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 * build fix: fixed the dependence on *LOAD-PATHNAME* and
17 *COMPILE-FILE-PATHNAME* being absolute pathnames.
18 * fixed some bugs revealed by Paul Dietz' test suite:
19 ** the FORMATTER-generated functions for ~V[ conditionals require
20 the correct number of arguments.
22 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
23 * new feature: reloading changed shared object files with
24 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
25 * new feature: references to foreign variables and functions
26 can now be compiled and loaded before the corresponding shared
27 object file is loaded, as long as the foreign definitions are
29 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
30 supported. (thanks to Dan Debertin)
31 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
32 just keywords. (thanks to Vincent Arkesteijn)
33 * fixed bug #331: structure-class instances corresponding to
34 DEFSTRUCT forms are now created eagerly.
35 * fixed bug #345: backtraces from calls to undefined functions work
36 on x86 as well. Related bug #61 is now also partially fixed on x86
37 (backtraces from throws to unknown catch tags.)
38 * bug fix: lambda-list parsing is now stricter vrt. order and number
39 of lambda-list keywords.
40 * bug fix: as specified by AMOP, an error is signalled if a
41 class-option appears multiple times in a DEFCLASS form. (reported
43 * bug fix: RANDOM can be compiled when the compiler derives the type
44 of its numeric argument as a disjoint set of small integers.
45 (reported by Paul Dietz)
46 * bug fix: starting a core saved with shared objects loaded when
47 those objects are not available no longer causes threaded SBCL to
48 hang. (reported by Sean Ross)
49 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
51 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
52 argument, which must always be NIL. (reported by Kalle Niemitalo)
53 * bug fix: printing 1.0d+23 no longer results in an error.
54 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
55 * bug fix: structure slot setters preserve evaluation order. (thanks
57 * fixed some bugs related to Unicode integration:
58 ** RUN-PROGRAM can allow its child to take input from a Lisp
59 stream. (reported by Stefan Scholl)
60 ** ASDF-INSTALL successfully downloads .tar.gz files in all
61 locales. (reported by Ken Causey)
62 * fixed some bugs revealed by Paul Dietz' test suite:
63 ** INCF, DECF and REMF evaluate their place form as specified in
65 ** FORMATTER expands ~{ ~} iteration directives with V or #
67 ** FORMATTER deals with the ~@[ ~] conditional directive where the
68 consequent uses no arguments correctly.
69 ** the system has a partial understanding of the (COMPLEX RATIO)
71 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
74 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
75 * new feature: a build-time option (controlled by the :SB-UNICODE
76 keyword feature, enabled by default) for building the system with
77 support for the entire 21-bit character space defined by the
79 * new feature: the system now has rudimentary external-format
80 support; the primary user-visible change at this time is that
81 characters with the high bit set (such as lower-case-e-acute) will
82 print correctly to a terminal in a UTF-8 environment.
83 * minor incompatible change: BASE-CHAR no longer names a class;
84 however, CHARACTER continues to do so, as required by ANSI.
85 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
86 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
87 has been moved to the SB-EXT package (temporarily re-exported from
89 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
90 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
92 * fixed bug #351: better error-handlind and reporting for malformed
94 * fixed bug #350: bignum-printing is now more memory-efficient,
95 allowing printing of very large bignums, eg. (expt 2 10000000).
96 (reported by Bruno Haible)
97 * fixed bug #302: better primitive-type selection for intersection
99 * fixed bug #308: non-graphic characters now all have names, as
100 required. (reported by Bruno Haible)
101 * bug fix: redefining a class with different superclasses now correctly
102 removes it from the direct-subclasses of its previous superclasses.
103 (reported by David Morse)
104 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
105 new value now works. (reported by Bruno Haible)
106 * bug fix: correct canonicalization of multiple non-standard slot
107 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
108 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
109 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
110 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
111 now exists, an signals an error.
112 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
113 during macroexpansion for non-iteration variables as well. (reported
114 by Bruno Haible for CMUCL)
115 * bug fix: Cyclic structures and unprintable objects in compiler
116 messages no longer cause errors. (reported by Bruno Haible)
117 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
118 and READ-FROM-STRING all now return a primary value of NIL if
119 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
120 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
121 (reported by Bruno Haible for CMUCL)
122 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
123 do not designate a condition. (reported by Bruno Haible for
125 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
126 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
127 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
128 permitting resolution in favour of any of the conflicting symbols.
129 (reported by Bruno Haible for CMUCL)
130 * bug fix: EQUAL compiler optimizations is less aggressive on
131 strings which can potentially compare true despite having distinct
132 specialized array element types.
133 * bug fix: unit enumerations can be defined without dividing by
134 zero. (thanks to Vincent Arkesteijn)
135 * FORMAT compile-time argument count checking has been enhanced.
136 (report from Bruno Haible for CMUCL)
137 * a partial workaround for the bug 262: the compiler does not try to
138 inline-expand a local function doing RETURN-FROM from a deleted
139 BLOCK. (thanks to Peter Denno for the bug report and to David
140 Wragg for the simple test case)
141 * fixed some bugs revealed by Paul Dietz' test suite:
142 ** RENAME-PACKAGE allows all package designators as new package
144 ** constraint propagation and lambda variable substitution are
145 more cautious in dealing with partially deleted code.
146 ** compiler sometimes lost reoptimization passes.
147 ** CERROR, given a condition as condition designator, passes its
148 remaining arguments to the continue format control without
150 ** Case-altering FORMAT directives work correctly on non-ASCII
152 ** The REAL type specifier handles bounds outside the floating
153 point ranges without signalling FLOATING-POINT-OVERFLOW.
154 ** Functions with IR1-transformations can create intercomponent
155 references to global functions.
156 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
158 ** FORMAT ~R treats a nil value for its first parameter correctly.
160 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
161 * enhancement: saving cores with foreign code loaded is now
162 supported on x86/NetBSD and sparc/Linux in addition to the previously
164 * bug fix: on some platforms repeated installations caused multiple
165 copies of HTML documentation to be installed -- should not happen
166 any more. (reported by Stefan Scholl)
167 * bug fix: parsing self-recursive alien record types multiple times
168 no longer causes infinite recursion. (reported by Thomas F. Burdick,
169 original patch by Helmut Eller for CMUCL)
170 * bug fix: stack-exhaustion detection works now on NetBSD as well.
171 (thanks to Richard Kreuter)
172 * bug fix: defining classes whose accessors are methods on existing
173 generic functions in other (locked) packages no longer signals
174 bogus package lock violations. (reported by François-René Rideau)
175 * bug fix: special variables as DEFMETHOD parameters no longer have
176 associated bogus type declarations. (reported by David Wragg and
178 * bug fix: read-write consistency on streams of element-type
179 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
180 * bug fix: redefiniton of the only method of a generic function with
181 no DEFGENERIC no longer emits a full WARNING. In addition,
182 redefinition of generic functions with no DEFGENERIC to an
183 incompatible lambda list now signals an error. (thanks to Zach
185 * bug fix: DEFGENERIC now works even when there's a function of the
186 same name in an enclosing lexical environment. (thanks to Zach
188 * fixed compiler failure, caused by instrumenting code during
189 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
190 * optimization: added loop analysis and improved register allocation
191 * fixed some bugs revealed by Paul Dietz' test suite:
192 ** POSITION on displaced vectors with non-zero displacement
193 returns the right answer.
194 ** (SIMPLE-STRING) is a valid type specifier for sequence
196 ** *PRINT-LEVEL* handling for slotless structures is pedantically
198 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
199 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
201 ** The justification version of the FORMAT ~< directive treats
202 non-zero minpad parameter correctly.
204 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
205 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
206 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
207 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
208 the supported interface.
209 * new feature: Single-stepping of code compiled with DEBUG 2 or
210 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
211 * new feature: saving cores with foreign code loaded is now
212 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
213 Timothy Moore's work for CMUCL)
214 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
215 parameters to * instead of NIL if no initform is supplied.
216 (reported by Johan Bockgaard)
217 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
218 correctly when FUNCALL appears as the car of the form. Note:
219 despite this FUNCALL forms are not currently subject to
220 compiler-macro expansion. (port of Raymond Toy's fix for the
221 same from CMUCL, reported by Johan Bockgaard)
222 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
223 (thanks for Teemu Kalvas)
224 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
225 case instead of terminating the iteration (thanks for Julian
226 Squires, Sean Champ and Raymond Toy)
227 * bug fix: incorrect expansion of defgeneric that caused a style
228 warning. (thanks for Zach Beane)
229 * bug fix: slot accessor effective method computation works properly
230 for classes with multiple non-standard applicable methods on
231 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
232 * on x86 compiler supports stack allocation of results of LIST and
233 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
234 CMUCL implementation by Gerd Moellmann)
235 * fixed some bugs revealed by Paul Dietz' test suite:
236 ** FORMAT strings with both the ~<~:;~> form of the justification
237 directive and pretty-printing directives cause an error.
239 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
240 * incompatible change: the internal functions
241 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
242 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
243 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
244 instead of the old functions.
245 * new feature: on platforms where "dladdr" is available foreign
246 function names now appear in backtraces. (based on Helmut Eller's
248 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
249 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
251 * documentation: ASDF manual is now distributed with SBCL.
252 * bug fix: SBCL can now load its contributed modules with REQUIRE
253 even if the system-provided entries have been removed from
254 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
256 * bug fix: backtraces involving undefined functions or assembly
257 routines are more informative. (thanks to Brian Downing)
258 * bug fix: mutually referent alien structures now work correctly.
259 (reported by Rick Taube)
260 * bug fix: structures defined by WITH-ALIEN can be referred to
261 within other definitions in the same WITH-ALIEN.
262 * bug fix: division operators (MOD, TRUNCATE and the like) with
263 constant zero divisors and integer dividends no longer generate
265 * bug fix: provide default methods for INPUT-STREAM-P and
266 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
267 * bug fix: improve the detection and resolution of MOP metacycles
268 (where effective method computation depends on the generic function
269 having its effective method computed). (reported by Bruno Haible)
270 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
271 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
272 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
273 * bug fix: pathnames with relative directory components can be
274 represented relative to default pathnames.
275 * optimization: in taking the GCD of bignums, reduce the two bignums
276 to approximately the same size (using Euclid's algorithm) before
277 applying the more sophisticated binary GCD. (thanks to Juho
279 * optimization: COUNT on bitvectors now operates word-at-a-time.
280 * optimization: ASH with a positive, but not necessarily constant,
281 (leftwards) shift, when in a modular context, is compiled to a
283 * fixed some bugs revealed by Paul Dietz' test suite:
284 ** FORMAT variable parameters ("~V<char>") are defaulted properly
285 if the corresponding argument is NIL.
286 ** FORMAT directives accepting a minpad parameter treat negative
288 ** literal commas as character parameters to FORMAT directives are
290 ** literal spaces directly after ~<Newline> directives within a
291 format-logical-block (~:< ~@:>) do not induce :FILL-style
292 conditional newlines.
293 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
294 as specified: it no longer includes conditional newlines).
295 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
297 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
298 :PER-LINE-PREFIX argument does not evaluate to a string.
300 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
301 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
302 the manual for details; for now, package locks can be disabled by
303 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
304 if no major problems are found then it is likely that they will be
305 compiled in unconditionally.
306 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
307 unsupported operators on all platforms. To load a shared library
308 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
309 object file, link it into a shared library outside of SBCL and
310 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
312 * bug workaround: ROOM T and the GC were not working together
313 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
314 reporting facility aren't preserved. That reporting has been
315 disabled, so now until and unless someone figures out how to make it
316 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
317 * minor incompatible change: as threatened around sbcl-0.8.0, the
318 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
319 The symbols are also exported from SB-PCL for backwards
320 compatibility, but more so than before SB-PCL should be treated as
321 an implementation-internal package.
322 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
324 * the SB-SPROF contrib now works on (most) non-x86 architectures.
325 It is known as of this release not to work on the Alpha, however.
326 * fixed bug #167: errors signalled due to illegal syntax in method
327 bodies are now more legible.
328 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
329 designators and can hence be used with TYPEP.
330 * fixed bug #333: CHECK-TYPE now ensures that the type error
331 signalled, if any, has the right object to be accessed by
332 TYPE-ERROR-DATUM. (reported by Tony Martinez)
333 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
334 5.1.2.3 for multiple-value place subforms. (reported by Kalle
336 * fixed bug #334: programmatic addition of slots using specialized
337 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
338 and :ALLOCATION :CLASS slots.
339 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
340 float exponents. (rereported by Peter Seibel)
341 * fixed a bug: #\Space (and other whitespace characters) are no
342 longer considered to be macro characters in standard syntax by
344 * fixed bug: initialization of condition class metaobjects no longer
345 causes an instance of the condition to be created. (reported by
347 * fixed bug: it is now possible to have slots such that
348 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
349 description is neither :INSTANCE nor :CLASS.
350 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
351 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
352 SLOT-VALUE-USING-CLASS), and no longer causes errors with
353 non-standard SLOT-DEFINITION-ALLOCATION values.
354 * fixed bugs: various race conditions handling exiting threads.
355 CL-PPCRE's thread torture test now passes.
356 * fixed arguably-a-bug: GC time proportional to number of threads in
357 system even when most of them are idle
358 * optimization: improved performance of BIT and SBIT on bit-vectors.
359 * .fasl file incompatibility: The fasl file version number has
360 been incremented because of changes associated with package locks.
362 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
363 * minor incompatible change: the system no longer provides
364 optimization hints (conditions which are TYPEP
365 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
366 mode; these hints are still emitted when higher SPEED optimization
368 * new contrib module: a sampling profiler (profiling by statistical
369 sampling, rather than by instrumenting functions) is available as
370 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
371 * the behaviour of the standard function ED is now customizeable by
372 third parties through a hook variable: see ED's documentation
373 string for information on the protocol.
374 * the compiler no longer emits efficiency notes for (FUNCALL X)
375 when the type of X is uncertain under default optimization
377 * fixed bug 276: mutating a binding of a specialized parameter to a
378 method to something that is not TYPEP the specializer is now
380 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
381 exhibit the required behaviour on the x86 platform. (thanks to
382 Peter van Eynde, Eric Marsden and Bruno Haible)
383 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
384 even for difficult arguments. (reported by Peter Graves)
385 * fixed bug 141a: the backquote printer now descends quoted
387 * fixed another bug in backquote printing: no more destructive
388 modification of the form's list structure. (reported by Brian
390 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
391 move between its address being taken and the call to
392 interrupt_thread, fixing a crashing race condition.
