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 * enhancement: the statistical profiler now walks deeper into the
9 call stack for more meaningful call-graphs and accrued time
10 reports (x86/x86-64 only). It also now reports time spent in
12 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
14 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
15 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
16 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
17 produces an error. (thanks to Vincent Arkesteijn)
18 * bug fix: NAMESTRING on pathnames with :WILD components in their
19 directories works correctly. (thanks to Artem V. Andreev)
20 * fixed bug 125: compiler preserves identity of closures. (reported
22 * build fix: fixed the dependence on *LOAD-PATHNAME* and
23 *COMPILE-FILE-PATHNAME* being absolute pathnames.
24 * on x86 compiler partially supports stack allocation of dynamic-extent
26 * fixed some bugs related to Unicode integration:
27 ** encoding and decoding errors are now much more robustly
28 handled; it should now be possible to recover even from invalid
29 input or output to the terminal. (thanks to Teemu Kalvas)
30 ** provided a first cut at implementing STRING-TO-OCTETS and
31 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
32 ** altered the SB-MD5 contributed module slightly, changing the
33 interface just enough for it to be supportable for builds where
34 lisp characters are not eight bits.
35 * fixed some bugs revealed by Paul Dietz' test suite:
36 ** the FORMATTER-generated functions for ~V[ conditionals require
37 the correct number of arguments.
39 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
40 * new feature: reloading changed shared object files with
41 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
42 * new feature: references to foreign variables and functions
43 can now be compiled and loaded before the corresponding shared
44 object file is loaded, as long as the foreign definitions are
46 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
47 supported. (thanks to Dan Debertin)
48 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
49 just keywords. (thanks to Vincent Arkesteijn)
50 * fixed bug #331: structure-class instances corresponding to
51 DEFSTRUCT forms are now created eagerly.
52 * fixed bug #345: backtraces from calls to undefined functions work
53 on x86 as well. Related bug #61 is now also partially fixed on x86
54 (backtraces from throws to unknown catch tags.)
55 * bug fix: lambda-list parsing is now stricter vrt. order and number
56 of lambda-list keywords.
57 * bug fix: as specified by AMOP, an error is signalled if a
58 class-option appears multiple times in a DEFCLASS form. (reported
60 * bug fix: RANDOM can be compiled when the compiler derives the type
61 of its numeric argument as a disjoint set of small integers.
62 (reported by Paul Dietz)
63 * bug fix: starting a core saved with shared objects loaded when
64 those objects are not available no longer causes threaded SBCL to
65 hang. (reported by Sean Ross)
66 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
68 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
69 argument, which must always be NIL. (reported by Kalle Niemitalo)
70 * bug fix: printing 1.0d+23 no longer results in an error.
71 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
72 * bug fix: structure slot setters preserve evaluation order. (thanks
74 * fixed some bugs related to Unicode integration:
75 ** RUN-PROGRAM can allow its child to take input from a Lisp
76 stream. (reported by Stefan Scholl)
77 ** ASDF-INSTALL successfully downloads .tar.gz files in all
78 locales. (reported by Ken Causey)
79 * fixed some bugs revealed by Paul Dietz' test suite:
80 ** INCF, DECF and REMF evaluate their place form as specified in
82 ** FORMATTER expands ~{ ~} iteration directives with V or #
84 ** FORMATTER deals with the ~@[ ~] conditional directive where the
85 consequent uses no arguments correctly.
86 ** the system has a partial understanding of the (COMPLEX RATIO)
88 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
91 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
92 * new feature: a build-time option (controlled by the :SB-UNICODE
93 keyword feature, enabled by default) for building the system with
94 support for the entire 21-bit character space defined by the
96 * new feature: the system now has rudimentary external-format
97 support; the primary user-visible change at this time is that
98 characters with the high bit set (such as lower-case-e-acute) will
99 print correctly to a terminal in a UTF-8 environment.
100 * minor incompatible change: BASE-CHAR no longer names a class;
101 however, CHARACTER continues to do so, as required by ANSI.
102 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
103 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
104 has been moved to the SB-EXT package (temporarily re-exported from
106 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
107 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
109 * fixed bug #351: better error-handlind and reporting for malformed
111 * fixed bug #350: bignum-printing is now more memory-efficient,
112 allowing printing of very large bignums, eg. (expt 2 10000000).
113 (reported by Bruno Haible)
114 * fixed bug #302: better primitive-type selection for intersection
116 * fixed bug #308: non-graphic characters now all have names, as
117 required. (reported by Bruno Haible)
118 * bug fix: redefining a class with different superclasses now correctly
119 removes it from the direct-subclasses of its previous superclasses.
120 (reported by David Morse)
121 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
122 new value now works. (reported by Bruno Haible)
123 * bug fix: correct canonicalization of multiple non-standard slot
124 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
125 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
126 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
127 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
128 now exists, an signals an error.
129 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
130 during macroexpansion for non-iteration variables as well. (reported
131 by Bruno Haible for CMUCL)
132 * bug fix: Cyclic structures and unprintable objects in compiler
133 messages no longer cause errors. (reported by Bruno Haible)
134 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
135 and READ-FROM-STRING all now return a primary value of NIL if
136 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
137 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
138 (reported by Bruno Haible for CMUCL)
139 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
140 do not designate a condition. (reported by Bruno Haible for
142 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
143 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
144 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
145 permitting resolution in favour of any of the conflicting symbols.
146 (reported by Bruno Haible for CMUCL)
147 * bug fix: EQUAL compiler optimizations is less aggressive on
148 strings which can potentially compare true despite having distinct
149 specialized array element types.
150 * bug fix: unit enumerations can be defined without dividing by
151 zero. (thanks to Vincent Arkesteijn)
152 * FORMAT compile-time argument count checking has been enhanced.
153 (report from Bruno Haible for CMUCL)
154 * a partial workaround for the bug 262: the compiler does not try to
155 inline-expand a local function doing RETURN-FROM from a deleted
156 BLOCK. (thanks to Peter Denno for the bug report and to David
157 Wragg for the simple test case)
158 * fixed some bugs revealed by Paul Dietz' test suite:
159 ** RENAME-PACKAGE allows all package designators as new package
161 ** constraint propagation and lambda variable substitution are
162 more cautious in dealing with partially deleted code.
163 ** compiler sometimes lost reoptimization passes.
164 ** CERROR, given a condition as condition designator, passes its
165 remaining arguments to the continue format control without
167 ** Case-altering FORMAT directives work correctly on non-ASCII
169 ** The REAL type specifier handles bounds outside the floating
170 point ranges without signalling FLOATING-POINT-OVERFLOW.
171 ** Functions with IR1-transformations can create intercomponent
172 references to global functions.
173 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
175 ** FORMAT ~R treats a nil value for its first parameter correctly.
177 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
178 * enhancement: saving cores with foreign code loaded is now
179 supported on x86/NetBSD and sparc/Linux in addition to the previously
181 * bug fix: on some platforms repeated installations caused multiple
182 copies of HTML documentation to be installed -- should not happen
183 any more. (reported by Stefan Scholl)
184 * bug fix: parsing self-recursive alien record types multiple times
185 no longer causes infinite recursion. (reported by Thomas F. Burdick,
186 original patch by Helmut Eller for CMUCL)
187 * bug fix: stack-exhaustion detection works now on NetBSD as well.
188 (thanks to Richard Kreuter)
189 * bug fix: defining classes whose accessors are methods on existing
190 generic functions in other (locked) packages no longer signals
191 bogus package lock violations. (reported by François-René Rideau)
192 * bug fix: special variables as DEFMETHOD parameters no longer have
193 associated bogus type declarations. (reported by David Wragg and
195 * bug fix: read-write consistency on streams of element-type
196 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
197 * bug fix: redefiniton of the only method of a generic function with
198 no DEFGENERIC no longer emits a full WARNING. In addition,
199 redefinition of generic functions with no DEFGENERIC to an
200 incompatible lambda list now signals an error. (thanks to Zach
202 * bug fix: DEFGENERIC now works even when there's a function of the
203 same name in an enclosing lexical environment. (thanks to Zach
205 * fixed compiler failure, caused by instrumenting code during
206 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
207 * optimization: added loop analysis and improved register allocation
208 * fixed some bugs revealed by Paul Dietz' test suite:
209 ** POSITION on displaced vectors with non-zero displacement
210 returns the right answer.
211 ** (SIMPLE-STRING) is a valid type specifier for sequence
213 ** *PRINT-LEVEL* handling for slotless structures is pedantically
215 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
216 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
218 ** The justification version of the FORMAT ~< directive treats
219 non-zero minpad parameter correctly.
221 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
222 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
223 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
224 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
225 the supported interface.
226 * new feature: Single-stepping of code compiled with DEBUG 2 or
227 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
228 * new feature: saving cores with foreign code loaded is now
229 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
230 Timothy Moore's work for CMUCL)
231 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
232 parameters to * instead of NIL if no initform is supplied.
233 (reported by Johan Bockgaard)
234 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
235 correctly when FUNCALL appears as the car of the form. Note:
236 despite this FUNCALL forms are not currently subject to
237 compiler-macro expansion. (port of Raymond Toy's fix for the
238 same from CMUCL, reported by Johan Bockgaard)
239 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
240 (thanks for Teemu Kalvas)
241 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
242 case instead of terminating the iteration (thanks for Julian
243 Squires, Sean Champ and Raymond Toy)
244 * bug fix: incorrect expansion of defgeneric that caused a style
245 warning. (thanks for Zach Beane)
246 * bug fix: slot accessor effective method computation works properly
247 for classes with multiple non-standard applicable methods on
248 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
249 * on x86 compiler supports stack allocation of results of LIST and
250 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
251 CMUCL implementation by Gerd Moellmann)
252 * fixed some bugs revealed by Paul Dietz' test suite:
253 ** FORMAT strings with both the ~<~:;~> form of the justification
254 directive and pretty-printing directives cause an error.
256 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
257 * incompatible change: the internal functions
258 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
259 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
260 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
261 instead of the old functions.
262 * new feature: on platforms where "dladdr" is available foreign
263 function names now appear in backtraces. (based on Helmut Eller's
265 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
266 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
268 * documentation: ASDF manual is now distributed with SBCL.
269 * bug fix: SBCL can now load its contributed modules with REQUIRE
270 even if the system-provided entries have been removed from
271 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
273 * bug fix: backtraces involving undefined functions or assembly
274 routines are more informative. (thanks to Brian Downing)
275 * bug fix: mutually referent alien structures now work correctly.
276 (reported by Rick Taube)
277 * bug fix: structures defined by WITH-ALIEN can be referred to
278 within other definitions in the same WITH-ALIEN.
279 * bug fix: division operators (MOD, TRUNCATE and the like) with
280 constant zero divisors and integer dividends no longer generate
282 * bug fix: provide default methods for INPUT-STREAM-P and
283 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
284 * bug fix: improve the detection and resolution of MOP metacycles
285 (where effective method computation depends on the generic function
286 having its effective method computed). (reported by Bruno Haible)
287 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
288 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
289 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
290 * bug fix: pathnames with relative directory components can be
291 represented relative to default pathnames.
292 * optimization: in taking the GCD of bignums, reduce the two bignums
293 to approximately the same size (using Euclid's algorithm) before
294 applying the more sophisticated binary GCD. (thanks to Juho
296 * optimization: COUNT on bitvectors now operates word-at-a-time.
297 * optimization: ASH with a positive, but not necessarily constant,
298 (leftwards) shift, when in a modular context, is compiled to a
300 * fixed some bugs revealed by Paul Dietz' test suite:
301 ** FORMAT variable parameters ("~V<char>") are defaulted properly
302 if the corresponding argument is NIL.
303 ** FORMAT directives accepting a minpad parameter treat negative
305 ** literal commas as character parameters to FORMAT directives are
307 ** literal spaces directly after ~<Newline> directives within a
308 format-logical-block (~:< ~@:>) do not induce :FILL-style
309 conditional newlines.
310 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
311 as specified: it no longer includes conditional newlines).
312 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
314 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
315 :PER-LINE-PREFIX argument does not evaluate to a string.
317 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
318 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
319 the manual for details; for now, package locks can be disabled by
320 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
321 if no major problems are found then it is likely that they will be
322 compiled in unconditionally.
323 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
324 unsupported operators on all platforms. To load a shared library
325 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
326 object file, link it into a shared library outside of SBCL and
327 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
329 * bug workaround: ROOM T and the GC were not working together
330 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
331 reporting facility aren't preserved. That reporting has been
332 disabled, so now until and unless someone figures out how to make it
333 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
334 * minor incompatible change: as threatened around sbcl-0.8.0, the
335 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
336 The symbols are also exported from SB-PCL for backwards
337 compatibility, but more so than before SB-PCL should be treated as
338 an implementation-internal package.
339 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
341 * the SB-SPROF contrib now works on (most) non-x86 architectures.
342 It is known as of this release not to work on the Alpha, however.
343 * fixed bug #167: errors signalled due to illegal syntax in method
344 bodies are now more legible.
345 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
346 designators and can hence be used with TYPEP.
347 * fixed bug #333: CHECK-TYPE now ensures that the type error
348 signalled, if any, has the right object to be accessed by
349 TYPE-ERROR-DATUM. (reported by Tony Martinez)
350 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
351 5.1.2.3 for multiple-value place subforms. (reported by Kalle
353 * fixed bug #334: programmatic addition of slots using specialized
354 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
355 and :ALLOCATION :CLASS slots.
356 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
357 float exponents. (rereported by Peter Seibel)
358 * fixed a bug: #\Space (and other whitespace characters) are no
359 longer considered to be macro characters in standard syntax by
361 * fixed bug: initialization of condition class metaobjects no longer
362 causes an instance of the condition to be created. (reported by
364 * fixed bug: it is now possible to have slots such that
365 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
366 description is neither :INSTANCE nor :CLASS.
367 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
368 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
369 SLOT-VALUE-USING-CLASS), and no longer causes errors with
370 non-standard SLOT-DEFINITION-ALLOCATION values.
371 * fixed bugs: various race conditions handling exiting threads.
372 CL-PPCRE's thread torture test now passes.
373 * fixed arguably-a-bug: GC time proportional to number of threads in
374 system even when most of them are idle
375 * optimization: improved performance of BIT and SBIT on bit-vectors.
376 * .fasl file incompatibility: The fasl file version number has
377 been incremented because of changes associated with package locks.
