1 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
3 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
5 * fixed loading of multiply forward-referenced layouts.
6 (thanks to Cheuksan Wang)
7 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
8 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
9 (see COMPILE-FILE documentation for details of :PRINT option)
10 * fixed bugs 19 and 317: fixed-format floating point printing is
11 more accurate. This also fixes a bug reported by Adam Warner
12 related to the ~@F format directive.
13 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
15 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
16 dispatch table if the from-char is a dispatch macro character.
17 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
18 by the wordsize no longer ignore the last word. (reported by Lutz
20 * fixed bug in type checking of dynamic-extent variables. (reported
22 * optimization: sequence traversal functions use their freedom to
23 coerce function designators to functions.
24 * optimization: code with many calls to CLOS methods specialized on
25 CLOS classes has had redundant type checks removed.
26 * optimization: added declarations to speed up operations that access
27 the internal character database (for example STRING-UPCASE)
28 * optimization: comparison operations between floats and sufficiently small
29 fixnums no longer create extra rationals
30 * fixed some bugs related to Unicode integration:
31 ** portions of multibyte characters at the end of buffers for
32 character-based file input are correctly transferred to the
33 start of the buffer at the next read.
34 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
35 passing it through to OPEN.
36 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
37 argument given to internal calls to OPEN.
38 ** the built-in comment readers (introduced by character sequences
39 ";" and "#|") are more forgiving to encoding errors; they will
40 STYLE-WARN and then attempt to resync the stream at a character
41 boundary. (thanks to Teemu Kalvas)
42 * fixed some bugs in the x86-64 port:
43 ** Negative short int return values from c-calls are sign-extended
45 ** The stack is aligned to 16-bytes for c-calls, as required by
46 the ABI. (thanks to Cheuksan Wang)
47 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
48 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
49 for enums. (thanks to Vincent Arkesteijn)
50 ** Multiple small optimizations and bugfixes for floating point
52 * fixed some bugs revealed by Paul Dietz' test suite:
53 ** Space, Tab, Linefeed, Return and Page have the invalid
54 secondary constituent character trait.
55 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
57 ** WITH-INPUT-FROM-STRING should only update the index place on
59 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
61 ** Bit-array operations (BIT-AND and similar) worked incorrectly
62 with one-dimensional arrays with fill pointers.
63 ** TYPE-OF failed on a complex with an integer realpart and a
65 ** compiler failure during type inference for the code of form
66 (IF (EQL X (THE ...)) ...) (MISC.535).
68 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
69 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
70 platform. The port supports 61-bit fixnums, large memory spaces
71 and reloading shared object files.
72 * enhancement: saving cores with foreign code loaded is now
73 supported on ppc/Darwin in addition to the previously supported
75 * enhancement: the statistical profiler now walks deeper into the
76 call stack for more meaningful call-graphs and accrued time
77 reports (x86/x86-64 only). It also now reports time spent in
79 * enhancement: it is now possible to trace most individual methods
80 of a generic function in addition to tracing the generic function
82 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
84 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
85 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
86 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
87 produces an error. (thanks to Vincent Arkesteijn)
88 * bug fix: NAMESTRING on pathnames with :WILD components in their
89 directories works correctly. (thanks to Artem V. Andreev)
90 * fixed bug 125: compiler preserves identity of closures. (reported
92 * bug fixed: functions with &REST arguments sometimes failed with
93 "Undefined function" when compiled with (DEBUG 3). (reported by
95 * bug fix: overflow during compiling of setting element of a bit
96 vector with constant index and value. (reported by Timmy Douglas)
97 * build fix: fixed the dependence on *LOAD-PATHNAME* and
98 *COMPILE-FILE-PATHNAME* being absolute pathnames.
99 * on x86 compiler partially supports stack allocation of dynamic-extent
101 * GO and RETURN-FROM do not check the extent of their exit points
102 when compiled with SAFETY 0.
103 * fixed some bugs related to Unicode integration:
104 ** encoding and decoding errors are now much more robustly
105 handled; it should now be possible to recover even from invalid
106 input or output to the terminal. (thanks to Teemu Kalvas)
107 ** provided a first cut at implementing STRING-TO-OCTETS and
108 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
109 ** altered the SB-MD5 contributed module slightly, changing the
110 interface just enough for it to be supportable for builds where
111 lisp characters are not eight bits.
112 * fixed some bugs revealed by Paul Dietz' test suite:
113 ** the FORMATTER-generated functions for ~V[ conditionals require
114 the correct number of arguments.
115 ** READ-FROM-STRING returns the mandated second value when applied
116 to displaced strings.
117 ** the #\Rubout and #\Backspace characters are treated as invalid
118 constituent characters by the tokenizer.
120 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
121 * new feature: reloading changed shared object files with
122 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
123 * new feature: references to foreign variables and functions
124 can now be compiled and loaded before the corresponding shared
125 object file is loaded, as long as the foreign definitions are
126 available at runtime.
127 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
128 supported. (thanks to Dan Debertin)
129 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
130 just keywords. (thanks to Vincent Arkesteijn)
131 * fixed bug #331: structure-class instances corresponding to
132 DEFSTRUCT forms are now created eagerly.
133 * fixed bug #345: backtraces from calls to undefined functions work
134 on x86 as well. Related bug #61 is now also partially fixed on x86
135 (backtraces from throws to unknown catch tags.)
136 * bug fix: lambda-list parsing is now stricter vrt. order and number
137 of lambda-list keywords.
138 * bug fix: as specified by AMOP, an error is signalled if a
139 class-option appears multiple times in a DEFCLASS form. (reported
141 * bug fix: RANDOM can be compiled when the compiler derives the type
142 of its numeric argument as a disjoint set of small integers.
143 (reported by Paul Dietz)
144 * bug fix: starting a core saved with shared objects loaded when
145 those objects are not available no longer causes threaded SBCL to
146 hang. (reported by Sean Ross)
147 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
149 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
150 argument, which must always be NIL. (reported by Kalle Niemitalo)
151 * bug fix: printing 1.0d+23 no longer results in an error.