393 * the SB-POSIX contrib implementation has been adjusted so that it
394 no longer exhibits ridiculously poor performance when constructing
395 instances corresponding to C structs.
397 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
398 * minor incompatible change: the sb-grovel contrib now treats C
399 structures as alien (in the sense of SB-ALIEN) objects rather than
400 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
401 has implications for memory management of client code
402 (sb-grovel-returned objects must now be manually managed) and for
403 type safety (alien objects now have full types).
404 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
405 used to control emission of compiler diagnostics, rather than the
406 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
407 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
408 quality should be considered deprecated.
409 * install.sh now installs the user manual as well
410 * (not quite a new documentable feature, but worth considering in
411 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
412 beginnings of a semantically meaningful condition hierarchy is
413 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
415 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
416 designator as the defaults argument.
417 * fixed bug: Displaced arrays whose displaced-to array has become
418 too small now cause ARRAY-DIMENSION to signal an error, providing
419 for safer bounds-checking. (reported by Bruno Haible)
420 * fixed bug: DEFCLASS slot definitions with identical :READER and
421 :WRITER names now signal a reasonable error. (reported by Thomas
423 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
425 * fixed bug: Invalid slot specification errors now print correctly.
426 (thanks to Zach Beane)
427 * fixed bug 320: Shared to local slot value transfers in class
428 redefinitions now happen corrently from superclasses as
429 well. (reported by Bruno Haible)
430 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
432 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
433 improper lists correctly. (reported by Bruno Haible)
434 * fixed bug 313: source-transform for <fun-name> was erroneously
435 applied to a call of a value of a variable with name <fun-name>.
436 (reported by Antonio Menezes Leitao)
437 * fixed bug 307: The obsolete instance protocol ensures that
438 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
439 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
440 remove dead unknown-values globs from the middle of the stack.
441 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
442 easier to resume long computations after using *BREAK-ON-SIGNALS*
443 to diagnose and fix failures. (thanks to Nikodemus Siivola)
444 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
445 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
446 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
447 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
448 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
450 * fixed a (fairly theoretical) bug in string printing: if
451 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
452 does not have array-element-type equal to the most general string
454 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
455 function name. (thanks to Nikodemus Siivola)
456 * fixed bug: compile-time format string checker failed on
457 non-closed ~{. (reported by Thomas F Burdick)
458 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
459 don't warn on legal loop code involving more than one aggregate
460 boolean. (thanks to Nikodemus Siivola)
461 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
462 in keyword position to MAKE-PACKAGE was not being reported as
464 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
465 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
467 * fixed bugs 280 and 312: the checking for multiple definitions in a
468 file is less likely to become confused by uses of inline
470 * fixed bug: the #S reader macro performs the keyword coercion
471 specified for slot names. (reported by Kalle Niemitalo)
472 * fixed bug: lambda lists may contain symbols whose names start with
473 & but are not lambda-list-keywords; their occurrence triggers a
475 * fixed bug 321: define-method-combination argument lambda lists do
476 not cause invalid code to be generated when &optional or &aux
477 variables have default values. (reported by Bruno Haible)
478 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
479 have CLOS classes; this appears also to have fixed reported
480 instability in stack exhaustion detection.
481 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
482 MUFFLE-WARNING when no associated restart is present is now
484 * optimization: rearranged the expansion of various defining macros
485 so that each expands into only one top-level form in a
486 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
488 * optimization: used a previously little-used slot in symbols to
489 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
491 * fixed some bugs revealed by Paul Dietz' test suite:
492 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
494 ** Fixed an optimization bug related to inheritance of initforms
495 from local to shared slots.
496 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
497 ** CAST optimizer forgot to flush argument derived type.
498 ** print/read consistency on floats is now orders of magnitude
499 more likely. (thanks also to Bruno Haible for a similar report
501 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
502 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
503 for objects of type REAL. Make it so.
504 ** SXHASH is specified (infelicitously) to respect similarity,
505 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
506 it so. (thanks to Markus Ziegler)
507 ** on the Alpha, the compiler succeeds in compiling functions
508 returning a known number of arguments greater than 63.
509 ** fixed handling of invalid NIL arguments in keyword position
511 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
513 ** provided a readably-printable representation for RANDOM-STATE
515 ** ensured that pathnames, where they have a namestring, always
516 print using #P"..." syntax.
518 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
519 * Support for the forthcoming 2.0 version of the NetBSD kernel
520 running on x86 hardware has been added. (thanks to Perry
521 E. Metzger most immediately, and others for their past work)
522 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
523 current release version; SBCL's OpenBSD support had been broken
524 since about the time of OpenBSD's switch to ELF binary format.)
525 * [placeholder for DX summary]
526 ** user code with &REST lists declared dynamic-extent, under high
527 speed or space and low safety and debug optimization policy.
528 * The manual has been converted to Texinfo format and the debugger
529 chapter from the cmucl manual has been added.
530 * A facility has been added to extract documentation strings from
531 sbcl and store them as Texinfo-formatted snippets for inclusion in
532 the manual (via Texinfo's @include directive)
533 * bug fix: compiler emitted division in optimized DEREF. (thanks for
534 the test case to Dave Roberts)
535 * bug fix: multidimensional simple arrays loaded from FASLs had fill
536 pointers. (reported by Sean Ross)
537 * bug fix: PROFILE output is printed nicely even for large numerical
538 values. (thanks to Zach Beane)
539 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
540 greater than 32 handle EOF correctly.
541 * bug fix: on X86 an immediate argument of the IMUL instruction is
542 correctly printed in disassembly. (thanks to Lutz Euler)
543 * bug fix: class slots in redefined classes preserve their old
544 values. (thanks to Bruno Haible and Nikodemus Siivola)
545 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
546 succeeds. (reported by Marco Baringer)
547 * fixed some bugs revealed by Paul Dietz' test suite:
548 ** READ-SEQUENCE now works on ECHO-STREAMs.
549 ** RATIONALIZE works more according to its specification. (thanks
552 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
553 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
554 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
555 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
556 should matter to you only if you rebind the printer control
557 variables and then find you want different bindings in the
558 debugger than in the ordinary execution of your program.)
559 * The runtime build system has been tweaked to support building
560 (on SPARC/SunOS) using a C compiler which invokes Sun's own
561 assembler and linker. (thanks to Nikodemus Siivola)
562 * Unbound, undefined, undeclared variables now trigger full
563 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
564 more often programmer error than deliberate exploitation of undefined
566 * optimization: the hash algorithm for strings has changed to one
567 that is less vulnerable to spurious collisions. (thanks to Juho
569 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
570 needless bounds checking. (thanks to Juho Snellman)
571 * optimization: implemented multiplication as a modular
572 (UNSIGNED-BYTE 32) operation on the PPC backend.
573 * fixed some bugs revealed by Paul Dietz' test suite:
574 ** ADJUST-ARRAY now copies the datum in a zero rank array if
576 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
577 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
578 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
580 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
582 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
584 ** ECHO-STREAMs no longer attempt to echo the end of file value to
585 their output stream on EOF from read.
586 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
587 have been read to end-of-file.
588 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
590 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
591 description of determination of which consecutive characters
593 ** Printing the "Space" character with escaping on now yields "#\\ ",
594 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
595 ** Reading floating-point numbers with *READ-BASE* set to a number
596 less than 10 works correctly.
597 ** Reading floating-point numbers with *READ-BASE* set to a number
598 more than 10 works correctly.
599 ** Printing with *PRINT-READABLY* targets the standard readtable, not
600 the readtable currently in effect.
602 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
603 * minor incompatible change: parsing of namestrings on a physical
604 (Unix) host has changed; numbers after the final #\. in a
605 namestring are no longer interpreted as a version field. This is
606 intented to be largely invisible to the user, except that the
607 meaning of the namestring "*.*.*" has changed: it now refers to a
608 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
609 should usually be replaced by
610 (make-pathname :name :wild :type :wild :version :wild)
611 with the added benefit that this is more likely to be portable.
612 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
613 option to OPEN now signals an error if the file being opened
614 exists; this may have an impact on existing code.
615 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
616 signals received, on the PowerPC platforms (both Linux and
617 Darwin). (thanks to Pierre Mai for pointing out the location of
619 * several fixes on OS X: The system now builds and runs cleanly on
620 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
621 (thanks to Brian Mastenbrook)
622 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
623 second-resolution: integer multiples of 1/3600 between -24 and 24.
624 (thanks to Vincent Arkesteijn)
625 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
626 type when called with 1 argument; PEEK-CHAR checked type of
627 PEEK-TYPE only after having read first character from a
628 stream. (reported by Peter Graves)
629 * bug fix: the garbage collector now has much better locality
630 behaviour, and in particular no longer treats all memory as being
631 exhausted when many small objects point to each other in a deeply
633 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
634 recognized as being TYPEP their class.
635 * bug fix: the PUSHNEW documentation string has been corrected.
636 (thanks to Vincent Arkesteijn)
637 * bug fix: defaulting of the value for the last of an atypically
638 large number of multiple values being bound was not being
639 performed correctly on the Alpha or PPC platforms
640 * optimization: implemented multiplication as a modular
641 (UNSIGNED-BYTE 32) operation on the x86 backend.
642 * optimization: SEARCH on simple-base-strings can now be open-coded.
643 (see also contrib/compiler-extras.lisp for inspiration for
644 teaching the compiler about the Boyer-Moore algorithm).
645 * value, returned by MAX (and MIN) called with several EQUALP, but
646 not EQL, arguments now does not depend on compiler settings.
647 * fixed some bugs revealed by Paul Dietz' test suite:
648 ** in stack analysis liveness information is propagated from
649 non-local entry points.
650 ** pathwise CAST removing failed when the CAST node did not start
652 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
653 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
655 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
656 host is already defined.
657 ** RENAME-FILE works on streams instead of signalling an internal
659 ** PEEK-CHAR uses the current readtable when determining whether
660 or not a character is whitespace.
661 ** MERGE-PATHNAMES handles the case when the pathname does not
662 specify a name while the default-pathname specifies a version.
663 ** Pathnames now stand a better chance of respecting print/read
665 ** Attempting to use standardized file system operators with a
666 pathname with invalid :DIRECTORY components signals a
668 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
670 ** DIRECTORY on logical pathnames is more correct.
671 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
672 signal a TYPE-ERROR if their argument is not a stream.
673 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
674 designator argument does not designate a stream.
675 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
676 examining the synonym.
677 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
679 ** OPEN and WITH-OPEN-STREAM allow opening streams with
680 element-type larger than ([UN]SIGNED-BYTE 32).
682 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
683 * When built with the :SB-FUTEX feature, threaded builds now take
684 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
685 for faster/more reliable mutex and condition variable support.
686 * Incompatible change (but one you probably shouldn't have been using
687 anyway): the interface and code for arbitrating between multiple
688 threads in the same user session has been redesigned.
689 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
690 error when called without an explicit environment argument.
691 (thanks to Nikodemus Siivola)
692 * bug fix: buffered :DIRECTION :IO streams are less likely to become
693 confused about their position. (thanks to Adam Warner and Gerd
695 * bug fix: Pretty printing backquoted forms with unquotations in the
696 argument list position of various code constructs such as LAMBDA
697 now works correctly. (reported by Paul Dietz)
698 * bug fix: Pretty printing unquotations no longer loses all
699 stream position information.
700 * optimization: performance of string output streams is now less
701 poor for multiple small sequence writes.
702 * optimization: performance of CSUBTYPEP in the presence of complex
703 expressions involving CONS and NOT many times has been improved.
704 (reported by Paul Dietz)
705 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
707 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
708 streams for the REPL, for the debugger, and so on. (thanks to
710 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
711 * fixed some bugs revealed by Paul Dietz' test suite:
712 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
713 is ignored if it is too small, rather than propagating through
714 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
715 ** extremely complex negations of CONS types were not being
716 sufficiently canonicalized, leading to inconsistencies in
718 ** VALUES tranformer lost derived type.
720 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
721 * fixed a bootstrapping bug: the build process no longer assumes
722 that the various BOOLE-related constants have the same value in
723 host and target lisps. (noted by Paul Dietz' test suite on an
724 SBCL binary built from CLISP)
725 * The system can now be dynamically linked on the MIPS platform,
726 which enables dynamic loading of foreign code from Lisp. (thanks
727 to Ralf Baechle for discussions on the MIPS ABI)
728 * The system now records debugging information for its own source
729 files in a filesystem-position-independent manner, relative to
730 the "SYS" logical host.
731 * fixed a compiler bug: MV-LET convertion did not check references
732 to the "max args" entry point. (reported by Brian Downing)
733 * tweaked disassembly notes to be less confident about proclaiming
734 some instruction as an LRA. (thanks to Brian Downing)
735 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
736 now each have their own history, command character, and other
737 characteristics. (thanks to David Lichteblau)
738 * fixed some bugs revealed by Paul Dietz' test suite:
739 ** compiler failure in compiling LOGAND expressions including a
741 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
742 shift greater than 32.
743 ** FLUSH-DEST did not mark blocks for type check regeneration.
744 ** HANDLER-CASE failed to accept declarations in handler clauses
745 in some circumstances.
747 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
748 * New code in contrib/sb-introspect (still probably not entirely
749 stable yet) provides some support for smart Lisp development
750 environments like SLIME.
751 * The conditions signalled for errors occurring when loading .fasl
752 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
753 in a way which should help ASDF recover gracefully.
754 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
755 stuff has been cleaned up. If you code contrib/ stuff, this might
756 affect you, and you can look at contrib/README, contrib/STANDARDS,
757 and/or the 0.8.4.27 diff to check.
758 * In full calls the compiler now does not generate checks for declared
759 argument types for all arguments.
760 * various threading fixes
761 ** and some experimental patches which didn't make it into
762 the main tree for this release, but which are shipped in
763 contrib/experimental-thread.patch as a possible fix for some
764 failures (deadlock, spinning...) in GC-intensive multithreaded
766 * fixed PPC build problem (source code incompatibility of different
767 library versions): added offsetof() hackery which attempts to divine
768 where glibc maintainers put uc_mcontext today
769 * fixed bug 282: compiler does not trust type assertions while passing
770 arguments to a full call.
