1 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
3 platform. The port supports 61-bit fixnums, large memory spaces
4 and reloading shared object files.
5 * enhancement: saving cores with foreign code loaded is now
6 supported on ppc/Darwin in addition to the previously supported
8 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
10 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
11 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
12 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
13 produces an error. (thanks to Vincent Arkesteijn)
14 * bug fix: NAMESTRING on pathnames with :WILD components in their
15 directories works correctly. (thanks to Artem V. Andreev)
16 * fixed bug 125: compiler preserves identity of closures. (reported
18 * build fix: fixed the dependence on *LOAD-PATHNAME* and
19 *COMPILE-FILE-PATHNAME* being absolute pathnames.
20 * on x86 compiler partially supports stack allocation of dynamic-extent
22 * fixed some bugs related to Unicode integration:
23 ** encoding and decoding errors are now much more robustly
24 handled; it should now be possible to recover even from invalid
25 input or output to the terminal. (thanks to Teemu Kalvas)
26 ** provided a first cut at implementing STRING-TO-OCTETS and
27 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
28 ** altered the SB-MD5 contributed module slightly, changing the
29 interface just enough for it to be supportable for builds where
30 lisp characters are not eight bits.
31 * fixed some bugs revealed by Paul Dietz' test suite:
32 ** the FORMATTER-generated functions for ~V[ conditionals require
33 the correct number of arguments.
35 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
36 * new feature: reloading changed shared object files with
37 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
38 * new feature: references to foreign variables and functions
39 can now be compiled and loaded before the corresponding shared
40 object file is loaded, as long as the foreign definitions are
42 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
43 supported. (thanks to Dan Debertin)
44 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
45 just keywords. (thanks to Vincent Arkesteijn)
46 * fixed bug #331: structure-class instances corresponding to
47 DEFSTRUCT forms are now created eagerly.
48 * fixed bug #345: backtraces from calls to undefined functions work
49 on x86 as well. Related bug #61 is now also partially fixed on x86
50 (backtraces from throws to unknown catch tags.)
51 * bug fix: lambda-list parsing is now stricter vrt. order and number
52 of lambda-list keywords.
53 * bug fix: as specified by AMOP, an error is signalled if a
54 class-option appears multiple times in a DEFCLASS form. (reported
56 * bug fix: RANDOM can be compiled when the compiler derives the type
57 of its numeric argument as a disjoint set of small integers.
58 (reported by Paul Dietz)
59 * bug fix: starting a core saved with shared objects loaded when
60 those objects are not available no longer causes threaded SBCL to
61 hang. (reported by Sean Ross)
62 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
64 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
65 argument, which must always be NIL. (reported by Kalle Niemitalo)
66 * bug fix: printing 1.0d+23 no longer results in an error.
67 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
68 * bug fix: structure slot setters preserve evaluation order. (thanks
70 * fixed some bugs related to Unicode integration:
71 ** RUN-PROGRAM can allow its child to take input from a Lisp
72 stream. (reported by Stefan Scholl)
73 ** ASDF-INSTALL successfully downloads .tar.gz files in all
74 locales. (reported by Ken Causey)
75 * fixed some bugs revealed by Paul Dietz' test suite:
76 ** INCF, DECF and REMF evaluate their place form as specified in
78 ** FORMATTER expands ~{ ~} iteration directives with V or #
80 ** FORMATTER deals with the ~@[ ~] conditional directive where the
81 consequent uses no arguments correctly.
82 ** the system has a partial understanding of the (COMPLEX RATIO)
84 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
87 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
88 * new feature: a build-time option (controlled by the :SB-UNICODE
89 keyword feature, enabled by default) for building the system with
90 support for the entire 21-bit character space defined by the
92 * new feature: the system now has rudimentary external-format
93 support; the primary user-visible change at this time is that
94 characters with the high bit set (such as lower-case-e-acute) will
95 print correctly to a terminal in a UTF-8 environment.
96 * minor incompatible change: BASE-CHAR no longer names a class;
97 however, CHARACTER continues to do so, as required by ANSI.
98 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
99 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
100 has been moved to the SB-EXT package (temporarily re-exported from
102 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
103 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
105 * fixed bug #351: better error-handlind and reporting for malformed
107 * fixed bug #350: bignum-printing is now more memory-efficient,
108 allowing printing of very large bignums, eg. (expt 2 10000000).
109 (reported by Bruno Haible)
110 * fixed bug #302: better primitive-type selection for intersection
112 * fixed bug #308: non-graphic characters now all have names, as
113 required. (reported by Bruno Haible)
114 * bug fix: redefining a class with different superclasses now correctly
115 removes it from the direct-subclasses of its previous superclasses.
116 (reported by David Morse)
117 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
118 new value now works. (reported by Bruno Haible)
119 * bug fix: correct canonicalization of multiple non-standard slot
120 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
121 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
122 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
123 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
124 now exists, an signals an error.
125 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
126 during macroexpansion for non-iteration variables as well. (reported
127 by Bruno Haible for CMUCL)
128 * bug fix: Cyclic structures and unprintable objects in compiler
129 messages no longer cause errors. (reported by Bruno Haible)
130 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
131 and READ-FROM-STRING all now return a primary value of NIL if
132 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
133 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
134 (reported by Bruno Haible for CMUCL)
135 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
136 do not designate a condition. (reported by Bruno Haible for
138 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
139 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
140 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
141 permitting resolution in favour of any of the conflicting symbols.
142 (reported by Bruno Haible for CMUCL)
143 * bug fix: EQUAL compiler optimizations is less aggressive on
144 strings which can potentially compare true despite having distinct
145 specialized array element types.
146 * bug fix: unit enumerations can be defined without dividing by
147 zero. (thanks to Vincent Arkesteijn)
148 * FORMAT compile-time argument count checking has been enhanced.
149 (report from Bruno Haible for CMUCL)
150 * a partial workaround for the bug 262: the compiler does not try to
151 inline-expand a local function doing RETURN-FROM from a deleted
152 BLOCK. (thanks to Peter Denno for the bug report and to David
153 Wragg for the simple test case)
154 * fixed some bugs revealed by Paul Dietz' test suite:
155 ** RENAME-PACKAGE allows all package designators as new package
157 ** constraint propagation and lambda variable substitution are
158 more cautious in dealing with partially deleted code.
159 ** compiler sometimes lost reoptimization passes.
160 ** CERROR, given a condition as condition designator, passes its
161 remaining arguments to the continue format control without
163 ** Case-altering FORMAT directives work correctly on non-ASCII
165 ** The REAL type specifier handles bounds outside the floating
166 point ranges without signalling FLOATING-POINT-OVERFLOW.
167 ** Functions with IR1-transformations can create intercomponent
168 references to global functions.
169 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
171 ** FORMAT ~R treats a nil value for its first parameter correctly.
173 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
174 * enhancement: saving cores with foreign code loaded is now
175 supported on x86/NetBSD and sparc/Linux in addition to the previously
177 * bug fix: on some platforms repeated installations caused multiple
178 copies of HTML documentation to be installed -- should not happen
179 any more. (reported by Stefan Scholl)
180 * bug fix: parsing self-recursive alien record types multiple times
181 no longer causes infinite recursion. (reported by Thomas F. Burdick,
182 original patch by Helmut Eller for CMUCL)
183 * bug fix: stack-exhaustion detection works now on NetBSD as well.
184 (thanks to Richard Kreuter)
185 * bug fix: defining classes whose accessors are methods on existing
186 generic functions in other (locked) packages no longer signals
187 bogus package lock violations. (reported by François-René Rideau)
188 * bug fix: special variables as DEFMETHOD parameters no longer have
189 associated bogus type declarations. (reported by David Wragg and
191 * bug fix: read-write consistency on streams of element-type
192 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
193 * bug fix: redefiniton of the only method of a generic function with
194 no DEFGENERIC no longer emits a full WARNING. In addition,
195 redefinition of generic functions with no DEFGENERIC to an
196 incompatible lambda list now signals an error. (thanks to Zach
198 * bug fix: DEFGENERIC now works even when there's a function of the
199 same name in an enclosing lexical environment. (thanks to Zach
201 * fixed compiler failure, caused by instrumenting code during
202 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
203 * optimization: added loop analysis and improved register allocation
204 * fixed some bugs revealed by Paul Dietz' test suite:
205 ** POSITION on displaced vectors with non-zero displacement
206 returns the right answer.
207 ** (SIMPLE-STRING) is a valid type specifier for sequence
209 ** *PRINT-LEVEL* handling for slotless structures is pedantically
211 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
212 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
214 ** The justification version of the FORMAT ~< directive treats
215 non-zero minpad parameter correctly.
217 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
218 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
219 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
220 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
221 the supported interface.
222 * new feature: Single-stepping of code compiled with DEBUG 2 or
223 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
224 * new feature: saving cores with foreign code loaded is now
225 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
226 Timothy Moore's work for CMUCL)
227 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
228 parameters to * instead of NIL if no initform is supplied.
229 (reported by Johan Bockgaard)
230 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
231 correctly when FUNCALL appears as the car of the form. Note:
232 despite this FUNCALL forms are not currently subject to
233 compiler-macro expansion. (port of Raymond Toy's fix for the
234 same from CMUCL, reported by Johan Bockgaard)
235 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
236 (thanks for Teemu Kalvas)
237 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
238 case instead of terminating the iteration (thanks for Julian
239 Squires, Sean Champ and Raymond Toy)
240 * bug fix: incorrect expansion of defgeneric that caused a style
241 warning. (thanks for Zach Beane)
242 * bug fix: slot accessor effective method computation works properly
243 for classes with multiple non-standard applicable methods on
244 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
245 * on x86 compiler supports stack allocation of results of LIST and
246 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
247 CMUCL implementation by Gerd Moellmann)
248 * fixed some bugs revealed by Paul Dietz' test suite:
249 ** FORMAT strings with both the ~<~:;~> form of the justification
250 directive and pretty-printing directives cause an error.
252 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
253 * incompatible change: the internal functions
254 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
255 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
256 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
257 instead of the old functions.
258 * new feature: on platforms where "dladdr" is available foreign
259 function names now appear in backtraces. (based on Helmut Eller's
261 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
262 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
264 * documentation: ASDF manual is now distributed with SBCL.
265 * bug fix: SBCL can now load its contributed modules with REQUIRE
266 even if the system-provided entries have been removed from
267 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
269 * bug fix: backtraces involving undefined functions or assembly
270 routines are more informative. (thanks to Brian Downing)
271 * bug fix: mutually referent alien structures now work correctly.
272 (reported by Rick Taube)
273 * bug fix: structures defined by WITH-ALIEN can be referred to
274 within other definitions in the same WITH-ALIEN.
275 * bug fix: division operators (MOD, TRUNCATE and the like) with
276 constant zero divisors and integer dividends no longer generate
278 * bug fix: provide default methods for INPUT-STREAM-P and
279 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
280 * bug fix: improve the detection and resolution of MOP metacycles
281 (where effective method computation depends on the generic function
282 having its effective method computed). (reported by Bruno Haible)
283 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
284 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
285 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
286 * bug fix: pathnames with relative directory components can be
287 represented relative to default pathnames.
288 * optimization: in taking the GCD of bignums, reduce the two bignums
289 to approximately the same size (using Euclid's algorithm) before
290 applying the more sophisticated binary GCD. (thanks to Juho
292 * optimization: COUNT on bitvectors now operates word-at-a-time.
293 * optimization: ASH with a positive, but not necessarily constant,
294 (leftwards) shift, when in a modular context, is compiled to a
296 * fixed some bugs revealed by Paul Dietz' test suite:
297 ** FORMAT variable parameters ("~V<char>") are defaulted properly
298 if the corresponding argument is NIL.
299 ** FORMAT directives accepting a minpad parameter treat negative
301 ** literal commas as character parameters to FORMAT directives are
303 ** literal spaces directly after ~<Newline> directives within a
304 format-logical-block (~:< ~@:>) do not induce :FILL-style
305 conditional newlines.
306 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
307 as specified: it no longer includes conditional newlines).
308 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
310 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
311 :PER-LINE-PREFIX argument does not evaluate to a string.
313 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
314 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
315 the manual for details; for now, package locks can be disabled by
316 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
317 if no major problems are found then it is likely that they will be
318 compiled in unconditionally.
319 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
320 unsupported operators on all platforms. To load a shared library
321 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
322 object file, link it into a shared library outside of SBCL and
323 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
325 * bug workaround: ROOM T and the GC were not working together
326 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
327 reporting facility aren't preserved. That reporting has been
328 disabled, so now until and unless someone figures out how to make it
329 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
330 * minor incompatible change: as threatened around sbcl-0.8.0, the
331 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
332 The symbols are also exported from SB-PCL for backwards
333 compatibility, but more so than before SB-PCL should be treated as
334 an implementation-internal package.
335 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
337 * the SB-SPROF contrib now works on (most) non-x86 architectures.
338 It is known as of this release not to work on the Alpha, however.
339 * fixed bug #167: errors signalled due to illegal syntax in method
340 bodies are now more legible.
341 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
342 designators and can hence be used with TYPEP.
343 * fixed bug #333: CHECK-TYPE now ensures that the type error
344 signalled, if any, has the right object to be accessed by
345 TYPE-ERROR-DATUM. (reported by Tony Martinez)
346 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
347 5.1.2.3 for multiple-value place subforms. (reported by Kalle
349 * fixed bug #334: programmatic addition of slots using specialized
350 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
351 and :ALLOCATION :CLASS slots.
352 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
353 float exponents. (rereported by Peter Seibel)
354 * fixed a bug: #\Space (and other whitespace characters) are no
355 longer considered to be macro characters in standard syntax by
357 * fixed bug: initialization of condition class metaobjects no longer
358 causes an instance of the condition to be created. (reported by
360 * fixed bug: it is now possible to have slots such that
361 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
362 description is neither :INSTANCE nor :CLASS.
363 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
364 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
365 SLOT-VALUE-USING-CLASS), and no longer causes errors with
366 non-standard SLOT-DEFINITION-ALLOCATION values.
367 * fixed bugs: various race conditions handling exiting threads.
368 CL-PPCRE's thread torture test now passes.
369 * fixed arguably-a-bug: GC time proportional to number of threads in
370 system even when most of them are idle
371 * optimization: improved performance of BIT and SBIT on bit-vectors.
372 * .fasl file incompatibility: The fasl file version number has
373 been incremented because of changes associated with package locks.