152 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
153 * bug fix: structure slot setters preserve evaluation order. (thanks
155 * fixed some bugs related to Unicode integration:
156 ** RUN-PROGRAM can allow its child to take input from a Lisp
157 stream. (reported by Stefan Scholl)
158 ** ASDF-INSTALL successfully downloads .tar.gz files in all
159 locales. (reported by Ken Causey)
160 * fixed some bugs revealed by Paul Dietz' test suite:
161 ** INCF, DECF and REMF evaluate their place form as specified in
163 ** FORMATTER expands ~{ ~} iteration directives with V or #
164 parameters correctly.
165 ** FORMATTER deals with the ~@[ ~] conditional directive where the
166 consequent uses no arguments correctly.
167 ** the system has a partial understanding of the (COMPLEX RATIO)
169 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
172 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
173 * new feature: a build-time option (controlled by the :SB-UNICODE
174 keyword feature, enabled by default) for building the system with
175 support for the entire 21-bit character space defined by the
177 * new feature: the system now has rudimentary external-format
178 support; the primary user-visible change at this time is that
179 characters with the high bit set (such as lower-case-e-acute) will
180 print correctly to a terminal in a UTF-8 environment.
181 * minor incompatible change: BASE-CHAR no longer names a class;
182 however, CHARACTER continues to do so, as required by ANSI.
183 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
184 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
185 has been moved to the SB-EXT package (temporarily re-exported from
187 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
188 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
190 * fixed bug #351: better error-handlind and reporting for malformed
192 * fixed bug #350: bignum-printing is now more memory-efficient,
193 allowing printing of very large bignums, eg. (expt 2 10000000).
194 (reported by Bruno Haible)
195 * fixed bug #302: better primitive-type selection for intersection
197 * fixed bug #308: non-graphic characters now all have names, as
198 required. (reported by Bruno Haible)
199 * bug fix: redefining a class with different superclasses now correctly
200 removes it from the direct-subclasses of its previous superclasses.
201 (reported by David Morse)
202 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
203 new value now works. (reported by Bruno Haible)
204 * bug fix: correct canonicalization of multiple non-standard slot
205 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
206 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
207 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
208 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
209 now exists, an signals an error.
210 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
211 during macroexpansion for non-iteration variables as well. (reported
212 by Bruno Haible for CMUCL)
213 * bug fix: Cyclic structures and unprintable objects in compiler
214 messages no longer cause errors. (reported by Bruno Haible)
215 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
216 and READ-FROM-STRING all now return a primary value of NIL if
217 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
218 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
219 (reported by Bruno Haible for CMUCL)
220 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
221 do not designate a condition. (reported by Bruno Haible for
223 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
224 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
225 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
226 permitting resolution in favour of any of the conflicting symbols.
227 (reported by Bruno Haible for CMUCL)
228 * bug fix: EQUAL compiler optimizations is less aggressive on
229 strings which can potentially compare true despite having distinct
230 specialized array element types.
231 * bug fix: unit enumerations can be defined without dividing by
232 zero. (thanks to Vincent Arkesteijn)
233 * FORMAT compile-time argument count checking has been enhanced.
234 (report from Bruno Haible for CMUCL)
235 * a partial workaround for the bug 262: the compiler does not try to
236 inline-expand a local function doing RETURN-FROM from a deleted
237 BLOCK. (thanks to Peter Denno for the bug report and to David
238 Wragg for the simple test case)
239 * fixed some bugs revealed by Paul Dietz' test suite:
240 ** RENAME-PACKAGE allows all package designators as new package
242 ** constraint propagation and lambda variable substitution are
243 more cautious in dealing with partially deleted code.
244 ** compiler sometimes lost reoptimization passes.
245 ** CERROR, given a condition as condition designator, passes its
246 remaining arguments to the continue format control without
248 ** Case-altering FORMAT directives work correctly on non-ASCII
250 ** The REAL type specifier handles bounds outside the floating
251 point ranges without signalling FLOATING-POINT-OVERFLOW.
252 ** Functions with IR1-transformations can create intercomponent
253 references to global functions.
254 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
256 ** FORMAT ~R treats a nil value for its first parameter correctly.
258 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
259 * enhancement: saving cores with foreign code loaded is now
260 supported on x86/NetBSD and sparc/Linux in addition to the previously
262 * bug fix: on some platforms repeated installations caused multiple
263 copies of HTML documentation to be installed -- should not happen
264 any more. (reported by Stefan Scholl)
265 * bug fix: parsing self-recursive alien record types multiple times
266 no longer causes infinite recursion. (reported by Thomas F. Burdick,
267 original patch by Helmut Eller for CMUCL)
268 * bug fix: stack-exhaustion detection works now on NetBSD as well.
269 (thanks to Richard Kreuter)
270 * bug fix: defining classes whose accessors are methods on existing
271 generic functions in other (locked) packages no longer signals
272 bogus package lock violations. (reported by François-René Rideau)
273 * bug fix: special variables as DEFMETHOD parameters no longer have
274 associated bogus type declarations. (reported by David Wragg and
276 * bug fix: read-write consistency on streams of element-type
277 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
278 * bug fix: redefiniton of the only method of a generic function with
279 no DEFGENERIC no longer emits a full WARNING. In addition,
280 redefinition of generic functions with no DEFGENERIC to an
281 incompatible lambda list now signals an error. (thanks to Zach
283 * bug fix: DEFGENERIC now works even when there's a function of the
284 same name in an enclosing lexical environment. (thanks to Zach
286 * fixed compiler failure, caused by instrumenting code during
287 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
288 * optimization: added loop analysis and improved register allocation
289 * fixed some bugs revealed by Paul Dietz' test suite:
290 ** POSITION on displaced vectors with non-zero displacement
291 returns the right answer.
292 ** (SIMPLE-STRING) is a valid type specifier for sequence
294 ** *PRINT-LEVEL* handling for slotless structures is pedantically
296 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
297 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
299 ** The justification version of the FORMAT ~< directive treats
300 non-zero minpad parameter correctly.
302 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
303 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
304 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
305 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
306 the supported interface.
307 * new feature: Single-stepping of code compiled with DEBUG 2 or
308 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
309 * new feature: saving cores with foreign code loaded is now
310 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
311 Timothy Moore's work for CMUCL)
312 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
313 parameters to * instead of NIL if no initform is supplied.