379 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
380 * minor incompatible change: the system no longer provides
381 optimization hints (conditions which are TYPEP
382 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
383 mode; these hints are still emitted when higher SPEED optimization
385 * new contrib module: a sampling profiler (profiling by statistical
386 sampling, rather than by instrumenting functions) is available as
387 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
388 * the behaviour of the standard function ED is now customizeable by
389 third parties through a hook variable: see ED's documentation
390 string for information on the protocol.
391 * the compiler no longer emits efficiency notes for (FUNCALL X)
392 when the type of X is uncertain under default optimization
394 * fixed bug 276: mutating a binding of a specialized parameter to a
395 method to something that is not TYPEP the specializer is now
397 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
398 exhibit the required behaviour on the x86 platform. (thanks to
399 Peter van Eynde, Eric Marsden and Bruno Haible)
400 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
401 even for difficult arguments. (reported by Peter Graves)
402 * fixed bug 141a: the backquote printer now descends quoted
404 * fixed another bug in backquote printing: no more destructive
405 modification of the form's list structure. (reported by Brian
407 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
408 move between its address being taken and the call to
409 interrupt_thread, fixing a crashing race condition.
410 * the SB-POSIX contrib implementation has been adjusted so that it
411 no longer exhibits ridiculously poor performance when constructing
412 instances corresponding to C structs.
414 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
415 * minor incompatible change: the sb-grovel contrib now treats C
416 structures as alien (in the sense of SB-ALIEN) objects rather than
417 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
418 has implications for memory management of client code
419 (sb-grovel-returned objects must now be manually managed) and for
420 type safety (alien objects now have full types).
421 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
422 used to control emission of compiler diagnostics, rather than the
423 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
424 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
425 quality should be considered deprecated.
426 * install.sh now installs the user manual as well
427 * (not quite a new documentable feature, but worth considering in
428 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
429 beginnings of a semantically meaningful condition hierarchy is
430 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
432 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
433 designator as the defaults argument.
434 * fixed bug: Displaced arrays whose displaced-to array has become
435 too small now cause ARRAY-DIMENSION to signal an error, providing
436 for safer bounds-checking. (reported by Bruno Haible)
437 * fixed bug: DEFCLASS slot definitions with identical :READER and
438 :WRITER names now signal a reasonable error. (reported by Thomas
440 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
442 * fixed bug: Invalid slot specification errors now print correctly.
443 (thanks to Zach Beane)
444 * fixed bug 320: Shared to local slot value transfers in class
445 redefinitions now happen corrently from superclasses as
446 well. (reported by Bruno Haible)
447 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
449 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
450 improper lists correctly. (reported by Bruno Haible)
451 * fixed bug 313: source-transform for <fun-name> was erroneously
452 applied to a call of a value of a variable with name <fun-name>.
453 (reported by Antonio Menezes Leitao)
454 * fixed bug 307: The obsolete instance protocol ensures that
455 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
456 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
457 remove dead unknown-values globs from the middle of the stack.
458 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
459 easier to resume long computations after using *BREAK-ON-SIGNALS*
460 to diagnose and fix failures. (thanks to Nikodemus Siivola)
461 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
462 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
463 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
464 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
465 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
467 * fixed a (fairly theoretical) bug in string printing: if
468 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
469 does not have array-element-type equal to the most general string
471 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
472 function name. (thanks to Nikodemus Siivola)
473 * fixed bug: compile-time format string checker failed on
474 non-closed ~{. (reported by Thomas F Burdick)
475 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
476 don't warn on legal loop code involving more than one aggregate
477 boolean. (thanks to Nikodemus Siivola)
478 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
479 in keyword position to MAKE-PACKAGE was not being reported as
481 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
482 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
484 * fixed bugs 280 and 312: the checking for multiple definitions in a
485 file is less likely to become confused by uses of inline
487 * fixed bug: the #S reader macro performs the keyword coercion
488 specified for slot names. (reported by Kalle Niemitalo)
489 * fixed bug: lambda lists may contain symbols whose names start with
490 & but are not lambda-list-keywords; their occurrence triggers a
492 * fixed bug 321: define-method-combination argument lambda lists do
493 not cause invalid code to be generated when &optional or &aux
494 variables have default values. (reported by Bruno Haible)
495 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
496 have CLOS classes; this appears also to have fixed reported
497 instability in stack exhaustion detection.
498 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
499 MUFFLE-WARNING when no associated restart is present is now
501 * optimization: rearranged the expansion of various defining macros
502 so that each expands into only one top-level form in a
503 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
505 * optimization: used a previously little-used slot in symbols to
506 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
508 * fixed some bugs revealed by Paul Dietz' test suite:
509 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
511 ** Fixed an optimization bug related to inheritance of initforms
512 from local to shared slots.
513 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
514 ** CAST optimizer forgot to flush argument derived type.
515 ** print/read consistency on floats is now orders of magnitude
516 more likely. (thanks also to Bruno Haible for a similar report
518 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
519 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
520 for objects of type REAL. Make it so.
521 ** SXHASH is specified (infelicitously) to respect similarity,
522 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
523 it so. (thanks to Markus Ziegler)
524 ** on the Alpha, the compiler succeeds in compiling functions
525 returning a known number of arguments greater than 63.
526 ** fixed handling of invalid NIL arguments in keyword position
528 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
530 ** provided a readably-printable representation for RANDOM-STATE
532 ** ensured that pathnames, where they have a namestring, always
533 print using #P"..." syntax.
535 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
536 * Support for the forthcoming 2.0 version of the NetBSD kernel
537 running on x86 hardware has been added. (thanks to Perry
538 E. Metzger most immediately, and others for their past work)
539 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
540 current release version; SBCL's OpenBSD support had been broken
541 since about the time of OpenBSD's switch to ELF binary format.)
542 * [placeholder for DX summary]
543 ** user code with &REST lists declared dynamic-extent, under high
544 speed or space and low safety and debug optimization policy.
545 * The manual has been converted to Texinfo format and the debugger
546 chapter from the cmucl manual has been added.
547 * A facility has been added to extract documentation strings from
548 sbcl and store them as Texinfo-formatted snippets for inclusion in
549 the manual (via Texinfo's @include directive)
550 * bug fix: compiler emitted division in optimized DEREF. (thanks for
551 the test case to Dave Roberts)
552 * bug fix: multidimensional simple arrays loaded from FASLs had fill
553 pointers. (reported by Sean Ross)
554 * bug fix: PROFILE output is printed nicely even for large numerical
555 values. (thanks to Zach Beane)
556 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
557 greater than 32 handle EOF correctly.
558 * bug fix: on X86 an immediate argument of the IMUL instruction is
559 correctly printed in disassembly. (thanks to Lutz Euler)
560 * bug fix: class slots in redefined classes preserve their old
561 values. (thanks to Bruno Haible and Nikodemus Siivola)
562 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
563 succeeds. (reported by Marco Baringer)
564 * fixed some bugs revealed by Paul Dietz' test suite:
565 ** READ-SEQUENCE now works on ECHO-STREAMs.
566 ** RATIONALIZE works more according to its specification. (thanks
569 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
570 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
571 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
572 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
573 should matter to you only if you rebind the printer control
574 variables and then find you want different bindings in the
575 debugger than in the ordinary execution of your program.)
576 * The runtime build system has been tweaked to support building
577 (on SPARC/SunOS) using a C compiler which invokes Sun's own
578 assembler and linker. (thanks to Nikodemus Siivola)
579 * Unbound, undefined, undeclared variables now trigger full
580 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
581 more often programmer error than deliberate exploitation of undefined
583 * optimization: the hash algorithm for strings has changed to one
584 that is less vulnerable to spurious collisions. (thanks to Juho
586 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
587 needless bounds checking. (thanks to Juho Snellman)
588 * optimization: implemented multiplication as a modular
589 (UNSIGNED-BYTE 32) operation on the PPC backend.
590 * fixed some bugs revealed by Paul Dietz' test suite:
591 ** ADJUST-ARRAY now copies the datum in a zero rank array if
593 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
594 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
595 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
597 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
599 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
601 ** ECHO-STREAMs no longer attempt to echo the end of file value to
602 their output stream on EOF from read.
603 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
604 have been read to end-of-file.
605 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
607 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
608 description of determination of which consecutive characters
610 ** Printing the "Space" character with escaping on now yields "#\\ ",
611 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
612 ** Reading floating-point numbers with *READ-BASE* set to a number
613 less than 10 works correctly.
614 ** Reading floating-point numbers with *READ-BASE* set to a number
615 more than 10 works correctly.
616 ** Printing with *PRINT-READABLY* targets the standard readtable, not
617 the readtable currently in effect.
619 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
620 * minor incompatible change: parsing of namestrings on a physical
621 (Unix) host has changed; numbers after the final #\. in a
622 namestring are no longer interpreted as a version field. This is
623 intented to be largely invisible to the user, except that the
624 meaning of the namestring "*.*.*" has changed: it now refers to a
625 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
626 should usually be replaced by
627 (make-pathname :name :wild :type :wild :version :wild)
628 with the added benefit that this is more likely to be portable.
629 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
630 option to OPEN now signals an error if the file being opened
631 exists; this may have an impact on existing code.
632 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
633 signals received, on the PowerPC platforms (both Linux and
634 Darwin). (thanks to Pierre Mai for pointing out the location of
636 * several fixes on OS X: The system now builds and runs cleanly on
637 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
638 (thanks to Brian Mastenbrook)
639 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
640 second-resolution: integer multiples of 1/3600 between -24 and 24.
641 (thanks to Vincent Arkesteijn)
642 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
643 type when called with 1 argument; PEEK-CHAR checked type of
644 PEEK-TYPE only after having read first character from a
645 stream. (reported by Peter Graves)
646 * bug fix: the garbage collector now has much better locality
647 behaviour, and in particular no longer treats all memory as being
648 exhausted when many small objects point to each other in a deeply
650 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
651 recognized as being TYPEP their class.
652 * bug fix: the PUSHNEW documentation string has been corrected.
653 (thanks to Vincent Arkesteijn)
654 * bug fix: defaulting of the value for the last of an atypically
655 large number of multiple values being bound was not being
656 performed correctly on the Alpha or PPC platforms
657 * optimization: implemented multiplication as a modular
658 (UNSIGNED-BYTE 32) operation on the x86 backend.
659 * optimization: SEARCH on simple-base-strings can now be open-coded.
660 (see also contrib/compiler-extras.lisp for inspiration for
661 teaching the compiler about the Boyer-Moore algorithm).
662 * value, returned by MAX (and MIN) called with several EQUALP, but
663 not EQL, arguments now does not depend on compiler settings.
664 * fixed some bugs revealed by Paul Dietz' test suite:
665 ** in stack analysis liveness information is propagated from
666 non-local entry points.
667 ** pathwise CAST removing failed when the CAST node did not start
669 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
670 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
672 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
673 host is already defined.
674 ** RENAME-FILE works on streams instead of signalling an internal
676 ** PEEK-CHAR uses the current readtable when determining whether
677 or not a character is whitespace.
678 ** MERGE-PATHNAMES handles the case when the pathname does not
679 specify a name while the default-pathname specifies a version.
680 ** Pathnames now stand a better chance of respecting print/read
682 ** Attempting to use standardized file system operators with a
683 pathname with invalid :DIRECTORY components signals a
685 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
687 ** DIRECTORY on logical pathnames is more correct.
688 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
689 signal a TYPE-ERROR if their argument is not a stream.
690 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
691 designator argument does not designate a stream.
692 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
693 examining the synonym.
694 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
696 ** OPEN and WITH-OPEN-STREAM allow opening streams with
697 element-type larger than ([UN]SIGNED-BYTE 32).
699 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
700 * When built with the :SB-FUTEX feature, threaded builds now take
701 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
702 for faster/more reliable mutex and condition variable support.
703 * Incompatible change (but one you probably shouldn't have been using
704 anyway): the interface and code for arbitrating between multiple
705 threads in the same user session has been redesigned.
706 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
707 error when called without an explicit environment argument.
708 (thanks to Nikodemus Siivola)
709 * bug fix: buffered :DIRECTION :IO streams are less likely to become
710 confused about their position. (thanks to Adam Warner and Gerd
712 * bug fix: Pretty printing backquoted forms with unquotations in the
713 argument list position of various code constructs such as LAMBDA
714 now works correctly. (reported by Paul Dietz)
715 * bug fix: Pretty printing unquotations no longer loses all
716 stream position information.
717 * optimization: performance of string output streams is now less
718 poor for multiple small sequence writes.
719 * optimization: performance of CSUBTYPEP in the presence of complex
720 expressions involving CONS and NOT many times has been improved.
721 (reported by Paul Dietz)
722 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
724 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
725 streams for the REPL, for the debugger, and so on. (thanks to
727 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
728 * fixed some bugs revealed by Paul Dietz' test suite:
729 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
730 is ignored if it is too small, rather than propagating through
731 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
732 ** extremely complex negations of CONS types were not being
733 sufficiently canonicalized, leading to inconsistencies in
735 ** VALUES tranformer lost derived type.
737 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
738 * fixed a bootstrapping bug: the build process no longer assumes
739 that the various BOOLE-related constants have the same value in
740 host and target lisps. (noted by Paul Dietz' test suite on an
741 SBCL binary built from CLISP)
742 * The system can now be dynamically linked on the MIPS platform,
743 which enables dynamic loading of foreign code from Lisp. (thanks
744 to Ralf Baechle for discussions on the MIPS ABI)
745 * The system now records debugging information for its own source
746 files in a filesystem-position-independent manner, relative to
747 the "SYS" logical host.
748 * fixed a compiler bug: MV-LET convertion did not check references
749 to the "max args" entry point. (reported by Brian Downing)
750 * tweaked disassembly notes to be less confident about proclaiming
751 some instruction as an LRA. (thanks to Brian Downing)
752 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
753 now each have their own history, command character, and other
754 characteristics. (thanks to David Lichteblau)
755 * fixed some bugs revealed by Paul Dietz' test suite:
756 ** compiler failure in compiling LOGAND expressions including a
758 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
759 shift greater than 32.
760 ** FLUSH-DEST did not mark blocks for type check regeneration.
761 ** HANDLER-CASE failed to accept declarations in handler clauses
762 in some circumstances.
764 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
765 * New code in contrib/sb-introspect (still probably not entirely
766 stable yet) provides some support for smart Lisp development
767 environments like SLIME.
768 * The conditions signalled for errors occurring when loading .fasl
769 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
770 in a way which should help ASDF recover gracefully.