375 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
376 * minor incompatible change: the system no longer provides
377 optimization hints (conditions which are TYPEP
378 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
379 mode; these hints are still emitted when higher SPEED optimization
381 * new contrib module: a sampling profiler (profiling by statistical
382 sampling, rather than by instrumenting functions) is available as
383 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
384 * the behaviour of the standard function ED is now customizeable by
385 third parties through a hook variable: see ED's documentation
386 string for information on the protocol.
387 * the compiler no longer emits efficiency notes for (FUNCALL X)
388 when the type of X is uncertain under default optimization
390 * fixed bug 276: mutating a binding of a specialized parameter to a
391 method to something that is not TYPEP the specializer is now
393 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
394 exhibit the required behaviour on the x86 platform. (thanks to
395 Peter van Eynde, Eric Marsden and Bruno Haible)
396 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
397 even for difficult arguments. (reported by Peter Graves)
398 * fixed bug 141a: the backquote printer now descends quoted
400 * fixed another bug in backquote printing: no more destructive
401 modification of the form's list structure. (reported by Brian
403 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
404 move between its address being taken and the call to
405 interrupt_thread, fixing a crashing race condition.
406 * the SB-POSIX contrib implementation has been adjusted so that it
407 no longer exhibits ridiculously poor performance when constructing
408 instances corresponding to C structs.
410 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
411 * minor incompatible change: the sb-grovel contrib now treats C
412 structures as alien (in the sense of SB-ALIEN) objects rather than
413 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
414 has implications for memory management of client code
415 (sb-grovel-returned objects must now be manually managed) and for
416 type safety (alien objects now have full types).
417 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
418 used to control emission of compiler diagnostics, rather than the
419 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
420 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
421 quality should be considered deprecated.
422 * install.sh now installs the user manual as well
423 * (not quite a new documentable feature, but worth considering in
424 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
425 beginnings of a semantically meaningful condition hierarchy is
426 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
428 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
429 designator as the defaults argument.
430 * fixed bug: Displaced arrays whose displaced-to array has become
431 too small now cause ARRAY-DIMENSION to signal an error, providing
432 for safer bounds-checking. (reported by Bruno Haible)
433 * fixed bug: DEFCLASS slot definitions with identical :READER and
434 :WRITER names now signal a reasonable error. (reported by Thomas
436 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
438 * fixed bug: Invalid slot specification errors now print correctly.
439 (thanks to Zach Beane)
440 * fixed bug 320: Shared to local slot value transfers in class
441 redefinitions now happen corrently from superclasses as
442 well. (reported by Bruno Haible)
443 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
445 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
446 improper lists correctly. (reported by Bruno Haible)
447 * fixed bug 313: source-transform for <fun-name> was erroneously
448 applied to a call of a value of a variable with name <fun-name>.
449 (reported by Antonio Menezes Leitao)
450 * fixed bug 307: The obsolete instance protocol ensures that
451 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
452 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
453 remove dead unknown-values globs from the middle of the stack.
454 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
455 easier to resume long computations after using *BREAK-ON-SIGNALS*
456 to diagnose and fix failures. (thanks to Nikodemus Siivola)
457 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
458 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
459 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
460 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
461 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
463 * fixed a (fairly theoretical) bug in string printing: if
464 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
465 does not have array-element-type equal to the most general string
467 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
468 function name. (thanks to Nikodemus Siivola)
469 * fixed bug: compile-time format string checker failed on
470 non-closed ~{. (reported by Thomas F Burdick)
471 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
472 don't warn on legal loop code involving more than one aggregate
473 boolean. (thanks to Nikodemus Siivola)
474 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
475 in keyword position to MAKE-PACKAGE was not being reported as
477 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
478 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
480 * fixed bugs 280 and 312: the checking for multiple definitions in a
481 file is less likely to become confused by uses of inline
483 * fixed bug: the #S reader macro performs the keyword coercion
484 specified for slot names. (reported by Kalle Niemitalo)
485 * fixed bug: lambda lists may contain symbols whose names start with
486 & but are not lambda-list-keywords; their occurrence triggers a
488 * fixed bug 321: define-method-combination argument lambda lists do
489 not cause invalid code to be generated when &optional or &aux
490 variables have default values. (reported by Bruno Haible)
491 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
492 have CLOS classes; this appears also to have fixed reported
493 instability in stack exhaustion detection.
494 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
495 MUFFLE-WARNING when no associated restart is present is now
497 * optimization: rearranged the expansion of various defining macros
498 so that each expands into only one top-level form in a
499 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
501 * optimization: used a previously little-used slot in symbols to
502 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
504 * fixed some bugs revealed by Paul Dietz' test suite:
505 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
507 ** Fixed an optimization bug related to inheritance of initforms
508 from local to shared slots.
509 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
510 ** CAST optimizer forgot to flush argument derived type.
511 ** print/read consistency on floats is now orders of magnitude
512 more likely. (thanks also to Bruno Haible for a similar report
514 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
515 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
516 for objects of type REAL. Make it so.
517 ** SXHASH is specified (infelicitously) to respect similarity,
518 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
519 it so. (thanks to Markus Ziegler)
520 ** on the Alpha, the compiler succeeds in compiling functions
521 returning a known number of arguments greater than 63.
522 ** fixed handling of invalid NIL arguments in keyword position
524 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
526 ** provided a readably-printable representation for RANDOM-STATE
528 ** ensured that pathnames, where they have a namestring, always
529 print using #P"..." syntax.
531 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
532 * Support for the forthcoming 2.0 version of the NetBSD kernel
533 running on x86 hardware has been added. (thanks to Perry
534 E. Metzger most immediately, and others for their past work)
535 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
536 current release version; SBCL's OpenBSD support had been broken
537 since about the time of OpenBSD's switch to ELF binary format.)
538 * [placeholder for DX summary]
539 ** user code with &REST lists declared dynamic-extent, under high
540 speed or space and low safety and debug optimization policy.
541 * The manual has been converted to Texinfo format and the debugger
542 chapter from the cmucl manual has been added.
543 * A facility has been added to extract documentation strings from
544 sbcl and store them as Texinfo-formatted snippets for inclusion in
545 the manual (via Texinfo's @include directive)
546 * bug fix: compiler emitted division in optimized DEREF. (thanks for
547 the test case to Dave Roberts)
548 * bug fix: multidimensional simple arrays loaded from FASLs had fill
549 pointers. (reported by Sean Ross)
550 * bug fix: PROFILE output is printed nicely even for large numerical
551 values. (thanks to Zach Beane)
552 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
553 greater than 32 handle EOF correctly.
554 * bug fix: on X86 an immediate argument of the IMUL instruction is
555 correctly printed in disassembly. (thanks to Lutz Euler)
556 * bug fix: class slots in redefined classes preserve their old
557 values. (thanks to Bruno Haible and Nikodemus Siivola)
558 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
559 succeeds. (reported by Marco Baringer)
560 * fixed some bugs revealed by Paul Dietz' test suite:
561 ** READ-SEQUENCE now works on ECHO-STREAMs.
562 ** RATIONALIZE works more according to its specification. (thanks
565 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
566 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
567 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
568 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
569 should matter to you only if you rebind the printer control
570 variables and then find you want different bindings in the
571 debugger than in the ordinary execution of your program.)
572 * The runtime build system has been tweaked to support building
573 (on SPARC/SunOS) using a C compiler which invokes Sun's own
574 assembler and linker. (thanks to Nikodemus Siivola)
575 * Unbound, undefined, undeclared variables now trigger full
576 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
577 more often programmer error than deliberate exploitation of undefined
579 * optimization: the hash algorithm for strings has changed to one
580 that is less vulnerable to spurious collisions. (thanks to Juho
582 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
583 needless bounds checking. (thanks to Juho Snellman)
584 * optimization: implemented multiplication as a modular
585 (UNSIGNED-BYTE 32) operation on the PPC backend.
586 * fixed some bugs revealed by Paul Dietz' test suite:
587 ** ADJUST-ARRAY now copies the datum in a zero rank array if
589 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
590 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
591 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
593 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
595 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
597 ** ECHO-STREAMs no longer attempt to echo the end of file value to
598 their output stream on EOF from read.
599 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
600 have been read to end-of-file.
601 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
603 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
604 description of determination of which consecutive characters
606 ** Printing the "Space" character with escaping on now yields "#\\ ",
607 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
608 ** Reading floating-point numbers with *READ-BASE* set to a number
609 less than 10 works correctly.
610 ** Reading floating-point numbers with *READ-BASE* set to a number
611 more than 10 works correctly.
612 ** Printing with *PRINT-READABLY* targets the standard readtable, not
613 the readtable currently in effect.
615 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
616 * minor incompatible change: parsing of namestrings on a physical
617 (Unix) host has changed; numbers after the final #\. in a
618 namestring are no longer interpreted as a version field. This is
619 intented to be largely invisible to the user, except that the
620 meaning of the namestring "*.*.*" has changed: it now refers to a
621 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
622 should usually be replaced by
623 (make-pathname :name :wild :type :wild :version :wild)
624 with the added benefit that this is more likely to be portable.
625 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
626 option to OPEN now signals an error if the file being opened
627 exists; this may have an impact on existing code.
628 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
629 signals received, on the PowerPC platforms (both Linux and
630 Darwin). (thanks to Pierre Mai for pointing out the location of
632 * several fixes on OS X: The system now builds and runs cleanly on
633 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
634 (thanks to Brian Mastenbrook)
635 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
636 second-resolution: integer multiples of 1/3600 between -24 and 24.
637 (thanks to Vincent Arkesteijn)
638 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
639 type when called with 1 argument; PEEK-CHAR checked type of
640 PEEK-TYPE only after having read first character from a
641 stream. (reported by Peter Graves)
642 * bug fix: the garbage collector now has much better locality
643 behaviour, and in particular no longer treats all memory as being
644 exhausted when many small objects point to each other in a deeply
646 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
647 recognized as being TYPEP their class.
648 * bug fix: the PUSHNEW documentation string has been corrected.
649 (thanks to Vincent Arkesteijn)
650 * bug fix: defaulting of the value for the last of an atypically
651 large number of multiple values being bound was not being
652 performed correctly on the Alpha or PPC platforms
653 * optimization: implemented multiplication as a modular
654 (UNSIGNED-BYTE 32) operation on the x86 backend.
655 * optimization: SEARCH on simple-base-strings can now be open-coded.
656 (see also contrib/compiler-extras.lisp for inspiration for
657 teaching the compiler about the Boyer-Moore algorithm).
658 * value, returned by MAX (and MIN) called with several EQUALP, but
659 not EQL, arguments now does not depend on compiler settings.
660 * fixed some bugs revealed by Paul Dietz' test suite:
661 ** in stack analysis liveness information is propagated from
662 non-local entry points.
663 ** pathwise CAST removing failed when the CAST node did not start
665 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
666 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
668 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
669 host is already defined.
670 ** RENAME-FILE works on streams instead of signalling an internal
672 ** PEEK-CHAR uses the current readtable when determining whether
673 or not a character is whitespace.
674 ** MERGE-PATHNAMES handles the case when the pathname does not
675 specify a name while the default-pathname specifies a version.
676 ** Pathnames now stand a better chance of respecting print/read
678 ** Attempting to use standardized file system operators with a
679 pathname with invalid :DIRECTORY components signals a
681 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
683 ** DIRECTORY on logical pathnames is more correct.
684 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
685 signal a TYPE-ERROR if their argument is not a stream.
686 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
687 designator argument does not designate a stream.
688 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
689 examining the synonym.
690 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
692 ** OPEN and WITH-OPEN-STREAM allow opening streams with
693 element-type larger than ([UN]SIGNED-BYTE 32).
695 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
696 * When built with the :SB-FUTEX feature, threaded builds now take
697 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
698 for faster/more reliable mutex and condition variable support.
699 * Incompatible change (but one you probably shouldn't have been using
700 anyway): the interface and code for arbitrating between multiple
701 threads in the same user session has been redesigned.
702 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
703 error when called without an explicit environment argument.
704 (thanks to Nikodemus Siivola)
705 * bug fix: buffered :DIRECTION :IO streams are less likely to become
706 confused about their position. (thanks to Adam Warner and Gerd
708 * bug fix: Pretty printing backquoted forms with unquotations in the
709 argument list position of various code constructs such as LAMBDA
710 now works correctly. (reported by Paul Dietz)
711 * bug fix: Pretty printing unquotations no longer loses all
712 stream position information.
713 * optimization: performance of string output streams is now less
714 poor for multiple small sequence writes.
715 * optimization: performance of CSUBTYPEP in the presence of complex
716 expressions involving CONS and NOT many times has been improved.
717 (reported by Paul Dietz)
718 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
720 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
721 streams for the REPL, for the debugger, and so on. (thanks to
723 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
724 * fixed some bugs revealed by Paul Dietz' test suite:
725 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
726 is ignored if it is too small, rather than propagating through
727 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
728 ** extremely complex negations of CONS types were not being
729 sufficiently canonicalized, leading to inconsistencies in
731 ** VALUES tranformer lost derived type.
733 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
734 * fixed a bootstrapping bug: the build process no longer assumes
735 that the various BOOLE-related constants have the same value in
736 host and target lisps. (noted by Paul Dietz' test suite on an
737 SBCL binary built from CLISP)
738 * The system can now be dynamically linked on the MIPS platform,
739 which enables dynamic loading of foreign code from Lisp. (thanks
740 to Ralf Baechle for discussions on the MIPS ABI)
741 * The system now records debugging information for its own source
742 files in a filesystem-position-independent manner, relative to
743 the "SYS" logical host.
744 * fixed a compiler bug: MV-LET convertion did not check references
745 to the "max args" entry point. (reported by Brian Downing)
746 * tweaked disassembly notes to be less confident about proclaiming
747 some instruction as an LRA. (thanks to Brian Downing)
748 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
749 now each have their own history, command character, and other
750 characteristics. (thanks to David Lichteblau)
751 * fixed some bugs revealed by Paul Dietz' test suite:
752 ** compiler failure in compiling LOGAND expressions including a
754 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
755 shift greater than 32.
756 ** FLUSH-DEST did not mark blocks for type check regeneration.
757 ** HANDLER-CASE failed to accept declarations in handler clauses
758 in some circumstances.
760 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
761 * New code in contrib/sb-introspect (still probably not entirely
762 stable yet) provides some support for smart Lisp development
763 environments like SLIME.
764 * The conditions signalled for errors occurring when loading .fasl
765 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
766 in a way which should help ASDF recover gracefully.