314 (reported by Johan Bockgaard)
315 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
316 correctly when FUNCALL appears as the car of the form. Note:
317 despite this FUNCALL forms are not currently subject to
318 compiler-macro expansion. (port of Raymond Toy's fix for the
319 same from CMUCL, reported by Johan Bockgaard)
320 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
321 (thanks for Teemu Kalvas)
322 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
323 case instead of terminating the iteration (thanks for Julian
324 Squires, Sean Champ and Raymond Toy)
325 * bug fix: incorrect expansion of defgeneric that caused a style
326 warning. (thanks for Zach Beane)
327 * bug fix: slot accessor effective method computation works properly
328 for classes with multiple non-standard applicable methods on
329 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
330 * on x86 compiler supports stack allocation of results of LIST and
331 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
332 CMUCL implementation by Gerd Moellmann)
333 * fixed some bugs revealed by Paul Dietz' test suite:
334 ** FORMAT strings with both the ~<~:;~> form of the justification
335 directive and pretty-printing directives cause an error.
337 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
338 * incompatible change: the internal functions
339 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
340 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
341 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
342 instead of the old functions.
343 * new feature: on platforms where "dladdr" is available foreign
344 function names now appear in backtraces. (based on Helmut Eller's
346 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
347 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
349 * documentation: ASDF manual is now distributed with SBCL.
350 * bug fix: SBCL can now load its contributed modules with REQUIRE
351 even if the system-provided entries have been removed from
352 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
354 * bug fix: backtraces involving undefined functions or assembly
355 routines are more informative. (thanks to Brian Downing)
356 * bug fix: mutually referent alien structures now work correctly.
357 (reported by Rick Taube)
358 * bug fix: structures defined by WITH-ALIEN can be referred to
359 within other definitions in the same WITH-ALIEN.
360 * bug fix: division operators (MOD, TRUNCATE and the like) with
361 constant zero divisors and integer dividends no longer generate
363 * bug fix: provide default methods for INPUT-STREAM-P and
364 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
365 * bug fix: improve the detection and resolution of MOP metacycles
366 (where effective method computation depends on the generic function
367 having its effective method computed). (reported by Bruno Haible)
368 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
369 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
370 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
371 * bug fix: pathnames with relative directory components can be
372 represented relative to default pathnames.
373 * optimization: in taking the GCD of bignums, reduce the two bignums
374 to approximately the same size (using Euclid's algorithm) before
375 applying the more sophisticated binary GCD. (thanks to Juho
377 * optimization: COUNT on bitvectors now operates word-at-a-time.
378 * optimization: ASH with a positive, but not necessarily constant,
379 (leftwards) shift, when in a modular context, is compiled to a
381 * fixed some bugs revealed by Paul Dietz' test suite:
382 ** FORMAT variable parameters ("~V<char>") are defaulted properly
383 if the corresponding argument is NIL.
384 ** FORMAT directives accepting a minpad parameter treat negative
386 ** literal commas as character parameters to FORMAT directives are
388 ** literal spaces directly after ~<Newline> directives within a
389 format-logical-block (~:< ~@:>) do not induce :FILL-style
390 conditional newlines.
391 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
392 as specified: it no longer includes conditional newlines).
393 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
395 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
396 :PER-LINE-PREFIX argument does not evaluate to a string.
398 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
399 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
400 the manual for details; for now, package locks can be disabled by
401 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
402 if no major problems are found then it is likely that they will be
403 compiled in unconditionally.
404 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
405 unsupported operators on all platforms. To load a shared library
406 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
407 object file, link it into a shared library outside of SBCL and
408 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
410 * bug workaround: ROOM T and the GC were not working together
411 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
412 reporting facility aren't preserved. That reporting has been
413 disabled, so now until and unless someone figures out how to make it
414 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
415 * minor incompatible change: as threatened around sbcl-0.8.0, the
416 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
417 The symbols are also exported from SB-PCL for backwards
418 compatibility, but more so than before SB-PCL should be treated as
419 an implementation-internal package.
420 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
422 * the SB-SPROF contrib now works on (most) non-x86 architectures.
423 It is known as of this release not to work on the Alpha, however.
424 * fixed bug #167: errors signalled due to illegal syntax in method
425 bodies are now more legible.
426 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
427 designators and can hence be used with TYPEP.
428 * fixed bug #333: CHECK-TYPE now ensures that the type error
429 signalled, if any, has the right object to be accessed by
430 TYPE-ERROR-DATUM. (reported by Tony Martinez)
431 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
432 5.1.2.3 for multiple-value place subforms. (reported by Kalle
434 * fixed bug #334: programmatic addition of slots using specialized
435 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
436 and :ALLOCATION :CLASS slots.
437 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
438 float exponents. (rereported by Peter Seibel)
439 * fixed a bug: #\Space (and other whitespace characters) are no
440 longer considered to be macro characters in standard syntax by
442 * fixed bug: initialization of condition class metaobjects no longer
443 causes an instance of the condition to be created. (reported by
445 * fixed bug: it is now possible to have slots such that
446 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
447 description is neither :INSTANCE nor :CLASS.
448 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
449 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
450 SLOT-VALUE-USING-CLASS), and no longer causes errors with
451 non-standard SLOT-DEFINITION-ALLOCATION values.
452 * fixed bugs: various race conditions handling exiting threads.
453 CL-PPCRE's thread torture test now passes.
454 * fixed arguably-a-bug: GC time proportional to number of threads in
455 system even when most of them are idle
456 * optimization: improved performance of BIT and SBIT on bit-vectors.
457 * .fasl file incompatibility: The fasl file version number has
458 been incremented because of changes associated with package locks.
460 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
461 * minor incompatible change: the system no longer provides
462 optimization hints (conditions which are TYPEP
463 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
464 mode; these hints are still emitted when higher SPEED optimization
466 * new contrib module: a sampling profiler (profiling by statistical
467 sampling, rather than by instrumenting functions) is available as
468 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
469 * the behaviour of the standard function ED is now customizeable by
470 third parties through a hook variable: see ED's documentation
471 string for information on the protocol.
472 * the compiler no longer emits efficiency notes for (FUNCALL X)
473 when the type of X is uncertain under default optimization
475 * fixed bug 276: mutating a binding of a specialized parameter to a
476 method to something that is not TYPEP the specializer is now
478 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
479 exhibit the required behaviour on the x86 platform. (thanks to
480 Peter van Eynde, Eric Marsden and Bruno Haible)
481 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
482 even for difficult arguments. (reported by Peter Graves)
483 * fixed bug 141a: the backquote printer now descends quoted
485 * fixed another bug in backquote printing: no more destructive
486 modification of the form's list structure. (reported by Brian
488 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
489 move between its address being taken and the call to
490 interrupt_thread, fixing a crashing race condition.