771 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
772 stuff has been cleaned up. If you code contrib/ stuff, this might
773 affect you, and you can look at contrib/README, contrib/STANDARDS,
774 and/or the 0.8.4.27 diff to check.
775 * In full calls the compiler now does not generate checks for declared
776 argument types for all arguments.
777 * various threading fixes
778 ** and some experimental patches which didn't make it into
779 the main tree for this release, but which are shipped in
780 contrib/experimental-thread.patch as a possible fix for some
781 failures (deadlock, spinning...) in GC-intensive multithreaded
783 * fixed PPC build problem (source code incompatibility of different
784 library versions): added offsetof() hackery which attempts to divine
785 where glibc maintainers put uc_mcontext today
786 * fixed bug 282: compiler does not trust type assertions while passing
787 arguments to a full call.
788 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
789 &OPTIONAL VALUES type parameter.
790 * fix bug 214: algorithm for noting rejected templates is now more
791 similar to that of template seletion. (also reported by rydis on
793 * fixed bug 141b: printing backquoted information readably and prettily
794 inserts a space where necessary.
795 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
796 (CHARACTER 10) are now reported as errors, rather than propagated
797 as unknown types. (reported by piso on #lisp)
798 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
799 with values NIL and :ERROR. (thanks to Milan Zamazal)
800 * fixed bug 191c: CLOS now does proper keyword argument checking as
801 described in CLHS 7.6.5 and 7.6.5.1.
802 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
803 longer raise an error; further, using a list as a for-as-arithmetic
804 counter now raises a meaningful error.
805 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
806 now understood by sequence creation functions such as MAKE-SEQUENCE
808 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
809 a STRING-INPUT-STREAM.
810 * compiler enhancement: SIGNUM is now better able to derive the type
812 * type declarations inside WITH-SLOTS are checked. (reported by
814 * fixed some bugs revealed by Paul Dietz' test suite:
815 ** incorrect optimization of TRUNCATE for a positive first
816 argument and negative second.
817 ** compiler failure in let-convertion during flushing dead code.
818 ** compiler failure while deriving type of TRUNCATE on an
819 interval, containing 0.
820 ** ASH of a negative bignum by a negative bignum count now returns
822 ** intersection of CONS types now canonicalizes properly, fixing
823 inconsistencies in SUBTYPEP.
825 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
826 * incompatible change: The --disable-debugger command line
827 option now clobbers the debugger at a more fundamental
828 level, by redefining #'INVOKE-DEBUGGER instead of by
829 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
830 is specified by ANSI to ignore *DEBUGGER-HOOK* and
831 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
832 enter the debugger REPL and then suffer recursive errors
833 because *DEBUG-IO* is also messed up in --disable-debugger mode;
834 while under the new system, BREAK in --disable-debugger mode
835 terminates the system just as an unhandled error would.
836 * fixed compiler performance when processing loops with a step >1;
837 * bug fix: DOCUMENTATION now retrieves generic function
838 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
839 support has been systematized, and now supports the methods
840 specified by ANSI, along with a default method and a method for
841 slot documentation. (reported by Nathan Froyd)
842 * bug fix: effective methods associated with a generic function are
843 no longer cached over a change of that generic function's method
844 combination. (reported by Andreas Fuchs)
845 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
846 names a non-existent directory. (thanks to Andreas Fuchs)
847 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
848 the PPC platform, lead to stack corruption; this has been fixed.
849 (reported by Rainer Joswig)
850 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
851 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
852 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
853 arithmetic in general, is now much more reliable on the PPC
855 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
856 platform now returns the right answer.
857 * optimization: restored some effective method precomputation in
858 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
859 precomputation is now tunable.
860 * optimization: compiler-internal data structure use has been
861 reviewed, and changes have been made that should improve the
862 performance of the compiler by about 20%.
863 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
864 simple-base-strings and simple-bit-vectors is improved.
865 * optimization: the optimization of 32-bit logical and arithmetic
866 functions introduced in version 0.8.3 on the x86 has been
867 implemented on the mips, ppc and sparc platforms; an
868 implementation of the same facility, but for 64-bit arithmetic,
869 has been added for the alpha.
870 * microoptimization: the compiler is better able to make use of the
871 x86 LEA instruction for multiplication by constants.
872 * bug fix: in some situations compiler did not report usage of
873 generic arithmetic in (SPEED 3) policy.
874 * bug 145b fix: compiler used wrong type specifier while converting
875 MEMBER-types to numeric.
876 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
878 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
879 index argument now works properly on the Alpha platform.
880 * bug fix: floating point exception treatment on the Alpha platform
882 * bug fix: FILE-POSITION works much better on string input and
883 output streams. (thanks to Nikodemus Siivola)
884 * bug fix: many threading/garbage collection symptoms sorted.
885 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
886 might be pseudo-atomic.
887 * internal change: Stopping for GC is now done with signals not
888 ptrace. GC is now done in whichever thread wanted it, instead of
890 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
892 * bug fix: result form in DO is not contained in the implicit
894 * incompatible change: ICR structure is changed; the value part of
895 CONTINUATION is now called LVAR; corresponding functions are
896 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
897 * added type deriver for ISQRT (thanks to Robert E. Brown).
898 * bug fix: better support for loading from the command line when an
899 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
901 * fixed some bugs revealed by Paul Dietz' test suite:
902 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
903 ** ROUND and FROUND now give the right answer when given very
904 small float arguments.
905 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
907 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
908 ** (GCD 0 <negative-integer>) returned <negative-integer>.
909 ** LCM should return a non-negative integer.
910 ** PARSE-INTEGER returned the index of a terminator instead of the
911 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
912 ** PARSE-INTEGER returned an incorrect index being applied to a
914 ** LCM with two arguments of 0 returns 0 rather than signalling
916 ** unsigned addition of a 32-bit constant with the high bit set no
917 longer causes an internal compiler error on the x86.
918 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
919 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
920 with negative last argument.
921 ** byte specifiers with zero size and position no longer cause
922 an error during type derivation.
923 ** bignum multiplication on the Alpha platform now returns the
925 * porting: The system now builds on SuSE AMD64, although it still
926 generates a 32-bit binary.
927 * .fasl file incompatibility: The fasl file version number has
928 been incremented (because of the changes to internal compiler
929 data structures referred to above).
931 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
932 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
933 more accurately, on the Darwin kernel running on PowerPC hardware.
934 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
935 * Compiler code deletion notes now signal a condition of type
936 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
937 an associated MUFFLE-WARNING restart.
938 * The compiler now performs limited argument count validation of
939 constant format strings in FORMAT, and where appropriate in ERROR,
940 CERROR and WARN. (thanks to Gerd Moellmann)
941 * New ASDF-INSTALL contrib can be used for automatic download and
942 installation of third-party Lisp code from CCLAN or other sites
944 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
945 which forces another thread to execute a function supplied by the
947 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
948 now accept and act upon their :ELEMENT-TYPE keyword argument.
949 (reported by Martin Atzmueller, Edi Weitz)
950 * bug fix: FILE-POSITION now accepts position designators up to
951 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
952 is the greater. (thanks to Patrik Nordebo)
953 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
954 Kalle Olavi Niemitalo)
955 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
956 printable. (reported by Eric Marsden)
957 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
958 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
959 * x86 bug fix in control stack exhaustion checking: now shows backtrace
960 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
961 (thanks to Stig Sandoe)
962 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
963 UNREAD-CHAR now correctly includes the unread character in the
964 target sequence. (thanks to Gerd Moellmann)
965 * bug fix in threaded builds: the system can now be suspended and
966 resumed by shell job control with minimal disruption.
967 * bug fixes in times and timezones >2038AD
968 * better handling of "where is GNU make?" problem in build scripts
969 (thanks to Nikodemus Siivola)
970 * new optimization: inside a named function any reference to a
971 function with the same name is considered to be a self-reference;
972 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
973 optimization quality.
974 * new optimization on x86: logical functions and + now have
975 optimized (UNSIGNED-BYTE 32) versions, which are automatically
976 used when the result is truncated to 32 bits.
977 * VALUES declaration is partially enabled.
978 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
979 * fixed some bugs revealed by Paul Dietz' test suite:
980 ** The system now obeys the constraint imposed by
981 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
982 types form a lattice under type intersection.
983 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
984 ** ASSOC now ignores NIL elements in an alist.
985 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
986 and (1+ MOST-POSITIVE-FIXNUM) answers.
987 ** The addition of a method with invalid qualifiers to a generic
988 function does not cause an error to be signalled immediately;
989 a warning is signalled, and the error is generated only on
990 calling the generic function.
991 * changed .fasl file version number, in part to add type codes for
992 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
993 obscure ANSI requirements
995 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
996 * fixed bug 148: failure to inline-expand a local function left
997 garbage, confusing the compiler.
998 * fixed bugs 3cd: structure slot readers perform type check if the
999 slot can have an invalid value (i.e. it is either not initialized
1000 or can be written with a less specific slot writer).
1001 * bug fix: the compiler now traps array references to elements off
1002 the end of an array; previously, the bounds checking in some
1003 circumstances could go off-by-one.
1004 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1006 * type declarations for array element types now obey the description
1007 on the CLHS page "Declaration TYPE", as per discussions on
1008 sbcl-help around 2003-05-08. This means that a declaration
1009 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1010 declaration, all references to BAR will be asserted or assumed
1011 (with THE, so dependent on compiler policy) to involve objects of
1012 type FOO. Note that no such declaration is implied in
1013 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1014 * declared types of functions from the "Conditions"
1015 chapter. (reported by Paul Dietz)
1016 * bug fix: CERROR accepts a function as its first argument.
1017 * bug fix: NTH an NTHCDR accept a bignum as index
1018 arguments. (reported by Adam Warner)
1019 * optimization: character compare routines now optimize comparing
1020 against a constant character. (reported by Gilbert Baumann)
1021 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1022 argument now works properly on the MIPS platform.
1023 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1024 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1025 declared to have a functional type, does not cause an error
1027 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1028 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1029 * fixed bug: if last continuation of a deleted block has a
1030 destination, this destination should be deleted too. (reported by
1032 * fixed a bug in the bootstrap process: the host compiler's values
1033 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1034 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1035 test case from Patrik Nordebo)
1036 * improved the ability of the disassembler on the PPC platform to
1037 provide helpful disassembly notes.
1038 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1039 the class in more cases than previously.
1040 * bug fix: FILE-POSITION now understands :START and :END for
1041 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1042 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1043 8) data. (thanks to David Lichteblau)
1044 * bug fix: it is possible to add a method to a generic function
1045 without lambda list.
1046 * bug fix: reader failed to signal END-OF-FILE inside an
1047 object representation. (reported by Nikodemus Siivola)
1048 * fixed some bugs revealed by Paul Dietz' test suite:
1049 ** LAST and [N]BUTLAST should accept a bignum.
1050 ** condition slot accessors are methods.
1051 ** (VECTOR NIL) is a subtype of STRING.
1053 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1054 * minor incompatible change: some nonsensical specialized lambda
1055 lists (used in DEFMETHOD) which were previously ignored now signal
1057 * minor incompatible change: the system is now aware of the types of
1058 variables in the COMMON-LISP package, and will signal errors for
1059 most violations of these type constraints (where previously they
1060 were silently accepted).
1061 * minor incompatible change: COMPILE-FILE now uses the freedom
1062 afforded (ANSI 3.2.2.3) to use derived function types for
1063 functions defined in the same file. This also permits the system
1064 to warn on static type mismatches and function
1065 redefinition. (Currently it does not work with high DEBUG level.)
1066 * minor incompatible change: VALUES declaration is disabled.
1067 * When issuing notes, the compiler now signals a condition of type
1068 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1069 restart for use in user handlers. It is expected that the
1070 COMPILER-NOTE condition will eventually become a condition
1071 supertype to a hierarchy of note types, which will then be
1072 handleable in a similar fashion. However, at the moment, no such
1073 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1074 but it's an implementation detail, not a classification for the
1076 * Changes in type checking closed the following bugs:
1077 ** type checking of unused values (192b, 194d, 203);
1078 ** template selection based on unsafe type assertions (192c, 236);
1079 ** type checking in branches (194bc).
1080 * A short form of VALUES type specifier has ANSI meaning (it has
1081 increased the number of situations when SBCL cannot perform type
1083 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1084 keywords or constants is permissible.
1085 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1086 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1087 classes. (thanks to Antonio Martinez)
1088 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1089 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1090 argument to a pretty-print stream. Instead, it leaves any such
1091 operations to DESCRIBE-OBJECT methods.
1092 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1094 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1095 lambda lists. (thanks to David Lichteblau)
1096 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1097 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1098 (thanks to Gerd Moellmann)
1099 * fixed bug 246: increased compilation speed of long
1100 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1102 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1104 * DEFCONSTANT now throws a condition of type
1105 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1106 constant to a non-EQL value; CONTINUE and ABORT restarts
1107 respectively change and preserve the value.
1108 * fixed bug 63: The code walker, part of the implementation of CLOS,
1109 is now better at handling symbol macros.
1110 * bug fix: There is no longer an internal implementation type named
1111 CL:LENGTH. (reported by Raymond Toy)
1112 * bug fix: In macro-like defining macros/special operators the
1113 implicit block does not enclose the lambda list.
1114 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1115 enclosed in parentheses) are not suitable as type specifiers, and
1116 their use properly signals an error now.
1117 * bug fix: An argument count mismatch for a type specifier in code
1118 being compiled no longer causes an unhandled error at compile
1119 time, but signals a compile-time warning.
1120 * fixed simple vector readable printing
1121 * bug fix: DESCRIBE takes more care over whether the class
1122 precedence list slot of a class is bound before accessing it.
1123 (reported by Markus Krummenacker)
1124 * bug fix: FORMATTER can successfully compile pretty-printer format
1125 strings which use variants of the ~* directive inside.
1126 * bug fix: SEARCH now applies its TEST predicate to the elements of
1127 the arguments in the correct order. (thanks to Wolfhard Buss)
1128 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1129 in type inference. (thanks to Robert E. Brown)
1130 * bug fix: Reading in symbols with an explicit package name of ""
1131 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1132 * fixed some bugs revealed by Paul Dietz' test suite:
1133 ** NIL is now allowed as a structure slot name.
1134 ** Arbitrary numbers, not just REALs, are allowed in certain
1135 circumstances in LOOP for-as-arithmetic clauses.
1136 ** Multiple class redefinitions before slot access no longer
1137 causes a type error.