767 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
768 stuff has been cleaned up. If you code contrib/ stuff, this might
769 affect you, and you can look at contrib/README, contrib/STANDARDS,
770 and/or the 0.8.4.27 diff to check.
771 * In full calls the compiler now does not generate checks for declared
772 argument types for all arguments.
773 * various threading fixes
774 ** and some experimental patches which didn't make it into
775 the main tree for this release, but which are shipped in
776 contrib/experimental-thread.patch as a possible fix for some
777 failures (deadlock, spinning...) in GC-intensive multithreaded
779 * fixed PPC build problem (source code incompatibility of different
780 library versions): added offsetof() hackery which attempts to divine
781 where glibc maintainers put uc_mcontext today
782 * fixed bug 282: compiler does not trust type assertions while passing
783 arguments to a full call.
784 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
785 &OPTIONAL VALUES type parameter.
786 * fix bug 214: algorithm for noting rejected templates is now more
787 similar to that of template seletion. (also reported by rydis on
789 * fixed bug 141b: printing backquoted information readably and prettily
790 inserts a space where necessary.
791 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
792 (CHARACTER 10) are now reported as errors, rather than propagated
793 as unknown types. (reported by piso on #lisp)
794 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
795 with values NIL and :ERROR. (thanks to Milan Zamazal)
796 * fixed bug 191c: CLOS now does proper keyword argument checking as
797 described in CLHS 7.6.5 and 7.6.5.1.
798 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
799 longer raise an error; further, using a list as a for-as-arithmetic
800 counter now raises a meaningful error.
801 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
802 now understood by sequence creation functions such as MAKE-SEQUENCE
804 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
805 a STRING-INPUT-STREAM.
806 * compiler enhancement: SIGNUM is now better able to derive the type
808 * type declarations inside WITH-SLOTS are checked. (reported by
810 * fixed some bugs revealed by Paul Dietz' test suite:
811 ** incorrect optimization of TRUNCATE for a positive first
812 argument and negative second.
813 ** compiler failure in let-convertion during flushing dead code.
814 ** compiler failure while deriving type of TRUNCATE on an
815 interval, containing 0.
816 ** ASH of a negative bignum by a negative bignum count now returns
818 ** intersection of CONS types now canonicalizes properly, fixing
819 inconsistencies in SUBTYPEP.
821 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
822 * incompatible change: The --disable-debugger command line
823 option now clobbers the debugger at a more fundamental
824 level, by redefining #'INVOKE-DEBUGGER instead of by
825 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
826 is specified by ANSI to ignore *DEBUGGER-HOOK* and
827 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
828 enter the debugger REPL and then suffer recursive errors
829 because *DEBUG-IO* is also messed up in --disable-debugger mode;
830 while under the new system, BREAK in --disable-debugger mode
831 terminates the system just as an unhandled error would.
832 * fixed compiler performance when processing loops with a step >1;
833 * bug fix: DOCUMENTATION now retrieves generic function
834 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
835 support has been systematized, and now supports the methods
836 specified by ANSI, along with a default method and a method for
837 slot documentation. (reported by Nathan Froyd)
838 * bug fix: effective methods associated with a generic function are
839 no longer cached over a change of that generic function's method
840 combination. (reported by Andreas Fuchs)
841 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
842 names a non-existent directory. (thanks to Andreas Fuchs)
843 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
844 the PPC platform, lead to stack corruption; this has been fixed.
845 (reported by Rainer Joswig)
846 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
847 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
848 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
849 arithmetic in general, is now much more reliable on the PPC
851 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
852 platform now returns the right answer.
853 * optimization: restored some effective method precomputation in
854 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
855 precomputation is now tunable.
856 * optimization: compiler-internal data structure use has been
857 reviewed, and changes have been made that should improve the
858 performance of the compiler by about 20%.
859 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
860 simple-base-strings and simple-bit-vectors is improved.
861 * optimization: the optimization of 32-bit logical and arithmetic
862 functions introduced in version 0.8.3 on the x86 has been
863 implemented on the mips, ppc and sparc platforms; an
864 implementation of the same facility, but for 64-bit arithmetic,
865 has been added for the alpha.
866 * microoptimization: the compiler is better able to make use of the
867 x86 LEA instruction for multiplication by constants.
868 * bug fix: in some situations compiler did not report usage of
869 generic arithmetic in (SPEED 3) policy.
870 * bug 145b fix: compiler used wrong type specifier while converting
871 MEMBER-types to numeric.
872 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
874 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
875 index argument now works properly on the Alpha platform.
876 * bug fix: floating point exception treatment on the Alpha platform
878 * bug fix: FILE-POSITION works much better on string input and
879 output streams. (thanks to Nikodemus Siivola)
880 * bug fix: many threading/garbage collection symptoms sorted.
881 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
882 might be pseudo-atomic.
883 * internal change: Stopping for GC is now done with signals not
884 ptrace. GC is now done in whichever thread wanted it, instead of
886 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
888 * bug fix: result form in DO is not contained in the implicit
890 * incompatible change: ICR structure is changed; the value part of
891 CONTINUATION is now called LVAR; corresponding functions are
892 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
893 * added type deriver for ISQRT (thanks to Robert E. Brown).
894 * bug fix: better support for loading from the command line when an
895 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
897 * fixed some bugs revealed by Paul Dietz' test suite:
898 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
899 ** ROUND and FROUND now give the right answer when given very
900 small float arguments.
901 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
903 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
904 ** (GCD 0 <negative-integer>) returned <negative-integer>.
905 ** LCM should return a non-negative integer.
906 ** PARSE-INTEGER returned the index of a terminator instead of the
907 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
908 ** PARSE-INTEGER returned an incorrect index being applied to a
910 ** LCM with two arguments of 0 returns 0 rather than signalling
912 ** unsigned addition of a 32-bit constant with the high bit set no
913 longer causes an internal compiler error on the x86.
914 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
915 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
916 with negative last argument.
917 ** byte specifiers with zero size and position no longer cause
918 an error during type derivation.
919 ** bignum multiplication on the Alpha platform now returns the
921 * porting: The system now builds on SuSE AMD64, although it still
922 generates a 32-bit binary.
923 * .fasl file incompatibility: The fasl file version number has
924 been incremented (because of the changes to internal compiler
925 data structures referred to above).
927 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
928 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
929 more accurately, on the Darwin kernel running on PowerPC hardware.
930 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
931 * Compiler code deletion notes now signal a condition of type
932 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
933 an associated MUFFLE-WARNING restart.
934 * The compiler now performs limited argument count validation of
935 constant format strings in FORMAT, and where appropriate in ERROR,
936 CERROR and WARN. (thanks to Gerd Moellmann)
937 * New ASDF-INSTALL contrib can be used for automatic download and
938 installation of third-party Lisp code from CCLAN or other sites
940 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
941 which forces another thread to execute a function supplied by the
943 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
944 now accept and act upon their :ELEMENT-TYPE keyword argument.
945 (reported by Martin Atzmueller, Edi Weitz)
946 * bug fix: FILE-POSITION now accepts position designators up to
947 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
948 is the greater. (thanks to Patrik Nordebo)
949 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
950 Kalle Olavi Niemitalo)
951 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
952 printable. (reported by Eric Marsden)
953 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
954 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
955 * x86 bug fix in control stack exhaustion checking: now shows backtrace
956 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
957 (thanks to Stig Sandoe)
958 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
959 UNREAD-CHAR now correctly includes the unread character in the
960 target sequence. (thanks to Gerd Moellmann)
961 * bug fix in threaded builds: the system can now be suspended and
962 resumed by shell job control with minimal disruption.
963 * bug fixes in times and timezones >2038AD
964 * better handling of "where is GNU make?" problem in build scripts
965 (thanks to Nikodemus Siivola)
966 * new optimization: inside a named function any reference to a
967 function with the same name is considered to be a self-reference;
968 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
969 optimization quality.
970 * new optimization on x86: logical functions and + now have
971 optimized (UNSIGNED-BYTE 32) versions, which are automatically
972 used when the result is truncated to 32 bits.
973 * VALUES declaration is partially enabled.
974 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
975 * fixed some bugs revealed by Paul Dietz' test suite:
976 ** The system now obeys the constraint imposed by
977 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
978 types form a lattice under type intersection.
979 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
980 ** ASSOC now ignores NIL elements in an alist.
981 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
982 and (1+ MOST-POSITIVE-FIXNUM) answers.
983 ** The addition of a method with invalid qualifiers to a generic
984 function does not cause an error to be signalled immediately;
985 a warning is signalled, and the error is generated only on
986 calling the generic function.
987 * changed .fasl file version number, in part to add type codes for
988 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
989 obscure ANSI requirements
991 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
992 * fixed bug 148: failure to inline-expand a local function left
993 garbage, confusing the compiler.
994 * fixed bugs 3cd: structure slot readers perform type check if the
995 slot can have an invalid value (i.e. it is either not initialized
996 or can be written with a less specific slot writer).
997 * bug fix: the compiler now traps array references to elements off
998 the end of an array; previously, the bounds checking in some
999 circumstances could go off-by-one.
1000 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1002 * type declarations for array element types now obey the description
1003 on the CLHS page "Declaration TYPE", as per discussions on
1004 sbcl-help around 2003-05-08. This means that a declaration
1005 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1006 declaration, all references to BAR will be asserted or assumed
1007 (with THE, so dependent on compiler policy) to involve objects of
1008 type FOO. Note that no such declaration is implied in
1009 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1010 * declared types of functions from the "Conditions"
1011 chapter. (reported by Paul Dietz)
1012 * bug fix: CERROR accepts a function as its first argument.
1013 * bug fix: NTH an NTHCDR accept a bignum as index
1014 arguments. (reported by Adam Warner)
1015 * optimization: character compare routines now optimize comparing
1016 against a constant character. (reported by Gilbert Baumann)
1017 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1018 argument now works properly on the MIPS platform.
1019 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1020 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1021 declared to have a functional type, does not cause an error
1023 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1024 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1025 * fixed bug: if last continuation of a deleted block has a
1026 destination, this destination should be deleted too. (reported by
1028 * fixed a bug in the bootstrap process: the host compiler's values
1029 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1030 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1031 test case from Patrik Nordebo)
1032 * improved the ability of the disassembler on the PPC platform to
1033 provide helpful disassembly notes.
1034 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1035 the class in more cases than previously.
1036 * bug fix: FILE-POSITION now understands :START and :END for
1037 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1038 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1039 8) data. (thanks to David Lichteblau)
1040 * bug fix: it is possible to add a method to a generic function
1041 without lambda list.
1042 * bug fix: reader failed to signal END-OF-FILE inside an
1043 object representation. (reported by Nikodemus Siivola)
1044 * fixed some bugs revealed by Paul Dietz' test suite:
1045 ** LAST and [N]BUTLAST should accept a bignum.
1046 ** condition slot accessors are methods.
1047 ** (VECTOR NIL) is a subtype of STRING.
1049 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1050 * minor incompatible change: some nonsensical specialized lambda
1051 lists (used in DEFMETHOD) which were previously ignored now signal
1053 * minor incompatible change: the system is now aware of the types of
1054 variables in the COMMON-LISP package, and will signal errors for
1055 most violations of these type constraints (where previously they
1056 were silently accepted).
1057 * minor incompatible change: COMPILE-FILE now uses the freedom
1058 afforded (ANSI 3.2.2.3) to use derived function types for
1059 functions defined in the same file. This also permits the system
1060 to warn on static type mismatches and function
1061 redefinition. (Currently it does not work with high DEBUG level.)
1062 * minor incompatible change: VALUES declaration is disabled.
1063 * When issuing notes, the compiler now signals a condition of type
1064 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1065 restart for use in user handlers. It is expected that the
1066 COMPILER-NOTE condition will eventually become a condition
1067 supertype to a hierarchy of note types, which will then be
1068 handleable in a similar fashion. However, at the moment, no such
1069 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1070 but it's an implementation detail, not a classification for the
1072 * Changes in type checking closed the following bugs:
1073 ** type checking of unused values (192b, 194d, 203);
1074 ** template selection based on unsafe type assertions (192c, 236);
1075 ** type checking in branches (194bc).
1076 * A short form of VALUES type specifier has ANSI meaning (it has
1077 increased the number of situations when SBCL cannot perform type
1079 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1080 keywords or constants is permissible.
1081 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1082 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1083 classes. (thanks to Antonio Martinez)
1084 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1085 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1086 argument to a pretty-print stream. Instead, it leaves any such
1087 operations to DESCRIBE-OBJECT methods.
1088 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1090 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1091 lambda lists. (thanks to David Lichteblau)
1092 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1093 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1094 (thanks to Gerd Moellmann)
1095 * fixed bug 246: increased compilation speed of long
1096 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1098 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1100 * DEFCONSTANT now throws a condition of type
1101 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1102 constant to a non-EQL value; CONTINUE and ABORT restarts
1103 respectively change and preserve the value.
1104 * fixed bug 63: The code walker, part of the implementation of CLOS,
1105 is now better at handling symbol macros.
1106 * bug fix: There is no longer an internal implementation type named
1107 CL:LENGTH. (reported by Raymond Toy)
1108 * bug fix: In macro-like defining macros/special operators the
1109 implicit block does not enclose the lambda list.
1110 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1111 enclosed in parentheses) are not suitable as type specifiers, and
1112 their use properly signals an error now.
1113 * bug fix: An argument count mismatch for a type specifier in code
1114 being compiled no longer causes an unhandled error at compile
1115 time, but signals a compile-time warning.
1116 * fixed simple vector readable printing
1117 * bug fix: DESCRIBE takes more care over whether the class
1118 precedence list slot of a class is bound before accessing it.
1119 (reported by Markus Krummenacker)
1120 * bug fix: FORMATTER can successfully compile pretty-printer format
1121 strings which use variants of the ~* directive inside.
1122 * bug fix: SEARCH now applies its TEST predicate to the elements of
1123 the arguments in the correct order. (thanks to Wolfhard Buss)
1124 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1125 in type inference. (thanks to Robert E. Brown)
1126 * bug fix: Reading in symbols with an explicit package name of ""
1127 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1128 * fixed some bugs revealed by Paul Dietz' test suite:
1129 ** NIL is now allowed as a structure slot name.
1130 ** Arbitrary numbers, not just REALs, are allowed in certain
1131 circumstances in LOOP for-as-arithmetic clauses.
1132 ** Multiple class redefinitions before slot access no longer
1133 causes a type error.