491 * the SB-POSIX contrib implementation has been adjusted so that it
492 no longer exhibits ridiculously poor performance when constructing
493 instances corresponding to C structs.
495 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
496 * minor incompatible change: the sb-grovel contrib now treats C
497 structures as alien (in the sense of SB-ALIEN) objects rather than
498 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
499 has implications for memory management of client code
500 (sb-grovel-returned objects must now be manually managed) and for
501 type safety (alien objects now have full types).
502 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
503 used to control emission of compiler diagnostics, rather than the
504 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
505 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
506 quality should be considered deprecated.
507 * install.sh now installs the user manual as well
508 * (not quite a new documentable feature, but worth considering in
509 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
510 beginnings of a semantically meaningful condition hierarchy is
511 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
513 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
514 designator as the defaults argument.
515 * fixed bug: Displaced arrays whose displaced-to array has become
516 too small now cause ARRAY-DIMENSION to signal an error, providing
517 for safer bounds-checking. (reported by Bruno Haible)
518 * fixed bug: DEFCLASS slot definitions with identical :READER and
519 :WRITER names now signal a reasonable error. (reported by Thomas
521 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
523 * fixed bug: Invalid slot specification errors now print correctly.
524 (thanks to Zach Beane)
525 * fixed bug 320: Shared to local slot value transfers in class
526 redefinitions now happen corrently from superclasses as
527 well. (reported by Bruno Haible)
528 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
530 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
531 improper lists correctly. (reported by Bruno Haible)
532 * fixed bug 313: source-transform for <fun-name> was erroneously
533 applied to a call of a value of a variable with name <fun-name>.
534 (reported by Antonio Menezes Leitao)
535 * fixed bug 307: The obsolete instance protocol ensures that
536 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
537 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
538 remove dead unknown-values globs from the middle of the stack.
539 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
540 easier to resume long computations after using *BREAK-ON-SIGNALS*
541 to diagnose and fix failures. (thanks to Nikodemus Siivola)
542 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
543 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
544 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
545 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
546 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
548 * fixed a (fairly theoretical) bug in string printing: if
549 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
550 does not have array-element-type equal to the most general string
552 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
553 function name. (thanks to Nikodemus Siivola)
554 * fixed bug: compile-time format string checker failed on
555 non-closed ~{. (reported by Thomas F Burdick)
556 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
557 don't warn on legal loop code involving more than one aggregate
558 boolean. (thanks to Nikodemus Siivola)
559 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
560 in keyword position to MAKE-PACKAGE was not being reported as
562 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
563 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
565 * fixed bugs 280 and 312: the checking for multiple definitions in a
566 file is less likely to become confused by uses of inline
568 * fixed bug: the #S reader macro performs the keyword coercion
569 specified for slot names. (reported by Kalle Niemitalo)
570 * fixed bug: lambda lists may contain symbols whose names start with
571 & but are not lambda-list-keywords; their occurrence triggers a
573 * fixed bug 321: define-method-combination argument lambda lists do
574 not cause invalid code to be generated when &optional or &aux
575 variables have default values. (reported by Bruno Haible)
576 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
577 have CLOS classes; this appears also to have fixed reported
578 instability in stack exhaustion detection.
579 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
580 MUFFLE-WARNING when no associated restart is present is now
582 * optimization: rearranged the expansion of various defining macros
583 so that each expands into only one top-level form in a
584 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
586 * optimization: used a previously little-used slot in symbols to
587 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
589 * fixed some bugs revealed by Paul Dietz' test suite:
590 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
592 ** Fixed an optimization bug related to inheritance of initforms
593 from local to shared slots.
594 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
595 ** CAST optimizer forgot to flush argument derived type.
596 ** print/read consistency on floats is now orders of magnitude
597 more likely. (thanks also to Bruno Haible for a similar report
599 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
600 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
601 for objects of type REAL. Make it so.
602 ** SXHASH is specified (infelicitously) to respect similarity,
603 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
604 it so. (thanks to Markus Ziegler)
605 ** on the Alpha, the compiler succeeds in compiling functions
606 returning a known number of arguments greater than 63.
607 ** fixed handling of invalid NIL arguments in keyword position
609 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
611 ** provided a readably-printable representation for RANDOM-STATE
613 ** ensured that pathnames, where they have a namestring, always
614 print using #P"..." syntax.
616 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
617 * Support for the forthcoming 2.0 version of the NetBSD kernel
618 running on x86 hardware has been added. (thanks to Perry
619 E. Metzger most immediately, and others for their past work)
620 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
621 current release version; SBCL's OpenBSD support had been broken
622 since about the time of OpenBSD's switch to ELF binary format.)
623 * [placeholder for DX summary]
624 ** user code with &REST lists declared dynamic-extent, under high
625 speed or space and low safety and debug optimization policy.
626 * The manual has been converted to Texinfo format and the debugger
627 chapter from the cmucl manual has been added.
628 * A facility has been added to extract documentation strings from
629 sbcl and store them as Texinfo-formatted snippets for inclusion in
630 the manual (via Texinfo's @include directive)
631 * bug fix: compiler emitted division in optimized DEREF. (thanks for
632 the test case to Dave Roberts)
633 * bug fix: multidimensional simple arrays loaded from FASLs had fill
634 pointers. (reported by Sean Ross)
635 * bug fix: PROFILE output is printed nicely even for large numerical
636 values. (thanks to Zach Beane)
637 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
638 greater than 32 handle EOF correctly.
639 * bug fix: on X86 an immediate argument of the IMUL instruction is
640 correctly printed in disassembly. (thanks to Lutz Euler)
641 * bug fix: class slots in redefined classes preserve their old
642 values. (thanks to Bruno Haible and Nikodemus Siivola)
643 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
644 succeeds. (reported by Marco Baringer)
645 * fixed some bugs revealed by Paul Dietz' test suite:
646 ** READ-SEQUENCE now works on ECHO-STREAMs.