1138 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1139 association between the name and a class.
1140 ** Generic functions with non-standard method-combination and over
1141 five methods all of which return constants no longer return NIL
1142 after the first few invocations. (thanks to Gerd Moellmann)
1143 ** CALL-NEXT-METHOD with no arguments now passes the original
1144 values of the arguments, even in the presence of assignment.
1145 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1147 ** DISASSEMBLE works with closures and funcallable instances.
1148 ** ADD-METHOD now returns the generic function, not the new method.
1149 ** FIND-METHOD signals an error if the lengths of the specializers
1150 is incompatible with the generic function, even if the ERRORP
1152 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1153 which its argument is a member.
1154 ** DEFCLASS only redefines the class named by its class-name
1155 argument if that name is the proper name of the class;
1156 otherwise, it creates a new class.
1157 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1158 of the UNBOUND-SLOT condition to the name of the slot.
1159 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1160 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1161 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1162 SLOT-MAKUNBOUND in the specified fashion.
1164 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1165 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1166 cross-compilation host. As a consequence, we can now bootstrap our
1167 way up to SBCL starting with a bare gcc toolchain and human-readable
1168 source code (first the source to CLISP, then the source to SBCL).
1169 * A contributed module containing a partial implementation of the
1170 simple-streams interface has been included. (thanks to Rudi
1172 * A contributed module implementing the RFC1321 Message Digest
1173 Algorithm, known as MD5, has been included.
1174 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1175 no longer has any effect, as the code controlled by this feature
1176 has been deleted. (As far as we know, no-one has ever built using
1177 this feature, and its semantics were confused in any case).
1178 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1179 work on conditions (as required by the ANSI specification),
1180 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1181 expected behaviour on conditions. Users should note, however,
1182 that such behaviour is not required by the ANSI specification,
1183 and so use of this behaviour may render their code unportable.
1184 * fixed some bugs revealed by Paul Dietz' test suite:
1185 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1187 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1188 ** SXHASH on condition objects no longer returns NIL.
1189 ** :ALLOCATION :CLASS slots are better treated; their values are
1190 updated on class redefinition, and initforms inherited from
1191 superclasses are applied.
1192 ** REMOVE-METHOD returns its generic function argument even when
1193 no method was removed.
1194 ** SHARED-INITIALIZE now initializes the values of the requested
1195 slots, including those with :ALLOCATION :CLASS.
1196 ** ALLOCATE-INSTANCE now works on structure classes defined via
1197 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1199 ** SLOT-EXISTS-P now works on conditions, as well as structures
1201 ** MAKE-LOAD-FORM now has the required methods on
1202 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1203 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1204 keyword, and now implements the SLOT-NAMES argument.
1205 ** methods with &OPTIONAL arguments no longer allow too many
1206 arguments to be passed in the call without error.
1207 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1208 option is consistent with the required arguments of the generic
1209 function lambda list.
1210 * bug fix: REQUIRE accepts a string designator. (Thanks to
1212 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1213 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1214 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1215 * bug fix: The compiler now checks for duplicated variables in macro
1217 * bug fix: SETQ on globals returns the correct value.
1218 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1219 not a COMPILER-ERROR (followed by some other strange error on
1220 choosing the CONTINUE restart).
1221 * bug fix: make.sh and friends are now more consistent in the way that
1222 they look for GNU "make".
1224 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1225 * experimental native threads support (on x86 Linux >=2.4 only).
1226 This is not compiled in by default: you need to add :SB-THREAD to
1227 the target features. See the "Beyond ANSI" chapter of the manual
1229 * fix for longstanding nonANSIism: The old distinction between
1230 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1231 The return value from CL:FIND-CLASS is now a CLOS class, and
1232 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1233 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1234 * An interface to the MetaObject Protocol, as described in Kiczales,
1235 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1236 MIT Press, 1991, is available from the SB-MOP package.
1237 * incompatible change: the SB-PCL package should now be considered
1238 a private implementation detail, and no longer a semi-private MOP
1240 * minor incompatible change: due to rearrangement for threads, the
1241 control stack and binding stack are now allocated at arbitrary
1242 addresses instead of being hardcoded per-port. Users affected by
1243 this probably have to be doing advanced things with shared
1244 libraries, and will know who they are.
1245 * minor incompatible change: Previously, all --eval forms used were
1246 processed with READ before any of them were processed with EVAL.
1247 Now each --eval form is processed with both READ and EVAL before
1248 the next --eval form is processed. (Thus package operations like
1249 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1250 work as the user might reasonably expect.)
1251 * minor incompatible change: *STANDARD-INPUT* is now only an
1252 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1254 * minor incompatible change: Y-OR-N-P is now character-oriented,
1255 not line oriented. Also, YES-OR-NO-P now works without errors.
1256 (thanks to Antonio Martinez)
1257 * sb-aclrepl module improvements: an integrated inspector, added
1258 repl features, and a bug fix to :trace command.
1259 * Known functions, which cannot be open coded by the backend, are
1260 considered to be able to check types of their arguments. (fixing
1261 a bug report by Nathan J. Froyd)
1262 * fixed a bug in computing method discriminating functions: It is
1263 now possible to define methods specialized on classes which have
1264 forward-referenced superclasses. (thanks to Gerd Moellmann)
1265 * fixed evaluation order in optional entries (reported by Gilbert
1267 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1268 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1269 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1270 required-by-AMOP NAME argument, as well as CLASS and
1271 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1272 * fixed bug 20: DEFMETHOD can define methods using names that are
1273 not the proper names of classes to designate class specializers.
1274 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1275 Unix file descriptors, instead of blowing up. (thanks to Antonio
1277 * Garbage collection refactoring: user-visible change is that a
1278 call to the GC function during WITHOUT-GCING will not do garbage
1279 collection until the end of the WITHOUT-GCING. If you were doing
1280 this you were probably losing anyway.
1281 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1282 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1283 TYPEP the latter but not the former.
1284 * The compiler issues a full WARNING for calls to undefined functions
1285 with names from the CL package.
1286 * MAP-INTO for a vector destination is open coded. (reported by
1287 Brian Downing on c.l.l)
1288 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1289 documentation string.
1290 * fixed some bugs revealed by Paul Dietz' test suite:
1291 ** COPY-ALIST now signals an error if its argument is a dotted
1293 ** Condition slots are now accessed more correctly in the presence
1294 of multiple initargs for a given slot.
1295 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1296 exclude restarts of the same name associated with a different
1298 ** DEFCLASS of forward-referenced classes with another
1299 forward-referenced class in the superclasses list no longer
1301 ** Condition slots are now initialized once each, not multiple
1302 times. (thanks to Gerd Moellmann)
1303 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1305 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1306 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1307 arguments contain duplicated elements.
1308 ** RESTART-CASE understands local macros.
1309 ** RESTART-CASE associates exactly its own restarts with a condition.
1310 ** ENDP in safe mode checks its argument to be of type LIST.
1311 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1312 in question is unbound.
1313 ** Optimized MAKE-INSTANCE functions no longer cause internal
1314 assertion failures in the presence of duplicate initargs.
1315 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1316 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1317 propagated even in the ctor optimized implementation of
1319 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1321 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1322 * a better implementation of SXHASH on (simple) bit vectors,
1323 measured both in execution speed and in distribution of results
1324 over the positive fixnums, has been installed. Likewise, a better
1325 implementation of EQUAL for simple bit vectors is now available.
1326 * fixed CEILING optimization for a divisor of form 2^k.
1327 * fixed bug 240 (emitting extra style warnings "using the lexical
1328 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1329 by Antonio Martinez)
1330 * fixed SXHASH, giving different results for NIL depending on type
1331 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1332 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1333 variable. (found by Rolf Wester)
1334 * SBCL does not ignore type declarations for special
1335 variables. (reported by rif on c.l.l 2003-03-05)
1336 * some bug fixes in contrib/sb-aclrepl/
1337 * fixed some bugs revealed by Paul Dietz' test suite:
1338 ** a bug in the CONS type specifier, whereby the CAR and CDR
1339 types got intertwined, has been fixed;
1340 ** the type system is now able to reason about the interaction
1341 between INTEGER and RATIO types more completely;
1342 ** APPEND, [N]REVERSE and NRECONC check that those their
1343 arguments, which must be proper lists, are really so;
1344 ** An array specialized to be unable to hold elements has been
1345 implemented, as required -- yes, really -- by ANSI;
1346 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1347 on malformed property lists;
1349 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1350 * incompatible packaging change: in line with Unix convention,
1351 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1352 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1353 * REQUIRE and PROVIDE are now optionally capable of doing something
1354 useful. See the documentation string for REQUIRE.
1355 * infrastructure for a managed SBCL contrib system: contributed
1356 modules in this release include:
1357 ** the ASDF system definition facility;
1358 ** an interface to the BSD Sockets API;
1359 ** an ACL-like convenience interface to the repl;
1360 (thanks to Kevin Rosenberg)
1361 ** an implementation of ROTATE-BYTE, with efficient implementation
1363 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1364 gives a closer approximation to the right answer than previously.
1365 (thanks to Raymond Toy)
1366 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1367 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1368 optional environment arguments, as required by ANSI.
1369 * fixed bugs in other functions taking environment objects, allowing
1370 calls with an explicit NIL environment argument to be compiled
1372 * fixed bug 228: primary return values from
1373 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1374 COMPILE or FUNCTION.
1375 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1376 :TYPE will no longer signal errors on innocuous objects.
1377 * fixed bug 231b: SETQ is better at respecting type declarations in
1378 the lexical environment.
1379 * fixed a bug in DEFCLASS: classes named by symbols with no or
1380 unprintable packages can now be defined.
1381 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1382 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1383 * fixed compiler failure related to checking types of functions.
1384 (reported by Robert E. Brown)
1385 * the compiler is now much more consistent in its error-checking
1386 treatment of bounding index arguments to sequence functions: in
1387 (SAFETY 3) code, errors will be signalled in almost all cases if
1388 invalid sequence bounding indices are passed to functions defined
1389 by ANSI to operate on sequences.
1390 * fixed a bug in the build procedure: documentation of SBCL-specific
1391 packages is now preserved and available in the final Lisp image.
1392 * lifted FDEFINITION lookup out of loops in the implementation of
1393 many list operations. (thanks to Robert E. Brown)
1394 * fixed a bug in the reader: the #n# reader macro now works for
1395 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1396 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1397 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1398 E. Brown and Tony Martinez respectively)
1399 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1400 SLOT-BOUNDP: in particular, it is now able to optimize them much
1401 better, and is now not vulnerable to having packages renamed.
1402 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1403 always called when a slot is not present in an instance. (thanks
1405 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1406 longer signal an error on unreading a character following EOF on
1407 the previous constituent stream. (thanks to Tony Martinez)
1408 * fixed some bugs revealed by Paul Dietz' test suite:
1409 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1410 not just nonnegative fixnums;
1411 ** the logical bit-array operators such as BIT-AND now accept an
1412 explicit NIL for their "opt-arg" argument (to indicate a
1413 freshly-consed result bit-array);
1414 ** ELT now signals an error on an invalid sequence index in safe
1416 ** the type system is now cleverer about negations of numeric
1417 types, and consequently understands the BIGNUM and RATIO types
1419 ** the type system is now cleverer about the interaction between
1420 INTEGER and RATIO types: while bugs still remain, many more
1421 cases are accurately computed;
1422 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1423 if it is in the last clause;
1424 ** CONSTANTLY now correctly returns a side-effect-free function in
1426 ** DECLARE is no longer treated as a special-operator; in
1427 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1428 * incremented fasl file version number due to the change in the
1429 DEFSTRUCT-SLOT-DESCRIPTION structure.
1431 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1432 * minor incompatible change: code processed by the "interpreter" or
1433 EVAL now has a compilation optimization policy of (DEBUG 2)
1434 (changed from (DEBUG 1)) to improve debuggability of interactive
1435 development, and to allow the use of the debug RETURN command in
1437 * an experimental implementation of the RETURN command for the
1438 debugger has been included. (thanks to Frederik Kuivinen)
1439 * fixed bug 62: constraints were not propagated into a loop.
1440 * fixed bug in embedded calls of SORT (reported and investigated by
1442 * fixed some bugs revealed by Paul Dietz' test suite:
1443 ** printing and reading of arrays with some dimensions having
1444 length 0 (thanks to Gerd Moellmann);
1445 ** BOA constructor with &AUX argument without a default value does
1446 not cause a type error;
1447 ** CONSTANTP now returns true for all self-evaluating objects.
1449 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1450 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1451 accessors that are related by inheritance, as specified in the
1452 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1453 Valtteri Vuorikoski)
1454 * The compiler is now able to inline functions that were defined in
1455 a complex lexical environment (e.g. inside a MACROLET).
1456 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1458 * Support for the upcoming FreeBSD-5.0 release has been included.
1459 (thanks to Dag-Erling Smorgrav)
1460 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1461 effect when it is not in a toplevel context.
1462 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1463 stand a better chance of being correct. (thanks to Gerd
1465 * fixed bug in COERCE, which now signals an error on coercing a
1466 rational to a bounded real type which excludes the expected
1468 * The compiler is now able to derive types more accurately from the
1469 COERCE and COMPILE functions.
1470 * fixed bug 223: functional binding is considered to be constant
1471 only for symbols in the CL package.
1472 * fixed bug 231: SETQ did not check the type of a variable being set
1473 (reported by Robert E. Brown)
1474 * A new optimization for MAKE-INSTANCE has been included, fixing
1475 various bugs (including relating to :ALLOCATION :CLASS slots and
1476 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1478 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1479 ** As required by ANSI, LOOP now disallows anonymous collection
1480 clauses such as COLLECT I in conjunction with aggregate boolean
1481 clauses such as THEREIS (= I 1);
1482 ** LOOP now signals an error when any variable is reused in the
1483 same loop (including the potentially useful construct analogous
1484 to WITH A = 1 WITH A = (1+ A);
1485 ** IT is only a special loop symbol within the first clause of a
1486 conditional loop clause;
1487 ** LOOP with a typed iteration variable over a hashtable now
1488 signals a type error iff it should.
1489 * fixed some other bugs revealed by Paul Dietz' test suite:
1490 ** FILE-STREAM now names the class previously known as FD-STREAM;
1491 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1492 argument) no longer signals an error;
1493 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1494 of the slot symbol, rather than using the current package
1495 ((:CONC-NAME "") continues to intern the slot's name in the
1497 * incremented fasl file version number, because of the incompatible
1498 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1499 of the new implementation of DEFINE-COMPILER-MACRO.