1134 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1135 association between the name and a class.
1136 ** Generic functions with non-standard method-combination and over
1137 five methods all of which return constants no longer return NIL
1138 after the first few invocations. (thanks to Gerd Moellmann)
1139 ** CALL-NEXT-METHOD with no arguments now passes the original
1140 values of the arguments, even in the presence of assignment.
1141 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1143 ** DISASSEMBLE works with closures and funcallable instances.
1144 ** ADD-METHOD now returns the generic function, not the new method.
1145 ** FIND-METHOD signals an error if the lengths of the specializers
1146 is incompatible with the generic function, even if the ERRORP
1148 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1149 which its argument is a member.
1150 ** DEFCLASS only redefines the class named by its class-name
1151 argument if that name is the proper name of the class;
1152 otherwise, it creates a new class.
1153 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1154 of the UNBOUND-SLOT condition to the name of the slot.
1155 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1156 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1157 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1158 SLOT-MAKUNBOUND in the specified fashion.
1160 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1161 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1162 cross-compilation host. As a consequence, we can now bootstrap our
1163 way up to SBCL starting with a bare gcc toolchain and human-readable
1164 source code (first the source to CLISP, then the source to SBCL).
1165 * A contributed module containing a partial implementation of the
1166 simple-streams interface has been included. (thanks to Rudi
1168 * A contributed module implementing the RFC1321 Message Digest
1169 Algorithm, known as MD5, has been included.
1170 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1171 no longer has any effect, as the code controlled by this feature
1172 has been deleted. (As far as we know, no-one has ever built using
1173 this feature, and its semantics were confused in any case).
1174 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1175 work on conditions (as required by the ANSI specification),
1176 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1177 expected behaviour on conditions. Users should note, however,
1178 that such behaviour is not required by the ANSI specification,
1179 and so use of this behaviour may render their code unportable.
1180 * fixed some bugs revealed by Paul Dietz' test suite:
1181 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1183 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1184 ** SXHASH on condition objects no longer returns NIL.
1185 ** :ALLOCATION :CLASS slots are better treated; their values are
1186 updated on class redefinition, and initforms inherited from
1187 superclasses are applied.
1188 ** REMOVE-METHOD returns its generic function argument even when
1189 no method was removed.
1190 ** SHARED-INITIALIZE now initializes the values of the requested
1191 slots, including those with :ALLOCATION :CLASS.
1192 ** ALLOCATE-INSTANCE now works on structure classes defined via
1193 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1195 ** SLOT-EXISTS-P now works on conditions, as well as structures
1197 ** MAKE-LOAD-FORM now has the required methods on
1198 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1199 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1200 keyword, and now implements the SLOT-NAMES argument.
1201 ** methods with &OPTIONAL arguments no longer allow too many
1202 arguments to be passed in the call without error.
1203 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1204 option is consistent with the required arguments of the generic
1205 function lambda list.
1206 * bug fix: REQUIRE accepts a string designator. (Thanks to
1208 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1209 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1210 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1211 * bug fix: The compiler now checks for duplicated variables in macro
1213 * bug fix: SETQ on globals returns the correct value.
1214 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1215 not a COMPILER-ERROR (followed by some other strange error on
1216 choosing the CONTINUE restart).
1217 * bug fix: make.sh and friends are now more consistent in the way that
1218 they look for GNU "make".
1220 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1221 * experimental native threads support (on x86 Linux >=2.4 only).
1222 This is not compiled in by default: you need to add :SB-THREAD to
1223 the target features. See the "Beyond ANSI" chapter of the manual
1225 * fix for longstanding nonANSIism: The old distinction between
1226 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1227 The return value from CL:FIND-CLASS is now a CLOS class, and
1228 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1229 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1230 * An interface to the MetaObject Protocol, as described in Kiczales,
1231 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1232 MIT Press, 1991, is available from the SB-MOP package.
1233 * incompatible change: the SB-PCL package should now be considered
1234 a private implementation detail, and no longer a semi-private MOP
1236 * minor incompatible change: due to rearrangement for threads, the
1237 control stack and binding stack are now allocated at arbitrary
1238 addresses instead of being hardcoded per-port. Users affected by
1239 this probably have to be doing advanced things with shared
1240 libraries, and will know who they are.
1241 * minor incompatible change: Previously, all --eval forms used were
1242 processed with READ before any of them were processed with EVAL.
1243 Now each --eval form is processed with both READ and EVAL before
1244 the next --eval form is processed. (Thus package operations like
1245 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1246 work as the user might reasonably expect.)
1247 * minor incompatible change: *STANDARD-INPUT* is now only an
1248 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1250 * minor incompatible change: Y-OR-N-P is now character-oriented,
1251 not line oriented. Also, YES-OR-NO-P now works without errors.
1252 (thanks to Antonio Martinez)
1253 * sb-aclrepl module improvements: an integrated inspector, added
1254 repl features, and a bug fix to :trace command.
1255 * Known functions, which cannot be open coded by the backend, are
1256 considered to be able to check types of their arguments. (fixing
1257 a bug report by Nathan J. Froyd)
1258 * fixed a bug in computing method discriminating functions: It is
1259 now possible to define methods specialized on classes which have
1260 forward-referenced superclasses. (thanks to Gerd Moellmann)
1261 * fixed evaluation order in optional entries (reported by Gilbert
1263 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1264 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1265 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1266 required-by-AMOP NAME argument, as well as CLASS and
1267 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1268 * fixed bug 20: DEFMETHOD can define methods using names that are
1269 not the proper names of classes to designate class specializers.
1270 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1271 Unix file descriptors, instead of blowing up. (thanks to Antonio
1273 * Garbage collection refactoring: user-visible change is that a
1274 call to the GC function during WITHOUT-GCING will not do garbage
1275 collection until the end of the WITHOUT-GCING. If you were doing
1276 this you were probably losing anyway.
1277 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1278 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1279 TYPEP the latter but not the former.
1280 * The compiler issues a full WARNING for calls to undefined functions
1281 with names from the CL package.
1282 * MAP-INTO for a vector destination is open coded. (reported by
1283 Brian Downing on c.l.l)
1284 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1285 documentation string.
1286 * fixed some bugs revealed by Paul Dietz' test suite:
1287 ** COPY-ALIST now signals an error if its argument is a dotted
1289 ** Condition slots are now accessed more correctly in the presence
1290 of multiple initargs for a given slot.
1291 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1292 exclude restarts of the same name associated with a different
1294 ** DEFCLASS of forward-referenced classes with another
1295 forward-referenced class in the superclasses list no longer
1297 ** Condition slots are now initialized once each, not multiple
1298 times. (thanks to Gerd Moellmann)
1299 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1301 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1302 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1303 arguments contain duplicated elements.
1304 ** RESTART-CASE understands local macros.
1305 ** RESTART-CASE associates exactly its own restarts with a condition.
1306 ** ENDP in safe mode checks its argument to be of type LIST.
1307 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1308 in question is unbound.
1309 ** Optimized MAKE-INSTANCE functions no longer cause internal
1310 assertion failures in the presence of duplicate initargs.
1311 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1312 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1313 propagated even in the ctor optimized implementation of
1315 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1317 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1318 * a better implementation of SXHASH on (simple) bit vectors,
1319 measured both in execution speed and in distribution of results
1320 over the positive fixnums, has been installed. Likewise, a better
1321 implementation of EQUAL for simple bit vectors is now available.
1322 * fixed CEILING optimization for a divisor of form 2^k.
1323 * fixed bug 240 (emitting extra style warnings "using the lexical
1324 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1325 by Antonio Martinez)
1326 * fixed SXHASH, giving different results for NIL depending on type
1327 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1328 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1329 variable. (found by Rolf Wester)
1330 * SBCL does not ignore type declarations for special
1331 variables. (reported by rif on c.l.l 2003-03-05)
1332 * some bug fixes in contrib/sb-aclrepl/
1333 * fixed some bugs revealed by Paul Dietz' test suite:
1334 ** a bug in the CONS type specifier, whereby the CAR and CDR
1335 types got intertwined, has been fixed;
1336 ** the type system is now able to reason about the interaction
1337 between INTEGER and RATIO types more completely;
1338 ** APPEND, [N]REVERSE and NRECONC check that those their
1339 arguments, which must be proper lists, are really so;
1340 ** An array specialized to be unable to hold elements has been
1341 implemented, as required -- yes, really -- by ANSI;
1342 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1343 on malformed property lists;
1345 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1346 * incompatible packaging change: in line with Unix convention,
1347 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1348 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1349 * REQUIRE and PROVIDE are now optionally capable of doing something
1350 useful. See the documentation string for REQUIRE.
1351 * infrastructure for a managed SBCL contrib system: contributed
1352 modules in this release include:
1353 ** the ASDF system definition facility;
1354 ** an interface to the BSD Sockets API;
1355 ** an ACL-like convenience interface to the repl;
1356 (thanks to Kevin Rosenberg)
1357 ** an implementation of ROTATE-BYTE, with efficient implementation
1359 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1360 gives a closer approximation to the right answer than previously.
1361 (thanks to Raymond Toy)
1362 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1363 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1364 optional environment arguments, as required by ANSI.
1365 * fixed bugs in other functions taking environment objects, allowing
1366 calls with an explicit NIL environment argument to be compiled
1368 * fixed bug 228: primary return values from
1369 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1370 COMPILE or FUNCTION.
1371 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1372 :TYPE will no longer signal errors on innocuous objects.
1373 * fixed bug 231b: SETQ is better at respecting type declarations in
1374 the lexical environment.
1375 * fixed a bug in DEFCLASS: classes named by symbols with no or
1376 unprintable packages can now be defined.
1377 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1378 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1379 * fixed compiler failure related to checking types of functions.
1380 (reported by Robert E. Brown)
1381 * the compiler is now much more consistent in its error-checking
1382 treatment of bounding index arguments to sequence functions: in
1383 (SAFETY 3) code, errors will be signalled in almost all cases if
1384 invalid sequence bounding indices are passed to functions defined
1385 by ANSI to operate on sequences.
1386 * fixed a bug in the build procedure: documentation of SBCL-specific
1387 packages is now preserved and available in the final Lisp image.
1388 * lifted FDEFINITION lookup out of loops in the implementation of
1389 many list operations. (thanks to Robert E. Brown)
1390 * fixed a bug in the reader: the #n# reader macro now works for
1391 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1392 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1393 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1394 E. Brown and Tony Martinez respectively)
1395 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1396 SLOT-BOUNDP: in particular, it is now able to optimize them much
1397 better, and is now not vulnerable to having packages renamed.
1398 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1399 always called when a slot is not present in an instance. (thanks
1401 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1402 longer signal an error on unreading a character following EOF on
1403 the previous constituent stream. (thanks to Tony Martinez)
1404 * fixed some bugs revealed by Paul Dietz' test suite:
1405 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1406 not just nonnegative fixnums;
1407 ** the logical bit-array operators such as BIT-AND now accept an
1408 explicit NIL for their "opt-arg" argument (to indicate a
1409 freshly-consed result bit-array);
1410 ** ELT now signals an error on an invalid sequence index in safe
1412 ** the type system is now cleverer about negations of numeric
1413 types, and consequently understands the BIGNUM and RATIO types
1415 ** the type system is now cleverer about the interaction between
1416 INTEGER and RATIO types: while bugs still remain, many more
1417 cases are accurately computed;
1418 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1419 if it is in the last clause;
1420 ** CONSTANTLY now correctly returns a side-effect-free function in
1422 ** DECLARE is no longer treated as a special-operator; in
1423 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1424 * incremented fasl file version number due to the change in the
1425 DEFSTRUCT-SLOT-DESCRIPTION structure.
1427 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1428 * minor incompatible change: code processed by the "interpreter" or
1429 EVAL now has a compilation optimization policy of (DEBUG 2)
1430 (changed from (DEBUG 1)) to improve debuggability of interactive
1431 development, and to allow the use of the debug RETURN command in
1433 * an experimental implementation of the RETURN command for the
1434 debugger has been included. (thanks to Frederik Kuivinen)
1435 * fixed bug 62: constraints were not propagated into a loop.
1436 * fixed bug in embedded calls of SORT (reported and investigated by
1438 * fixed some bugs revealed by Paul Dietz' test suite:
1439 ** printing and reading of arrays with some dimensions having
1440 length 0 (thanks to Gerd Moellmann);
1441 ** BOA constructor with &AUX argument without a default value does
1442 not cause a type error;
1443 ** CONSTANTP now returns true for all self-evaluating objects.
1445 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1446 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1447 accessors that are related by inheritance, as specified in the
1448 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1449 Valtteri Vuorikoski)
1450 * The compiler is now able to inline functions that were defined in
1451 a complex lexical environment (e.g. inside a MACROLET).
1452 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1454 * Support for the upcoming FreeBSD-5.0 release has been included.
1455 (thanks to Dag-Erling Smorgrav)
1456 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1457 effect when it is not in a toplevel context.
1458 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1459 stand a better chance of being correct. (thanks to Gerd
1461 * fixed bug in COERCE, which now signals an error on coercing a
1462 rational to a bounded real type which excludes the expected
1464 * The compiler is now able to derive types more accurately from the
1465 COERCE and COMPILE functions.
1466 * fixed bug 223: functional binding is considered to be constant
1467 only for symbols in the CL package.
1468 * fixed bug 231: SETQ did not check the type of a variable being set
1469 (reported by Robert E. Brown)
1470 * A new optimization for MAKE-INSTANCE has been included, fixing
1471 various bugs (including relating to :ALLOCATION :CLASS slots and
1472 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1474 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1475 ** As required by ANSI, LOOP now disallows anonymous collection
1476 clauses such as COLLECT I in conjunction with aggregate boolean
1477 clauses such as THEREIS (= I 1);
1478 ** LOOP now signals an error when any variable is reused in the
1479 same loop (including the potentially useful construct analogous
1480 to WITH A = 1 WITH A = (1+ A);
1481 ** IT is only a special loop symbol within the first clause of a
1482 conditional loop clause;
1483 ** LOOP with a typed iteration variable over a hashtable now
1484 signals a type error iff it should.
1485 * fixed some other bugs revealed by Paul Dietz' test suite:
1486 ** FILE-STREAM now names the class previously known as FD-STREAM;
1487 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1488 argument) no longer signals an error;
1489 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1490 of the slot symbol, rather than using the current package
1491 ((:CONC-NAME "") continues to intern the slot's name in the
1493 * incremented fasl file version number, because of the incompatible
1494 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1495 of the new implementation of DEFINE-COMPILER-MACRO.