771 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
772 &OPTIONAL VALUES type parameter.
773 * fix bug 214: algorithm for noting rejected templates is now more
774 similar to that of template seletion. (also reported by rydis on
776 * fixed bug 141b: printing backquoted information readably and prettily
777 inserts a space where necessary.
778 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
779 (CHARACTER 10) are now reported as errors, rather than propagated
780 as unknown types. (reported by piso on #lisp)
781 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
782 with values NIL and :ERROR. (thanks to Milan Zamazal)
783 * fixed bug 191c: CLOS now does proper keyword argument checking as
784 described in CLHS 7.6.5 and 7.6.5.1.
785 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
786 longer raise an error; further, using a list as a for-as-arithmetic
787 counter now raises a meaningful error.
788 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
789 now understood by sequence creation functions such as MAKE-SEQUENCE
791 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
792 a STRING-INPUT-STREAM.
793 * compiler enhancement: SIGNUM is now better able to derive the type
795 * type declarations inside WITH-SLOTS are checked. (reported by
797 * fixed some bugs revealed by Paul Dietz' test suite:
798 ** incorrect optimization of TRUNCATE for a positive first
799 argument and negative second.
800 ** compiler failure in let-convertion during flushing dead code.
801 ** compiler failure while deriving type of TRUNCATE on an
802 interval, containing 0.
803 ** ASH of a negative bignum by a negative bignum count now returns
805 ** intersection of CONS types now canonicalizes properly, fixing
806 inconsistencies in SUBTYPEP.
808 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
809 * incompatible change: The --disable-debugger command line
810 option now clobbers the debugger at a more fundamental
811 level, by redefining #'INVOKE-DEBUGGER instead of by
812 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
813 is specified by ANSI to ignore *DEBUGGER-HOOK* and
814 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
815 enter the debugger REPL and then suffer recursive errors
816 because *DEBUG-IO* is also messed up in --disable-debugger mode;
817 while under the new system, BREAK in --disable-debugger mode
818 terminates the system just as an unhandled error would.
819 * fixed compiler performance when processing loops with a step >1;
820 * bug fix: DOCUMENTATION now retrieves generic function
821 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
822 support has been systematized, and now supports the methods
823 specified by ANSI, along with a default method and a method for
824 slot documentation. (reported by Nathan Froyd)
825 * bug fix: effective methods associated with a generic function are
826 no longer cached over a change of that generic function's method
827 combination. (reported by Andreas Fuchs)
828 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
829 names a non-existent directory. (thanks to Andreas Fuchs)
830 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
831 the PPC platform, lead to stack corruption; this has been fixed.
832 (reported by Rainer Joswig)
833 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
834 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
835 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
836 arithmetic in general, is now much more reliable on the PPC
838 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
839 platform now returns the right answer.
840 * optimization: restored some effective method precomputation in
841 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
842 precomputation is now tunable.
843 * optimization: compiler-internal data structure use has been
844 reviewed, and changes have been made that should improve the
845 performance of the compiler by about 20%.
846 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
847 simple-base-strings and simple-bit-vectors is improved.
848 * optimization: the optimization of 32-bit logical and arithmetic
849 functions introduced in version 0.8.3 on the x86 has been
850 implemented on the mips, ppc and sparc platforms; an
851 implementation of the same facility, but for 64-bit arithmetic,
852 has been added for the alpha.
853 * microoptimization: the compiler is better able to make use of the
854 x86 LEA instruction for multiplication by constants.
855 * bug fix: in some situations compiler did not report usage of
856 generic arithmetic in (SPEED 3) policy.
857 * bug 145b fix: compiler used wrong type specifier while converting
858 MEMBER-types to numeric.
859 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
861 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
862 index argument now works properly on the Alpha platform.
863 * bug fix: floating point exception treatment on the Alpha platform
865 * bug fix: FILE-POSITION works much better on string input and
866 output streams. (thanks to Nikodemus Siivola)
867 * bug fix: many threading/garbage collection symptoms sorted.
868 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
869 might be pseudo-atomic.
870 * internal change: Stopping for GC is now done with signals not
871 ptrace. GC is now done in whichever thread wanted it, instead of
873 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
875 * bug fix: result form in DO is not contained in the implicit
877 * incompatible change: ICR structure is changed; the value part of
878 CONTINUATION is now called LVAR; corresponding functions are
879 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
880 * added type deriver for ISQRT (thanks to Robert E. Brown).
881 * bug fix: better support for loading from the command line when an
882 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
884 * fixed some bugs revealed by Paul Dietz' test suite:
885 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
886 ** ROUND and FROUND now give the right answer when given very
887 small float arguments.
888 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
890 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
891 ** (GCD 0 <negative-integer>) returned <negative-integer>.
892 ** LCM should return a non-negative integer.
893 ** PARSE-INTEGER returned the index of a terminator instead of the
894 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
895 ** PARSE-INTEGER returned an incorrect index being applied to a
897 ** LCM with two arguments of 0 returns 0 rather than signalling
899 ** unsigned addition of a 32-bit constant with the high bit set no
900 longer causes an internal compiler error on the x86.
901 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
902 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
903 with negative last argument.
904 ** byte specifiers with zero size and position no longer cause
905 an error during type derivation.
906 ** bignum multiplication on the Alpha platform now returns the
908 * porting: The system now builds on SuSE AMD64, although it still
909 generates a 32-bit binary.
910 * .fasl file incompatibility: The fasl file version number has
911 been incremented (because of the changes to internal compiler
912 data structures referred to above).
914 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
915 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
916 more accurately, on the Darwin kernel running on PowerPC hardware.
917 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
918 * Compiler code deletion notes now signal a condition of type
919 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
920 an associated MUFFLE-WARNING restart.
921 * The compiler now performs limited argument count validation of
922 constant format strings in FORMAT, and where appropriate in ERROR,
923 CERROR and WARN. (thanks to Gerd Moellmann)
924 * New ASDF-INSTALL contrib can be used for automatic download and
925 installation of third-party Lisp code from CCLAN or other sites
927 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
928 which forces another thread to execute a function supplied by the
930 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
931 now accept and act upon their :ELEMENT-TYPE keyword argument.
932 (reported by Martin Atzmueller, Edi Weitz)
933 * bug fix: FILE-POSITION now accepts position designators up to
934 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
935 is the greater. (thanks to Patrik Nordebo)
936 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
937 Kalle Olavi Niemitalo)
938 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
939 printable. (reported by Eric Marsden)
940 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
941 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
942 * x86 bug fix in control stack exhaustion checking: now shows backtrace
943 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
944 (thanks to Stig Sandoe)
945 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
946 UNREAD-CHAR now correctly includes the unread character in the
947 target sequence. (thanks to Gerd Moellmann)
948 * bug fix in threaded builds: the system can now be suspended and
949 resumed by shell job control with minimal disruption.
950 * bug fixes in times and timezones >2038AD
951 * better handling of "where is GNU make?" problem in build scripts
952 (thanks to Nikodemus Siivola)
953 * new optimization: inside a named function any reference to a
954 function with the same name is considered to be a self-reference;
955 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
956 optimization quality.
957 * new optimization on x86: logical functions and + now have
958 optimized (UNSIGNED-BYTE 32) versions, which are automatically
959 used when the result is truncated to 32 bits.
960 * VALUES declaration is partially enabled.
961 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
962 * fixed some bugs revealed by Paul Dietz' test suite:
963 ** The system now obeys the constraint imposed by
964 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
965 types form a lattice under type intersection.
966 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
967 ** ASSOC now ignores NIL elements in an alist.
968 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
969 and (1+ MOST-POSITIVE-FIXNUM) answers.
970 ** The addition of a method with invalid qualifiers to a generic
971 function does not cause an error to be signalled immediately;
972 a warning is signalled, and the error is generated only on
973 calling the generic function.
974 * changed .fasl file version number, in part to add type codes for
975 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
976 obscure ANSI requirements
978 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
979 * fixed bug 148: failure to inline-expand a local function left
980 garbage, confusing the compiler.
981 * fixed bugs 3cd: structure slot readers perform type check if the
982 slot can have an invalid value (i.e. it is either not initialized
983 or can be written with a less specific slot writer).
984 * bug fix: the compiler now traps array references to elements off
985 the end of an array; previously, the bounds checking in some
986 circumstances could go off-by-one.
987 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
989 * type declarations for array element types now obey the description
990 on the CLHS page "Declaration TYPE", as per discussions on
991 sbcl-help around 2003-05-08. This means that a declaration
992 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
993 declaration, all references to BAR will be asserted or assumed
994 (with THE, so dependent on compiler policy) to involve objects of
995 type FOO. Note that no such declaration is implied in
996 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
997 * declared types of functions from the "Conditions"
998 chapter. (reported by Paul Dietz)
999 * bug fix: CERROR accepts a function as its first argument.
1000 * bug fix: NTH an NTHCDR accept a bignum as index
1001 arguments. (reported by Adam Warner)
1002 * optimization: character compare routines now optimize comparing
1003 against a constant character. (reported by Gilbert Baumann)
1004 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1005 argument now works properly on the MIPS platform.
1006 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1007 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1008 declared to have a functional type, does not cause an error
1010 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1011 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1012 * fixed bug: if last continuation of a deleted block has a
1013 destination, this destination should be deleted too. (reported by
1015 * fixed a bug in the bootstrap process: the host compiler's values
1016 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1017 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1018 test case from Patrik Nordebo)
1019 * improved the ability of the disassembler on the PPC platform to
1020 provide helpful disassembly notes.
1021 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1022 the class in more cases than previously.
1023 * bug fix: FILE-POSITION now understands :START and :END for
1024 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1025 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1026 8) data. (thanks to David Lichteblau)
1027 * bug fix: it is possible to add a method to a generic function
1028 without lambda list.
1029 * bug fix: reader failed to signal END-OF-FILE inside an
1030 object representation. (reported by Nikodemus Siivola)
1031 * fixed some bugs revealed by Paul Dietz' test suite:
1032 ** LAST and [N]BUTLAST should accept a bignum.
1033 ** condition slot accessors are methods.
1034 ** (VECTOR NIL) is a subtype of STRING.
1036 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1037 * minor incompatible change: some nonsensical specialized lambda
1038 lists (used in DEFMETHOD) which were previously ignored now signal
1040 * minor incompatible change: the system is now aware of the types of
1041 variables in the COMMON-LISP package, and will signal errors for
1042 most violations of these type constraints (where previously they
1043 were silently accepted).
1044 * minor incompatible change: COMPILE-FILE now uses the freedom
1045 afforded (ANSI 3.2.2.3) to use derived function types for
1046 functions defined in the same file. This also permits the system
1047 to warn on static type mismatches and function
1048 redefinition. (Currently it does not work with high DEBUG level.)
1049 * minor incompatible change: VALUES declaration is disabled.
1050 * When issuing notes, the compiler now signals a condition of type
1051 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1052 restart for use in user handlers. It is expected that the
1053 COMPILER-NOTE condition will eventually become a condition
1054 supertype to a hierarchy of note types, which will then be
1055 handleable in a similar fashion. However, at the moment, no such
1056 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1057 but it's an implementation detail, not a classification for the
1059 * Changes in type checking closed the following bugs:
1060 ** type checking of unused values (192b, 194d, 203);
1061 ** template selection based on unsafe type assertions (192c, 236);
1062 ** type checking in branches (194bc).
1063 * A short form of VALUES type specifier has ANSI meaning (it has
1064 increased the number of situations when SBCL cannot perform type
1066 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1067 keywords or constants is permissible.
1068 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1069 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1070 classes. (thanks to Antonio Martinez)
1071 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1072 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1073 argument to a pretty-print stream. Instead, it leaves any such
1074 operations to DESCRIBE-OBJECT methods.
1075 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1077 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1078 lambda lists. (thanks to David Lichteblau)
1079 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1080 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1081 (thanks to Gerd Moellmann)
1082 * fixed bug 246: increased compilation speed of long
1083 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1085 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1087 * DEFCONSTANT now throws a condition of type
1088 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1089 constant to a non-EQL value; CONTINUE and ABORT restarts
1090 respectively change and preserve the value.
1091 * fixed bug 63: The code walker, part of the implementation of CLOS,
1092 is now better at handling symbol macros.
1093 * bug fix: There is no longer an internal implementation type named
1094 CL:LENGTH. (reported by Raymond Toy)
1095 * bug fix: In macro-like defining macros/special operators the
1096 implicit block does not enclose the lambda list.
1097 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1098 enclosed in parentheses) are not suitable as type specifiers, and
1099 their use properly signals an error now.
1100 * bug fix: An argument count mismatch for a type specifier in code
1101 being compiled no longer causes an unhandled error at compile
1102 time, but signals a compile-time warning.
1103 * fixed simple vector readable printing
1104 * bug fix: DESCRIBE takes more care over whether the class
1105 precedence list slot of a class is bound before accessing it.
1106 (reported by Markus Krummenacker)
1107 * bug fix: FORMATTER can successfully compile pretty-printer format
1108 strings which use variants of the ~* directive inside.
1109 * bug fix: SEARCH now applies its TEST predicate to the elements of
1110 the arguments in the correct order. (thanks to Wolfhard Buss)
1111 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1112 in type inference. (thanks to Robert E. Brown)
1113 * bug fix: Reading in symbols with an explicit package name of ""
1114 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1115 * fixed some bugs revealed by Paul Dietz' test suite:
1116 ** NIL is now allowed as a structure slot name.
1117 ** Arbitrary numbers, not just REALs, are allowed in certain
1118 circumstances in LOOP for-as-arithmetic clauses.
1119 ** Multiple class redefinitions before slot access no longer
1120 causes a type error.
1121 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1122 association between the name and a class.
1123 ** Generic functions with non-standard method-combination and over
1124 five methods all of which return constants no longer return NIL
1125 after the first few invocations. (thanks to Gerd Moellmann)
1126 ** CALL-NEXT-METHOD with no arguments now passes the original
1127 values of the arguments, even in the presence of assignment.
1128 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1130 ** DISASSEMBLE works with closures and funcallable instances.
1131 ** ADD-METHOD now returns the generic function, not the new method.
1132 ** FIND-METHOD signals an error if the lengths of the specializers
1133 is incompatible with the generic function, even if the ERRORP
1135 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1136 which its argument is a member.