1501 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1502 * Support for building SBCL for MIPS platforms running in
1503 little-endian mode has now been checked in, and basic
1504 functionality on said platforms verified.
1505 * minor incompatible change: PCL now records the pathname of a file
1506 in which methods and the like are defined, rather than its
1508 * minor incompatible change: TRUENAME now considers the truename of
1509 a file naming a directory to be the pathname with :DIRECTORY
1510 component indicating that directory.
1511 * minor incompatible change: a NAMED clause in the extended form of
1512 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1513 reason for the previous behaviour is unclear.
1514 * more systematization and improvement of CLOS and MOP conformance
1515 in PCL (thanks to Gerd Moellman and Pierre Mai):
1516 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1518 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1519 primary methods with no specializers;
1520 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1522 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1523 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1524 FORWARD-REFERENCED-CLASSes; error reporting on
1525 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1527 ** SXHASH on CLOS instances now uses a slot internal to the
1528 instance to return different numbers on distinct instances,
1529 while preserving the same return value through invocations of
1531 ** DEFMETHOD signals errors when methods with longer incongruent
1532 lambda lists are added to generic functions;
1533 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1534 CLASS, as specified in AMOP;
1535 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1536 on the order returned by the primary method for classes of
1537 class STANDARD-CLASS;
1538 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1539 * fixed some bugs shown by Paul Dietz' test suite:
1540 ** DOLIST puts its body in TAGBODY;
1541 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1543 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1544 value producing form;
1545 ** if more variables are given to PROGV than values, extra
1546 variables are bound and made to have no value;
1547 ** NSUBSTITUTE on list arguments gets the right answer with
1549 ** ELT signals an error of type TYPE-ERROR when the index argument
1550 is not a valid sequence index;
1551 ** LOOP signals (at macroexpansion time) an error of type
1552 PROGRAM-ERROR when duplicate variable names are found;
1553 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1554 ** FUNCALL of special-operators now cause an error of type
1556 ** PSETQ now works as required in the presence of side-effecting
1557 symbol-macro places;
1558 ** NCONC accepts any object as its last argument;
1559 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1561 ** loop-for-as-package does not require a package to be explicitely
1563 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1565 * fixed bug 166: compiler preserves "there is a way to go"
1566 invariant when deleting code.
1567 * fixed bug 172: macro lambda lists with required arguments after
1568 &REST arguments now cause an error to be signalled. (thanks to
1570 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1571 bug. (thanks to Matthew Danish)
1572 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1574 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1575 arguments are lexically rebound. (thanks to Gerd Moellmann and
1577 * fixed bug 194: error messages are now more informative when there
1578 is no primary method applicable in a call to a generic
1579 function. (thanks to Gerd Moellmann)
1580 * fixed bug in command line argument checking (thanks to Julian
1582 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1583 (thanks to Matthew Danish)
1584 * incremented fasl file version number, because of the
1585 SXHASH-related changes in the layout of CLOS data structures
1587 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1588 * minor incompatible change: The runtime (the Unix executable named
1589 "sbcl") is now much pickier about the .core files it will load.
1590 Essentially it now requires .core files to descend from the same
1591 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1592 as the runtime does. (The intent is to prevent the crashes which
1593 can occur, and which can even be reported as mysterious failures,
1594 when people patch the sources or change the build parameters
1595 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1596 sbcl and .core files.)
1597 * fixed bug: VALUES-LIST is no longer optimized away.
1598 * fixed bug 142: The FFI conversion of C string values to Lisp
1599 string values no longer conses excessively. (thanks to Nathan
1600 Froyd porting Raymond Toy's fix to CMU CL)
1601 * began to systematize and improve MOP conformance in PCL (thanks to
1602 Nathan Froyd, Gerd Moellman and Pierre Mai):
1603 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1605 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1606 ** FINALIZE-INHERITANCE is now called on class finalization;
1607 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1608 argument precedence order.
1609 * fixed bug 202: The compiler no longer fails on functions whose
1610 derived types contradict their declared type.
1611 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1612 so it can be non-toplevel.
1613 * The fasl file version number has changed (because of the new
1614 implementation of DEFMACRO).
1615 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1616 safe code, that any length requirement by their type-specifier
1617 argument is valid. The exceptions to this are described in bug
1619 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1620 functions now check the types of their inputs as required by ANSI.
1621 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1622 introduced symbol is DECLAREd to be SPECIAL.
1623 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1624 * fixed bug: :COUNT argument to sequence functions may be negative
1625 * fixed bug: body of DO-SYMBOLS may contain declarations
1626 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1627 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1628 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1629 symbol macro only once
1630 * fixed printing of call frame when argument list is unavailable
1631 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1632 * compiler no longer signals WARNING on unknown keyword
1635 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1636 * A beta-quality port to the mips architecture running Linux,
1637 based on the old CMUCL backend, has been made. It has been tested
1638 on a big-endian kernel, and works sufficiently well to be able to
1639 rebuild itself; it has not been tested in little-endian mode.
1640 * fixed an inconsistency between gencgc.c and purify.c which made
1641 dumping/loading .core files unreliable
1642 * fixed bug 120a: The compiler now deals correctly with IFs where
1643 the consequent is the same as the alternative, instead of
1644 misderiving the return type. (thanks to Alexey Dejneka)
1645 * fixed bug 113: Logical pathnames are now dumpable (the logical
1646 host is resolved at load-time, throwing an error if it is not
1648 * fixed bug 174: FORMAT's error message is slightly clearer when a
1649 non-printing character is used in a format directive.
1650 * fixed several bugs in compiler checking of type declarations, i.e.
1651 violations of the Python "declarations are assertions" principle
1652 (thanks to Alexey Dejneka)
1653 * fixed several bugs in PCL's error checking (thanks to Gerd
1655 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1657 * fixed bug in compilation of functions as first class values
1658 (thanks to Antonio Martinez-Shotton)
1659 * The compiler's handling TYPE-ERRORs which it can prove will
1660 inevitably happen at runtime has been cleaned up and corrected
1661 in several ways. (thanks to Alexey Dejneka)
1662 * improved argument type checking for various basic arithmetic
1663 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1664 lavished on them in the past that they can be compiled in many
1665 ways in different special cases
1666 * fixed bug 181: compiler checks validity of user supplied type
1668 * cleaned up code flushing in optimization: Function calls which
1669 should signal errors for safety purposes (e.g. which ANSI says
1670 should signal errors when their arguments are of incorrect type)
1671 are no longer optimized away.
1672 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1673 * incremented fasl file version number, because changes in the
1674 implementation of sequence functions like COERCE caused
1675 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1676 in old inline expansions) to become undefined. (Actually these
1677 changes were later undone, so we might very well be binary
1678 compatible with 0.7.7 after all, but leaving the version number
1679 incremented seemed like the simplest and most conservative
1682 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1683 * An alpha-quality port to the parisc architecture running Linux,
1684 based on the old CMUCL backend, has been made. This, even more so
1685 than the other backends, should be considered still a work in
1686 progress; known problems include that the Linux kernel in 64-bit
1687 mode does not propagate the correct sigcontext structure to
1688 userspace, and consequently SBCL on a parisc64 kernel will not
1690 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1691 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1692 suppressed.) Also now that the compiler is looking at declarations
1693 in the environment, it checks optimization declarations as well,
1694 and suppresses inlining when (> DEBUG SPEED).
1695 * More fixes have been made to treatment of floating point exception
1696 treatment and other Unix signals. In particular, floating point
1697 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1698 * The detection and handling of control stack exhaustion (infinite
1699 or very deeply nested recursion) has changed. Stack exhaustion
1700 detection is now done by write-protecting pages at the OS level
1701 and applies at all optimization settings; when found, a
1702 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1703 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1704 be caught using IGNORE-ERRORS.
1705 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1706 correctly when transforming recursive LABELS functions to LETs.
1707 (thanks to Alexey Dejneka)
1708 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1709 that are names of constants or global variables.
1710 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1711 alien routines with docstrings.
1712 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1713 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1715 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1716 (thanks to Eric Marsden)
1717 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1718 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1719 * Bug fix: The SPARC backend can now compile functions involving
1720 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1721 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1722 to a functions expecting a PATHNAME-DESIGNATOR.
1723 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1724 lambda lists. (thanks to Alexey Dejneka)
1725 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1726 functions so that EXPORTing the name of the class doesn't cause
1727 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1728 to Antonio Martinez for reporting this)
1729 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1730 constant-folded without failing an assertion. (thanks to Einar
1731 Floystad Dorum for reporting this)
1732 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1733 types are now dealt with more correctly by the compiler.
1734 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1735 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1736 behaviour analogous to RENAME-FILE. This puts its behaviour more
1737 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1739 * The fasl file version number has changed again. (because of the
1740 bug fix involving the names of PCL MAKE-INSTANCE functions)
1742 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1743 * bug fix: Floating point exceptions are treated much more
1744 consistently on the x86/Linux and PPC/Linux platforms.
1745 * Array initialization with :INITIAL-ELEMENT is now much faster for
1746 cases when the compiler cannot open code the array creation, but
1747 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1748 array accesses have also seen a speed increase.
1749 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1750 specified. (This was at the root of some bad interactions between
1751 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1753 * bug fix: Internal error arguments for undefined functions are now
1754 computed correctly on the PPC/Linux platform.
1755 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1756 Raymond Toy's patch for CMU CL)
1757 * Support for the Solaris 9 operating environment has been included
1758 (thanks to Daniel Merritt)
1759 * A very ugly but hopefully complete draft of the missing FFI chapter
1760 of the manual has been created by reformatting the corresponding
1761 CMU CL manual chapter into (currently very ugly and incoherent)
1762 DocBook and bringing it up to date for SBCL behavior. Thus, the
1763 manual is now essentially complete, at least by my extreme
1764 once-and-only-once standards, whereby it's acceptable to refer to
1765 the doc strings of SB-EXT functions as the primary documentation.
1766 * The fasl file version number has changed again, due to cleanup of
1767 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1768 is no longer a static symbol.)
1770 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1771 * SBCL now builds with OpenMCL (version 0.12) as the
1772 cross-compilation host; also, more progress has been made toward
1773 bootstrapping under CLISP.
1774 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1776 * bug 158 fixed: The compiler can now deal with integer loop
1777 increments different from 1; fixing this turned out also to fix
1779 * bug 169 fixed: no more bogus warnings about using lexical bindings
1780 despite the presence of perfectly good SPECIAL declarations (thanks
1781 to David Lichteblau)
1782 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1783 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1784 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1786 * bug 180 fixed: Method combination specifications no longer ignore
1787 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1788 * bug fix: Structure type predicate functions now check their argument
1789 count as they should.
1790 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1791 correctly. (thanks to Pierre Mai)
1792 * minor incompatible change: The --noprogrammer option is deprecated
1793 in favor of the new --disable-debugger option, which is very similar.
1794 (The major difference is that it takes effect at a slightly different
1795 time at startup, causing handling of errors in --sysinit and
1796 --userinit files will be affected differently.) The
1797 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1798 been added to allow this functionality to be controlled from ordinary
1799 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1800 maintainers, who might want to run non-interactive scripts to
1801 build SBCL cores which will later be used interactively.)
1802 * minor incompatible change: The LOAD function no longer, when given
1803 a wild pathname to load, loads all files matching that pathname.
1804 Instead, an error of type FILE-ERROR is signalled.
1806 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1807 * bug 147 fixed: The compiler preserves its block link/count
1808 invariants more correctly now so that it doesn't crash. (thanks
1810 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1812 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1813 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1814 no-value-for-this-character as NIL (as specified by ANSI).
1815 * HOST-NAMESTRING on physical pathnames now returns a string that is
1816 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1817 (thanks to Christophe Rhodes)
1818 * The Alpha port handles icache flushing more correctly. (thanks to
1820 * More progress has been made toward bootstrapping under CLISP. (thanks
1821 to Christophe Rhodes)
1822 * The fasl file format has changed again, because dynamic loading
1823 on OpenBSD (which has non-ELF object files) motivated some cleanups
1824 in the way that foreign symbols are transformed and passed around.
1825 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1826 is no longer treated as whitespace by the reader, but instead as
1827 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1828 |A
\7fB|, instead of A as it used to.
1830 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1831 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1832 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1833 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1834 this as of 0.7.1.45, but was left out of the previous news section
1835 (thanks to Dan Barlow)
1836 * SBCL now runs on the Solaris operating system on SPARC architectures
1837 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1838 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1839 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1840 * SPARC backend cleanups, allowing builds of cores optimized for V8
1841 and V9 SPARCS, and also emission of code targeted to a particular
1842 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1844 * SBCL is closer to bootstrapping under CLISP, thanks to various
1845 fixes by Christophe Rhodes.
1846 * The fasl file format has changed again, to allow the compiler's
1847 INFO database to support symbol macros.
1848 * The user manual (in doc/) is formatted into HTML more nicely.
1849 (thanks to coreythomas)
1850 * The system is smarter about SUBTYPEP relationships, especially
1851 those involving NOT types (including types such as ATOM which are
1852 represented internally using NOT types). Thus SUBTYPEP is less
1853 likely to return (VALUES NIL NIL) in general, and in particular
1854 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1856 * The fasl file format has changed again, because the internal
1857 representation of types now includes a new slot to support the new
1858 SUBTYPEP-of-NOT-types logic.
1859 * (not a change in the main branch of SBCL, but a related prototype
1860 which can hopefully be merged into the main branch of SBCL in the
1861 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1862 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1863 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1864 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1865 dispatch macro character combinations. (thanks to Alexey Dejneka)
1866 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1867 that superficially look like logical namestrings correctly.
1868 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1870 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1871 default supplied-p) for &optional and &key arguments. (thanks to
1874 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1875 * incompatible change: The compiler is now less aggressive about
1876 tail call optimization, doing it only when (> SPACE DEBUG) or
1877 (> SPEED DEBUG). (This is an incompatible change because there are
1878 programs which relied on the old CMU-CL-style behavior to optimize
1879 away their unbounded recursion which will now die of stack overflow.)
1880 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1881 for non-GENCGC systems has been increased to 20M (since that
1882 seems much closer to the likely performance optimum for modern
1883 systems than the old 4M value was)
1884 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1885 and *DEBUG-PRINT-LEVEL*
1886 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1887 of CMU CL's support for SPARC, and various endianness and other
1888 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1889 * new syntactic sugar for the Unix command line: --load foo.bar is now
1890 an alternate notation for --eval '(load "foo.bar")'.