1497 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1498 * Support for building SBCL for MIPS platforms running in
1499 little-endian mode has now been checked in, and basic
1500 functionality on said platforms verified.
1501 * minor incompatible change: PCL now records the pathname of a file
1502 in which methods and the like are defined, rather than its
1504 * minor incompatible change: TRUENAME now considers the truename of
1505 a file naming a directory to be the pathname with :DIRECTORY
1506 component indicating that directory.
1507 * minor incompatible change: a NAMED clause in the extended form of
1508 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1509 reason for the previous behaviour is unclear.
1510 * more systematization and improvement of CLOS and MOP conformance
1511 in PCL (thanks to Gerd Moellman and Pierre Mai):
1512 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1514 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1515 primary methods with no specializers;
1516 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1518 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1519 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1520 FORWARD-REFERENCED-CLASSes; error reporting on
1521 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1523 ** SXHASH on CLOS instances now uses a slot internal to the
1524 instance to return different numbers on distinct instances,
1525 while preserving the same return value through invocations of
1527 ** DEFMETHOD signals errors when methods with longer incongruent
1528 lambda lists are added to generic functions;
1529 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1530 CLASS, as specified in AMOP;
1531 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1532 on the order returned by the primary method for classes of
1533 class STANDARD-CLASS;
1534 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1535 * fixed some bugs shown by Paul Dietz' test suite:
1536 ** DOLIST puts its body in TAGBODY;
1537 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1539 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1540 value producing form;
1541 ** if more variables are given to PROGV than values, extra
1542 variables are bound and made to have no value;
1543 ** NSUBSTITUTE on list arguments gets the right answer with
1545 ** ELT signals an error of type TYPE-ERROR when the index argument
1546 is not a valid sequence index;
1547 ** LOOP signals (at macroexpansion time) an error of type
1548 PROGRAM-ERROR when duplicate variable names are found;
1549 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1550 ** FUNCALL of special-operators now cause an error of type
1552 ** PSETQ now works as required in the presence of side-effecting
1553 symbol-macro places;
1554 ** NCONC accepts any object as its last argument;
1555 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1557 ** loop-for-as-package does not require a package to be explicitely
1559 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1561 * fixed bug 166: compiler preserves "there is a way to go"
1562 invariant when deleting code.
1563 * fixed bug 172: macro lambda lists with required arguments after
1564 &REST arguments now cause an error to be signalled. (thanks to
1566 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1567 bug. (thanks to Matthew Danish)
1568 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1570 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1571 arguments are lexically rebound. (thanks to Gerd Moellmann and
1573 * fixed bug 194: error messages are now more informative when there
1574 is no primary method applicable in a call to a generic
1575 function. (thanks to Gerd Moellmann)
1576 * fixed bug in command line argument checking (thanks to Julian
1578 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1579 (thanks to Matthew Danish)
1580 * incremented fasl file version number, because of the
1581 SXHASH-related changes in the layout of CLOS data structures
1583 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1584 * minor incompatible change: The runtime (the Unix executable named
1585 "sbcl") is now much pickier about the .core files it will load.
1586 Essentially it now requires .core files to descend from the same
1587 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1588 as the runtime does. (The intent is to prevent the crashes which
1589 can occur, and which can even be reported as mysterious failures,
1590 when people patch the sources or change the build parameters
1591 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1592 sbcl and .core files.)
1593 * fixed bug: VALUES-LIST is no longer optimized away.
1594 * fixed bug 142: The FFI conversion of C string values to Lisp
1595 string values no longer conses excessively. (thanks to Nathan
1596 Froyd porting Raymond Toy's fix to CMU CL)
1597 * began to systematize and improve MOP conformance in PCL (thanks to
1598 Nathan Froyd, Gerd Moellman and Pierre Mai):
1599 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1601 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1602 ** FINALIZE-INHERITANCE is now called on class finalization;
1603 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1604 argument precedence order.
1605 * fixed bug 202: The compiler no longer fails on functions whose
1606 derived types contradict their declared type.
1607 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1608 so it can be non-toplevel.
1609 * The fasl file version number has changed (because of the new
1610 implementation of DEFMACRO).
1611 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1612 safe code, that any length requirement by their type-specifier
1613 argument is valid. The exceptions to this are described in bug
1615 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1616 functions now check the types of their inputs as required by ANSI.
1617 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1618 introduced symbol is DECLAREd to be SPECIAL.
1619 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1620 * fixed bug: :COUNT argument to sequence functions may be negative
1621 * fixed bug: body of DO-SYMBOLS may contain declarations
1622 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1623 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1624 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1625 symbol macro only once
1626 * fixed printing of call frame when argument list is unavailable
1627 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1628 * compiler no longer signals WARNING on unknown keyword
1631 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1632 * A beta-quality port to the mips architecture running Linux,
1633 based on the old CMUCL backend, has been made. It has been tested
1634 on a big-endian kernel, and works sufficiently well to be able to
1635 rebuild itself; it has not been tested in little-endian mode.
1636 * fixed an inconsistency between gencgc.c and purify.c which made
1637 dumping/loading .core files unreliable
1638 * fixed bug 120a: The compiler now deals correctly with IFs where
1639 the consequent is the same as the alternative, instead of
1640 misderiving the return type. (thanks to Alexey Dejneka)
1641 * fixed bug 113: Logical pathnames are now dumpable (the logical
1642 host is resolved at load-time, throwing an error if it is not
1644 * fixed bug 174: FORMAT's error message is slightly clearer when a
1645 non-printing character is used in a format directive.
1646 * fixed several bugs in compiler checking of type declarations, i.e.
1647 violations of the Python "declarations are assertions" principle
1648 (thanks to Alexey Dejneka)
1649 * fixed several bugs in PCL's error checking (thanks to Gerd
1651 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1653 * fixed bug in compilation of functions as first class values
1654 (thanks to Antonio Martinez-Shotton)
1655 * The compiler's handling TYPE-ERRORs which it can prove will
1656 inevitably happen at runtime has been cleaned up and corrected
1657 in several ways. (thanks to Alexey Dejneka)
1658 * improved argument type checking for various basic arithmetic
1659 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1660 lavished on them in the past that they can be compiled in many
1661 ways in different special cases
1662 * fixed bug 181: compiler checks validity of user supplied type
1664 * cleaned up code flushing in optimization: Function calls which
1665 should signal errors for safety purposes (e.g. which ANSI says
1666 should signal errors when their arguments are of incorrect type)
1667 are no longer optimized away.
1668 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1669 * incremented fasl file version number, because changes in the
1670 implementation of sequence functions like COERCE caused
1671 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1672 in old inline expansions) to become undefined. (Actually these
1673 changes were later undone, so we might very well be binary
1674 compatible with 0.7.7 after all, but leaving the version number
1675 incremented seemed like the simplest and most conservative
1678 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1679 * An alpha-quality port to the parisc architecture running Linux,
1680 based on the old CMUCL backend, has been made. This, even more so
1681 than the other backends, should be considered still a work in
1682 progress; known problems include that the Linux kernel in 64-bit
1683 mode does not propagate the correct sigcontext structure to
1684 userspace, and consequently SBCL on a parisc64 kernel will not
1686 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1687 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1688 suppressed.) Also now that the compiler is looking at declarations
1689 in the environment, it checks optimization declarations as well,
1690 and suppresses inlining when (> DEBUG SPEED).
1691 * More fixes have been made to treatment of floating point exception
1692 treatment and other Unix signals. In particular, floating point
1693 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1694 * The detection and handling of control stack exhaustion (infinite
1695 or very deeply nested recursion) has changed. Stack exhaustion
1696 detection is now done by write-protecting pages at the OS level
1697 and applies at all optimization settings; when found, a
1698 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1699 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1700 be caught using IGNORE-ERRORS.
1701 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1702 correctly when transforming recursive LABELS functions to LETs.
1703 (thanks to Alexey Dejneka)
1704 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1705 that are names of constants or global variables.
1706 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1707 alien routines with docstrings.
1708 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1709 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1711 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1712 (thanks to Eric Marsden)
1713 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1714 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1715 * Bug fix: The SPARC backend can now compile functions involving
1716 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1717 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1718 to a functions expecting a PATHNAME-DESIGNATOR.
1719 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1720 lambda lists. (thanks to Alexey Dejneka)
1721 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1722 functions so that EXPORTing the name of the class doesn't cause
1723 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1724 to Antonio Martinez for reporting this)
1725 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1726 constant-folded without failing an assertion. (thanks to Einar
1727 Floystad Dorum for reporting this)
1728 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1729 types are now dealt with more correctly by the compiler.
1730 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1731 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1732 behaviour analogous to RENAME-FILE. This puts its behaviour more
1733 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1735 * The fasl file version number has changed again. (because of the
1736 bug fix involving the names of PCL MAKE-INSTANCE functions)
1738 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1739 * bug fix: Floating point exceptions are treated much more
1740 consistently on the x86/Linux and PPC/Linux platforms.
1741 * Array initialization with :INITIAL-ELEMENT is now much faster for
1742 cases when the compiler cannot open code the array creation, but
1743 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1744 array accesses have also seen a speed increase.
1745 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1746 specified. (This was at the root of some bad interactions between
1747 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1749 * bug fix: Internal error arguments for undefined functions are now
1750 computed correctly on the PPC/Linux platform.
1751 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1752 Raymond Toy's patch for CMU CL)
1753 * Support for the Solaris 9 operating environment has been included
1754 (thanks to Daniel Merritt)
1755 * A very ugly but hopefully complete draft of the missing FFI chapter
1756 of the manual has been created by reformatting the corresponding
1757 CMU CL manual chapter into (currently very ugly and incoherent)
1758 DocBook and bringing it up to date for SBCL behavior. Thus, the
1759 manual is now essentially complete, at least by my extreme
1760 once-and-only-once standards, whereby it's acceptable to refer to
1761 the doc strings of SB-EXT functions as the primary documentation.
1762 * The fasl file version number has changed again, due to cleanup of
1763 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1764 is no longer a static symbol.)
1766 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1767 * SBCL now builds with OpenMCL (version 0.12) as the
1768 cross-compilation host; also, more progress has been made toward
1769 bootstrapping under CLISP.
1770 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1772 * bug 158 fixed: The compiler can now deal with integer loop
1773 increments different from 1; fixing this turned out also to fix
1775 * bug 169 fixed: no more bogus warnings about using lexical bindings
1776 despite the presence of perfectly good SPECIAL declarations (thanks
1777 to David Lichteblau)
1778 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1779 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1780 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1782 * bug 180 fixed: Method combination specifications no longer ignore
1783 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1784 * bug fix: Structure type predicate functions now check their argument
1785 count as they should.
1786 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1787 correctly. (thanks to Pierre Mai)
1788 * minor incompatible change: The --noprogrammer option is deprecated
1789 in favor of the new --disable-debugger option, which is very similar.
1790 (The major difference is that it takes effect at a slightly different
1791 time at startup, causing handling of errors in --sysinit and
1792 --userinit files will be affected differently.) The
1793 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1794 been added to allow this functionality to be controlled from ordinary
1795 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1796 maintainers, who might want to run non-interactive scripts to
1797 build SBCL cores which will later be used interactively.)
1798 * minor incompatible change: The LOAD function no longer, when given
1799 a wild pathname to load, loads all files matching that pathname.
1800 Instead, an error of type FILE-ERROR is signalled.
1802 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1803 * bug 147 fixed: The compiler preserves its block link/count
1804 invariants more correctly now so that it doesn't crash. (thanks
1806 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1808 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1809 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1810 no-value-for-this-character as NIL (as specified by ANSI).
1811 * HOST-NAMESTRING on physical pathnames now returns a string that is
1812 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1813 (thanks to Christophe Rhodes)
1814 * The Alpha port handles icache flushing more correctly. (thanks to
1816 * More progress has been made toward bootstrapping under CLISP. (thanks
1817 to Christophe Rhodes)
1818 * The fasl file format has changed again, because dynamic loading
1819 on OpenBSD (which has non-ELF object files) motivated some cleanups
1820 in the way that foreign symbols are transformed and passed around.
1821 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1822 is no longer treated as whitespace by the reader, but instead as
1823 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1824 |A
\7fB|, instead of A as it used to.
1826 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1827 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1828 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1829 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1830 this as of 0.7.1.45, but was left out of the previous news section
1831 (thanks to Dan Barlow)
1832 * SBCL now runs on the Solaris operating system on SPARC architectures
1833 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1834 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1835 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1836 * SPARC backend cleanups, allowing builds of cores optimized for V8
1837 and V9 SPARCS, and also emission of code targeted to a particular
1838 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1840 * SBCL is closer to bootstrapping under CLISP, thanks to various
1841 fixes by Christophe Rhodes.
1842 * The fasl file format has changed again, to allow the compiler's
1843 INFO database to support symbol macros.
1844 * The user manual (in doc/) is formatted into HTML more nicely.
1845 (thanks to coreythomas)
1846 * The system is smarter about SUBTYPEP relationships, especially
1847 those involving NOT types (including types such as ATOM which are
1848 represented internally using NOT types). Thus SUBTYPEP is less
1849 likely to return (VALUES NIL NIL) in general, and in particular
1850 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1852 * The fasl file format has changed again, because the internal
1853 representation of types now includes a new slot to support the new
1854 SUBTYPEP-of-NOT-types logic.
1855 * (not a change in the main branch of SBCL, but a related prototype
1856 which can hopefully be merged into the main branch of SBCL in the
1857 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1858 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1859 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1860 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1861 dispatch macro character combinations. (thanks to Alexey Dejneka)
1862 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1863 that superficially look like logical namestrings correctly.
1864 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1866 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1867 default supplied-p) for &optional and &key arguments. (thanks to
1870 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1871 * incompatible change: The compiler is now less aggressive about
1872 tail call optimization, doing it only when (> SPACE DEBUG) or
1873 (> SPEED DEBUG). (This is an incompatible change because there are
1874 programs which relied on the old CMU-CL-style behavior to optimize
1875 away their unbounded recursion which will now die of stack overflow.)
1876 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1877 for non-GENCGC systems has been increased to 20M (since that
1878 seems much closer to the likely performance optimum for modern
1879 systems than the old 4M value was)
1880 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1881 and *DEBUG-PRINT-LEVEL*
1882 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1883 of CMU CL's support for SPARC, and various endianness and other
1884 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1885 * new syntactic sugar for the Unix command line: --load foo.bar is now
1886 an alternate notation for --eval '(load "foo.bar")'.