1137 ** DEFCLASS only redefines the class named by its class-name
1138 argument if that name is the proper name of the class;
1139 otherwise, it creates a new class.
1140 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1141 of the UNBOUND-SLOT condition to the name of the slot.
1142 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1143 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1144 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1145 SLOT-MAKUNBOUND in the specified fashion.
1147 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1148 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1149 cross-compilation host. As a consequence, we can now bootstrap our
1150 way up to SBCL starting with a bare gcc toolchain and human-readable
1151 source code (first the source to CLISP, then the source to SBCL).
1152 * A contributed module containing a partial implementation of the
1153 simple-streams interface has been included. (thanks to Rudi
1155 * A contributed module implementing the RFC1321 Message Digest
1156 Algorithm, known as MD5, has been included.
1157 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1158 no longer has any effect, as the code controlled by this feature
1159 has been deleted. (As far as we know, no-one has ever built using
1160 this feature, and its semantics were confused in any case).
1161 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1162 work on conditions (as required by the ANSI specification),
1163 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1164 expected behaviour on conditions. Users should note, however,
1165 that such behaviour is not required by the ANSI specification,
1166 and so use of this behaviour may render their code unportable.
1167 * fixed some bugs revealed by Paul Dietz' test suite:
1168 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1170 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1171 ** SXHASH on condition objects no longer returns NIL.
1172 ** :ALLOCATION :CLASS slots are better treated; their values are
1173 updated on class redefinition, and initforms inherited from
1174 superclasses are applied.
1175 ** REMOVE-METHOD returns its generic function argument even when
1176 no method was removed.
1177 ** SHARED-INITIALIZE now initializes the values of the requested
1178 slots, including those with :ALLOCATION :CLASS.
1179 ** ALLOCATE-INSTANCE now works on structure classes defined via
1180 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1182 ** SLOT-EXISTS-P now works on conditions, as well as structures
1184 ** MAKE-LOAD-FORM now has the required methods on
1185 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1186 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1187 keyword, and now implements the SLOT-NAMES argument.
1188 ** methods with &OPTIONAL arguments no longer allow too many
1189 arguments to be passed in the call without error.
1190 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1191 option is consistent with the required arguments of the generic
1192 function lambda list.
1193 * bug fix: REQUIRE accepts a string designator. (Thanks to
1195 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1196 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1197 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1198 * bug fix: The compiler now checks for duplicated variables in macro
1200 * bug fix: SETQ on globals returns the correct value.
1201 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1202 not a COMPILER-ERROR (followed by some other strange error on
1203 choosing the CONTINUE restart).
1204 * bug fix: make.sh and friends are now more consistent in the way that
1205 they look for GNU "make".
1207 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1208 * experimental native threads support (on x86 Linux >=2.4 only).
1209 This is not compiled in by default: you need to add :SB-THREAD to
1210 the target features. See the "Beyond ANSI" chapter of the manual
1212 * fix for longstanding nonANSIism: The old distinction between
1213 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1214 The return value from CL:FIND-CLASS is now a CLOS class, and
1215 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1216 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1217 * An interface to the MetaObject Protocol, as described in Kiczales,
1218 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1219 MIT Press, 1991, is available from the SB-MOP package.
1220 * incompatible change: the SB-PCL package should now be considered
1221 a private implementation detail, and no longer a semi-private MOP
1223 * minor incompatible change: due to rearrangement for threads, the
1224 control stack and binding stack are now allocated at arbitrary
1225 addresses instead of being hardcoded per-port. Users affected by
1226 this probably have to be doing advanced things with shared
1227 libraries, and will know who they are.
1228 * minor incompatible change: Previously, all --eval forms used were
1229 processed with READ before any of them were processed with EVAL.
1230 Now each --eval form is processed with both READ and EVAL before
1231 the next --eval form is processed. (Thus package operations like
1232 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1233 work as the user might reasonably expect.)
1234 * minor incompatible change: *STANDARD-INPUT* is now only an
1235 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1237 * minor incompatible change: Y-OR-N-P is now character-oriented,
1238 not line oriented. Also, YES-OR-NO-P now works without errors.
1239 (thanks to Antonio Martinez)
1240 * sb-aclrepl module improvements: an integrated inspector, added
1241 repl features, and a bug fix to :trace command.
1242 * Known functions, which cannot be open coded by the backend, are
1243 considered to be able to check types of their arguments. (fixing
1244 a bug report by Nathan J. Froyd)
1245 * fixed a bug in computing method discriminating functions: It is
1246 now possible to define methods specialized on classes which have
1247 forward-referenced superclasses. (thanks to Gerd Moellmann)
1248 * fixed evaluation order in optional entries (reported by Gilbert
1250 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1251 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1252 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1253 required-by-AMOP NAME argument, as well as CLASS and
1254 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1255 * fixed bug 20: DEFMETHOD can define methods using names that are
1256 not the proper names of classes to designate class specializers.
1257 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1258 Unix file descriptors, instead of blowing up. (thanks to Antonio
1260 * Garbage collection refactoring: user-visible change is that a
1261 call to the GC function during WITHOUT-GCING will not do garbage
1262 collection until the end of the WITHOUT-GCING. If you were doing
1263 this you were probably losing anyway.
1264 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1265 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1266 TYPEP the latter but not the former.
1267 * The compiler issues a full WARNING for calls to undefined functions
1268 with names from the CL package.
1269 * MAP-INTO for a vector destination is open coded. (reported by
1270 Brian Downing on c.l.l)
1271 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1272 documentation string.
1273 * fixed some bugs revealed by Paul Dietz' test suite:
1274 ** COPY-ALIST now signals an error if its argument is a dotted
1276 ** Condition slots are now accessed more correctly in the presence
1277 of multiple initargs for a given slot.
1278 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1279 exclude restarts of the same name associated with a different
1281 ** DEFCLASS of forward-referenced classes with another
1282 forward-referenced class in the superclasses list no longer
1284 ** Condition slots are now initialized once each, not multiple
1285 times. (thanks to Gerd Moellmann)
1286 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1288 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1289 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1290 arguments contain duplicated elements.
1291 ** RESTART-CASE understands local macros.
1292 ** RESTART-CASE associates exactly its own restarts with a condition.
1293 ** ENDP in safe mode checks its argument to be of type LIST.
1294 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1295 in question is unbound.
1296 ** Optimized MAKE-INSTANCE functions no longer cause internal
1297 assertion failures in the presence of duplicate initargs.
1298 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1299 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1300 propagated even in the ctor optimized implementation of
1302 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1304 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1305 * a better implementation of SXHASH on (simple) bit vectors,
1306 measured both in execution speed and in distribution of results
1307 over the positive fixnums, has been installed. Likewise, a better
1308 implementation of EQUAL for simple bit vectors is now available.
1309 * fixed CEILING optimization for a divisor of form 2^k.
1310 * fixed bug 240 (emitting extra style warnings "using the lexical
1311 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1312 by Antonio Martinez)
1313 * fixed SXHASH, giving different results for NIL depending on type
1314 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1315 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1316 variable. (found by Rolf Wester)
1317 * SBCL does not ignore type declarations for special
1318 variables. (reported by rif on c.l.l 2003-03-05)
1319 * some bug fixes in contrib/sb-aclrepl/
1320 * fixed some bugs revealed by Paul Dietz' test suite:
1321 ** a bug in the CONS type specifier, whereby the CAR and CDR
1322 types got intertwined, has been fixed;
1323 ** the type system is now able to reason about the interaction
1324 between INTEGER and RATIO types more completely;
1325 ** APPEND, [N]REVERSE and NRECONC check that those their
1326 arguments, which must be proper lists, are really so;
1327 ** An array specialized to be unable to hold elements has been
1328 implemented, as required -- yes, really -- by ANSI;
1329 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1330 on malformed property lists;
1332 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1333 * incompatible packaging change: in line with Unix convention,
1334 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1335 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1336 * REQUIRE and PROVIDE are now optionally capable of doing something
1337 useful. See the documentation string for REQUIRE.
1338 * infrastructure for a managed SBCL contrib system: contributed
1339 modules in this release include:
1340 ** the ASDF system definition facility;
1341 ** an interface to the BSD Sockets API;
1342 ** an ACL-like convenience interface to the repl;
1343 (thanks to Kevin Rosenberg)
1344 ** an implementation of ROTATE-BYTE, with efficient implementation
1346 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1347 gives a closer approximation to the right answer than previously.
1348 (thanks to Raymond Toy)
1349 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1350 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1351 optional environment arguments, as required by ANSI.
1352 * fixed bugs in other functions taking environment objects, allowing
1353 calls with an explicit NIL environment argument to be compiled
1355 * fixed bug 228: primary return values from
1356 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1357 COMPILE or FUNCTION.
1358 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1359 :TYPE will no longer signal errors on innocuous objects.
1360 * fixed bug 231b: SETQ is better at respecting type declarations in
1361 the lexical environment.
1362 * fixed a bug in DEFCLASS: classes named by symbols with no or
1363 unprintable packages can now be defined.
1364 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1365 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1366 * fixed compiler failure related to checking types of functions.
1367 (reported by Robert E. Brown)
1368 * the compiler is now much more consistent in its error-checking
1369 treatment of bounding index arguments to sequence functions: in
1370 (SAFETY 3) code, errors will be signalled in almost all cases if
1371 invalid sequence bounding indices are passed to functions defined
1372 by ANSI to operate on sequences.
1373 * fixed a bug in the build procedure: documentation of SBCL-specific
1374 packages is now preserved and available in the final Lisp image.
1375 * lifted FDEFINITION lookup out of loops in the implementation of
1376 many list operations. (thanks to Robert E. Brown)
1377 * fixed a bug in the reader: the #n# reader macro now works for
1378 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1379 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1380 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1381 E. Brown and Tony Martinez respectively)
1382 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1383 SLOT-BOUNDP: in particular, it is now able to optimize them much
1384 better, and is now not vulnerable to having packages renamed.
1385 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1386 always called when a slot is not present in an instance. (thanks
1388 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1389 longer signal an error on unreading a character following EOF on
1390 the previous constituent stream. (thanks to Tony Martinez)
1391 * fixed some bugs revealed by Paul Dietz' test suite:
1392 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1393 not just nonnegative fixnums;
1394 ** the logical bit-array operators such as BIT-AND now accept an
1395 explicit NIL for their "opt-arg" argument (to indicate a
1396 freshly-consed result bit-array);
1397 ** ELT now signals an error on an invalid sequence index in safe
1399 ** the type system is now cleverer about negations of numeric
1400 types, and consequently understands the BIGNUM and RATIO types
1402 ** the type system is now cleverer about the interaction between
1403 INTEGER and RATIO types: while bugs still remain, many more
1404 cases are accurately computed;
1405 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1406 if it is in the last clause;
1407 ** CONSTANTLY now correctly returns a side-effect-free function in
1409 ** DECLARE is no longer treated as a special-operator; in
1410 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1411 * incremented fasl file version number due to the change in the
1412 DEFSTRUCT-SLOT-DESCRIPTION structure.
1414 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1415 * minor incompatible change: code processed by the "interpreter" or
1416 EVAL now has a compilation optimization policy of (DEBUG 2)
1417 (changed from (DEBUG 1)) to improve debuggability of interactive
1418 development, and to allow the use of the debug RETURN command in
1420 * an experimental implementation of the RETURN command for the
1421 debugger has been included. (thanks to Frederik Kuivinen)
1422 * fixed bug 62: constraints were not propagated into a loop.
1423 * fixed bug in embedded calls of SORT (reported and investigated by
1425 * fixed some bugs revealed by Paul Dietz' test suite:
1426 ** printing and reading of arrays with some dimensions having
1427 length 0 (thanks to Gerd Moellmann);
1428 ** BOA constructor with &AUX argument without a default value does
1429 not cause a type error;
1430 ** CONSTANTP now returns true for all self-evaluating objects.
1432 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1433 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1434 accessors that are related by inheritance, as specified in the
1435 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1436 Valtteri Vuorikoski)
1437 * The compiler is now able to inline functions that were defined in
1438 a complex lexical environment (e.g. inside a MACROLET).
1439 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1441 * Support for the upcoming FreeBSD-5.0 release has been included.
1442 (thanks to Dag-Erling Smorgrav)
1443 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1444 effect when it is not in a toplevel context.
1445 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1446 stand a better chance of being correct. (thanks to Gerd
1448 * fixed bug in COERCE, which now signals an error on coercing a
1449 rational to a bounded real type which excludes the expected
1451 * The compiler is now able to derive types more accurately from the
1452 COERCE and COMPILE functions.
1453 * fixed bug 223: functional binding is considered to be constant
1454 only for symbols in the CL package.
1455 * fixed bug 231: SETQ did not check the type of a variable being set
1456 (reported by Robert E. Brown)
1457 * A new optimization for MAKE-INSTANCE has been included, fixing
1458 various bugs (including relating to :ALLOCATION :CLASS slots and
1459 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1461 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1462 ** As required by ANSI, LOOP now disallows anonymous collection
1463 clauses such as COLLECT I in conjunction with aggregate boolean
1464 clauses such as THEREIS (= I 1);
1465 ** LOOP now signals an error when any variable is reused in the
1466 same loop (including the potentially useful construct analogous
1467 to WITH A = 1 WITH A = (1+ A);
1468 ** IT is only a special loop symbol within the first clause of a
1469 conditional loop clause;
1470 ** LOOP with a typed iteration variable over a hashtable now
1471 signals a type error iff it should.
1472 * fixed some other bugs revealed by Paul Dietz' test suite:
1473 ** FILE-STREAM now names the class previously known as FD-STREAM;
1474 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1475 argument) no longer signals an error;
1476 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1477 of the slot symbol, rather than using the current package
1478 ((:CONC-NAME "") continues to intern the slot's name in the
1480 * incremented fasl file version number, because of the incompatible
1481 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1482 of the new implementation of DEFINE-COMPILER-MACRO.
1484 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1485 * Support for building SBCL for MIPS platforms running in
1486 little-endian mode has now been checked in, and basic
1487 functionality on said platforms verified.
1488 * minor incompatible change: PCL now records the pathname of a file
1489 in which methods and the like are defined, rather than its
1491 * minor incompatible change: TRUENAME now considers the truename of
1492 a file naming a directory to be the pathname with :DIRECTORY
1493 component indicating that directory.