1892 ** The system now detects stack overflow and handles it gracefully,
1893 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1894 optimization settings. (This is a good thing in general, and
1895 its introduction in this version should be particularly timely
1896 for anyone whose code fails because of suppression of tail
1898 ** The system now hunts for the C variable "environ" in a more
1899 devious way, to avoid segfaults when the C library version differs
1900 between compile time and run time. (thanks to Christophe Rhodes)
1901 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1902 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1903 ** The compiler no longer issues bogus style warnings for undefined
1904 classes in the same source file as the DEFCLASSes which defined
1905 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1907 ** fixes in CONDITION class precedence list for undefined function
1908 errors (thanks to Alexei Dejneka)
1909 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1910 correctly. (thanks to Dan Barlow)
1911 ** portability fixes aiming at bootstrapping under CLISP (thanks
1912 to Dave McDonald and Christophe Rhodes)
1913 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1914 ** fixes in type translation and and type inference (thanks to
1916 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1917 ** various fixes in the new ports (thanks to Dan Barlow)
1918 * several changes related to debugging:
1919 ** suppression of tail recursion, as noted above
1920 ** stack overflow detection, as noted above
1921 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1922 is now the default. (For some time encapsulation has been more
1923 reliable than the breakpoint-based :ENCAPSULATE NIL
1924 implementation, at least on X86 systems; and I just noticed that
1925 encapsulation also seems closer to the spirit of the ANSI
1928 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1930 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1931 up properly again. (There was a packaging bug in 0.7.0 which
1932 left their definitions in SB-SYS::LOAD-FOREIGN and
1933 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1934 vital for most things which interface to C-level interfaces,
1935 like extensions working with sockets or databases or
1936 Perl-compatible regexes or whatever, and the need to fix
1937 this bug is the main reason that 0.7.1 was released so
1939 ** DEFGENERIC is now choosier about the methods it redefines, so that
1940 reLOADing a previously-LOADed file containing DEFGENERICs does
1941 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1942 cycle now works as it should. (thanks to Alexey Dejneka)
1943 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1944 fixed by Christophe Rhodes. (He actually submitted this patch
1945 months ago, and I delayed until after 0.7.0.)
1946 ** Bug 111 (internal compiler confusion about runtime checks on
1947 FUNCTION types) was fixed by Alexey Dejneka.
1948 * Some internal cleanups (getting rid of variables which aren't
1949 needed now that the byte interpreter is gone) caused the fasl
1950 file format number to change again.
1952 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1953 * major incompatible change: The default fasl file extension, i.e. the
1954 default extension for files produced by COMPILE-FILE, has changed
1955 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1957 ** There are many changes in the implementation of the compiler.
1958 SBCL is now essentially a compiler-only implementation of ANSI
1959 Common Lisp. EVAL still "interprets" a few special cases, but
1960 almost all the interesting cases are handled by creating
1961 a LAMBDA expression, calling COMPILE on it, then calling
1962 FUNCALL on the result.
1963 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1964 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1965 Since the code is newer, there might still be some new bugs
1966 (though not as many as before Martin Atzmueller's fixes:-). But
1967 the new code is substantially simpler and clearer, and hopefully
1968 any remaining bugs will be simpler, less fundamental, and more
1969 fixable then the bugs in the old code.
1970 ** The revised compiler is still a little unsteady on its feet.
1972 *** The debugging information it produces (particularly the names
1973 of FUNCTION objects) is sometimes much less useful than what
1974 the old compiler produced.
1975 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1976 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1977 within a MACROLET) has been temporarily weakened.
1978 ** There are new compiler optimizations for various functions:
1979 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1980 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1981 *** the math functions TRUNCATE, FLOOR, and CEILING
1982 *** the function-of-all-trades COERCE
1983 Mostly these should be transparent, but there's one
1984 potentially-annoying problem (bug 117): when the compiler
1985 inline-expands a function and does type analysis on the result,
1986 it can create control paths which have type mismatches, and
1987 when it can't prove that those control paths aren't taken,
1988 it will issue WARNINGs about the type mismatches. This is
1989 a particular problem in practice for the new sequence functions.
1990 It's not clear how this should be fixed, and for now, a
1991 workaround is given in the entry for 117 in the BUGS file.
1992 ** (Because of the interaction between the two previous items --
1993 occasional inlining problems and new inline expansions -- some
1994 of the new sequence function optimizations won't really kick in
1995 completely until debugging information, and then inlining, are
1996 straightened out in some future version.)
1997 * minor incompatible changes:
1998 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1999 DIRECTORY no longer implicitly promotes NIL slots of its
2000 pathname argument to :WILD. In particular, when you ask for the
2001 contents of a directory (which you used to be able to do without
2002 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2003 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2004 ** changes in behavior that ANSI explicitly defines to be
2005 implementation dependent:
2006 *** The new compiler-only implementation still conforms with ANSI,
2007 but acts a little different than before. Besides the obvious
2008 changes in performance tradeoffs (that the cost per form passed
2009 to EVAL has gone up, and the cost per form executed by EVAL
2010 has gone down), the behavior of the system changes a little
2011 because there are no longer any interpreted function objects.
2012 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2013 e.g. doing COMPILE on the output of interactive DEFUN is
2015 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2017 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2018 has changed from (:CL) to NIL.
2019 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2020 e.g. #\Newline and #\Tab, don't have names specified in the
2021 ANSI Common Lisp standard, is now based on their ASCII symbolic
2022 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2023 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2024 are no longer used for output.
2025 ** changes in internal implementation constants:
2026 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2027 4 million. (If your application spends a lot of time GCing and
2028 you have a lot of RAM, you might want to experiment with
2029 increasing it even more.)
2030 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2031 However, almost all old code should still continue to work without
2032 immediate update, as SB-C-CALL is now a (deprecated) nickname
2034 ** Old operator names in the style DEF-FOO are now deprecated in
2035 favor of new corresponding names DEFINE-FOO, for consistency with
2036 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2037 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2038 internal symbols, but a few supported extensions like
2039 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2040 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2041 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2042 etc. as you get deeper into recursive calls to the debugger
2043 command loop, instead of the old "5]", "5]]", "5]]]"
2044 sequence. (I was motivated to do this when squabbles between
2045 ILISP and SBCL left me very deeply nested in the debugger. In the
2046 short term, this change will probably provoke more ILISP/SBCL
2047 squabbles, but hopefully it will be an improvement in the long run.)
2048 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2049 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2050 function, since declaiming return types involving aliens is
2051 (1) annoyingly messy to do by hand and (2) vital to efficient
2052 compilation of code which calls such functions.
2053 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2054 longer reexported by the SB-EXT package. They're solely useful
2055 for alien code, so it seems more logical that you should get
2056 them from the SB-ALIEN package, not in SB-EXT.
2057 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2058 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2059 features. Instead, the code that they used to control is always
2060 built into the system.
2061 * many other bug fixes
2062 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2063 advantage of the new EVAL-WHEN stuff and to clean them up in
2064 general, and they are now more ANSI-compliant in a number of
2065 ways. Martin Atzmueller is responsible for a lot of this.
2066 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2068 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2069 *** correct ERROR type for various file operations
2070 *** some fixes for Lisp streams
2071 *** DEFMETHOD syntax checking
2072 *** changing old weird representation of debug information as
2073 strings (which, among their other deficiencies, don't transform
2074 correctly when you rename packages, and don't change their
2075 print representation when you change things like *PACKAGE*
2076 and *PRINT-LENGTH*) to symbols and lists of symbols
2077 He also made several improvements and fixed several bugs in DESCRIBE.
2078 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2080 *** misbehavior of WRITE-STRING/WRITE-LINE
2081 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2082 and several other LOOP problems as well
2083 *** DIRECTORY when similar filenames are present
2084 *** DEFGENERIC with :METHOD options
2085 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2086 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2087 *** argument ordering in FIND with :TEST option
2088 *** mishandled package designator argument in APROPOS-LIST
2089 *** various problems in the backquote readmacro
2090 *** a bug in APROPOS
2091 *** probably some others that I'm not describing very well here,
2092 since the CVS log documents them by reference to sbcl-devel
2093 messages, and the SourceForge archives aren't working well.:-(
2094 ** Dan Barlow improved the Alpha port (and is making progress on the
2095 PPC port, for those of you who think different).
2096 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2097 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2098 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2099 continues to work on the SPARC port (for those of you in a position
2100 to look down upon our little PC-compatible boxes from a great height).
2101 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2102 and uses that copy, rather than the current dynamic value, when
2103 it's trying to decide whether to truncate output. Thus e.g.
2104 (let ((*print-lines* 50))
2105 (pprint-logical-block (stream nil)
2107 (let ((*print-lines* 8))
2108 (print (aref possiblybigthings i) stream)))))
2109 should now truncate the logical block only at 50 lines, instead of
2110 often truncating it at 8 lines, as it did before.
2111 * The doc/cmucl/ directory, containing old CMU CL documentation
2112 from the time of the fork, is no longer part of the base system.
2113 SourceForge has shut down its anonymous FTP service, and with it
2114 my original plan for distributing the old CMU CL documentation
2115 there. For now, if you need these files you can download an old
2116 SBCL source release and extract them from it.
2117 * The fasl file version number changed again, for dozens of reasons,
2118 some of which are apparent above.
2120 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2121 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2122 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2123 patch, so that DISASSEMBLE output is much nicer.
2124 * The code in the SB-PROFILE package now seems reasonably stable.
2125 I still haven't decided what the final interface should look like
2126 (I'd like PROFILE to interact cleanly with TRACE, since both
2127 facilities use function encapsulation) but if you have a need
2128 for profiling now, you can probably use it successfully with
2129 the current CMU-CL-style interface.
2130 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2131 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2132 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2133 link now returns the dangling link itself, and for similar
2134 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2135 link itself. (In these cases the old code signalled an error and
2136 looped endlessly, respectively.) Thus, DIRECTORY now works even
2137 in the presence of dangling and cyclic symbolic links.
2138 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2139 is now a supported extension again, since the consensus on
2140 sbcl-devel was that it can be useful for ordinary development
2141 work, not just for debugging SBCL itself.
2142 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2143 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2144 that currently-defined functions might be redefined later with
2145 different return types.
2146 * Hash tables can be printed readably, as inspired by CMU CL code
2147 of Eric Marsden and SBCL code of Martin Atzmueller.
2148 * better error handling in CLOS method combination, thanks to
2149 Martin Atzmueller porting Pierre Mai's CMU CL patches
2150 * more overflow fixes for >16Mbyte I/O buffers
2151 * A bug in READ has been fixed, so that now a single Ctrl-D
2152 character suffices to cause end-of-file on character streams.
2153 In particular, now you only need one Ctrl-D at the command
2154 line (not two) to exit SBCL.
2155 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2157 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2158 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2159 does the right thing.
2160 * The implementation of some type tests, especially for CONDITION
2161 types, is now tidier and maybe faster, due to CMU CL code
2162 originally by Douglas Crosher, ported by Martin Atzmueller.
2163 * Some math functions have been fixed, and there are new
2164 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2165 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2166 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2167 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2168 likely to be built into sbcl-0.7.0.)
2169 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2170 CL, ported to SBCL by Martin Atzmueller)
2171 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2172 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2173 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2174 * There's a new slam.sh hack to shorten the edit/compile/debug
2175 cycle for low-level changes to SBCL itself, and a new
2176 :SB-AFTER-XC-CORE target feature to control the generation of
2177 the after-xc.core file needed by slam.sh.
2178 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2179 COMPILE-FILE is no longer supported, so that now every function
2180 gets an entry point, so that block compilation looks a little
2181 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2182 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2183 returns the number of bytes consed since the system started,
2184 rather than the number consed since the first time the function
2185 was called. (The new definition parallels ANSI functions like
2186 CL:GET-INTERNAL-RUN-TIME.)
2187 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2188 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2189 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2190 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2191 * Fasl file version numbers are now independent of the target CPU,
2192 since historically most system changes which required version
2193 number changes have affected all CPUs equally. Similarly,
2194 the byte fasl file version is now equal to the ordinary
2197 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2198 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2199 is no longer recognized. I apologize for this, because it was
2200 listed in SB-EXT as a supported extension, but I found that
2201 its existing behavior was poorly specified, as well as incorrectly
2202 specified, and it looked like too much of a mess to straighten it
2203 out. I have enough on my hands trying to get ANSI stuff to work..
2204 * many patches ported from CMU CL by Martin Atzmueller, with
2205 half a dozen bug fixes in pretty-printing and the debugger, and
2206 half a dozen others elsewhere
2207 * fixed bug 13: Floating point infinities are now supported again.
2208 They might still be a little bit flaky, but thanks to bug reports
2209 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2210 as flaky as they were.
2211 * The --noprogrammer command line option is now supported. (Its
2212 behavior is slightly different in detail from what the old man
2213 page claimed it would do, but it's still appropriate under the
2214 same circumstances that the man page talks about.)
2215 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2216 are now supported, and enabled by default. Thus, the compiler can
2217 handle many floating point and complex operations much less
2218 inefficiently. (Thus e.g. you can implement a complex FFT
2220 * The compiler now detects type mismatches between DECLAIM FTYPE
2221 and DEFUN better, and implements CHECK-TYPE more correctly, and
2222 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2223 thanks to patches from Martin Atzmueller.
2224 * various fixes to make the cross-compiler more portable to
2225 ANSI-conforming-but-different cross-compilation hosts (notably
2226 Lispworks for Windows, following bug reports from Arthur Lemmens)
2227 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2228 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2229 fixed thanks to Pierre Mai's CMU CL patches.
2230 * fixes to keep the system from overflowing internal counters when
2231 it tries to use i/o buffers larger than 16M bytes
2232 * fixed bug 45a: Various internal functions required to support
2233 complex special functions have been merged from CMU CL sources.
2234 (When I was first setting up SBCL, I misunderstood a compile-time
2235 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2236 * improved support for type intersection and union, fixing bug 12
2237 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2238 more obscure bugs as well
2239 * some steps toward byte-compiling non-performance-critical
2240 parts of the system, courtesy of patches from Martin Atzmueller
2241 * Christophe Rhodes has made some debian packages of sbcl at
2242 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2243 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2244 stable, but are nonetheless usable. When he's ready, I'd be happy
2245 to add them to the SourceForge "File Releases" section. (And if
2246 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2247 * new fasl file format version number (because of changes in
2248 internal representation of (OR ..) types to accommodate the new
2249 support for (AND ..) types, among other things)
2251 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2252 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2254 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2256 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2257 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2258 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2259 * DESCRIBE now works on structure objects again.