1888 ** The system now detects stack overflow and handles it gracefully,
1889 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1890 optimization settings. (This is a good thing in general, and
1891 its introduction in this version should be particularly timely
1892 for anyone whose code fails because of suppression of tail
1894 ** The system now hunts for the C variable "environ" in a more
1895 devious way, to avoid segfaults when the C library version differs
1896 between compile time and run time. (thanks to Christophe Rhodes)
1897 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1898 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1899 ** The compiler no longer issues bogus style warnings for undefined
1900 classes in the same source file as the DEFCLASSes which defined
1901 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1903 ** fixes in CONDITION class precedence list for undefined function
1904 errors (thanks to Alexei Dejneka)
1905 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1906 correctly. (thanks to Dan Barlow)
1907 ** portability fixes aiming at bootstrapping under CLISP (thanks
1908 to Dave McDonald and Christophe Rhodes)
1909 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1910 ** fixes in type translation and and type inference (thanks to
1912 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1913 ** various fixes in the new ports (thanks to Dan Barlow)
1914 * several changes related to debugging:
1915 ** suppression of tail recursion, as noted above
1916 ** stack overflow detection, as noted above
1917 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1918 is now the default. (For some time encapsulation has been more
1919 reliable than the breakpoint-based :ENCAPSULATE NIL
1920 implementation, at least on X86 systems; and I just noticed that
1921 encapsulation also seems closer to the spirit of the ANSI
1924 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1926 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1927 up properly again. (There was a packaging bug in 0.7.0 which
1928 left their definitions in SB-SYS::LOAD-FOREIGN and
1929 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1930 vital for most things which interface to C-level interfaces,
1931 like extensions working with sockets or databases or
1932 Perl-compatible regexes or whatever, and the need to fix
1933 this bug is the main reason that 0.7.1 was released so
1935 ** DEFGENERIC is now choosier about the methods it redefines, so that
1936 reLOADing a previously-LOADed file containing DEFGENERICs does
1937 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1938 cycle now works as it should. (thanks to Alexey Dejneka)
1939 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1940 fixed by Christophe Rhodes. (He actually submitted this patch
1941 months ago, and I delayed until after 0.7.0.)
1942 ** Bug 111 (internal compiler confusion about runtime checks on
1943 FUNCTION types) was fixed by Alexey Dejneka.
1944 * Some internal cleanups (getting rid of variables which aren't
1945 needed now that the byte interpreter is gone) caused the fasl
1946 file format number to change again.
1948 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1949 * major incompatible change: The default fasl file extension, i.e. the
1950 default extension for files produced by COMPILE-FILE, has changed
1951 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1953 ** There are many changes in the implementation of the compiler.
1954 SBCL is now essentially a compiler-only implementation of ANSI
1955 Common Lisp. EVAL still "interprets" a few special cases, but
1956 almost all the interesting cases are handled by creating
1957 a LAMBDA expression, calling COMPILE on it, then calling
1958 FUNCALL on the result.
1959 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1960 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1961 Since the code is newer, there might still be some new bugs
1962 (though not as many as before Martin Atzmueller's fixes:-). But
1963 the new code is substantially simpler and clearer, and hopefully
1964 any remaining bugs will be simpler, less fundamental, and more
1965 fixable then the bugs in the old code.
1966 ** The revised compiler is still a little unsteady on its feet.
1968 *** The debugging information it produces (particularly the names
1969 of FUNCTION objects) is sometimes much less useful than what
1970 the old compiler produced.
1971 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1972 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1973 within a MACROLET) has been temporarily weakened.
1974 ** There are new compiler optimizations for various functions:
1975 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1976 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1977 *** the math functions TRUNCATE, FLOOR, and CEILING
1978 *** the function-of-all-trades COERCE
1979 Mostly these should be transparent, but there's one
1980 potentially-annoying problem (bug 117): when the compiler
1981 inline-expands a function and does type analysis on the result,
1982 it can create control paths which have type mismatches, and
1983 when it can't prove that those control paths aren't taken,
1984 it will issue WARNINGs about the type mismatches. This is
1985 a particular problem in practice for the new sequence functions.
1986 It's not clear how this should be fixed, and for now, a
1987 workaround is given in the entry for 117 in the BUGS file.
1988 ** (Because of the interaction between the two previous items --
1989 occasional inlining problems and new inline expansions -- some
1990 of the new sequence function optimizations won't really kick in
1991 completely until debugging information, and then inlining, are
1992 straightened out in some future version.)
1993 * minor incompatible changes:
1994 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1995 DIRECTORY no longer implicitly promotes NIL slots of its
1996 pathname argument to :WILD. In particular, when you ask for the
1997 contents of a directory (which you used to be able to do without
1998 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1999 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2000 ** changes in behavior that ANSI explicitly defines to be
2001 implementation dependent:
2002 *** The new compiler-only implementation still conforms with ANSI,
2003 but acts a little different than before. Besides the obvious
2004 changes in performance tradeoffs (that the cost per form passed
2005 to EVAL has gone up, and the cost per form executed by EVAL
2006 has gone down), the behavior of the system changes a little
2007 because there are no longer any interpreted function objects.
2008 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2009 e.g. doing COMPILE on the output of interactive DEFUN is
2011 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2013 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2014 has changed from (:CL) to NIL.
2015 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2016 e.g. #\Newline and #\Tab, don't have names specified in the
2017 ANSI Common Lisp standard, is now based on their ASCII symbolic
2018 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2019 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2020 are no longer used for output.
2021 ** changes in internal implementation constants:
2022 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2023 4 million. (If your application spends a lot of time GCing and
2024 you have a lot of RAM, you might want to experiment with
2025 increasing it even more.)
2026 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2027 However, almost all old code should still continue to work without
2028 immediate update, as SB-C-CALL is now a (deprecated) nickname
2030 ** Old operator names in the style DEF-FOO are now deprecated in
2031 favor of new corresponding names DEFINE-FOO, for consistency with
2032 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2033 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2034 internal symbols, but a few supported extensions like
2035 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2036 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2037 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2038 etc. as you get deeper into recursive calls to the debugger
2039 command loop, instead of the old "5]", "5]]", "5]]]"
2040 sequence. (I was motivated to do this when squabbles between
2041 ILISP and SBCL left me very deeply nested in the debugger. In the
2042 short term, this change will probably provoke more ILISP/SBCL
2043 squabbles, but hopefully it will be an improvement in the long run.)
2044 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2045 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2046 function, since declaiming return types involving aliens is
2047 (1) annoyingly messy to do by hand and (2) vital to efficient
2048 compilation of code which calls such functions.
2049 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2050 longer reexported by the SB-EXT package. They're solely useful
2051 for alien code, so it seems more logical that you should get
2052 them from the SB-ALIEN package, not in SB-EXT.
2053 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2054 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2055 features. Instead, the code that they used to control is always
2056 built into the system.
2057 * many other bug fixes
2058 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2059 advantage of the new EVAL-WHEN stuff and to clean them up in
2060 general, and they are now more ANSI-compliant in a number of
2061 ways. Martin Atzmueller is responsible for a lot of this.
2062 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2064 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2065 *** correct ERROR type for various file operations
2066 *** some fixes for Lisp streams
2067 *** DEFMETHOD syntax checking
2068 *** changing old weird representation of debug information as
2069 strings (which, among their other deficiencies, don't transform
2070 correctly when you rename packages, and don't change their
2071 print representation when you change things like *PACKAGE*
2072 and *PRINT-LENGTH*) to symbols and lists of symbols
2073 He also made several improvements and fixed several bugs in DESCRIBE.
2074 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2076 *** misbehavior of WRITE-STRING/WRITE-LINE
2077 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2078 and several other LOOP problems as well
2079 *** DIRECTORY when similar filenames are present
2080 *** DEFGENERIC with :METHOD options
2081 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2082 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2083 *** argument ordering in FIND with :TEST option
2084 *** mishandled package designator argument in APROPOS-LIST
2085 *** various problems in the backquote readmacro
2086 *** a bug in APROPOS
2087 *** probably some others that I'm not describing very well here,
2088 since the CVS log documents them by reference to sbcl-devel
2089 messages, and the SourceForge archives aren't working well.:-(
2090 ** Dan Barlow improved the Alpha port (and is making progress on the
2091 PPC port, for those of you who think different).
2092 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2093 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2094 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2095 continues to work on the SPARC port (for those of you in a position
2096 to look down upon our little PC-compatible boxes from a great height).
2097 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2098 and uses that copy, rather than the current dynamic value, when
2099 it's trying to decide whether to truncate output. Thus e.g.
2100 (let ((*print-lines* 50))
2101 (pprint-logical-block (stream nil)
2103 (let ((*print-lines* 8))
2104 (print (aref possiblybigthings i) stream)))))
2105 should now truncate the logical block only at 50 lines, instead of
2106 often truncating it at 8 lines, as it did before.
2107 * The doc/cmucl/ directory, containing old CMU CL documentation
2108 from the time of the fork, is no longer part of the base system.
2109 SourceForge has shut down its anonymous FTP service, and with it
2110 my original plan for distributing the old CMU CL documentation
2111 there. For now, if you need these files you can download an old
2112 SBCL source release and extract them from it.
2113 * The fasl file version number changed again, for dozens of reasons,
2114 some of which are apparent above.
2116 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2117 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2118 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2119 patch, so that DISASSEMBLE output is much nicer.
2120 * The code in the SB-PROFILE package now seems reasonably stable.
2121 I still haven't decided what the final interface should look like
2122 (I'd like PROFILE to interact cleanly with TRACE, since both
2123 facilities use function encapsulation) but if you have a need
2124 for profiling now, you can probably use it successfully with
2125 the current CMU-CL-style interface.
2126 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2127 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2128 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2129 link now returns the dangling link itself, and for similar
2130 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2131 link itself. (In these cases the old code signalled an error and
2132 looped endlessly, respectively.) Thus, DIRECTORY now works even
2133 in the presence of dangling and cyclic symbolic links.
2134 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2135 is now a supported extension again, since the consensus on
2136 sbcl-devel was that it can be useful for ordinary development
2137 work, not just for debugging SBCL itself.
2138 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2139 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2140 that currently-defined functions might be redefined later with
2141 different return types.
2142 * Hash tables can be printed readably, as inspired by CMU CL code
2143 of Eric Marsden and SBCL code of Martin Atzmueller.
2144 * better error handling in CLOS method combination, thanks to
2145 Martin Atzmueller porting Pierre Mai's CMU CL patches
2146 * more overflow fixes for >16Mbyte I/O buffers
2147 * A bug in READ has been fixed, so that now a single Ctrl-D
2148 character suffices to cause end-of-file on character streams.
2149 In particular, now you only need one Ctrl-D at the command
2150 line (not two) to exit SBCL.
2151 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2153 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2154 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2155 does the right thing.
2156 * The implementation of some type tests, especially for CONDITION
2157 types, is now tidier and maybe faster, due to CMU CL code
2158 originally by Douglas Crosher, ported by Martin Atzmueller.
2159 * Some math functions have been fixed, and there are new
2160 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2161 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2162 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2163 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2164 likely to be built into sbcl-0.7.0.)
2165 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2166 CL, ported to SBCL by Martin Atzmueller)
2167 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2168 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2169 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2170 * There's a new slam.sh hack to shorten the edit/compile/debug
2171 cycle for low-level changes to SBCL itself, and a new
2172 :SB-AFTER-XC-CORE target feature to control the generation of
2173 the after-xc.core file needed by slam.sh.
2174 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2175 COMPILE-FILE is no longer supported, so that now every function
2176 gets an entry point, so that block compilation looks a little
2177 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2178 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2179 returns the number of bytes consed since the system started,
2180 rather than the number consed since the first time the function
2181 was called. (The new definition parallels ANSI functions like
2182 CL:GET-INTERNAL-RUN-TIME.)
2183 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2184 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2185 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2186 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2187 * Fasl file version numbers are now independent of the target CPU,
2188 since historically most system changes which required version
2189 number changes have affected all CPUs equally. Similarly,
2190 the byte fasl file version is now equal to the ordinary
2193 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2194 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2195 is no longer recognized. I apologize for this, because it was
2196 listed in SB-EXT as a supported extension, but I found that
2197 its existing behavior was poorly specified, as well as incorrectly
2198 specified, and it looked like too much of a mess to straighten it
2199 out. I have enough on my hands trying to get ANSI stuff to work..
2200 * many patches ported from CMU CL by Martin Atzmueller, with
2201 half a dozen bug fixes in pretty-printing and the debugger, and
2202 half a dozen others elsewhere
2203 * fixed bug 13: Floating point infinities are now supported again.
2204 They might still be a little bit flaky, but thanks to bug reports
2205 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2206 as flaky as they were.
2207 * The --noprogrammer command line option is now supported. (Its
2208 behavior is slightly different in detail from what the old man
2209 page claimed it would do, but it's still appropriate under the
2210 same circumstances that the man page talks about.)
2211 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2212 are now supported, and enabled by default. Thus, the compiler can
2213 handle many floating point and complex operations much less
2214 inefficiently. (Thus e.g. you can implement a complex FFT
2216 * The compiler now detects type mismatches between DECLAIM FTYPE
2217 and DEFUN better, and implements CHECK-TYPE more correctly, and
2218 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2219 thanks to patches from Martin Atzmueller.
2220 * various fixes to make the cross-compiler more portable to
2221 ANSI-conforming-but-different cross-compilation hosts (notably
2222 Lispworks for Windows, following bug reports from Arthur Lemmens)
2223 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2224 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2225 fixed thanks to Pierre Mai's CMU CL patches.
2226 * fixes to keep the system from overflowing internal counters when
2227 it tries to use i/o buffers larger than 16M bytes
2228 * fixed bug 45a: Various internal functions required to support
2229 complex special functions have been merged from CMU CL sources.
2230 (When I was first setting up SBCL, I misunderstood a compile-time
2231 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2232 * improved support for type intersection and union, fixing bug 12
2233 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2234 more obscure bugs as well
2235 * some steps toward byte-compiling non-performance-critical
2236 parts of the system, courtesy of patches from Martin Atzmueller
2237 * Christophe Rhodes has made some debian packages of sbcl at
2238 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2239 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2240 stable, but are nonetheless usable. When he's ready, I'd be happy
2241 to add them to the SourceForge "File Releases" section. (And if
2242 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2243 * new fasl file format version number (because of changes in
2244 internal representation of (OR ..) types to accommodate the new
2245 support for (AND ..) types, among other things)
2247 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2248 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2250 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2252 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2253 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2254 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2255 * DESCRIBE now works on structure objects again.