1494 * minor incompatible change: a NAMED clause in the extended form of
1495 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1496 reason for the previous behaviour is unclear.
1497 * more systematization and improvement of CLOS and MOP conformance
1498 in PCL (thanks to Gerd Moellman and Pierre Mai):
1499 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1501 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1502 primary methods with no specializers;
1503 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1505 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1506 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1507 FORWARD-REFERENCED-CLASSes; error reporting on
1508 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1510 ** SXHASH on CLOS instances now uses a slot internal to the
1511 instance to return different numbers on distinct instances,
1512 while preserving the same return value through invocations of
1514 ** DEFMETHOD signals errors when methods with longer incongruent
1515 lambda lists are added to generic functions;
1516 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1517 CLASS, as specified in AMOP;
1518 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1519 on the order returned by the primary method for classes of
1520 class STANDARD-CLASS;
1521 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1522 * fixed some bugs shown by Paul Dietz' test suite:
1523 ** DOLIST puts its body in TAGBODY;
1524 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1526 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1527 value producing form;
1528 ** if more variables are given to PROGV than values, extra
1529 variables are bound and made to have no value;
1530 ** NSUBSTITUTE on list arguments gets the right answer with
1532 ** ELT signals an error of type TYPE-ERROR when the index argument
1533 is not a valid sequence index;
1534 ** LOOP signals (at macroexpansion time) an error of type
1535 PROGRAM-ERROR when duplicate variable names are found;
1536 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1537 ** FUNCALL of special-operators now cause an error of type
1539 ** PSETQ now works as required in the presence of side-effecting
1540 symbol-macro places;
1541 ** NCONC accepts any object as its last argument;
1542 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1544 ** loop-for-as-package does not require a package to be explicitely
1546 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1548 * fixed bug 166: compiler preserves "there is a way to go"
1549 invariant when deleting code.
1550 * fixed bug 172: macro lambda lists with required arguments after
1551 &REST arguments now cause an error to be signalled. (thanks to
1553 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1554 bug. (thanks to Matthew Danish)
1555 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1557 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1558 arguments are lexically rebound. (thanks to Gerd Moellmann and
1560 * fixed bug 194: error messages are now more informative when there
1561 is no primary method applicable in a call to a generic
1562 function. (thanks to Gerd Moellmann)
1563 * fixed bug in command line argument checking (thanks to Julian
1565 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1566 (thanks to Matthew Danish)
1567 * incremented fasl file version number, because of the
1568 SXHASH-related changes in the layout of CLOS data structures
1570 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1571 * minor incompatible change: The runtime (the Unix executable named
1572 "sbcl") is now much pickier about the .core files it will load.
1573 Essentially it now requires .core files to descend from the same
1574 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1575 as the runtime does. (The intent is to prevent the crashes which
1576 can occur, and which can even be reported as mysterious failures,
1577 when people patch the sources or change the build parameters
1578 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1579 sbcl and .core files.)
1580 * fixed bug: VALUES-LIST is no longer optimized away.
1581 * fixed bug 142: The FFI conversion of C string values to Lisp
1582 string values no longer conses excessively. (thanks to Nathan
1583 Froyd porting Raymond Toy's fix to CMU CL)
1584 * began to systematize and improve MOP conformance in PCL (thanks to
1585 Nathan Froyd, Gerd Moellman and Pierre Mai):
1586 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1588 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1589 ** FINALIZE-INHERITANCE is now called on class finalization;
1590 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1591 argument precedence order.
1592 * fixed bug 202: The compiler no longer fails on functions whose
1593 derived types contradict their declared type.
1594 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1595 so it can be non-toplevel.
1596 * The fasl file version number has changed (because of the new
1597 implementation of DEFMACRO).
1598 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1599 safe code, that any length requirement by their type-specifier
1600 argument is valid. The exceptions to this are described in bug
1602 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1603 functions now check the types of their inputs as required by ANSI.
1604 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1605 introduced symbol is DECLAREd to be SPECIAL.
1606 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1607 * fixed bug: :COUNT argument to sequence functions may be negative
1608 * fixed bug: body of DO-SYMBOLS may contain declarations
1609 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1610 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1611 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1612 symbol macro only once
1613 * fixed printing of call frame when argument list is unavailable
1614 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1615 * compiler no longer signals WARNING on unknown keyword
1618 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1619 * A beta-quality port to the mips architecture running Linux,
1620 based on the old CMUCL backend, has been made. It has been tested
1621 on a big-endian kernel, and works sufficiently well to be able to
1622 rebuild itself; it has not been tested in little-endian mode.
1623 * fixed an inconsistency between gencgc.c and purify.c which made
1624 dumping/loading .core files unreliable
1625 * fixed bug 120a: The compiler now deals correctly with IFs where
1626 the consequent is the same as the alternative, instead of
1627 misderiving the return type. (thanks to Alexey Dejneka)
1628 * fixed bug 113: Logical pathnames are now dumpable (the logical
1629 host is resolved at load-time, throwing an error if it is not
1631 * fixed bug 174: FORMAT's error message is slightly clearer when a
1632 non-printing character is used in a format directive.
1633 * fixed several bugs in compiler checking of type declarations, i.e.
1634 violations of the Python "declarations are assertions" principle
1635 (thanks to Alexey Dejneka)
1636 * fixed several bugs in PCL's error checking (thanks to Gerd
1638 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1640 * fixed bug in compilation of functions as first class values
1641 (thanks to Antonio Martinez-Shotton)
1642 * The compiler's handling TYPE-ERRORs which it can prove will
1643 inevitably happen at runtime has been cleaned up and corrected
1644 in several ways. (thanks to Alexey Dejneka)
1645 * improved argument type checking for various basic arithmetic
1646 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1647 lavished on them in the past that they can be compiled in many
1648 ways in different special cases
1649 * fixed bug 181: compiler checks validity of user supplied type
1651 * cleaned up code flushing in optimization: Function calls which
1652 should signal errors for safety purposes (e.g. which ANSI says
1653 should signal errors when their arguments are of incorrect type)
1654 are no longer optimized away.
1655 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1656 * incremented fasl file version number, because changes in the
1657 implementation of sequence functions like COERCE caused
1658 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1659 in old inline expansions) to become undefined. (Actually these
1660 changes were later undone, so we might very well be binary
1661 compatible with 0.7.7 after all, but leaving the version number
1662 incremented seemed like the simplest and most conservative
1665 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1666 * An alpha-quality port to the parisc architecture running Linux,
1667 based on the old CMUCL backend, has been made. This, even more so
1668 than the other backends, should be considered still a work in
1669 progress; known problems include that the Linux kernel in 64-bit
1670 mode does not propagate the correct sigcontext structure to
1671 userspace, and consequently SBCL on a parisc64 kernel will not
1673 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1674 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1675 suppressed.) Also now that the compiler is looking at declarations
1676 in the environment, it checks optimization declarations as well,
1677 and suppresses inlining when (> DEBUG SPEED).
1678 * More fixes have been made to treatment of floating point exception
1679 treatment and other Unix signals. In particular, floating point
1680 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1681 * The detection and handling of control stack exhaustion (infinite
1682 or very deeply nested recursion) has changed. Stack exhaustion
1683 detection is now done by write-protecting pages at the OS level
1684 and applies at all optimization settings; when found, a
1685 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1686 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1687 be caught using IGNORE-ERRORS.
1688 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1689 correctly when transforming recursive LABELS functions to LETs.
1690 (thanks to Alexey Dejneka)
1691 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1692 that are names of constants or global variables.
1693 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1694 alien routines with docstrings.
1695 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1696 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1698 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1699 (thanks to Eric Marsden)
1700 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1701 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1702 * Bug fix: The SPARC backend can now compile functions involving
1703 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1704 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1705 to a functions expecting a PATHNAME-DESIGNATOR.
1706 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1707 lambda lists. (thanks to Alexey Dejneka)
1708 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1709 functions so that EXPORTing the name of the class doesn't cause
1710 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1711 to Antonio Martinez for reporting this)
1712 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1713 constant-folded without failing an assertion. (thanks to Einar
1714 Floystad Dorum for reporting this)
1715 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1716 types are now dealt with more correctly by the compiler.
1717 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1718 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1719 behaviour analogous to RENAME-FILE. This puts its behaviour more
1720 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1722 * The fasl file version number has changed again. (because of the
1723 bug fix involving the names of PCL MAKE-INSTANCE functions)
1725 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1726 * bug fix: Floating point exceptions are treated much more
1727 consistently on the x86/Linux and PPC/Linux platforms.
1728 * Array initialization with :INITIAL-ELEMENT is now much faster for
1729 cases when the compiler cannot open code the array creation, but
1730 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1731 array accesses have also seen a speed increase.
1732 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1733 specified. (This was at the root of some bad interactions between
1734 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1736 * bug fix: Internal error arguments for undefined functions are now
1737 computed correctly on the PPC/Linux platform.
1738 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1739 Raymond Toy's patch for CMU CL)
1740 * Support for the Solaris 9 operating environment has been included
1741 (thanks to Daniel Merritt)
1742 * A very ugly but hopefully complete draft of the missing FFI chapter
1743 of the manual has been created by reformatting the corresponding
1744 CMU CL manual chapter into (currently very ugly and incoherent)
1745 DocBook and bringing it up to date for SBCL behavior. Thus, the
1746 manual is now essentially complete, at least by my extreme
1747 once-and-only-once standards, whereby it's acceptable to refer to
1748 the doc strings of SB-EXT functions as the primary documentation.
1749 * The fasl file version number has changed again, due to cleanup of
1750 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1751 is no longer a static symbol.)
1753 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1754 * SBCL now builds with OpenMCL (version 0.12) as the
1755 cross-compilation host; also, more progress has been made toward
1756 bootstrapping under CLISP.
1757 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1759 * bug 158 fixed: The compiler can now deal with integer loop
1760 increments different from 1; fixing this turned out also to fix
1762 * bug 169 fixed: no more bogus warnings about using lexical bindings
1763 despite the presence of perfectly good SPECIAL declarations (thanks
1764 to David Lichteblau)
1765 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1766 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1767 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1769 * bug 180 fixed: Method combination specifications no longer ignore
1770 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1771 * bug fix: Structure type predicate functions now check their argument
1772 count as they should.
1773 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1774 correctly. (thanks to Pierre Mai)
1775 * minor incompatible change: The --noprogrammer option is deprecated
1776 in favor of the new --disable-debugger option, which is very similar.
1777 (The major difference is that it takes effect at a slightly different
1778 time at startup, causing handling of errors in --sysinit and
1779 --userinit files will be affected differently.) The
1780 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1781 been added to allow this functionality to be controlled from ordinary
1782 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1783 maintainers, who might want to run non-interactive scripts to
1784 build SBCL cores which will later be used interactively.)
1785 * minor incompatible change: The LOAD function no longer, when given
1786 a wild pathname to load, loads all files matching that pathname.
1787 Instead, an error of type FILE-ERROR is signalled.
1789 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1790 * bug 147 fixed: The compiler preserves its block link/count
1791 invariants more correctly now so that it doesn't crash. (thanks
1793 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1795 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1796 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1797 no-value-for-this-character as NIL (as specified by ANSI).
1798 * HOST-NAMESTRING on physical pathnames now returns a string that is
1799 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1800 (thanks to Christophe Rhodes)
1801 * The Alpha port handles icache flushing more correctly. (thanks to
1803 * More progress has been made toward bootstrapping under CLISP. (thanks
1804 to Christophe Rhodes)
1805 * The fasl file format has changed again, because dynamic loading
1806 on OpenBSD (which has non-ELF object files) motivated some cleanups
1807 in the way that foreign symbols are transformed and passed around.
1808 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1809 is no longer treated as whitespace by the reader, but instead as
1810 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1811 |A
\7fB|, instead of A as it used to.
1813 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1814 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1815 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1816 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1817 this as of 0.7.1.45, but was left out of the previous news section
1818 (thanks to Dan Barlow)
1819 * SBCL now runs on the Solaris operating system on SPARC architectures
1820 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1821 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1822 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1823 * SPARC backend cleanups, allowing builds of cores optimized for V8
1824 and V9 SPARCS, and also emission of code targeted to a particular
1825 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1827 * SBCL is closer to bootstrapping under CLISP, thanks to various
1828 fixes by Christophe Rhodes.
1829 * The fasl file format has changed again, to allow the compiler's
1830 INFO database to support symbol macros.
1831 * The user manual (in doc/) is formatted into HTML more nicely.
1832 (thanks to coreythomas)
1833 * The system is smarter about SUBTYPEP relationships, especially
1834 those involving NOT types (including types such as ATOM which are
1835 represented internally using NOT types). Thus SUBTYPEP is less
1836 likely to return (VALUES NIL NIL) in general, and in particular
1837 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1839 * The fasl file format has changed again, because the internal
1840 representation of types now includes a new slot to support the new
1841 SUBTYPEP-of-NOT-types logic.
1842 * (not a change in the main branch of SBCL, but a related prototype
1843 which can hopefully be merged into the main branch of SBCL in the
1844 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1845 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1846 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1847 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1848 dispatch macro character combinations. (thanks to Alexey Dejneka)
1849 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1850 that superficially look like logical namestrings correctly.
1851 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1853 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1854 default supplied-p) for &optional and &key arguments. (thanks to
1857 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1858 * incompatible change: The compiler is now less aggressive about
1859 tail call optimization, doing it only when (> SPACE DEBUG) or
1860 (> SPEED DEBUG). (This is an incompatible change because there are
1861 programs which relied on the old CMU-CL-style behavior to optimize
1862 away their unbounded recursion which will now die of stack overflow.)
1863 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1864 for non-GENCGC systems has been increased to 20M (since that
1865 seems much closer to the likely performance optimum for modern
1866 systems than the old 4M value was)
1867 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1868 and *DEBUG-PRINT-LEVEL*
1869 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1870 of CMU CL's support for SPARC, and various endianness and other
1871 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1872 * new syntactic sugar for the Unix command line: --load foo.bar is now
1873 an alternate notation for --eval '(load "foo.bar")'.