647 ** RATIONALIZE works more according to its specification. (thanks
650 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
651 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
652 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
653 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
654 should matter to you only if you rebind the printer control
655 variables and then find you want different bindings in the
656 debugger than in the ordinary execution of your program.)
657 * The runtime build system has been tweaked to support building
658 (on SPARC/SunOS) using a C compiler which invokes Sun's own
659 assembler and linker. (thanks to Nikodemus Siivola)
660 * Unbound, undefined, undeclared variables now trigger full
661 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
662 more often programmer error than deliberate exploitation of undefined
664 * optimization: the hash algorithm for strings has changed to one
665 that is less vulnerable to spurious collisions. (thanks to Juho
667 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
668 needless bounds checking. (thanks to Juho Snellman)
669 * optimization: implemented multiplication as a modular
670 (UNSIGNED-BYTE 32) operation on the PPC backend.
671 * fixed some bugs revealed by Paul Dietz' test suite:
672 ** ADJUST-ARRAY now copies the datum in a zero rank array if
674 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
675 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
676 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
678 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
680 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
682 ** ECHO-STREAMs no longer attempt to echo the end of file value to
683 their output stream on EOF from read.
684 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
685 have been read to end-of-file.
686 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
688 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
689 description of determination of which consecutive characters
691 ** Printing the "Space" character with escaping on now yields "#\\ ",
692 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
693 ** Reading floating-point numbers with *READ-BASE* set to a number
694 less than 10 works correctly.
695 ** Reading floating-point numbers with *READ-BASE* set to a number
696 more than 10 works correctly.
697 ** Printing with *PRINT-READABLY* targets the standard readtable, not
698 the readtable currently in effect.
700 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
701 * minor incompatible change: parsing of namestrings on a physical
702 (Unix) host has changed; numbers after the final #\. in a
703 namestring are no longer interpreted as a version field. This is
704 intented to be largely invisible to the user, except that the
705 meaning of the namestring "*.*.*" has changed: it now refers to a
706 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
707 should usually be replaced by
708 (make-pathname :name :wild :type :wild :version :wild)
709 with the added benefit that this is more likely to be portable.
710 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
711 option to OPEN now signals an error if the file being opened
712 exists; this may have an impact on existing code.
713 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
714 signals received, on the PowerPC platforms (both Linux and
715 Darwin). (thanks to Pierre Mai for pointing out the location of
717 * several fixes on OS X: The system now builds and runs cleanly on
718 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
719 (thanks to Brian Mastenbrook)
720 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
721 second-resolution: integer multiples of 1/3600 between -24 and 24.
722 (thanks to Vincent Arkesteijn)
723 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
724 type when called with 1 argument; PEEK-CHAR checked type of
725 PEEK-TYPE only after having read first character from a
726 stream. (reported by Peter Graves)
727 * bug fix: the garbage collector now has much better locality
728 behaviour, and in particular no longer treats all memory as being
729 exhausted when many small objects point to each other in a deeply
731 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
732 recognized as being TYPEP their class.
733 * bug fix: the PUSHNEW documentation string has been corrected.
734 (thanks to Vincent Arkesteijn)
735 * bug fix: defaulting of the value for the last of an atypically
736 large number of multiple values being bound was not being
737 performed correctly on the Alpha or PPC platforms
738 * optimization: implemented multiplication as a modular
739 (UNSIGNED-BYTE 32) operation on the x86 backend.
740 * optimization: SEARCH on simple-base-strings can now be open-coded.
741 (see also contrib/compiler-extras.lisp for inspiration for
742 teaching the compiler about the Boyer-Moore algorithm).
743 * value, returned by MAX (and MIN) called with several EQUALP, but
744 not EQL, arguments now does not depend on compiler settings.
745 * fixed some bugs revealed by Paul Dietz' test suite:
746 ** in stack analysis liveness information is propagated from
747 non-local entry points.
748 ** pathwise CAST removing failed when the CAST node did not start
750 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
751 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
753 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
754 host is already defined.
755 ** RENAME-FILE works on streams instead of signalling an internal
757 ** PEEK-CHAR uses the current readtable when determining whether
758 or not a character is whitespace.
759 ** MERGE-PATHNAMES handles the case when the pathname does not
760 specify a name while the default-pathname specifies a version.
761 ** Pathnames now stand a better chance of respecting print/read
763 ** Attempting to use standardized file system operators with a
764 pathname with invalid :DIRECTORY components signals a
766 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
768 ** DIRECTORY on logical pathnames is more correct.
769 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
770 signal a TYPE-ERROR if their argument is not a stream.
771 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
772 designator argument does not designate a stream.
773 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
774 examining the synonym.
775 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
777 ** OPEN and WITH-OPEN-STREAM allow opening streams with
778 element-type larger than ([UN]SIGNED-BYTE 32).
780 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
781 * When built with the :SB-FUTEX feature, threaded builds now take
782 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
783 for faster/more reliable mutex and condition variable support.
784 * Incompatible change (but one you probably shouldn't have been using
785 anyway): the interface and code for arbitrating between multiple
786 threads in the same user session has been redesigned.
787 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
788 error when called without an explicit environment argument.
789 (thanks to Nikodemus Siivola)
790 * bug fix: buffered :DIRECTION :IO streams are less likely to become
791 confused about their position. (thanks to Adam Warner and Gerd
793 * bug fix: Pretty printing backquoted forms with unquotations in the
794 argument list position of various code constructs such as LAMBDA
795 now works correctly. (reported by Paul Dietz)
796 * bug fix: Pretty printing unquotations no longer loses all
797 stream position information.
798 * optimization: performance of string output streams is now less
799 poor for multiple small sequence writes.
800 * optimization: performance of CSUBTYPEP in the presence of complex
801 expressions involving CONS and NOT many times has been improved.
802 (reported by Paul Dietz)
803 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
805 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
806 streams for the REPL, for the debugger, and so on. (thanks to
808 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
809 * fixed some bugs revealed by Paul Dietz' test suite:
810 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
811 is ignored if it is too small, rather than propagating through
812 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
813 ** extremely complex negations of CONS types were not being
814 sufficiently canonicalized, leading to inconsistencies in
816 ** VALUES tranformer lost derived type.