2256 * Most function call argument type mismatches are now handled as
2257 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2258 know whether the function will be redefined before the call is
2259 executed. (The compiler could flag local calls with full WARNINGs,
2260 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2261 it doesn't keep track of enough information to know whether calls
2262 are local in this sense.)
2263 * Compiler output is now more verbose, with messages truncated
2264 later than before. (There should be some supported way for users
2265 to override the default verbosity, but I haven't decided how to
2266 provide it yet, so this behavior is still controlled by the internal
2267 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2268 src/compiler/ir1util.lisp.)
2269 * Fasl file format version numbers have increased again, because
2270 support for the Gray streams extension changes the layout of the
2271 system's STREAM objects.
2272 * The Gray subclassable streams extension now works, thanks to a
2273 patch from Martin Atzmueller.
2274 * The full LOAD-FOREIGN extension (not just the primitive
2275 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2276 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2277 but like most other programs, it defaults to copying the Unix
2278 environment from the original process instead of starting the
2279 new process in an empty environment.
2280 * Extensions which manipulate the Unix environment now support
2281 an :ENVIRONMENT keyword option which doesn't smash case or
2282 do other bad things. The CMU-CL-style :ENV option is retained
2283 for porting convenience.
2284 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2285 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2287 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2289 * A patch from Martin Atzmueller seems to have solved the SIGINT
2290 problem, and as far as we know, signal-handling now works cleanly.
2291 (If you find any new bugs, please report them!)
2292 * The system no longer defaults Lisp source file names to types
2293 ".l", ".cl", or ".lsp", but only to ".lisp".
2294 * The compiler no longer uses special default file extensions for
2295 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2296 seems to expect a single default extension for all compiled code,
2297 and there's no compelling reason to try to stretch the standard
2298 to allow two different extensions.) Instead, byte-compiled files
2299 default to the same extension as native-compiled files.
2300 * Fasl file format version numbers have increased again, because
2301 a rearrangement of internal implementation packages made some
2302 dumped symbols in old fasl files unreadable in new cores.
2303 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2304 many fewer weird special cases.
2305 * Bug #17 (differing COMPILE-FILE behavior between logical and
2306 physical pathnames) has been fixed, and some related misbehavior too,
2307 thanks to a patch from Martin Atzmueller.
2308 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2309 by Tim Moore, ported to SBCL by Martin Atzmueller.
2310 * Martin Atzmueller fixed several filesystem-related problems,
2311 including bug #36, in part by porting CMU CL patches, which were
2312 written in part by Paul Werkowski.
2313 * More compiler warnings in src/runtime/ are gone, thanks to
2314 more patches from Martin Atzmueller.
2315 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2318 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2320 * DESCRIBE now works on CONDITION objects.
2321 * The debugger now handles errors which arise when trying to print
2322 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2324 * The build system now uses an additional file, customize-target-features.lisp,
2325 to allow local modifications to the target *FEATURES* list. (The point of
2326 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2327 debugging features enabled, without having to worry about propagating it
2328 into everyone's system when I do a "cvs update".) When no
2329 customize-target-features.lisp file exists, the target *FEATURES* list
2330 should be constructed the same way as before.
2331 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2332 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2333 warn about behavior which is undefined under the ANSI spec. Note: This
2334 is specified by ANSI, but it's not very popular with programmers.
2335 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2336 macro in the SBCL sources for an example of a workaround which you
2337 might use to make portable ANSI-standard code which does what you want.
2338 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2339 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2340 when it's used as a non-toplevel form. (This is required in order
2341 to implement the DEFCONSTANT-EQX macro.)
2342 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2343 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2344 funny things, due to bugs in the implementation of EVAL-WHEN
2345 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2346 non-toplevel DEFCONSTANTs are uncommon.)
2347 * The core file version number and fasl file version number have been
2348 incremented, because the old noncompliant DEFCONSTANT behavior involved
2349 calling functions which no longer exist, and because I also took the
2350 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2351 * fixed bug 1 (error handling before read-eval-print loop starts), and
2352 redid debugger restarts and related debugger commands somewhat while
2354 ** The QUIT debugger command is gone, since it did something
2355 rather different than the SB-EXT:QUIT command, and since it never
2356 worked properly outside the main toplevel read/eval/print loop.
2357 Invoking the new TOPLEVEL restart provides the same functionality.
2358 ** The GO debugger command is also gone, since you can just invoke
2359 the CONTINUE restart directly instead.
2360 ** The TOP debugger command is also gone, since it's redundant with the
2361 FRAME 0 command, and since it interfered with abbreviations for the
2363 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2365 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2366 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2367 some porting work by Martin Atzmueller)
2368 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2369 with a boatload of patches which he ported to SBCL. Now that those
2371 ** The system tries to make sure that its low-priority messages
2372 are prefixed by semicolons, to help people who like to use
2373 syntax highlighting in their ILISP buffer. (This patch
2374 was originally due to Raymond Toy.)
2375 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2376 patches originally written by Raymond Toy.
2377 ** The compiler understands coercion between single-value and
2378 multiple-VALUES type expressions better, getting rid of some very
2379 weird behavior, thanks to patches originally by Robert MacLachlan
2380 and Douglas Crosher.
2381 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2382 lambda lists, thanks to a patch originally by Pierre Mai.
2383 ** The system no longer bogusly warns about "abbreviated type
2385 ** The compiler gets less confused by inlining and RETURN-FROM,
2386 thanks to some patches originally by Tim Moore.
2387 ** The system no longer hangs when dumping circular lists to fasl
2388 files, thanks to a patch originally from Douglas Crosher.
2389 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2390 undefined function error.
2391 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2392 in structure slot initforms) for now, documented workaround instead:-|
2393 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2394 * fixed bug 5: added stubs for various Gray stream functions called
2395 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2396 installed, at least appropriate type errors are generated
2397 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2398 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2399 consistently in DEFMETHOD forms.
2400 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2401 it doesn't seem to affect SBCL after all
2402 * The C runtime system now builds with better optimization and many
2403 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2405 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2407 * The system is now under CVS at SourceForge (instead of the
2408 CVS repository on my home machine).
2409 * The new signal handling code has been tweaked to treat register
2410 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2411 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2412 downstream, e.g. in debug-int.lisp, has implicit dependencies
2413 on the unsignedness of integer representation of machine words,
2414 and that caused the system to bomb out with infinite regress
2415 when trying to recover from type errors involving signed values,
2416 e.g. (BUTLAST '(1 2 3) -1).)
2417 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2418 (This was one of the bugs Peter Van Eynde reported back in July.)
2419 * The system now uses code inspired by Colin Walters' O(N)
2420 implementation of MAP (from the cmucl-imp@cons.org mailing
2421 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2422 inline the MAP operation, and there is more than one
2423 sequence argument to the MAP call (so that it can't just
2424 do ETYPECASE once and for all based on the type of the
2425 single sequence argument). (The old non-inline implementation
2426 of the general M-argument sequence-of-length-N case required
2427 O(M*N*N) time when any of the sequence arguments were LISTs.)
2428 * The QUIT :UNIX-CODE keyword argument has been renamed to
2429 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2430 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2432 * Raymond Wiker's patches to port dynamic loading from Linux to
2433 FreeBSD have been added.
2434 * The BUGS file is now more nearly up to date, thanks in large part
2435 to Martin Atzmueller's review of it.
2436 * The debugger now flushes standard output streams before it begins
2437 its output ("debugger invoked" and so forth).
2438 * The core version number and fasl file version number have both
2439 been incremented, because of incompatible changes in the layout
2441 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2442 used to not be called for a saved Lisp image.)
2443 * Martin Atzmueller's version of a patch to fix a compiler crash,
2444 as posted on sbcl-devel 13 September 2000, has been installed.
2445 * Instead of installing Martin Atzmueller's patch for the
2446 compiler transform for SUBSEQ, I deleted the compiler transform,
2447 and transforms for some similar consing operations.
2448 * A bug in signal handling which kept TRACE from working on OpenBSD
2450 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2451 away by constant folding
2452 * The system now defines its address space constants in one place
2453 (in the Lisp sources), and propagates them automatically elsewhere
2454 (through GENESIS and the sbcl.h file). Therefore, patching the
2455 address map is less unnecessarily tedious and error-prone. The
2456 Lisp names of address space constants have also been systematized.
2457 * CVS tags like dollar-Header-dollar have been removed from
2458 the sources, because they have never saved me trouble and
2459 they've been source of trouble working with patches and other
2460 diff-related operations.
2461 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2462 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2464 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2466 * The system has been ported to OpenBSD.
2467 * The system now compiles with a simple "sh make.sh" on the systems
2468 that it's supported on. I.e., now you no longer need to tweak
2469 text in the target-features.lisp-expr and symlinks in src/runtime/
2470 by hand, the make.sh takes care of it for you.
2471 * The system is no longer so grossly inefficient when compiling code
2472 involving vectors implemented as general (not simple) vectors (VECTOR T),
2473 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2474 which dares to use the various sequence functions on non-simple
2475 vectors, takes less of a performance hit.
2476 * There is now a primitive type predicate VECTOR-T-P
2477 to test for the (VECTOR T) type, so that e.g.
2478 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2479 can now be compiled with some semblance of efficiency. (The old code
2480 turned the type declaration into a full call to %TYPEP at runtime!)
2481 * AREF on (VECTOR T) is still not fast, since it's still compiled
2482 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2483 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2484 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2485 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2486 without making full calls to SUBTYPEP at runtime.
2487 (Some analogous efficiency issues for non-simple vectors specialized to
2488 element types other than T, or for non-simple multidimensional arrays,
2489 have not been addressed. They could almost certainly be handled the
2490 same way if anyone is motivated to do so.)
2491 * The changes in array handling break binary compatibility, so
2492 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2493 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2494 instead of (VALUES T T).
2495 * By following the instructions that Dan Barlow posted to sbcl-devel
2496 on 2 July 2000, I was able to enable primitive dynamic object
2497 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2498 functionality is not implemented (since it calls ld to resolve
2499 library references automatically, requiring RUN-PROGRAM for its
2500 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2501 try to resolve library references) is now supported.
2502 * The system now flushes the standard output streams when it terminates,
2503 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2504 them at several other probably-convenient times, e.g. in each pass of
2505 the toplevel read-eval-print loop, and after evaluating a form given
2506 as an "--eval" command-line option. (These changes were motivated by a
2507 discussion of stream flushing issues on cmucl-imp in August 2000.)
2508 * The source transform for TYPEP of array types no longer assumes
2509 that an array whose element type is a not-yet-defined type
2510 is implemented as an array of T, but instead punts, so that the
2511 type will be interpreted at runtime.
2512 * There is now some support for cross-compiling in make.sh: each of
2513 the phases of make.sh has its own script. (This should be transparent
2514 to people doing ordinary, non-cross-compile builds.)
2515 * Since my laptop doesn't have hundreds of megabytes of memory like
2516 my desktop machine, I became more motivated to do some items on
2517 my to-do list in order to reduce the size of the system a little:
2518 ** Arrange for various needed-only-at-cold-init things to be
2519 uninterned after cold init. To support this, those things have
2520 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2521 symbols with such names are now uninterned after cold init).
2522 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2523 fasl files for cold load.
2524 ** Remove the old compiler structure pooling code (which used to
2525 be conditional on the target feature :SB-ALLOC) completely.
2526 ** Redo the representation of some data in cold init to be more compact.
2527 (I also looked into supporting byte compiled code at bootstrap time,
2528 which would probably reduce the size of the system a lot, but that
2529 looked too complicated, so I punted for now.)
2530 * The maximum signal nesting depth in the src/runtime/ support code has
2531 been reduced from 4096 to 256. (I don't know any reason for the very
2532 large old value. If the new smaller value turns out to break something,
2533 I'll probably just bump it back up.)
2534 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2536 * Many, many bugs reported by Peter Van Eynde have been added to
2537 the BUGS list; some have even been fixed.
2538 * While enabling dynamic object file loading, I tried to make the
2539 code easier to understand, renaming various functions and variables
2540 with less ambiguous names, and changing some function calling
2541 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2542 * While trying to figure out how to do the OpenBSD port, I tried to
2543 clean up some of the code in src/runtime/. In particular, I dropped
2544 support for non-POSIX signal handling, added various comments,
2545 tweaked the code to reduce the number of compilation warnings, and
2546 renamed some files to increase consistency.
2547 * To support the new automatic configuration functionality in make.sh,
2548 the source file target-features.lisp-expr has been replaced with the
2549 source file base-target-features.lisp-expr and the machine-generated
2550 file local-target-features.lisp-expr.
2551 * fixed a stupid quoting error in make.sh so that using CMU CL
2552 "lisp -batch" as cross-compilation host works again
2554 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2556 * DESCRIBE no longer tries to call itself recursively to describe
2557 bound/fbound values, so that it no longer fails on symbols which are
2558 bound to themselves (like keywords, T, and NIL).
2559 * DESCRIBE now works on generic functions.
2560 * The printer now prints less-screwed-up representations of closures
2561 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2562 * A private symbol is used instead of the :EMPTY keyword previously
2563 used to mark empty slots in hash tables. Thus
2564 (DEFVAR *HT* (MAKE-HASH-TABLE))
2565 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2566 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2567 now does what ANSI says that it should. (You can still get
2568 similar noncompliant behavior if bang on the hash table
2569 implementation with all the symbols you get back from
2570 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2571 This breaks binary compatibility, since tests for equality to
2572 :EMPTY are wired into things like the macroexpansion of
2573 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2575 * There's now a minimal placeholder implementation for CL:STEP,
2576 as required by ANSI.
2577 * An obscure bug in the interaction of the normal compiler, the byte
2578 compiler, inlining, and structure predicates has been patched
2579 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2580 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2582 * Missing ordinary arguments in a macro call are now detected even
2583 when the macro lambda list contains &KEY or &REST.
2584 * The debugger no longer complains about encountering the top of the
2585 stack when you type "FRAME 0" to explicitly instruct it to go to
2586 the top of the stack. And it now prints the frame you request even
2587 if it's the current frame (instead of saying "You are here.").
2588 * As specified by ANSI, the system now always prints keywords
2589 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2591 * The default initial SIZE of HASH-TABLEs is now smaller.