1875 ** The system now detects stack overflow and handles it gracefully,
1876 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1877 optimization settings. (This is a good thing in general, and
1878 its introduction in this version should be particularly timely
1879 for anyone whose code fails because of suppression of tail
1881 ** The system now hunts for the C variable "environ" in a more
1882 devious way, to avoid segfaults when the C library version differs
1883 between compile time and run time. (thanks to Christophe Rhodes)
1884 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1885 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1886 ** The compiler no longer issues bogus style warnings for undefined
1887 classes in the same source file as the DEFCLASSes which defined
1888 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1890 ** fixes in CONDITION class precedence list for undefined function
1891 errors (thanks to Alexei Dejneka)
1892 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1893 correctly. (thanks to Dan Barlow)
1894 ** portability fixes aiming at bootstrapping under CLISP (thanks
1895 to Dave McDonald and Christophe Rhodes)
1896 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1897 ** fixes in type translation and and type inference (thanks to
1899 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1900 ** various fixes in the new ports (thanks to Dan Barlow)
1901 * several changes related to debugging:
1902 ** suppression of tail recursion, as noted above
1903 ** stack overflow detection, as noted above
1904 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1905 is now the default. (For some time encapsulation has been more
1906 reliable than the breakpoint-based :ENCAPSULATE NIL
1907 implementation, at least on X86 systems; and I just noticed that
1908 encapsulation also seems closer to the spirit of the ANSI
1911 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1913 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1914 up properly again. (There was a packaging bug in 0.7.0 which
1915 left their definitions in SB-SYS::LOAD-FOREIGN and
1916 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1917 vital for most things which interface to C-level interfaces,
1918 like extensions working with sockets or databases or
1919 Perl-compatible regexes or whatever, and the need to fix
1920 this bug is the main reason that 0.7.1 was released so
1922 ** DEFGENERIC is now choosier about the methods it redefines, so that
1923 reLOADing a previously-LOADed file containing DEFGENERICs does
1924 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
1925 cycle now works as it should. (thanks to Alexey Dejneka)
1926 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
1927 fixed by Christophe Rhodes. (He actually submitted this patch
1928 months ago, and I delayed until after 0.7.0.)
1929 ** Bug 111 (internal compiler confusion about runtime checks on
1930 FUNCTION types) was fixed by Alexey Dejneka.
1931 * Some internal cleanups (getting rid of variables which aren't
1932 needed now that the byte interpreter is gone) caused the fasl
1933 file format number to change again.
1935 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
1936 * major incompatible change: The default fasl file extension, i.e. the
1937 default extension for files produced by COMPILE-FILE, has changed
1938 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
1940 ** There are many changes in the implementation of the compiler.
1941 SBCL is now essentially a compiler-only implementation of ANSI
1942 Common Lisp. EVAL still "interprets" a few special cases, but
1943 almost all the interesting cases are handled by creating
1944 a LAMBDA expression, calling COMPILE on it, then calling
1945 FUNCALL on the result.
1946 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
1947 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
1948 Since the code is newer, there might still be some new bugs
1949 (though not as many as before Martin Atzmueller's fixes:-). But
1950 the new code is substantially simpler and clearer, and hopefully
1951 any remaining bugs will be simpler, less fundamental, and more
1952 fixable then the bugs in the old code.
1953 ** The revised compiler is still a little unsteady on its feet.
1955 *** The debugging information it produces (particularly the names
1956 of FUNCTION objects) is sometimes much less useful than what
1957 the old compiler produced.
1958 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
1959 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
1960 within a MACROLET) has been temporarily weakened.
1961 ** There are new compiler optimizations for various functions:
1962 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
1963 FIND-IF-NOT, POSITION-IF-NOT, and FILL
1964 *** the math functions TRUNCATE, FLOOR, and CEILING
1965 *** the function-of-all-trades COERCE
1966 Mostly these should be transparent, but there's one
1967 potentially-annoying problem (bug 117): when the compiler
1968 inline-expands a function and does type analysis on the result,
1969 it can create control paths which have type mismatches, and
1970 when it can't prove that those control paths aren't taken,
1971 it will issue WARNINGs about the type mismatches. This is
1972 a particular problem in practice for the new sequence functions.
1973 It's not clear how this should be fixed, and for now, a
1974 workaround is given in the entry for 117 in the BUGS file.
1975 ** (Because of the interaction between the two previous items --
1976 occasional inlining problems and new inline expansions -- some
1977 of the new sequence function optimizations won't really kick in
1978 completely until debugging information, and then inlining, are
1979 straightened out in some future version.)
1980 * minor incompatible changes:
1981 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
1982 DIRECTORY no longer implicitly promotes NIL slots of its
1983 pathname argument to :WILD. In particular, when you ask for the
1984 contents of a directory (which you used to be able to do without
1985 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
1986 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1987 ** changes in behavior that ANSI explicitly defines to be
1988 implementation dependent:
1989 *** The new compiler-only implementation still conforms with ANSI,
1990 but acts a little different than before. Besides the obvious
1991 changes in performance tradeoffs (that the cost per form passed
1992 to EVAL has gone up, and the cost per form executed by EVAL
1993 has gone down), the behavior of the system changes a little
1994 because there are no longer any interpreted function objects.
1995 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
1996 e.g. doing COMPILE on the output of interactive DEFUN is
1998 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2000 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2001 has changed from (:CL) to NIL.
2002 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2003 e.g. #\Newline and #\Tab, don't have names specified in the
2004 ANSI Common Lisp standard, is now based on their ASCII symbolic
2005 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2006 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2007 are no longer used for output.
2008 ** changes in internal implementation constants:
2009 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2010 4 million. (If your application spends a lot of time GCing and
2011 you have a lot of RAM, you might want to experiment with
2012 increasing it even more.)
2013 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2014 However, almost all old code should still continue to work without
2015 immediate update, as SB-C-CALL is now a (deprecated) nickname
2017 ** Old operator names in the style DEF-FOO are now deprecated in
2018 favor of new corresponding names DEFINE-FOO, for consistency with
2019 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2020 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2021 internal symbols, but a few supported extensions like
2022 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2023 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2024 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2025 etc. as you get deeper into recursive calls to the debugger
2026 command loop, instead of the old "5]", "5]]", "5]]]"
2027 sequence. (I was motivated to do this when squabbles between
2028 ILISP and SBCL left me very deeply nested in the debugger. In the
2029 short term, this change will probably provoke more ILISP/SBCL
2030 squabbles, but hopefully it will be an improvement in the long run.)
2031 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2032 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2033 function, since declaiming return types involving aliens is
2034 (1) annoyingly messy to do by hand and (2) vital to efficient
2035 compilation of code which calls such functions.
2036 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2037 longer reexported by the SB-EXT package. They're solely useful
2038 for alien code, so it seems more logical that you should get
2039 them from the SB-ALIEN package, not in SB-EXT.
2040 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2041 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2042 features. Instead, the code that they used to control is always
2043 built into the system.
2044 * many other bug fixes
2045 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2046 advantage of the new EVAL-WHEN stuff and to clean them up in
2047 general, and they are now more ANSI-compliant in a number of
2048 ways. Martin Atzmueller is responsible for a lot of this.
2049 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2051 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2052 *** correct ERROR type for various file operations
2053 *** some fixes for Lisp streams
2054 *** DEFMETHOD syntax checking
2055 *** changing old weird representation of debug information as
2056 strings (which, among their other deficiencies, don't transform
2057 correctly when you rename packages, and don't change their
2058 print representation when you change things like *PACKAGE*
2059 and *PRINT-LENGTH*) to symbols and lists of symbols
2060 He also made several improvements and fixed several bugs in DESCRIBE.
2061 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2063 *** misbehavior of WRITE-STRING/WRITE-LINE
2064 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2065 and several other LOOP problems as well
2066 *** DIRECTORY when similar filenames are present
2067 *** DEFGENERIC with :METHOD options
2068 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2069 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2070 *** argument ordering in FIND with :TEST option
2071 *** mishandled package designator argument in APROPOS-LIST
2072 *** various problems in the backquote readmacro
2073 *** a bug in APROPOS
2074 *** probably some others that I'm not describing very well here,
2075 since the CVS log documents them by reference to sbcl-devel
2076 messages, and the SourceForge archives aren't working well.:-(
2077 ** Dan Barlow improved the Alpha port (and is making progress on the
2078 PPC port, for those of you who think different).
2079 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2080 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2081 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2082 continues to work on the SPARC port (for those of you in a position
2083 to look down upon our little PC-compatible boxes from a great height).
2084 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2085 and uses that copy, rather than the current dynamic value, when
2086 it's trying to decide whether to truncate output. Thus e.g.
2087 (let ((*print-lines* 50))
2088 (pprint-logical-block (stream nil)
2090 (let ((*print-lines* 8))
2091 (print (aref possiblybigthings i) stream)))))
2092 should now truncate the logical block only at 50 lines, instead of
2093 often truncating it at 8 lines, as it did before.
2094 * The doc/cmucl/ directory, containing old CMU CL documentation
2095 from the time of the fork, is no longer part of the base system.
2096 SourceForge has shut down its anonymous FTP service, and with it
2097 my original plan for distributing the old CMU CL documentation
2098 there. For now, if you need these files you can download an old
2099 SBCL source release and extract them from it.
2100 * The fasl file version number changed again, for dozens of reasons,
2101 some of which are apparent above.
2103 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2104 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2105 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2106 patch, so that DISASSEMBLE output is much nicer.
2107 * The code in the SB-PROFILE package now seems reasonably stable.
2108 I still haven't decided what the final interface should look like
2109 (I'd like PROFILE to interact cleanly with TRACE, since both
2110 facilities use function encapsulation) but if you have a need
2111 for profiling now, you can probably use it successfully with
2112 the current CMU-CL-style interface.
2113 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2114 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2115 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2116 link now returns the dangling link itself, and for similar
2117 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2118 link itself. (In these cases the old code signalled an error and
2119 looped endlessly, respectively.) Thus, DIRECTORY now works even
2120 in the presence of dangling and cyclic symbolic links.
2121 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2122 is now a supported extension again, since the consensus on
2123 sbcl-devel was that it can be useful for ordinary development
2124 work, not just for debugging SBCL itself.
2125 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2126 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2127 that currently-defined functions might be redefined later with
2128 different return types.
2129 * Hash tables can be printed readably, as inspired by CMU CL code
2130 of Eric Marsden and SBCL code of Martin Atzmueller.
2131 * better error handling in CLOS method combination, thanks to
2132 Martin Atzmueller porting Pierre Mai's CMU CL patches
2133 * more overflow fixes for >16Mbyte I/O buffers
2134 * A bug in READ has been fixed, so that now a single Ctrl-D
2135 character suffices to cause end-of-file on character streams.
2136 In particular, now you only need one Ctrl-D at the command
2137 line (not two) to exit SBCL.
2138 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2140 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2141 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2142 does the right thing.
2143 * The implementation of some type tests, especially for CONDITION
2144 types, is now tidier and maybe faster, due to CMU CL code
2145 originally by Douglas Crosher, ported by Martin Atzmueller.
2146 * Some math functions have been fixed, and there are new
2147 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2148 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2149 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2150 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2151 likely to be built into sbcl-0.7.0.)
2152 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2153 CL, ported to SBCL by Martin Atzmueller)
2154 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2155 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2156 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2157 * There's a new slam.sh hack to shorten the edit/compile/debug
2158 cycle for low-level changes to SBCL itself, and a new
2159 :SB-AFTER-XC-CORE target feature to control the generation of
2160 the after-xc.core file needed by slam.sh.
2161 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2162 COMPILE-FILE is no longer supported, so that now every function
2163 gets an entry point, so that block compilation looks a little
2164 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2165 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2166 returns the number of bytes consed since the system started,
2167 rather than the number consed since the first time the function
2168 was called. (The new definition parallels ANSI functions like
2169 CL:GET-INTERNAL-RUN-TIME.)
2170 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2171 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2172 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2173 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2174 * Fasl file version numbers are now independent of the target CPU,
2175 since historically most system changes which required version
2176 number changes have affected all CPUs equally. Similarly,
2177 the byte fasl file version is now equal to the ordinary
2180 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2181 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2182 is no longer recognized. I apologize for this, because it was
2183 listed in SB-EXT as a supported extension, but I found that
2184 its existing behavior was poorly specified, as well as incorrectly
2185 specified, and it looked like too much of a mess to straighten it
2186 out. I have enough on my hands trying to get ANSI stuff to work..
2187 * many patches ported from CMU CL by Martin Atzmueller, with
2188 half a dozen bug fixes in pretty-printing and the debugger, and
2189 half a dozen others elsewhere
2190 * fixed bug 13: Floating point infinities are now supported again.
2191 They might still be a little bit flaky, but thanks to bug reports
2192 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2193 as flaky as they were.
2194 * The --noprogrammer command line option is now supported. (Its
2195 behavior is slightly different in detail from what the old man
2196 page claimed it would do, but it's still appropriate under the
2197 same circumstances that the man page talks about.)
2198 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2199 are now supported, and enabled by default. Thus, the compiler can
2200 handle many floating point and complex operations much less
2201 inefficiently. (Thus e.g. you can implement a complex FFT
2203 * The compiler now detects type mismatches between DECLAIM FTYPE
2204 and DEFUN better, and implements CHECK-TYPE more correctly, and
2205 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2206 thanks to patches from Martin Atzmueller.
2207 * various fixes to make the cross-compiler more portable to
2208 ANSI-conforming-but-different cross-compilation hosts (notably
2209 Lispworks for Windows, following bug reports from Arthur Lemmens)
2210 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2211 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2212 fixed thanks to Pierre Mai's CMU CL patches.
2213 * fixes to keep the system from overflowing internal counters when
2214 it tries to use i/o buffers larger than 16M bytes
2215 * fixed bug 45a: Various internal functions required to support
2216 complex special functions have been merged from CMU CL sources.
2217 (When I was first setting up SBCL, I misunderstood a compile-time
2218 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2219 * improved support for type intersection and union, fixing bug 12
2220 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2221 more obscure bugs as well
2222 * some steps toward byte-compiling non-performance-critical
2223 parts of the system, courtesy of patches from Martin Atzmueller
2224 * Christophe Rhodes has made some debian packages of sbcl at
2225 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2226 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2227 stable, but are nonetheless usable. When he's ready, I'd be happy
2228 to add them to the SourceForge "File Releases" section. (And if
2229 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2230 * new fasl file format version number (because of changes in
2231 internal representation of (OR ..) types to accommodate the new
2232 support for (AND ..) types, among other things)
2234 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2235 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2237 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2239 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2240 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2241 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2242 * DESCRIBE now works on structure objects again.
