object is already of the correct type. (This
2705 can be a win for machine generated code, including the output of other
2706 optimization transforms, such as the MAP transform above.)
2707 * Credit information has been moved from source file headers into CREDITS.
2708 * Source file headers have been made more standard.
2709 * The CASE macro now compiles without complaining even when it has
2712 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2714 * (Note that the way that the PCL macroexpansions were rewritten
2715 to accommodate the change in DEFGENERIC below breaks binary
2716 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2717 not run under sbcl-0.6.2. Once we get out of alpha releases,
2718 i.e. hit release 1.0.0, we'll probably try to maintain binary
2719 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2720 and sbcl-1.4.4. Until then, however, it might be fairly common
2721 for maintenance releases to break binary compatibility.)
2722 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2724 * The reporting of unhandled signals has been changed to print some
2725 explanatory text as well as the report form. (Previously only
2726 the report form was printed.)
2727 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2728 it defines, so that the compiler no longer issues undefined function
2729 warnings for compiled-but-not-yet-loaded generic functions.
2730 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2731 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2732 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2733 section of the ANSI spec are supported.
2734 * The "" nickname for the "KEYWORD" package has been removed.
2735 The reader still handles symbol tokens which begin with a package marker
2736 as keywords, but it doesn't expose its mechanism for doing so in the
2737 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2738 * The system now issues STYLE-WARNINGs for contradictory TYPE
2739 proclamations. (Warnings for contradictory FTYPE proclamations would
2740 be nice too, but those can't be done usefully unless the type system
2741 is made smarter about FUNCTION types.)
2742 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2743 systematized, so that "*target-*.lisp is supposed to exist only on the
2744 target and imply that there's a related file which exists on the
2745 host, and *host-*.lisp is supposed to exist only on the host and imply
2746 that there's a related file which exists on the target. This involves a
2747 lot of renaming. Hopefully the acute confusion caused by the renaming
2748 will be justified by the reduction in chronic confusion..
2749 ** runtime-type.lisp -> early-target-type.lisp
2750 ** target-type.lisp -> late-target-type.lisp
2751 ** early-host-format.lisp -> early-format.lisp
2752 ** late-host-format.lisp -> late-format.lisp
2753 ** host-error.lisp -> misc-error.lisp
2754 ** early-error.lisp -> early-target-error.lisp
2755 ** late-error.lisp -> late-target-error.lisp
2756 ** host-defboot.lisp -> early-defboot.lisp
2757 ** code/misc.lisp -> code/target-misc.lisp
2758 ** code/host-misc.lisp -> code/misc.lisp
2759 ** code/numbers.lisp -> code/target-numbers.lisp
2760 ** code/early-numbers.lisp -> numbers.lisp
2761 ** early-host-type.lisp -> early-type.lisp
2762 ** late-host-type.lisp -> late-type.lisp
2763 ** host-typep.lisp -> typep.lisp
2764 ** load.lisp -> target-load.lisp
2765 ** host-load.lisp -> load.lisp
2766 ** host-disassem.lisp -> disassem.lisp
2767 ** host-insts.lisp -> insts.lisp
2768 ** byte-comp.lisp -> target-byte-comp.lisp
2769 ** host-byte-comp.lisp -> byte-comp.lisp
2770 ** host-signal.lisp -> signal.lisp
2771 ** host-defstruct.lisp -> defstruct.lisp
2772 ** late-target-type.lisp -> deftypes-for-target.lisp
2773 Furthermore, several other previously target-only files foo.lisp (e.g.
2774 hash-table.lisp and random.lisp) have been split into a target-and-host
2775 foo.lisp file and a target-only target-foo.lisp file, with their key type
2776 definitions in the target-and-host part, so that the cross-compiler will
2777 know more about target types.
2778 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2779 gone away. In their place are various *BACKEND-FOO* variables
2780 corresponding to the slots of the old structure.
2781 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2782 into the target SBCL has been fixed.
2783 * The chill.lisp system for loading cold code into a running SBCL
2785 * Support for the CMU CL "scavenger hook" extension has been removed.
2786 (It was undocumented and unused in the CMU CL sources that SBCL was
2787 derived from, and stale in sbcl-0.6.1.)
2788 * Various errors in the cross-compiler type system were detected
2789 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2790 (enabling various consistency checks). Many of them were fixed,
2791 but some hard problems remain, so the compiler is back to
2792 running without *TYPE-SYSTEM-INITIALIZED* for now.
2793 * As part of the cross-compiler type system cleanup, I implemented
2794 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2795 * I have started adding UNCROSS calls throughout the type system
2796 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2797 on cross-compiler input files will be able to go away, and various
2799 * CONSTANTP now returns true for quoted forms (as explicitly required
2802 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2804 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2805 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2806 (MAKE-LIST -1). In the longer term, it ought to become true
2807 that declarations are assertions even at SAFETY 1. For now, it's not
2808 quite true even at SAFETY 3, but it's at least more nearly true..