818 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
819 * fixed a bootstrapping bug: the build process no longer assumes
820 that the various BOOLE-related constants have the same value in
821 host and target lisps. (noted by Paul Dietz' test suite on an
822 SBCL binary built from CLISP)
823 * The system can now be dynamically linked on the MIPS platform,
824 which enables dynamic loading of foreign code from Lisp. (thanks
825 to Ralf Baechle for discussions on the MIPS ABI)
826 * The system now records debugging information for its own source
827 files in a filesystem-position-independent manner, relative to
828 the "SYS" logical host.
829 * fixed a compiler bug: MV-LET convertion did not check references
830 to the "max args" entry point. (reported by Brian Downing)
831 * tweaked disassembly notes to be less confident about proclaiming
832 some instruction as an LRA. (thanks to Brian Downing)
833 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
834 now each have their own history, command character, and other
835 characteristics. (thanks to David Lichteblau)
836 * fixed some bugs revealed by Paul Dietz' test suite:
837 ** compiler failure in compiling LOGAND expressions including a
839 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
840 shift greater than 32.
841 ** FLUSH-DEST did not mark blocks for type check regeneration.
842 ** HANDLER-CASE failed to accept declarations in handler clauses
843 in some circumstances.
845 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
846 * New code in contrib/sb-introspect (still probably not entirely
847 stable yet) provides some support for smart Lisp development
848 environments like SLIME.
849 * The conditions signalled for errors occurring when loading .fasl
850 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
851 in a way which should help ASDF recover gracefully.
852 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
853 stuff has been cleaned up. If you code contrib/ stuff, this might
854 affect you, and you can look at contrib/README, contrib/STANDARDS,
855 and/or the 0.8.4.27 diff to check.
856 * In full calls the compiler now does not generate checks for declared
857 argument types for all arguments.
858 * various threading fixes
859 ** and some experimental patches which didn't make it into
860 the main tree for this release, but which are shipped in
861 contrib/experimental-thread.patch as a possible fix for some
862 failures (deadlock, spinning...) in GC-intensive multithreaded
864 * fixed PPC build problem (source code incompatibility of different
865 library versions): added offsetof() hackery which attempts to divine
866 where glibc maintainers put uc_mcontext today
867 * fixed bug 282: compiler does not trust type assertions while passing
868 arguments to a full call.
869 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
870 &OPTIONAL VALUES type parameter.
871 * fix bug 214: algorithm for noting rejected templates is now more
872 similar to that of template seletion. (also reported by rydis on
874 * fixed bug 141b: printing backquoted information readably and prettily
875 inserts a space where necessary.
876 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
877 (CHARACTER 10) are now reported as errors, rather than propagated
878 as unknown types. (reported by piso on #lisp)
879 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
880 with values NIL and :ERROR. (thanks to Milan Zamazal)
881 * fixed bug 191c: CLOS now does proper keyword argument checking as
882 described in CLHS 7.6.5 and 7.6.5.1.
883 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
884 longer raise an error; further, using a list as a for-as-arithmetic
885 counter now raises a meaningful error.
886 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
887 now understood by sequence creation functions such as MAKE-SEQUENCE
889 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
890 a STRING-INPUT-STREAM.
891 * compiler enhancement: SIGNUM is now better able to derive the type
893 * type declarations inside WITH-SLOTS are checked. (reported by
895 * fixed some bugs revealed by Paul Dietz' test suite:
896 ** incorrect optimization of TRUNCATE for a positive first
897 argument and negative second.
898 ** compiler failure in let-convertion during flushing dead code.
899 ** compiler failure while deriving type of TRUNCATE on an
900 interval, containing 0.
901 ** ASH of a negative bignum by a negative bignum count now returns
903 ** intersection of CONS types now canonicalizes properly, fixing
904 inconsistencies in SUBTYPEP.
906 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
907 * incompatible change: The --disable-debugger command line
908 option now clobbers the debugger at a more fundamental
909 level, by redefining #'INVOKE-DEBUGGER instead of by
910 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
911 is specified by ANSI to ignore *DEBUGGER-HOOK* and
912 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
913 enter the debugger REPL and then suffer recursive errors
914 because *DEBUG-IO* is also messed up in --disable-debugger mode;
915 while under the new system, BREAK in --disable-debugger mode
916 terminates the system just as an unhandled error would.
917 * fixed compiler performance when processing loops with a step >1;
918 * bug fix: DOCUMENTATION now retrieves generic function
919 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
920 support has been systematized, and now supports the methods
921 specified by ANSI, along with a default method and a method for
922 slot documentation. (reported by Nathan Froyd)
923 * bug fix: effective methods associated with a generic function are
924 no longer cached over a change of that generic function's method
925 combination. (reported by Andreas Fuchs)
926 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
927 names a non-existent directory. (thanks to Andreas Fuchs)
928 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
929 the PPC platform, lead to stack corruption; this has been fixed.
930 (reported by Rainer Joswig)
931 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
932 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
933 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
934 arithmetic in general, is now much more reliable on the PPC
936 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
937 platform now returns the right answer.
938 * optimization: restored some effective method precomputation in
939 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
940 precomputation is now tunable.
941 * optimization: compiler-internal data structure use has been
942 reviewed, and changes have been made that should improve the
943 performance of the compiler by about 20%.
944 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
945 simple-base-strings and simple-bit-vectors is improved.
946 * optimization: the optimization of 32-bit logical and arithmetic
947 functions introduced in version 0.8.3 on the x86 has been
948 implemented on the mips, ppc and sparc platforms; an
949 implementation of the same facility, but for 64-bit arithmetic,
950 has been added for the alpha.
951 * microoptimization: the compiler is better able to make use of the
952 x86 LEA instruction for multiplication by constants.
953 * bug fix: in some situations compiler did not report usage of
954 generic arithmetic in (SPEED 3) policy.
955 * bug 145b fix: compiler used wrong type specifier while converting
956 MEMBER-types to numeric.
957 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
959 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
960 index argument now works properly on the Alpha platform.
961 * bug fix: floating point exception treatment on the Alpha platform
963 * bug fix: FILE-POSITION works much better on string input and
964 output streams. (thanks to Nikodemus Siivola)
965 * bug fix: many threading/garbage collection symptoms sorted.
966 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
967 might be pseudo-atomic.