2243 * Most function call argument type mismatches are now handled as
2244 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2245 know whether the function will be redefined before the call is
2246 executed. (The compiler could flag local calls with full WARNINGs,
2247 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2248 it doesn't keep track of enough information to know whether calls
2249 are local in this sense.)
2250 * Compiler output is now more verbose, with messages truncated
2251 later than before. (There should be some supported way for users
2252 to override the default verbosity, but I haven't decided how to
2253 provide it yet, so this behavior is still controlled by the internal
2254 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2255 src/compiler/ir1util.lisp.)
2256 * Fasl file format version numbers have increased again, because
2257 support for the Gray streams extension changes the layout of the
2258 system's STREAM objects.
2259 * The Gray subclassable streams extension now works, thanks to a
2260 patch from Martin Atzmueller.
2261 * The full LOAD-FOREIGN extension (not just the primitive
2262 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2263 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2264 but like most other programs, it defaults to copying the Unix
2265 environment from the original process instead of starting the
2266 new process in an empty environment.
2267 * Extensions which manipulate the Unix environment now support
2268 an :ENVIRONMENT keyword option which doesn't smash case or
2269 do other bad things. The CMU-CL-style :ENV option is retained
2270 for porting convenience.
2271 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2272 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2274 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2276 * A patch from Martin Atzmueller seems to have solved the SIGINT
2277 problem, and as far as we know, signal-handling now works cleanly.
2278 (If you find any new bugs, please report them!)
2279 * The system no longer defaults Lisp source file names to types
2280 ".l", ".cl", or ".lsp", but only to ".lisp".
2281 * The compiler no longer uses special default file extensions for
2282 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2283 seems to expect a single default extension for all compiled code,
2284 and there's no compelling reason to try to stretch the standard
2285 to allow two different extensions.) Instead, byte-compiled files
2286 default to the same extension as native-compiled files.
2287 * Fasl file format version numbers have increased again, because
2288 a rearrangement of internal implementation packages made some
2289 dumped symbols in old fasl files unreadable in new cores.
2290 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2291 many fewer weird special cases.
2292 * Bug #17 (differing COMPILE-FILE behavior between logical and
2293 physical pathnames) has been fixed, and some related misbehavior too,
2294 thanks to a patch from Martin Atzmueller.
2295 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2296 by Tim Moore, ported to SBCL by Martin Atzmueller.
2297 * Martin Atzmueller fixed several filesystem-related problems,
2298 including bug #36, in part by porting CMU CL patches, which were
2299 written in part by Paul Werkowski.
2300 * More compiler warnings in src/runtime/ are gone, thanks to
2301 more patches from Martin Atzmueller.
2302 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2305 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2307 * DESCRIBE now works on CONDITION objects.
2308 * The debugger now handles errors which arise when trying to print
2309 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2311 * The build system now uses an additional file, customize-target-features.lisp,
2312 to allow local modifications to the target *FEATURES* list. (The point of
2313 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2314 debugging features enabled, without having to worry about propagating it
2315 into everyone's system when I do a "cvs update".) When no
2316 customize-target-features.lisp file exists, the target *FEATURES* list
2317 should be constructed the same way as before.
2318 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2319 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2320 warn about behavior which is undefined under the ANSI spec. Note: This
2321 is specified by ANSI, but it's not very popular with programmers.
2322 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2323 macro in the SBCL sources for an example of a workaround which you
2324 might use to make portable ANSI-standard code which does what you want.
2325 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2326 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2327 when it's used as a non-toplevel form. (This is required in order
2328 to implement the DEFCONSTANT-EQX macro.)
2329 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2330 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2331 funny things, due to bugs in the implementation of EVAL-WHEN
2332 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2333 non-toplevel DEFCONSTANTs are uncommon.)
2334 * The core file version number and fasl file version number have been
2335 incremented, because the old noncompliant DEFCONSTANT behavior involved
2336 calling functions which no longer exist, and because I also took the
2337 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2338 * fixed bug 1 (error handling before read-eval-print loop starts), and
2339 redid debugger restarts and related debugger commands somewhat while
2341 ** The QUIT debugger command is gone, since it did something
2342 rather different than the SB-EXT:QUIT command, and since it never
2343 worked properly outside the main toplevel read/eval/print loop.
2344 Invoking the new TOPLEVEL restart provides the same functionality.
2345 ** The GO debugger command is also gone, since you can just invoke
2346 the CONTINUE restart directly instead.
2347 ** The TOP debugger command is also gone, since it's redundant with the
2348 FRAME 0 command, and since it interfered with abbreviations for the
2350 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2352 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2353 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2354 some porting work by Martin Atzmueller)
2355 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2356 with a boatload of patches which he ported to SBCL. Now that those
2358 ** The system tries to make sure that its low-priority messages
2359 are prefixed by semicolons, to help people who like to use
2360 syntax highlighting in their ILISP buffer. (This patch
2361 was originally due to Raymond Toy.)
2362 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2363 patches originally written by Raymond Toy.
2364 ** The compiler understands coercion between single-value and
2365 multiple-VALUES type expressions better, getting rid of some very
2366 weird behavior, thanks to patches originally by Robert MacLachlan
2367 and Douglas Crosher.
2368 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2369 lambda lists, thanks to a patch originally by Pierre Mai.
2370 ** The system no longer bogusly warns about "abbreviated type
2372 ** The compiler gets less confused by inlining and RETURN-FROM,
2373 thanks to some patches originally by Tim Moore.
2374 ** The system no longer hangs when dumping circular lists to fasl
2375 files, thanks to a patch originally from Douglas Crosher.
2376 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2377 undefined function error.
2378 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2379 in structure slot initforms) for now, documented workaround instead:-|
2380 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2381 * fixed bug 5: added stubs for various Gray stream functions called
2382 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2383 installed, at least appropriate type errors are generated
2384 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2385 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2386 consistently in DEFMETHOD forms.
2387 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2388 it doesn't seem to affect SBCL after all
2389 * The C runtime system now builds with better optimization and many
2390 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2392 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2394 * The system is now under CVS at SourceForge (instead of the
2395 CVS repository on my home machine).
2396 * The new signal handling code has been tweaked to treat register
2397 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2398 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2399 downstream, e.g. in debug-int.lisp, has implicit dependencies
2400 on the unsignedness of integer representation of machine words,
2401 and that caused the system to bomb out with infinite regress
2402 when trying to recover from type errors involving signed values,
2403 e.g. (BUTLAST '(1 2 3) -1).)
2404 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2405 (This was one of the bugs Peter Van Eynde reported back in July.)
2406 * The system now uses code inspired by Colin Walters' O(N)
2407 implementation of MAP (from the cmucl-imp@cons.org mailing
2408 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2409 inline the MAP operation, and there is more than one
2410 sequence argument to the MAP call (so that it can't just
2411 do ETYPECASE once and for all based on the type of the
2412 single sequence argument). (The old non-inline implementation
2413 of the general M-argument sequence-of-length-N case required
2414 O(M*N*N) time when any of the sequence arguments were LISTs.)
2415 * The QUIT :UNIX-CODE keyword argument has been renamed to
2416 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2417 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2419 * Raymond Wiker's patches to port dynamic loading from Linux to
2420 FreeBSD have been added.
2421 * The BUGS file is now more nearly up to date, thanks in large part
2422 to Martin Atzmueller's review of it.
2423 * The debugger now flushes standard output streams before it begins
2424 its output ("debugger invoked" and so forth).
2425 * The core version number and fasl file version number have both
2426 been incremented, because of incompatible changes in the layout
2428 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2429 used to not be called for a saved Lisp image.)
2430 * Martin Atzmueller's version of a patch to fix a compiler crash,
2431 as posted on sbcl-devel 13 September 2000, has been installed.
2432 * Instead of installing Martin Atzmueller's patch for the
2433 compiler transform for SUBSEQ, I deleted the compiler transform,
2434 and transforms for some similar consing operations.
2435 * A bug in signal handling which kept TRACE from working on OpenBSD
2437 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2438 away by constant folding
2439 * The system now defines its address space constants in one place
2440 (in the Lisp sources), and propagates them automatically elsewhere
2441 (through GENESIS and the sbcl.h file). Therefore, patching the
2442 address map is less unnecessarily tedious and error-prone. The
2443 Lisp names of address space constants have also been systematized.
2444 * CVS tags like dollar-Header-dollar have been removed from
2445 the sources, because they have never saved me trouble and
2446 they've been source of trouble working with patches and other
2447 diff-related operations.
2448 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2449 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2451 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2453 * The system has been ported to OpenBSD.
2454 * The system now compiles with a simple "sh make.sh" on the systems
2455 that it's supported on. I.e., now you no longer need to tweak
2456 text in the target-features.lisp-expr and symlinks in src/runtime/
2457 by hand, the make.sh takes care of it for you.
2458 * The system is no longer so grossly inefficient when compiling code
2459 involving vectors implemented as general (not simple) vectors (VECTOR T),
2460 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2461 which dares to use the various sequence functions on non-simple
2462 vectors, takes less of a performance hit.
2463 * There is now a primitive type predicate VECTOR-T-P
2464 to test for the (VECTOR T) type, so that e.g.
2465 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2466 can now be compiled with some semblance of efficiency. (The old code
2467 turned the type declaration into a full call to %TYPEP at runtime!)
2468 * AREF on (VECTOR T) is still not fast, since it's still compiled
2469 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2470 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2471 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2472 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2473 without making full calls to SUBTYPEP at runtime.
2474 (Some analogous efficiency issues for non-simple vectors specialized to
2475 element types other than T, or for non-simple multidimensional arrays,
2476 have not been addressed. They could almost certainly be handled the
2477 same way if anyone is motivated to do so.)
2478 * The changes in array handling break binary compatibility, so
2479 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2480 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2481 instead of (VALUES T T).
2482 * By following the instructions that Dan Barlow posted to sbcl-devel
2483 on 2 July 2000, I was able to enable primitive dynamic object
2484 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2485 functionality is not implemented (since it calls ld to resolve
2486 library references automatically, requiring RUN-PROGRAM for its
2487 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2488 try to resolve library references) is now supported.
2489 * The system now flushes the standard output streams when it terminates,
2490 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2491 them at several other probably-convenient times, e.g. in each pass of
2492 the toplevel read-eval-print loop, and after evaluating a form given
2493 as an "--eval" command-line option. (These changes were motivated by a
2494 discussion of stream flushing issues on cmucl-imp in August 2000.)
2495 * The source transform for TYPEP of array types no longer assumes
2496 that an array whose element type is a not-yet-defined type
2497 is implemented as an array of T, but instead punts, so that the
2498 type will be interpreted at runtime.
2499 * There is now some support for cross-compiling in make.sh: each of
2500 the phases of make.sh has its own script. (This should be transparent
2501 to people doing ordinary, non-cross-compile builds.)
2502 * Since my laptop doesn't have hundreds of megabytes of memory like
2503 my desktop machine, I became more motivated to do some items on
2504 my to-do list in order to reduce the size of the system a little:
2505 ** Arrange for various needed-only-at-cold-init things to be
2506 uninterned after cold init. To support this, those things have
2507 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2508 symbols with such names are now uninterned after cold init).
2509 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2510 fasl files for cold load.
2511 ** Remove the old compiler structure pooling code (which used to
2512 be conditional on the target feature :SB-ALLOC) completely.
2513 ** Redo the representation of some data in cold init to be more compact.
2514 (I also looked into supporting byte compiled code at bootstrap time,
2515 which would probably reduce the size of the system a lot, but that
2516 looked too complicated, so I punted for now.)
2517 * The maximum signal nesting depth in the src/runtime/ support code has
2518 been reduced from 4096 to 256. (I don't know any reason for the very
2519 large old value. If the new smaller value turns out to break something,
2520 I'll probably just bump it back up.)
2521 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2523 * Many, many bugs reported by Peter Van Eynde have been added to
2524 the BUGS list; some have even been fixed.
2525 * While enabling dynamic object file loading, I tried to make the
2526 code easier to understand, renaming various functions and variables
2527 with less ambiguous names, and changing some function calling
2528 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2529 * While trying to figure out how to do the OpenBSD port, I tried to
2530 clean up some of the code in src/runtime/. In particular, I dropped
2531 support for non-POSIX signal handling, added various comments,
2532 tweaked the code to reduce the number of compilation warnings, and
2533 renamed some files to increase consistency.
2534 * To support the new automatic configuration functionality in make.sh,
2535 the source file target-features.lisp-expr has been replaced with the
2536 source file base-target-features.lisp-expr and the machine-generated
2537 file local-target-features.lisp-expr.
2538 * fixed a stupid quoting error in make.sh so that using CMU CL
2539 "lisp -batch" as cross-compilation host works again
2541 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2543 * DESCRIBE no longer tries to call itself recursively to describe
2544 bound/fbound values, so that it no longer fails on symbols which are
2545 bound to themselves (like keywords, T, and NIL).
2546 * DESCRIBE now works on generic functions.
2547 * The printer now prints less-screwed-up representations of closures
2548 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2549 * A private symbol is used instead of the :EMPTY keyword previously
2550 used to mark empty slots in hash tables. Thus
2551 (DEFVAR *HT* (MAKE-HASH-TABLE))
2552 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2553 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2554 now does what ANSI says that it should. (You can still get
2555 similar noncompliant behavior if bang on the hash table
2556 implementation with all the symbols you get back from
2557 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2558 This breaks binary compatibility, since tests for equality to
2559 :EMPTY are wired into things like the macroexpansion of
2560 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2562 * There's now a minimal placeholder implementation for CL:STEP,
2563 as required by ANSI.
2564 * An obscure bug in the interaction of the normal compiler, the byte
2565 compiler, inlining, and structure predicates has been patched
2566 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2567 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2569 * Missing ordinary arguments in a macro call are now detected even
2570 when the macro lambda list contains &KEY or &REST.
2571 * The debugger no longer complains about encountering the top of the
2572 stack when you type "FRAME 0" to explicitly instruct it to go to
2573 the top of the stack. And it now prints the frame you request even
2574 if it's the current frame (instead of saying "You are here.").
2575 * As specified by ANSI, the system now always prints keywords
2576 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2578 * The default initial SIZE of HASH-TABLEs is now smaller.
2579 * Type information from CLOS class dispatch is now propagated
2580 into DEFMETHOD bodies, so that e.g.