2592 * Type information from CLOS class dispatch is now propagated
2593 into DEFMETHOD bodies, so that e.g.
2594 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2596 is now basically equivalent to
2597 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2598 (DECLARE (TYPE SINGLE-FLOAT X))
2600 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2601 operation, without having to do run-time type dispatch.
2602 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2603 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2604 * Since I don't seem to be making big file reorganizations very often
2605 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2606 is overflowing my ability to conveniently back them up), I've finally
2607 checked the system into CVS. (The CVS repository is on my home system,
2608 not at SourceForge -- putting it on SourceForge might come later.)
2609 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2610 high-level GC-NOTIFY-FOO functions send their output. (There's
2611 still very little control of where low-level verbose GC functions
2612 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2613 less than it used to -- the GC-NOTIFY-FOO functions are now under
2614 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2615 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2616 in only one place in the source code, and propagates it automatically
2617 everywhere that it's needed. Thus e.g. when I bump the version from
2618 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2619 * The C source files now include boilerplate legalese and documentation
2620 at the head of each file (just as the Lisp source files already did).
2621 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2622 to his page will be replaced with a link to his new CLiki service.
2624 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2626 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2627 * The build process now looks for GNU make under the default name "gmake",
2628 instead of "make" as it used to. If GNU make is not available as "gmake"
2629 on your system, you can change this default behavior by setting the
2630 GNUMAKE environment variable.
2631 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2632 can build without error under CMU CL.
2634 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2636 * There is now a partial SBCL user manual (with some new text and some
2637 text cribbed from the CMU CL manual).
2638 * The beginnings of a profiler have been added (starting with the
2639 CMU CL profiler and simplifying and cleaning up). Eventually the
2640 main interface should be through the TRACE macro, but for now,
2641 it's still accessed through vaguely CMU-CL-style functions and macros
2642 exported from the package SB-PROFILE.
2643 * Some problems left over from porting CMU CL to the new
2644 cross-compilation bootstrap process have been cleaned up:
2645 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2646 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2647 part when getting rid of the compiler *BACKEND* stuff.)
2648 ** The value of *NULL-TYPE* was screwed up, because it was
2649 being initialized before the type system knew the final
2650 definition of the 'NULL type. This screwed up several key
2651 optimizations in the compiler, causing inefficiency in all sorts
2652 of places. (I found it because I wanted to understand why
2653 GET-INTERNAL-RUN-TIME was consing.)
2654 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2655 PROCLAIM FTYPE information. Unfortunately this broke binary
2656 compatibility again, since now the forms output by DEFGENERIC
2657 to refer to functions which didn't exist in 0.6.3.
2658 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2659 can use the new (as of 0.6.3) transform for SOME into MAP into
2661 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2662 values to (INTEGER 0 1000000), so that the time code will no longer
2663 occasionally get blown up by Linux returning 1000000 microseconds
2664 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2665 its output conform to the ANSI spec. (Alas, this makes its output
2666 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2667 * A full call to MAP NIL with a single sequence argument no longer conses.
2668 * fixes to problems pointed out by Martin Atzmueller:
2669 * The manual page no longer talks about multiprocessing as though
2670 it were currently supported.
2671 * The ILISP support patches have been removed from the distribution,
2672 because as of version 5.10.1, ILISP now supports SBCL without us
2673 having to maintain patches.
2674 * added a modified version of Raymond Toy's recent CMU CL patch for
2675 EQUALP comparison of HASH-TABLE
2677 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2679 * The system still can't cross-compile itself with
2680 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2681 entails), but at least it can compile more of itself that way
2682 than it used to be able to, and various buglets which were uncovered
2683 by trying to cross-compile itself that way have now been fixed.
2684 * This release breaks binary compatibility again. This time
2685 at least I've incremented the FASL file format version to 2, so that the
2686 problem can be detected reliably instead of just causing weird errors.
2687 * various new style warnings:
2688 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2689 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2690 ** using the lexical binding of a variable named in the *FOO* style
2691 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2692 as specified by ANSI.
2693 * *RANDOM-STATE* is no longer automatically initialized from
2694 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2695 default behavior of the system is to repeat its behavior every time
2696 it's run. If you'd like to change this behavior, you can always
2697 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2698 old convention there was no comparably easy way to get the system to
2699 repeat its behavior every time it was run.
2700 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2701 FTYPE declarations has been removed, in favor of their ANSI
2702 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2703 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2704 the types of their sequence arguments can be determined at compile time.
2705 This is done through a new open code expansion for MAP which eliminates
2706 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2707 argument types can be determined at compile time.
2708 * The optimizer now transforms COERCE into an identity operation when it
2709 can prove that the coerced object is already of the correct type. (This
2710 can be a win for machine generated code, including the output of other
2711 optimization transforms, such as the MAP transform above.)
2712 * Credit information has been moved from source file headers into CREDITS.
2713 * Source file headers have been made more standard.
2714 * The CASE macro now compiles without complaining even when it has
2717 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2719 * (Note that the way that the PCL macroexpansions were rewritten
2720 to accommodate the change in DEFGENERIC below breaks binary
2721 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2722 not run under sbcl-0.6.2. Once we get out of alpha releases,
2723 i.e. hit release 1.0.0, we'll probably try to maintain binary
2724 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2725 and sbcl-1.4.4. Until then, however, it might be fairly common
2726 for maintenance releases to break binary compatibility.)
2727 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2729 * The reporting of unhandled signals has been changed to print some
2730 explanatory text as well as the report form. (Previously only
2731 the report form was printed.)
2732 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2733 it defines, so that the compiler no longer issues undefined function
2734 warnings for compiled-but-not-yet-loaded generic functions.
2735 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2736 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2737 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2738 section of the ANSI spec are supported.
2739 * The "" nickname for the "KEYWORD" package has been removed.
2740 The reader still handles symbol tokens which begin with a package marker
2741 as keywords, but it doesn't expose its mechanism for doing so in the
2742 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2743 * The system now issues STYLE-WARNINGs for contradictory TYPE
2744 proclamations. (Warnings for contradictory FTYPE proclamations would
2745 be nice too, but those can't be done usefully unless the type system
2746 is made smarter about FUNCTION types.)
2747 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2748 systematized, so that "*target-*.lisp is supposed to exist only on the
2749 target and imply that there's a related file which exists on the
2750 host, and *host-*.lisp is supposed to exist only on the host and imply
2751 that there's a related file which exists on the target. This involves a
2752 lot of renaming. Hopefully the acute confusion caused by the renaming
2753 will be justified by the reduction in chronic confusion..
2754 ** runtime-type.lisp -> early-target-type.lisp
2755 ** target-type.lisp -> late-target-type.lisp
2756 ** early-host-format.lisp -> early-format.lisp
2757 ** late-host-format.lisp -> late-format.lisp
2758 ** host-error.lisp -> misc-error.lisp
2759 ** early-error.lisp -> early-target-error.lisp
2760 ** late-error.lisp -> late-target-error.lisp
2761 ** host-defboot.lisp -> early-defboot.lisp
2762 ** code/misc.lisp -> code/target-misc.lisp
2763 ** code/host-misc.lisp -> code/misc.lisp
2764 ** code/numbers.lisp -> code/target-numbers.lisp
2765 ** code/early-numbers.lisp -> numbers.lisp
2766 ** early-host-type.lisp -> early-type.lisp
2767 ** late-host-type.lisp -> late-type.lisp
2768 ** host-typep.lisp -> typep.lisp
2769 ** load.lisp -> target-load.lisp
2770 ** host-load.lisp -> load.lisp
2771 ** host-disassem.lisp -> disassem.lisp
2772 ** host-insts.lisp -> insts.lisp
2773 ** byte-comp.lisp -> target-byte-comp.lisp
2774 ** host-byte-comp.lisp -> byte-comp.lisp
2775 ** host-signal.lisp -> signal.lisp
2776 ** host-defstruct.lisp -> defstruct.lisp
2777 ** late-target-type.lisp -> deftypes-for-target.lisp
2778 Furthermore, several other previously target-only files foo.lisp (e.g.
2779 hash-table.lisp and random.lisp) have been split into a target-and-host
2780 foo.lisp file and a target-only target-foo.lisp file, with their key type
2781 definitions in the target-and-host part, so that the cross-compiler will
2782 know more about target types.
2783 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2784 gone away. In their place are various *BACKEND-FOO* variables
2785 corresponding to the slots of the old structure.
2786 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2787 into the target SBCL has been fixed.
2788 * The chill.lisp system for loading cold code into a running SBCL
2790 * Support for the CMU CL "scavenger hook" extension has been removed.
2791 (It was undocumented and unused in the CMU CL sources that SBCL was
2792 derived from, and stale in sbcl-0.6.1.)
2793 * Various errors in the cross-compiler type system were detected
2794 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2795 (enabling various consistency checks). Many of them were fixed,
2796 but some hard problems remain, so the compiler is back to
2797 running without *TYPE-SYSTEM-INITIALIZED* for now.
2798 * As part of the cross-compiler type system cleanup, I implemented
2799 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2800 * I have started adding UNCROSS calls throughout the type system
2801 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2802 on cross-compiler input files will be able to go away, and various
2804 * CONSTANTP now returns true for quoted forms (as explicitly required
2807 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2809 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2810 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2811 (MAKE-LIST -1). In the longer term, it ought to become true
2812 that declarations are assertions even at SAFETY 1. For now, it's not
2813 quite true even at SAFETY 3, but it's at least more nearly true..
2814 (Note that this change seems to increases the size of the system by
2815 O(5%) and to decrease the speed of the compiler by 20% or more.)
2816 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2817 for the problem of printing dozens of lines of distracting information
2818 about low-level system machinery as part of the top stack frame
2819 on entry to the debugger when an undefined function was called.
2820 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2821 is not reset to COMMON-LISP-USER.
2822 * Compilation of stuff related to dyncount.lisp has been made conditional
2823 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2824 smaller. The various dyncount-related symbols have been moved into
2825 a new "SB-DYNCOUNT" package.
2826 * tty-inspect.lisp has been renamed to inspect.lisp.
2827 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2828 feature has gone away. (When we eventually port to other flavors of
2829 libc and/or Unix, we'll try to make the differences between flavors
2830 invisible at the user level.)
2831 * Various other *FEATURES* tags, and/or their associated conditionals,
2832 have been removed if obsolescent, or given better documentation, or
2833 sometimes given more-mnemonic names.
2835 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2837 * tidied up "make.sh" script
2838 * tidied up system directory structure
2839 * better "clean.sh" behavior
2840 * added doc/FOR-CMUCL-DEVELOPERS
2841 * many many small tweaks to output format, e.g. removing possibly-confusing
2842 trailing #\. character in DESCRIBE-INSTANCE
2843 * (EQUALP #\A 'A) no longer signals an error.
2844 * new hashing code, including EQUALP hashing
2845 * tidied up Lisp initialization and toplevel
2846 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2847 * command line argument processing
2848 * added POSIX-GETENV function to deal with Unix-ish environment variables
2849 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2850 terminating SBCL on EOF
2851 * non-verbose GC by default
2852 * There is no more "sbcl" shell script; the sbcl file is now the C
2853 runtime executable (just like CMU CL).
2854 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2856 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2857 symbol and package information as Lisp native symbol and package objects
2858 instead of strings naming symbols and strings naming packages. This way,
2859 whenever packages are renamed (as in warm init), debug information is
2860 transformed along with everything else.
2861 * tweaked the optimization policy declarations which control the building
2862 of SBCL itself. Now, among other things, the system no longer saves
2863 source location debugging information. (This helps two problems at once
2864 by reducing SBCL size and by keeping SBCL from trying to look for its
2865 sources -- which may not exist -- when reporting errors.)
2866 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2867 debugging and testing purposes
2868 * cleaned up printing, making the printer call PRINT-OBJECT for
2869 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2870 methods, giving nearly-ANSI behavior
2871 * converted almost all special variables to use *FOO* naming convention
2872 * deleted PARSE-TIME functionality, since it can be done portably
2873 * moved some files out of cold init into warm init
2874 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2876 * regularized formatting of source files
2877 * added an install.sh script
2878 * fixed ridiculous memory usage of cross-compiler by making
2879 compiler/alloc.lisp not try to do pooling unless it can hook
2880 itself into the GC of the cross-compilation host. Now the system
2881 builds nicely on my old laptop.
2882 * added :SB-ALLOC in target-features.lisp-expr
2883 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2884 was not implemented)
2885 * re-did condition handling and note reporting in the compiler. Notes
2886 are no longer handled by signalling conditions. Style warnings
2887 and warnings are handled more correctly and reported in such a way
2888 that it's easy to find one or the other in your output (so that you
2889 can e.g. figure out which of many problems caused COMPILE-FILE to
2891 * changed the severity of several compiler warnings from full WARNING
2892 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2893 compiler note reporting so that it doesn't use the condition system
2894 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2895 COMPILE-FILE command)
2896 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2897 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2898 no longer supported, since their implementation was deeply intertwingled
2899 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2900 * removed bogus "support" for compiler macros named (SETF FOO), and
2901 removed the compiler macro for SETF INFO (but only after making a fool
2902 of myself on the cmucl-imp mailing list by posting a bogus patch for
2903 DEFINE-COMPILER-MACRO..)
2904 * Compiled files containing forms which have side effects on the Lisp
2905 reader (such as DEFPACKAGE forms) are now handled more correctly.
2906 (Compiler queuing of top level lambdas has been suppressed by setting
2907 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2908 * deleted various currently-unused source files, e.g. gengc.lisp. They
2909 may be added back at some point e.g. when porting to other architectures,
2910 but until they are it's distracting to distribute them and to try to
2912 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2913 there were so many of them they're just distractions, and UNCROSS is
2914 known to be able to handle the current sources
2915 * moved PROFILE functionality into TRACE, so that it will be clear
2916 how the wrapping and unwrapping of functions when you profile them
2917 interacts with the wrapping and unwrapping of functions when you
2918 trace them. (Actually, the functionality isn't there yet, but at least
2919 the interface specification is there. Hopefully, the functionality will
2920 arrive with some future maintenance release.)
2921 * removed host-oops.lisp
2922 * changed signature of QUIT function to allow UNIX-CODE argument
2923 * fixed READ-SEQUENCE bug
2924 * tweaked verbose GC output so that it looks more like the progress
2925 output that ANSI specifies for functions like LOAD
2926 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2927 * added <http://sbcl.sourceforge.com> to the banner information printed by