2809 (Note that this change seems to increases the size of the system by
2810 O(5%) and to decrease the speed of the compiler by 20% or more.)
2811 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2812 for the problem of printing dozens of lines of distracting information
2813 about low-level system machinery as part of the top stack frame
2814 on entry to the debugger when an undefined function was called.
2815 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2816 is not reset to COMMON-LISP-USER.
2817 * Compilation of stuff related to dyncount.lisp has been made conditional
2818 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2819 smaller. The various dyncount-related symbols have been moved into
2820 a new "SB-DYNCOUNT" package.
2821 * tty-inspect.lisp has been renamed to inspect.lisp.
2822 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2823 feature has gone away. (When we eventually port to other flavors of
2824 libc and/or Unix, we'll try to make the differences between flavors
2825 invisible at the user level.)
2826 * Various other *FEATURES* tags, and/or their associated conditionals,
2827 have been removed if obsolescent, or given better documentation, or
2828 sometimes given more-mnemonic names.
2830 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2832 * tidied up "make.sh" script
2833 * tidied up system directory structure
2834 * better "clean.sh" behavior
2835 * added doc/FOR-CMUCL-DEVELOPERS
2836 * many many small tweaks to output format, e.g. removing possibly-confusing
2837 trailing #\. character in DESCRIBE-INSTANCE
2838 * (EQUALP #\A 'A) no longer signals an error.
2839 * new hashing code, including EQUALP hashing
2840 * tidied up Lisp initialization and toplevel
2841 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2842 * command line argument processing
2843 * added POSIX-GETENV function to deal with Unix-ish environment variables
2844 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2845 terminating SBCL on EOF
2846 * non-verbose GC by default
2847 * There is no more "sbcl" shell script; the sbcl file is now the C
2848 runtime executable (just like CMU CL).
2849 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2851 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2852 symbol and package information as Lisp native symbol and package objects
2853 instead of strings naming symbols and strings naming packages. This way,
2854 whenever packages are renamed (as in warm init), debug information is
2855 transformed along with everything else.
2856 * tweaked the optimization policy declarations which control the building
2857 of SBCL itself. Now, among other things, the system no longer saves
2858 source location debugging information. (This helps two problems at once
2859 by reducing SBCL size and by keeping SBCL from trying to look for its
2860 sources -- which may not exist -- when reporting errors.)
2861 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2862 debugging and testing purposes
2863 * cleaned up printing, making the printer call PRINT-OBJECT for
2864 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2865 methods, giving nearly-ANSI behavior
2866 * converted almost all special variables to use *FOO* naming convention
2867 * deleted PARSE-TIME functionality, since it can be done portably
2868 * moved some files out of cold init into warm init
2869 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2871 * regularized formatting of source files
2872 * added an install.sh script
2873 * fixed ridiculous memory usage of cross-compiler by making
2874 compiler/alloc.lisp not try to do pooling unless it can hook
2875 itself into the GC of the cross-compilation host. Now the system
2876 builds nicely on my old laptop.
2877 * added :SB-ALLOC in target-features.lisp-expr
2878 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2879 was not implemented)
2880 * re-did condition handling and note reporting in the compiler. Notes
2881 are no longer handled by signalling conditions. Style warnings
2882 and warnings are handled more correctly and reported in such a way
2883 that it's easy to find one or the other in your output (so that you
2884 can e.g. figure out which of many problems caused COMPILE-FILE to
2886 * changed the severity of several compiler warnings from full WARNING
2887 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2888 compiler note reporting so that it doesn't use the condition system
2889 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2890 COMPILE-FILE command)
2891 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2892 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2893 no longer supported, since their implementation was deeply intertwingled
2894 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2895 * removed bogus "support" for compiler macros named (SETF FOO), and
2896 removed the compiler macro for SETF INFO (but only after making a fool
2897 of myself on the cmucl-imp mailing list by posting a bogus patch for
2898 DEFINE-COMPILER-MACRO..)
2899 * Compiled files containing forms which have side effects on the Lisp
2900 reader (such as DEFPACKAGE forms) are now handled more correctly.
2901 (Compiler queuing of top level lambdas has been suppressed by setting
2902 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2903 * deleted various currently-unused source files, e.g. gengc.lisp. They
2904 may be added back at some point e.g. when porting to other architectures,
2905 but until they are it's distracting to distribute them and to try to
2907 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2908 there were so many of them they're just distractions, and UNCROSS is
2909 known to be able to handle the current sources
2910 * moved PROFILE functionality into TRACE, so that it will be clear
2911 how the wrapping and unwrapping of functions when you profile them
2912 interacts with the wrapping and unwrapping of functions when you
2913 trace them. (Actually, the functionality isn't there yet, but at least
2914 the interface specification is there. Hopefully, the functionality will
2915 arrive with some future maintenance release.)
2916 * removed host-oops.lisp
2917 * changed signature of QUIT function to allow UNIX-CODE argument
2918 * fixed READ-SEQUENCE bug
2919 * tweaked verbose GC output so that it looks more like the progress
2920 output that ANSI specifies for functions like LOAD
2921 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2922 * added <http://sbcl.sourceforge.com> to the banner information printed by