968 * internal change: Stopping for GC is now done with signals not
969 ptrace. GC is now done in whichever thread wanted it, instead of
971 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
973 * bug fix: result form in DO is not contained in the implicit
975 * incompatible change: ICR structure is changed; the value part of
976 CONTINUATION is now called LVAR; corresponding functions are
977 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
978 * added type deriver for ISQRT (thanks to Robert E. Brown).
979 * bug fix: better support for loading from the command line when an
980 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
982 * fixed some bugs revealed by Paul Dietz' test suite:
983 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
984 ** ROUND and FROUND now give the right answer when given very
985 small float arguments.
986 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
988 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
989 ** (GCD 0 <negative-integer>) returned <negative-integer>.
990 ** LCM should return a non-negative integer.
991 ** PARSE-INTEGER returned the index of a terminator instead of the
992 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
993 ** PARSE-INTEGER returned an incorrect index being applied to a
995 ** LCM with two arguments of 0 returns 0 rather than signalling
997 ** unsigned addition of a 32-bit constant with the high bit set no
998 longer causes an internal compiler error on the x86.
999 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1000 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1001 with negative last argument.
1002 ** byte specifiers with zero size and position no longer cause
1003 an error during type derivation.
1004 ** bignum multiplication on the Alpha platform now returns the
1006 * porting: The system now builds on SuSE AMD64, although it still
1007 generates a 32-bit binary.
1008 * .fasl file incompatibility: The fasl file version number has
1009 been incremented (because of the changes to internal compiler
1010 data structures referred to above).
1012 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1013 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1014 more accurately, on the Darwin kernel running on PowerPC hardware.
1015 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1016 * Compiler code deletion notes now signal a condition of type
1017 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1018 an associated MUFFLE-WARNING restart.
1019 * The compiler now performs limited argument count validation of
1020 constant format strings in FORMAT, and where appropriate in ERROR,
1021 CERROR and WARN. (thanks to Gerd Moellmann)
1022 * New ASDF-INSTALL contrib can be used for automatic download and
1023 installation of third-party Lisp code from CCLAN or other sites
1025 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1026 which forces another thread to execute a function supplied by the
1028 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1029 now accept and act upon their :ELEMENT-TYPE keyword argument.
1030 (reported by Martin Atzmueller, Edi Weitz)
1031 * bug fix: FILE-POSITION now accepts position designators up to
1032 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1033 is the greater. (thanks to Patrik Nordebo)
1034 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1035 Kalle Olavi Niemitalo)
1036 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1037 printable. (reported by Eric Marsden)
1038 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1039 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1040 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1041 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1042 (thanks to Stig Sandoe)
1043 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1044 UNREAD-CHAR now correctly includes the unread character in the
1045 target sequence. (thanks to Gerd Moellmann)
1046 * bug fix in threaded builds: the system can now be suspended and
1047 resumed by shell job control with minimal disruption.
1048 * bug fixes in times and timezones >2038AD
1049 * better handling of "where is GNU make?" problem in build scripts
1050 (thanks to Nikodemus Siivola)
1051 * new optimization: inside a named function any reference to a
1052 function with the same name is considered to be a self-reference;
1053 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1054 optimization quality.
1055 * new optimization on x86: logical functions and + now have
1056 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1057 used when the result is truncated to 32 bits.
1058 * VALUES declaration is partially enabled.
1059 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1060 * fixed some bugs revealed by Paul Dietz' test suite:
1061 ** The system now obeys the constraint imposed by
1062 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1063 types form a lattice under type intersection.
1064 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1065 ** ASSOC now ignores NIL elements in an alist.
1066 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1067 and (1+ MOST-POSITIVE-FIXNUM) answers.
1068 ** The addition of a method with invalid qualifiers to a generic
1069 function does not cause an error to be signalled immediately;
1070 a warning is signalled, and the error is generated only on
1071 calling the generic function.
1072 * changed .fasl file version number, in part to add type codes for
1073 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1074 obscure ANSI requirements
1076 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1077 * fixed bug 148: failure to inline-expand a local function left
1078 garbage, confusing the compiler.
1079 * fixed bugs 3cd: structure slot readers perform type check if the
1080 slot can have an invalid value (i.e. it is either not initialized
1081 or can be written with a less specific slot writer).
1082 * bug fix: the compiler now traps array references to elements off
1083 the end of an array; previously, the bounds checking in some
1084 circumstances could go off-by-one.
1085 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1087 * type declarations for array element types now obey the description
1088 on the CLHS page "Declaration TYPE", as per discussions on
1089 sbcl-help around 2003-05-08. This means that a declaration
1090 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1091 declaration, all references to BAR will be asserted or assumed
1092 (with THE, so dependent on compiler policy) to involve objects of
1093 type FOO. Note that no such declaration is implied in
1094 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1095 * declared types of functions from the "Conditions"
1096 chapter. (reported by Paul Dietz)
1097 * bug fix: CERROR accepts a function as its first argument.
1098 * bug fix: NTH an NTHCDR accept a bignum as index
1099 arguments. (reported by Adam Warner)
1100 * optimization: character compare routines now optimize comparing
1101 against a constant character. (reported by Gilbert Baumann)
1102 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1103 argument now works properly on the MIPS platform.
1104 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1105 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1106 declared to have a functional type, does not cause an error
1108 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1109 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1110 * fixed bug: if last continuation of a deleted block has a
1111 destination, this destination should be deleted too. (reported by
1113 * fixed a bug in the bootstrap process: the host compiler's values
1114 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1115 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1116 test case from Patrik Nordebo)
1117 * improved the ability of the disassembler on the PPC platform to
1118 provide helpful disassembly notes.
1119 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1120 the class in more cases than previously.
1121 * bug fix: FILE-POSITION now understands :START and :END for
1122 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1123 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1124 8) data. (thanks to David Lichteblau)
1125 * bug fix: it is possible to add a method to a generic function
1126 without lambda list.
1127 * bug fix: reader failed to signal END-OF-FILE inside an
1128 object representation. (reported by Nikodemus Siivola)
1129 * fixed some bugs revealed by Paul Dietz' test suite:
1130 ** LAST and [N]BUTLAST should accept a bignum.
1131 ** condition slot accessors are methods.
1132 ** (VECTOR NIL) is a subtype of STRING.