2260 * Most function call argument type mismatches are now handled as
2261 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2262 know whether the function will be redefined before the call is
2263 executed. (The compiler could flag local calls with full WARNINGs,
2264 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2265 it doesn't keep track of enough information to know whether calls
2266 are local in this sense.)
2267 * Compiler output is now more verbose, with messages truncated
2268 later than before. (There should be some supported way for users
2269 to override the default verbosity, but I haven't decided how to
2270 provide it yet, so this behavior is still controlled by the internal
2271 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2272 src/compiler/ir1util.lisp.)
2273 * Fasl file format version numbers have increased again, because
2274 support for the Gray streams extension changes the layout of the
2275 system's STREAM objects.
2276 * The Gray subclassable streams extension now works, thanks to a
2277 patch from Martin Atzmueller.
2278 * The full LOAD-FOREIGN extension (not just the primitive
2279 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2280 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2281 but like most other programs, it defaults to copying the Unix
2282 environment from the original process instead of starting the
2283 new process in an empty environment.
2284 * Extensions which manipulate the Unix environment now support
2285 an :ENVIRONMENT keyword option which doesn't smash case or
2286 do other bad things. The CMU-CL-style :ENV option is retained
2287 for porting convenience.
2288 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2289 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2291 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2293 * A patch from Martin Atzmueller seems to have solved the SIGINT
2294 problem, and as far as we know, signal-handling now works cleanly.
2295 (If you find any new bugs, please report them!)
2296 * The system no longer defaults Lisp source file names to types
2297 ".l", ".cl", or ".lsp", but only to ".lisp".
2298 * The compiler no longer uses special default file extensions for
2299 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2300 seems to expect a single default extension for all compiled code,
2301 and there's no compelling reason to try to stretch the standard
2302 to allow two different extensions.) Instead, byte-compiled files
2303 default to the same extension as native-compiled files.
2304 * Fasl file format version numbers have increased again, because
2305 a rearrangement of internal implementation packages made some
2306 dumped symbols in old fasl files unreadable in new cores.
2307 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2308 many fewer weird special cases.
2309 * Bug #17 (differing COMPILE-FILE behavior between logical and
2310 physical pathnames) has been fixed, and some related misbehavior too,
2311 thanks to a patch from Martin Atzmueller.
2312 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2313 by Tim Moore, ported to SBCL by Martin Atzmueller.
2314 * Martin Atzmueller fixed several filesystem-related problems,
2315 including bug #36, in part by porting CMU CL patches, which were
2316 written in part by Paul Werkowski.
2317 * More compiler warnings in src/runtime/ are gone, thanks to
2318 more patches from Martin Atzmueller.
2319 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2322 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2324 * DESCRIBE now works on CONDITION objects.
2325 * The debugger now handles errors which arise when trying to print
2326 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2328 * The build system now uses an additional file, customize-target-features.lisp,
2329 to allow local modifications to the target *FEATURES* list. (The point of
2330 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2331 debugging features enabled, without having to worry about propagating it
2332 into everyone's system when I do a "cvs update".) When no
2333 customize-target-features.lisp file exists, the target *FEATURES* list
2334 should be constructed the same way as before.
2335 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2336 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2337 warn about behavior which is undefined under the ANSI spec. Note: This
2338 is specified by ANSI, but it's not very popular with programmers.
2339 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2340 macro in the SBCL sources for an example of a workaround which you
2341 might use to make portable ANSI-standard code which does what you want.
2342 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2343 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2344 when it's used as a non-toplevel form. (This is required in order
2345 to implement the DEFCONSTANT-EQX macro.)
2346 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2347 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2348 funny things, due to bugs in the implementation of EVAL-WHEN
2349 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2350 non-toplevel DEFCONSTANTs are uncommon.)
2351 * The core file version number and fasl file version number have been
2352 incremented, because the old noncompliant DEFCONSTANT behavior involved
2353 calling functions which no longer exist, and because I also took the
2354 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2355 * fixed bug 1 (error handling before read-eval-print loop starts), and
2356 redid debugger restarts and related debugger commands somewhat while
2358 ** The QUIT debugger command is gone, since it did something
2359 rather different than the SB-EXT:QUIT command, and since it never
2360 worked properly outside the main toplevel read/eval/print loop.
2361 Invoking the new TOPLEVEL restart provides the same functionality.
2362 ** The GO debugger command is also gone, since you can just invoke
2363 the CONTINUE restart directly instead.
2364 ** The TOP debugger command is also gone, since it's redundant with the
2365 FRAME 0 command, and since it interfered with abbreviations for the
2367 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2369 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2370 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2371 some porting work by Martin Atzmueller)
2372 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2373 with a boatload of patches which he ported to SBCL. Now that those
2375 ** The system tries to make sure that its low-priority messages
2376 are prefixed by semicolons, to help people who like to use
2377 syntax highlighting in their ILISP buffer. (This patch
2378 was originally due to Raymond Toy.)
2379 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2380 patches originally written by Raymond Toy.
2381 ** The compiler understands coercion between single-value and
2382 multiple-VALUES type expressions better, getting rid of some very
2383 weird behavior, thanks to patches originally by Robert MacLachlan
2384 and Douglas Crosher.
2385 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2386 lambda lists, thanks to a patch originally by Pierre Mai.
2387 ** The system no longer bogusly warns about "abbreviated type
2389 ** The compiler gets less confused by inlining and RETURN-FROM,
2390 thanks to some patches originally by Tim Moore.
2391 ** The system no longer hangs when dumping circular lists to fasl
2392 files, thanks to a patch originally from Douglas Crosher.
2393 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2394 undefined function error.
2395 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2396 in structure slot initforms) for now, documented workaround instead:-|
2397 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2398 * fixed bug 5: added stubs for various Gray stream functions called
2399 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2400 installed, at least appropriate type errors are generated
2401 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2402 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2403 consistently in DEFMETHOD forms.
2404 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2405 it doesn't seem to affect SBCL after all
2406 * The C runtime system now builds with better optimization and many
2407 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2409 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2411 * The system is now under CVS at SourceForge (instead of the
2412 CVS repository on my home machine).
2413 * The new signal handling code has been tweaked to treat register
2414 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2415 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2416 downstream, e.g. in debug-int.lisp, has implicit dependencies
2417 on the unsignedness of integer representation of machine words,
2418 and that caused the system to bomb out with infinite regress
2419 when trying to recover from type errors involving signed values,
2420 e.g. (BUTLAST '(1 2 3) -1).)
2421 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2422 (This was one of the bugs Peter Van Eynde reported back in July.)
2423 * The system now uses code inspired by Colin Walters' O(N)
2424 implementation of MAP (from the cmucl-imp@cons.org mailing
2425 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2426 inline the MAP operation, and there is more than one
2427 sequence argument to the MAP call (so that it can't just
2428 do ETYPECASE once and for all based on the type of the
2429 single sequence argument). (The old non-inline implementation
2430 of the general M-argument sequence-of-length-N case required
2431 O(M*N*N) time when any of the sequence arguments were LISTs.)
2432 * The QUIT :UNIX-CODE keyword argument has been renamed to
2433 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2434 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2436 * Raymond Wiker's patches to port dynamic loading from Linux to
2437 FreeBSD have been added.
2438 * The BUGS file is now more nearly up to date, thanks in large part
2439 to Martin Atzmueller's review of it.
2440 * The debugger now flushes standard output streams before it begins
2441 its output ("debugger invoked" and so forth).
2442 * The core version number and fasl file version number have both
2443 been incremented, because of incompatible changes in the layout
2445 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2446 used to not be called for a saved Lisp image.)
2447 * Martin Atzmueller's version of a patch to fix a compiler crash,
2448 as posted on sbcl-devel 13 September 2000, has been installed.
2449 * Instead of installing Martin Atzmueller's patch for the
2450 compiler transform for SUBSEQ, I deleted the compiler transform,
2451 and transforms for some similar consing operations.
2452 * A bug in signal handling which kept TRACE from working on OpenBSD
2454 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2455 away by constant folding
2456 * The system now defines its address space constants in one place
2457 (in the Lisp sources), and propagates them automatically elsewhere
2458 (through GENESIS and the sbcl.h file). Therefore, patching the
2459 address map is less unnecessarily tedious and error-prone. The
2460 Lisp names of address space constants have also been systematized.
2461 * CVS tags like dollar-Header-dollar have been removed from
2462 the sources, because they have never saved me trouble and
2463 they've been source of trouble working with patches and other
2464 diff-related operations.
2465 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2466 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2468 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2470 * The system has been ported to OpenBSD.
2471 * The system now compiles with a simple "sh make.sh" on the systems
2472 that it's supported on. I.e., now you no longer need to tweak
2473 text in the target-features.lisp-expr and symlinks in src/runtime/
2474 by hand, the make.sh takes care of it for you.
2475 * The system is no longer so grossly inefficient when compiling code
2476 involving vectors implemented as general (not simple) vectors (VECTOR T),
2477 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2478 which dares to use the various sequence functions on non-simple
2479 vectors, takes less of a performance hit.
2480 * There is now a primitive type predicate VECTOR-T-P
2481 to test for the (VECTOR T) type, so that e.g.
2482 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2483 can now be compiled with some semblance of efficiency. (The old code
2484 turned the type declaration into a full call to %TYPEP at runtime!)
2485 * AREF on (VECTOR T) is still not fast, since it's still compiled
2486 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2487 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2488 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2489 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2490 without making full calls to SUBTYPEP at runtime.
2491 (Some analogous efficiency issues for non-simple vectors specialized to
2492 element types other than T, or for non-simple multidimensional arrays,
2493 have not been addressed. They could almost certainly be handled the
2494 same way if anyone is motivated to do so.)
2495 * The changes in array handling break binary compatibility, so
2496 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2497 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2498 instead of (VALUES T T).
2499 * By following the instructions that Dan Barlow posted to sbcl-devel
2500 on 2 July 2000, I was able to enable primitive dynamic object
2501 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2502 functionality is not implemented (since it calls ld to resolve
2503 library references automatically, requiring RUN-PROGRAM for its
2504 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2505 try to resolve library references) is now supported.
2506 * The system now flushes the standard output streams when it terminates,
2507 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2508 them at several other probably-convenient times, e.g. in each pass of
2509 the toplevel read-eval-print loop, and after evaluating a form given
2510 as an "--eval" command-line option. (These changes were motivated by a
2511 discussion of stream flushing issues on cmucl-imp in August 2000.)
2512 * The source transform for TYPEP of array types no longer assumes
2513 that an array whose element type is a not-yet-defined type
2514 is implemented as an array of T, but instead punts, so that the
2515 type will be interpreted at runtime.
2516 * There is now some support for cross-compiling in make.sh: each of
2517 the phases of make.sh has its own script. (This should be transparent
2518 to people doing ordinary, non-cross-compile builds.)
2519 * Since my laptop doesn't have hundreds of megabytes of memory like
2520 my desktop machine, I became more motivated to do some items on
2521 my to-do list in order to reduce the size of the system a little:
2522 ** Arrange for various needed-only-at-cold-init things to be
2523 uninterned after cold init. To support this, those things have
2524 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2525 symbols with such names are now uninterned after cold init).
2526 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2527 fasl files for cold load.
2528 ** Remove the old compiler structure pooling code (which used to
2529 be conditional on the target feature :SB-ALLOC) completely.
2530 ** Redo the representation of some data in cold init to be more compact.
2531 (I also looked into supporting byte compiled code at bootstrap time,
2532 which would probably reduce the size of the system a lot, but that
2533 looked too complicated, so I punted for now.)
2534 * The maximum signal nesting depth in the src/runtime/ support code has
2535 been reduced from 4096 to 256. (I don't know any reason for the very
2536 large old value. If the new smaller value turns out to break something,
2537 I'll probably just bump it back up.)
2538 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2540 * Many, many bugs reported by Peter Van Eynde have been added to
2541 the BUGS list; some have even been fixed.
2542 * While enabling dynamic object file loading, I tried to make the
2543 code easier to understand, renaming various functions and variables
2544 with less ambiguous names, and changing some function calling
2545 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2546 * While trying to figure out how to do the OpenBSD port, I tried to
2547 clean up some of the code in src/runtime/. In particular, I dropped
2548 support for non-POSIX signal handling, added various comments,
2549 tweaked the code to reduce the number of compilation warnings, and
2550 renamed some files to increase consistency.
2551 * To support the new automatic configuration functionality in make.sh,
2552 the source file target-features.lisp-expr has been replaced with the
2553 source file base-target-features.lisp-expr and the machine-generated
2554 file local-target-features.lisp-expr.
2555 * fixed a stupid quoting error in make.sh so that using CMU CL
2556 "lisp -batch" as cross-compilation host works again
2558 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2560 * DESCRIBE no longer tries to call itself recursively to describe
2561 bound/fbound values, so that it no longer fails on symbols which are
2562 bound to themselves (like keywords, T, and NIL).
2563 * DESCRIBE now works on generic functions.
2564 * The printer now prints less-screwed-up representations of closures
2565 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2566 * A private symbol is used instead of the :EMPTY keyword previously
2567 used to mark empty slots in hash tables. Thus
2568 (DEFVAR *HT* (MAKE-HASH-TABLE))
2569 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2570 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2571 now does what ANSI says that it should. (You can still get
2572 similar noncompliant behavior if bang on the hash table
2573 implementation with all the symbols you get back from
2574 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2575 This breaks binary compatibility, since tests for equality to
2576 :EMPTY are wired into things like the macroexpansion of
2577 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2579 * There's now a minimal placeholder implementation for CL:STEP,
2580 as required by ANSI.
2581 * An obscure bug in the interaction of the normal compiler, the byte
2582 compiler, inlining, and structure predicates has been patched
2583 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2584 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2586 * Missing ordinary arguments in a macro call are now detected even
2587 when the macro lambda list contains &KEY or &REST.
2588 * The debugger no longer complains about encountering the top of the
2589 stack when you type "FRAME 0" to explicitly instruct it to go to
2590 the top of the stack. And it now prints the frame you request even
2591 if it's the current frame (instead of saying "You are here.").
2592 * As specified by ANSI, the system now always prints keywords
2593 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2595 * The default initial SIZE of HASH-TABLEs is now smaller.