2581 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2583 is now basically equivalent to
2584 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2585 (DECLARE (TYPE SINGLE-FLOAT X))
2587 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2588 operation, without having to do run-time type dispatch.
2589 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2590 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2591 * Since I don't seem to be making big file reorganizations very often
2592 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2593 is overflowing my ability to conveniently back them up), I've finally
2594 checked the system into CVS. (The CVS repository is on my home system,
2595 not at SourceForge -- putting it on SourceForge might come later.)
2596 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2597 high-level GC-NOTIFY-FOO functions send their output. (There's
2598 still very little control of where low-level verbose GC functions
2599 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2600 less than it used to -- the GC-NOTIFY-FOO functions are now under
2601 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2602 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2603 in only one place in the source code, and propagates it automatically
2604 everywhere that it's needed. Thus e.g. when I bump the version from
2605 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2606 * The C source files now include boilerplate legalese and documentation
2607 at the head of each file (just as the Lisp source files already did).
2608 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2609 to his page will be replaced with a link to his new CLiki service.
2611 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2613 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2614 * The build process now looks for GNU make under the default name "gmake",
2615 instead of "make" as it used to. If GNU make is not available as "gmake"
2616 on your system, you can change this default behavior by setting the
2617 GNUMAKE environment variable.
2618 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2619 can build without error under CMU CL.
2621 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2623 * There is now a partial SBCL user manual (with some new text and some
2624 text cribbed from the CMU CL manual).
2625 * The beginnings of a profiler have been added (starting with the
2626 CMU CL profiler and simplifying and cleaning up). Eventually the
2627 main interface should be through the TRACE macro, but for now,
2628 it's still accessed through vaguely CMU-CL-style functions and macros
2629 exported from the package SB-PROFILE.
2630 * Some problems left over from porting CMU CL to the new
2631 cross-compilation bootstrap process have been cleaned up:
2632 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2633 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2634 part when getting rid of the compiler *BACKEND* stuff.)
2635 ** The value of *NULL-TYPE* was screwed up, because it was
2636 being initialized before the type system knew the final
2637 definition of the 'NULL type. This screwed up several key
2638 optimizations in the compiler, causing inefficiency in all sorts
2639 of places. (I found it because I wanted to understand why
2640 GET-INTERNAL-RUN-TIME was consing.)
2641 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2642 PROCLAIM FTYPE information. Unfortunately this broke binary
2643 compatibility again, since now the forms output by DEFGENERIC
2644 to refer to functions which didn't exist in 0.6.3.
2645 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2646 can use the new (as of 0.6.3) transform for SOME into MAP into
2648 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2649 values to (INTEGER 0 1000000), so that the time code will no longer
2650 occasionally get blown up by Linux returning 1000000 microseconds
2651 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2652 its output conform to the ANSI spec. (Alas, this makes its output
2653 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2654 * A full call to MAP NIL with a single sequence argument no longer conses.
2655 * fixes to problems pointed out by Martin Atzmueller:
2656 * The manual page no longer talks about multiprocessing as though
2657 it were currently supported.
2658 * The ILISP support patches have been removed from the distribution,
2659 because as of version 5.10.1, ILISP now supports SBCL without us
2660 having to maintain patches.
2661 * added a modified version of Raymond Toy's recent CMU CL patch for
2662 EQUALP comparison of HASH-TABLE
2664 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2666 * The system still can't cross-compile itself with
2667 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2668 entails), but at least it can compile more of itself that way
2669 than it used to be able to, and various buglets which were uncovered
2670 by trying to cross-compile itself that way have now been fixed.
2671 * This release breaks binary compatibility again. This time
2672 at least I've incremented the FASL file format version to 2, so that the
2673 problem can be detected reliably instead of just causing weird errors.
2674 * various new style warnings:
2675 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2676 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2677 ** using the lexical binding of a variable named in the *FOO* style
2678 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2679 as specified by ANSI.
2680 * *RANDOM-STATE* is no longer automatically initialized from
2681 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2682 default behavior of the system is to repeat its behavior every time
2683 it's run. If you'd like to change this behavior, you can always
2684 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2685 old convention there was no comparably easy way to get the system to
2686 repeat its behavior every time it was run.
2687 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2688 FTYPE declarations has been removed, in favor of their ANSI
2689 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2690 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2691 the types of their sequence arguments can be determined at compile time.
2692 This is done through a new open code expansion for MAP which eliminates
2693 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2694 argument types can be determined at compile time.
2695 * The optimizer now transforms COERCE into an identity operation when it
2696 can prove that the coerced object is already of the correct type. (This
2697 can be a win for machine generated code, including the output of other
2698 optimization transforms, such as the MAP transform above.)
2699 * Credit information has been moved from source file headers into CREDITS.
2700 * Source file headers have been made more standard.
2701 * The CASE macro now compiles without complaining even when it has
2704 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2706 * (Note that the way that the PCL macroexpansions were rewritten
2707 to accommodate the change in DEFGENERIC below breaks binary
2708 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2709 not run under sbcl-0.6.2. Once we get out of alpha releases,
2710 i.e. hit release 1.0.0, we'll probably try to maintain binary
2711 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2712 and sbcl-1.4.4. Until then, however, it might be fairly common
2713 for maintenance releases to break binary compatibility.)
2714 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2716 * The reporting of unhandled signals has been changed to print some
2717 explanatory text as well as the report form. (Previously only
2718 the report form was printed.)
2719 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2720 it defines, so that the compiler no longer issues undefined function
2721 warnings for compiled-but-not-yet-loaded generic functions.
2722 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2723 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2724 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2725 section of the ANSI spec are supported.
2726 * The "" nickname for the "KEYWORD" package has been removed.
2727 The reader still handles symbol tokens which begin with a package marker
2728 as keywords, but it doesn't expose its mechanism for doing so in the
2729 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2730 * The system now issues STYLE-WARNINGs for contradictory TYPE
2731 proclamations. (Warnings for contradictory FTYPE proclamations would
2732 be nice too, but those can't be done usefully unless the type system
2733 is made smarter about FUNCTION types.)
2734 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2735 systematized, so that "*target-*.lisp is supposed to exist only on the
2736 target and imply that there's a related file which exists on the
2737 host, and *host-*.lisp is supposed to exist only on the host and imply
2738 that there's a related file which exists on the target. This involves a
2739 lot of renaming. Hopefully the acute confusion caused by the renaming
2740 will be justified by the reduction in chronic confusion..
2741 ** runtime-type.lisp -> early-target-type.lisp
2742 ** target-type.lisp -> late-target-type.lisp
2743 ** early-host-format.lisp -> early-format.lisp
2744 ** late-host-format.lisp -> late-format.lisp
2745 ** host-error.lisp -> misc-error.lisp
2746 ** early-error.lisp -> early-target-error.lisp
2747 ** late-error.lisp -> late-target-error.lisp
2748 ** host-defboot.lisp -> early-defboot.lisp
2749 ** code/misc.lisp -> code/target-misc.lisp
2750 ** code/host-misc.lisp -> code/misc.lisp
2751 ** code/numbers.lisp -> code/target-numbers.lisp
2752 ** code/early-numbers.lisp -> numbers.lisp
2753 ** early-host-type.lisp -> early-type.lisp
2754 ** late-host-type.lisp -> late-type.lisp
2755 ** host-typep.lisp -> typep.lisp
2756 ** load.lisp -> target-load.lisp
2757 ** host-load.lisp -> load.lisp
2758 ** host-disassem.lisp -> disassem.lisp
2759 ** host-insts.lisp -> insts.lisp
2760 ** byte-comp.lisp -> target-byte-comp.lisp
2761 ** host-byte-comp.lisp -> byte-comp.lisp
2762 ** host-signal.lisp -> signal.lisp
2763 ** host-defstruct.lisp -> defstruct.lisp
2764 ** late-target-type.lisp -> deftypes-for-target.lisp
2765 Furthermore, several other previously target-only files foo.lisp (e.g.
2766 hash-table.lisp and random.lisp) have been split into a target-and-host
2767 foo.lisp file and a target-only target-foo.lisp file, with their key type
2768 definitions in the target-and-host part, so that the cross-compiler will
2769 know more about target types.
2770 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2771 gone away. In their place are various *BACKEND-FOO* variables
2772 corresponding to the slots of the old structure.
2773 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2774 into the target SBCL has been fixed.
2775 * The chill.lisp system for loading cold code into a running SBCL
2777 * Support for the CMU CL "scavenger hook" extension has been removed.
2778 (It was undocumented and unused in the CMU CL sources that SBCL was
2779 derived from, and stale in sbcl-0.6.1.)
2780 * Various errors in the cross-compiler type system were detected
2781 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2782 (enabling various consistency checks). Many of them were fixed,
2783 but some hard problems remain, so the compiler is back to
2784 running without *TYPE-SYSTEM-INITIALIZED* for now.
2785 * As part of the cross-compiler type system cleanup, I implemented
2786 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2787 * I have started adding UNCROSS calls throughout the type system
2788 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2789 on cross-compiler input files will be able to go away, and various
2791 * CONSTANTP now returns true for quoted forms (as explicitly required
2794 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2796 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2797 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2798 (MAKE-LIST -1). In the longer term, it ought to become true
2799 that declarations are assertions even at SAFETY 1. For now, it's not
2800 quite true even at SAFETY 3, but it's at least more nearly true..
2801 (Note that this change seems to increases the size of the system by
2802 O(5%) and to decrease the speed of the compiler by 20% or more.)
2803 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2804 for the problem of printing dozens of lines of distracting information
2805 about low-level system machinery as part of the top stack frame
2806 on entry to the debugger when an undefined function was called.
2807 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2808 is not reset to COMMON-LISP-USER.
2809 * Compilation of stuff related to dyncount.lisp has been made conditional
2810 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2811 smaller. The various dyncount-related symbols have been moved into
2812 a new "SB-DYNCOUNT" package.
2813 * tty-inspect.lisp has been renamed to inspect.lisp.
2814 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2815 feature has gone away. (When we eventually port to other flavors of
2816 libc and/or Unix, we'll try to make the differences between flavors
2817 invisible at the user level.)
2818 * Various other *FEATURES* tags, and/or their associated conditionals,
2819 have been removed if obsolescent, or given better documentation, or
2820 sometimes given more-mnemonic names.
2822 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2824 * tidied up "make.sh" script
2825 * tidied up system directory structure
2826 * better "clean.sh" behavior
2827 * added doc/FOR-CMUCL-DEVELOPERS
2828 * many many small tweaks to output format, e.g. removing possibly-confusing
2829 trailing #\. character in DESCRIBE-INSTANCE
2830 * (EQUALP #\A 'A) no longer signals an error.
2831 * new hashing code, including EQUALP hashing
2832 * tidied up Lisp initialization and toplevel
2833 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2834 * command line argument processing
2835 * added POSIX-GETENV function to deal with Unix-ish environment variables
2836 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2837 terminating SBCL on EOF
2838 * non-verbose GC by default
2839 * There is no more "sbcl" shell script; the sbcl file is now the C
2840 runtime executable (just like CMU CL).
2841 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2843 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2844 symbol and package information as Lisp native symbol and package objects
2845 instead of strings naming symbols and strings naming packages. This way,
2846 whenever packages are renamed (as in warm init), debug information is
2847 transformed along with everything else.
2848 * tweaked the optimization policy declarations which control the building
2849 of SBCL itself. Now, among other things, the system no longer saves
2850 source location debugging information. (This helps two problems at once
2851 by reducing SBCL size and by keeping SBCL from trying to look for its
2852 sources -- which may not exist -- when reporting errors.)
2853 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2854 debugging and testing purposes
2855 * cleaned up printing, making the printer call PRINT-OBJECT for
2856 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2857 methods, giving nearly-ANSI behavior
2858 * converted almost all special variables to use *FOO* naming convention
2859 * deleted PARSE-TIME functionality, since it can be done portably
2860 * moved some files out of cold init into warm init
2861 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2863 * regularized formatting of source files
2864 * added an install.sh script
2865 * fixed ridiculous memory usage of cross-compiler by making
2866 compiler/alloc.lisp not try to do pooling unless it can hook
2867 itself into the GC of the cross-compilation host. Now the system
2868 builds nicely on my old laptop.
2869 * added :SB-ALLOC in target-features.lisp-expr
2870 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2871 was not implemented)
2872 * re-did condition handling and note reporting in the compiler. Notes
2873 are no longer handled by signalling conditions. Style warnings
2874 and warnings are handled more correctly and reported in such a way
2875 that it's easy to find one or the other in your output (so that you
2876 can e.g. figure out which of many problems caused COMPILE-FILE to
2878 * changed the severity of several compiler warnings from full WARNING
2879 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2880 compiler note reporting so that it doesn't use the condition system
2881 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2882 COMPILE-FILE command)
2883 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2884 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2885 no longer supported, since their implementation was deeply intertwingled
2886 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2887 * removed bogus "support" for compiler macros named (SETF FOO), and
2888 removed the compiler macro for SETF INFO (but only after making a fool
2889 of myself on the cmucl-imp mailing list by posting a bogus patch for
2890 DEFINE-COMPILER-MACRO..)
2891 * Compiled files containing forms which have side effects on the Lisp
2892 reader (such as DEFPACKAGE forms) are now handled more correctly.
2893 (Compiler queuing of top level lambdas has been suppressed by setting
2894 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2895 * deleted various currently-unused source files, e.g. gengc.lisp. They
2896 may be added back at some point e.g. when porting to other architectures,
2897 but until they are it's distracting to distribute them and to try to
2899 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2900 there were so many of them they're just distractions, and UNCROSS is
2901 known to be able to handle the current sources
2902 * moved PROFILE functionality into TRACE, so that it will be clear
2903 how the wrapping and unwrapping of functions when you profile them
2904 interacts with the wrapping and unwrapping of functions when you
2905 trace them. (Actually, the functionality isn't there yet, but at least
2906 the interface specification is there. Hopefully, the functionality will
2907 arrive with some future maintenance release.)
2908 * removed host-oops.lisp
2909 * changed signature of QUIT function to allow UNIX-CODE argument
2910 * fixed READ-SEQUENCE bug
2911 * tweaked verbose GC output so that it looks more like the progress
2912 output that ANSI specifies for functions like LOAD
2913 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2914 * added <http://sbcl.sourceforge.com> to the banner information printed by