2596 * Type information from CLOS class dispatch is now propagated
2597 into DEFMETHOD bodies, so that e.g.
2598 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2600 is now basically equivalent to
2601 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2602 (DECLARE (TYPE SINGLE-FLOAT X))
2604 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2605 operation, without having to do run-time type dispatch.
2606 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2607 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2608 * Since I don't seem to be making big file reorganizations very often
2609 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2610 is overflowing my ability to conveniently back them up), I've finally
2611 checked the system into CVS. (The CVS repository is on my home system,
2612 not at SourceForge -- putting it on SourceForge might come later.)
2613 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2614 high-level GC-NOTIFY-FOO functions send their output. (There's
2615 still very little control of where low-level verbose GC functions
2616 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2617 less than it used to -- the GC-NOTIFY-FOO functions are now under
2618 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2619 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2620 in only one place in the source code, and propagates it automatically
2621 everywhere that it's needed. Thus e.g. when I bump the version from
2622 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2623 * The C source files now include boilerplate legalese and documentation
2624 at the head of each file (just as the Lisp source files already did).
2625 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2626 to his page will be replaced with a link to his new CLiki service.
2628 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2630 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2631 * The build process now looks for GNU make under the default name "gmake",
2632 instead of "make" as it used to. If GNU make is not available as "gmake"
2633 on your system, you can change this default behavior by setting the
2634 GNUMAKE environment variable.
2635 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2636 can build without error under CMU CL.
2638 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2640 * There is now a partial SBCL user manual (with some new text and some
2641 text cribbed from the CMU CL manual).
2642 * The beginnings of a profiler have been added (starting with the
2643 CMU CL profiler and simplifying and cleaning up). Eventually the
2644 main interface should be through the TRACE macro, but for now,
2645 it's still accessed through vaguely CMU-CL-style functions and macros
2646 exported from the package SB-PROFILE.
2647 * Some problems left over from porting CMU CL to the new
2648 cross-compilation bootstrap process have been cleaned up:
2649 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2650 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2651 part when getting rid of the compiler *BACKEND* stuff.)
2652 ** The value of *NULL-TYPE* was screwed up, because it was
2653 being initialized before the type system knew the final
2654 definition of the 'NULL type. This screwed up several key
2655 optimizations in the compiler, causing inefficiency in all sorts
2656 of places. (I found it because I wanted to understand why
2657 GET-INTERNAL-RUN-TIME was consing.)
2658 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2659 PROCLAIM FTYPE information. Unfortunately this broke binary
2660 compatibility again, since now the forms output by DEFGENERIC
2661 to refer to functions which didn't exist in 0.6.3.
2662 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2663 can use the new (as of 0.6.3) transform for SOME into MAP into
2665 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2666 values to (INTEGER 0 1000000), so that the time code will no longer
2667 occasionally get blown up by Linux returning 1000000 microseconds
2668 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2669 its output conform to the ANSI spec. (Alas, this makes its output
2670 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2671 * A full call to MAP NIL with a single sequence argument no longer conses.
2672 * fixes to problems pointed out by Martin Atzmueller:
2673 * The manual page no longer talks about multiprocessing as though
2674 it were currently supported.
2675 * The ILISP support patches have been removed from the distribution,
2676 because as of version 5.10.1, ILISP now supports SBCL without us
2677 having to maintain patches.
2678 * added a modified version of Raymond Toy's recent CMU CL patch for
2679 EQUALP comparison of HASH-TABLE
2681 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2683 * The system still can't cross-compile itself with
2684 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2685 entails), but at least it can compile more of itself that way
2686 than it used to be able to, and various buglets which were uncovered
2687 by trying to cross-compile itself that way have now been fixed.
2688 * This release breaks binary compatibility again. This time
2689 at least I've incremented the FASL file format version to 2, so that the
2690 problem can be detected reliably instead of just causing weird errors.
2691 * various new style warnings:
2692 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2693 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2694 ** using the lexical binding of a variable named in the *FOO* style
2695 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2696 as specified by ANSI.
2697 * *RANDOM-STATE* is no longer automatically initialized from
2698 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2699 default behavior of the system is to repeat its behavior every time
2700 it's run. If you'd like to change this behavior, you can always
2701 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2702 old convention there was no comparably easy way to get the system to
2703 repeat its behavior every time it was run.
2704 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2705 FTYPE declarations has been removed, in favor of their ANSI
2706 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2707 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2708 the types of their sequence arguments can be determined at compile time.
2709 This is done through a new open code expansion for MAP which eliminates
2710 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2711 argument types can be determined at compile time.
2712 * The optimizer now transforms COERCE into an identity operation when it
2713 can prove that the coerced object is already of the correct type. (This
2714 can be a win for machine generated code, including the output of other
2715 optimization transforms, such as the MAP transform above.)
2716 * Credit information has been moved from source file headers into CREDITS.
2717 * Source file headers have been made more standard.
2718 * The CASE macro now compiles without complaining even when it has
2721 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2723 * (Note that the way that the PCL macroexpansions were rewritten
2724 to accommodate the change in DEFGENERIC below breaks binary
2725 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2726 not run under sbcl-0.6.2. Once we get out of alpha releases,
2727 i.e. hit release 1.0.0, we'll probably try to maintain binary
2728 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2729 and sbcl-1.4.4. Until then, however, it might be fairly common
2730 for maintenance releases to break binary compatibility.)
2731 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2733 * The reporting of unhandled signals has been changed to print some
2734 explanatory text as well as the report form. (Previously only
2735 the report form was printed.)
2736 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2737 it defines, so that the compiler no longer issues undefined function
2738 warnings for compiled-but-not-yet-loaded generic functions.
2739 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2740 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2741 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2742 section of the ANSI spec are supported.
2743 * The "" nickname for the "KEYWORD" package has been removed.
2744 The reader still handles symbol tokens which begin with a package marker
2745 as keywords, but it doesn't expose its mechanism for doing so in the
2746 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2747 * The system now issues STYLE-WARNINGs for contradictory TYPE
2748 proclamations. (Warnings for contradictory FTYPE proclamations would
2749 be nice too, but those can't be done usefully unless the type system
2750 is made smarter about FUNCTION types.)
2751 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2752 systematized, so that "*target-*.lisp is supposed to exist only on the
2753 target and imply that there's a related file which exists on the
2754 host, and *host-*.lisp is supposed to exist only on the host and imply
2755 that there's a related file which exists on the target. This involves a
2756 lot of renaming. Hopefully the acute confusion caused by the renaming
2757 will be justified by the reduction in chronic confusion..
2758 ** runtime-type.lisp -> early-target-type.lisp
2759 ** target-type.lisp -> late-target-type.lisp
2760 ** early-host-format.lisp -> early-format.lisp
2761 ** late-host-format.lisp -> late-format.lisp
2762 ** host-error.lisp -> misc-error.lisp
2763 ** early-error.lisp -> early-target-error.lisp
2764 ** late-error.lisp -> late-target-error.lisp
2765 ** host-defboot.lisp -> early-defboot.lisp
2766 ** code/misc.lisp -> code/target-misc.lisp
2767 ** code/host-misc.lisp -> code/misc.lisp
2768 ** code/numbers.lisp -> code/target-numbers.lisp
2769 ** code/early-numbers.lisp -> numbers.lisp
2770 ** early-host-type.lisp -> early-type.lisp
2771 ** late-host-type.lisp -> late-type.lisp
2772 ** host-typep.lisp -> typep.lisp
2773 ** load.lisp -> target-load.lisp
2774 ** host-load.lisp -> load.lisp
2775 ** host-disassem.lisp -> disassem.lisp
2776 ** host-insts.lisp -> insts.lisp
2777 ** byte-comp.lisp -> target-byte-comp.lisp
2778 ** host-byte-comp.lisp -> byte-comp.lisp
2779 ** host-signal.lisp -> signal.lisp
2780 ** host-defstruct.lisp -> defstruct.lisp
2781 ** late-target-type.lisp -> deftypes-for-target.lisp
2782 Furthermore, several other previously target-only files foo.lisp (e.g.
2783 hash-table.lisp and random.lisp) have been split into a target-and-host
2784 foo.lisp file and a target-only target-foo.lisp file, with their key type
2785 definitions in the target-and-host part, so that the cross-compiler will
2786 know more about target types.
2787 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2788 gone away. In their place are various *BACKEND-FOO* variables
2789 corresponding to the slots of the old structure.
2790 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2791 into the target SBCL has been fixed.
2792 * The chill.lisp system for loading cold code into a running SBCL
2794 * Support for the CMU CL "scavenger hook" extension has been removed.
2795 (It was undocumented and unused in the CMU CL sources that SBCL was
2796 derived from, and stale in sbcl-0.6.1.)
2797 * Various errors in the cross-compiler type system were detected
2798 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2799 (enabling various consistency checks). Many of them were fixed,
2800 but some hard problems remain, so the compiler is back to
2801 running without *TYPE-SYSTEM-INITIALIZED* for now.
2802 * As part of the cross-compiler type system cleanup, I implemented
2803 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2804 * I have started adding UNCROSS calls throughout the type system
2805 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2806 on cross-compiler input files will be able to go away, and various
2808 * CONSTANTP now returns true for quoted forms (as explicitly required
2811 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2813 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2814 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2815 (MAKE-LIST -1). In the longer term, it ought to become true
2816 that declarations are assertions even at SAFETY 1. For now, it's not
2817 quite true even at SAFETY 3, but it's at least more nearly true..
2818 (Note that this change seems to increases the size of the system by
2819 O(5%) and to decrease the speed of the compiler by 20% or more.)
2820 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2821 for the problem of printing dozens of lines of distracting information
2822 about low-level system machinery as part of the top stack frame
2823 on entry to the debugger when an undefined function was called.
2824 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2825 is not reset to COMMON-LISP-USER.
2826 * Compilation of stuff related to dyncount.lisp has been made conditional
2827 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2828 smaller. The various dyncount-related symbols have been moved into
2829 a new "SB-DYNCOUNT" package.
2830 * tty-inspect.lisp has been renamed to inspect.lisp.
2831 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2832 feature has gone away. (When we eventually port to other flavors of
2833 libc and/or Unix, we'll try to make the differences between flavors
2834 invisible at the user level.)
2835 * Various other *FEATURES* tags, and/or their associated conditionals,
2836 have been removed if obsolescent, or given better documentation, or
2837 sometimes given more-mnemonic names.
2839 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2841 * tidied up "make.sh" script
2842 * tidied up system directory structure
2843 * better "clean.sh" behavior
2844 * added doc/FOR-CMUCL-DEVELOPERS
2845 * many many small tweaks to output format, e.g. removing possibly-confusing
2846 trailing #\. character in DESCRIBE-INSTANCE
2847 * (EQUALP #\A 'A) no longer signals an error.
2848 * new hashing code, including EQUALP hashing
2849 * tidied up Lisp initialization and toplevel
2850 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2851 * command line argument processing
2852 * added POSIX-GETENV function to deal with Unix-ish environment variables
2853 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2854 terminating SBCL on EOF
2855 * non-verbose GC by default
2856 * There is no more "sbcl" shell script; the sbcl file is now the C
2857 runtime executable (just like CMU CL).
2858 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2860 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2861 symbol and package information as Lisp native symbol and package objects
2862 instead of strings naming symbols and strings naming packages. This way,
2863 whenever packages are renamed (as in warm init), debug information is
2864 transformed along with everything else.
2865 * tweaked the optimization policy declarations which control the building
2866 of SBCL itself. Now, among other things, the system no longer saves
2867 source location debugging information. (This helps two problems at once
2868 by reducing SBCL size and by keeping SBCL from trying to look for its
2869 sources -- which may not exist -- when reporting errors.)
2870 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2871 debugging and testing purposes
2872 * cleaned up printing, making the printer call PRINT-OBJECT for
2873 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2874 methods, giving nearly-ANSI behavior
2875 * converted almost all special variables to use *FOO* naming convention
2876 * deleted PARSE-TIME functionality, since it can be done portably
2877 * moved some files out of cold init into warm init
2878 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2880 * regularized formatting of source files
2881 * added an install.sh script
2882 * fixed ridiculous memory usage of cross-compiler by making
2883 compiler/alloc.lisp not try to do pooling unless it can hook
2884 itself into the GC of the cross-compilation host. Now the system
2885 builds nicely on my old laptop.
2886 * added :SB-ALLOC in target-features.lisp-expr
2887 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2888 was not implemented)
2889 * re-did condition handling and note reporting in the compiler. Notes
2890 are no longer handled by signalling conditions. Style warnings
2891 and warnings are handled more correctly and reported in such a way
2892 that it's easy to find one or the other in your output (so that you
2893 can e.g. figure out which of many problems caused COMPILE-FILE to
2895 * changed the severity of several compiler warnings from full WARNING
2896 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2897 compiler note reporting so that it doesn't use the condition system
2898 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2899 COMPILE-FILE command)
2900 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2901 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2902 no longer supported, since their implementation was deeply intertwingled
2903 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2904 * removed bogus "support" for compiler macros named (SETF FOO), and
2905 removed the compiler macro for SETF INFO (but only after making a fool
2906 of myself on the cmucl-imp mailing list by posting a bogus patch for
2907 DEFINE-COMPILER-MACRO..)
2908 * Compiled files containing forms which have side effects on the Lisp
2909 reader (such as DEFPACKAGE forms) are now handled more correctly.
2910 (Compiler queuing of top level lambdas has been suppressed by setting
2911 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2912 * deleted various currently-unused source files, e.g. gengc.lisp. They
2913 may be added back at some point e.g. when porting to other architectures,
2914 but until they are it's distracting to distribute them and to try to
2916 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2917 there were so many of them they're just distractions, and UNCROSS is
2918 known to be able to handle the current sources
2919 * moved PROFILE functionality into TRACE, so that it will be clear
2920 how the wrapping and unwrapping of functions when you profile them
2921 interacts with the wrapping and unwrapping of functions when you
2922 trace them. (Actually, the functionality isn't there yet, but at least
2923 the interface specification is there. Hopefully, the functionality will
2924 arrive with some future maintenance release.)
2925 * removed host-oops.lisp
2926 * changed signature of QUIT function to allow UNIX-CODE argument
2927 * fixed READ-SEQUENCE bug
2928 * tweaked verbose GC output so that it looks more like the progress
2929 output that ANSI specifies for functions like LOAD
2930 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2931 * added <http://sbcl.sourceforge.com> to the banner information printed by