1134 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1135 * minor incompatible change: some nonsensical specialized lambda
1136 lists (used in DEFMETHOD) which were previously ignored now signal
1138 * minor incompatible change: the system is now aware of the types of
1139 variables in the COMMON-LISP package, and will signal errors for
1140 most violations of these type constraints (where previously they
1141 were silently accepted).
1142 * minor incompatible change: COMPILE-FILE now uses the freedom
1143 afforded (ANSI 3.2.2.3) to use derived function types for
1144 functions defined in the same file. This also permits the system
1145 to warn on static type mismatches and function
1146 redefinition. (Currently it does not work with high DEBUG level.)
1147 * minor incompatible change: VALUES declaration is disabled.
1148 * When issuing notes, the compiler now signals a condition of type
1149 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1150 restart for use in user handlers. It is expected that the
1151 COMPILER-NOTE condition will eventually become a condition
1152 supertype to a hierarchy of note types, which will then be
1153 handleable in a similar fashion. However, at the moment, no such
1154 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1155 but it's an implementation detail, not a classification for the
1157 * Changes in type checking closed the following bugs:
1158 ** type checking of unused values (192b, 194d, 203);
1159 ** template selection based on unsafe type assertions (192c, 236);
1160 ** type checking in branches (194bc).
1161 * A short form of VALUES type specifier has ANSI meaning (it has
1162 increased the number of situations when SBCL cannot perform type
1164 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1165 keywords or constants is permissible.
1166 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1167 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1168 classes. (thanks to Antonio Martinez)
1169 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1170 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1171 argument to a pretty-print stream. Instead, it leaves any such
1172 operations to DESCRIBE-OBJECT methods.
1173 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1175 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1176 lambda lists. (thanks to David Lichteblau)
1177 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1178 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1179 (thanks to Gerd Moellmann)
1180 * fixed bug 246: increased compilation speed of long
1181 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1183 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1185 * DEFCONSTANT now throws a condition of type
1186 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1187 constant to a non-EQL value; CONTINUE and ABORT restarts
1188 respectively change and preserve the value.
1189 * fixed bug 63: The code walker, part of the implementation of CLOS,
1190 is now better at handling symbol macros.
1191 * bug fix: There is no longer an internal implementation type named
1192 CL:LENGTH. (reported by Raymond Toy)
1193 * bug fix: In macro-like defining macros/special operators the
1194 implicit block does not enclose the lambda list.
1195 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1196 enclosed in parentheses) are not suitable as type specifiers, and
1197 their use properly signals an error now.
1198 * bug fix: An argument count mismatch for a type specifier in code
1199 being compiled no longer causes an unhandled error at compile
1200 time, but signals a compile-time warning.
1201 * fixed simple vector readable printing
1202 * bug fix: DESCRIBE takes more care over whether the class
1203 precedence list slot of a class is bound before accessing it.
1204 (reported by Markus Krummenacker)
1205 * bug fix: FORMATTER can successfully compile pretty-printer format
1206 strings which use variants of the ~* directive inside.
1207 * bug fix: SEARCH now applies its TEST predicate to the elements of
1208 the arguments in the correct order. (thanks to Wolfhard Buss)
1209 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1210 in type inference. (thanks to Robert E. Brown)
1211 * bug fix: Reading in symbols with an explicit package name of ""
1212 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1213 * fixed some bugs revealed by Paul Dietz' test suite:
1214 ** NIL is now allowed as a structure slot name.
1215 ** Arbitrary numbers, not just REALs, are allowed in certain
1216 circumstances in LOOP for-as-arithmetic clauses.
1217 ** Multiple class redefinitions before slot access no longer
1218 causes a type error.
1219 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1220 association between the name and a class.
1221 ** Generic functions with non-standard method-combination and over
1222 five methods all of which return constants no longer return NIL
1223 after the first few invocations. (thanks to Gerd Moellmann)
1224 ** CALL-NEXT-METHOD with no arguments now passes the original
1225 values of the arguments, even in the presence of assignment.
1226 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1228 ** DISASSEMBLE works with closures and funcallable instances.
1229 ** ADD-METHOD now returns the generic function, not the new method.
1230 ** FIND-METHOD signals an error if the lengths of the specializers
1231 is incompatible with the generic function, even if the ERRORP
1233 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1234 which its argument is a member.
1235 ** DEFCLASS only redefines the class named by its class-name
1236 argument if that name is the proper name of the class;
1237 otherwise, it creates a new class.
1238 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1239 of the UNBOUND-SLOT condition to the name of the slot.
1240 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1241 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1242 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1243 SLOT-MAKUNBOUND in the specified fashion.
1245 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1246 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1247 cross-compilation host. As a consequence, we can now bootstrap our
1248 way up to SBCL starting with a bare gcc toolchain and human-readable
1249 source code (first the source to CLISP, then the source to SBCL).
1250 * A contributed module containing a partial implementation of the
1251 simple-streams interface has been included. (thanks to Rudi
1253 * A contributed module implementing the RFC1321 Message Digest
1254 Algorithm, known as MD5, has been included.
1255 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1256 no longer has any effect, as the code controlled by this feature
1257 has been deleted. (As far as we know, no-one has ever built using
1258 this feature, and its semantics were confused in any case).
1259 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1260 work on conditions (as required by the ANSI specification),
1261 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1262 expected behaviour on conditions. Users should note, however,
1263 that such behaviour is not required by the ANSI specification,
1264 and so use of this behaviour may render their code unportable.
1265 * fixed some bugs revealed by Paul Dietz' test suite:
1266 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1268 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1269 ** SXHASH on condition objects no longer returns NIL.
1270 ** :ALLOCATION :CLASS slots are better treated; their values are
1271 updated on class redefinition, and initforms inherited from
1272 superclasses are applied.
1273 ** REMOVE-METHOD returns its generic function argument even when
1274 no method was removed.
1275 ** SHARED-INITIALIZE now initializes the values of the requested
1276 slots, including those with :ALLOCATION :CLASS.
1277 ** ALLOCATE-INSTANCE now works on structure classes defined via
1278 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1280 ** SLOT-EXISTS-P now works on conditions, as well as structures
1282 ** MAKE-LOAD-FORM now has the required methods on
1283 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1284 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1285 keyword, and now implements the SLOT-NAMES argument.
1286 ** methods with &OPTIONAL arguments no longer allow too many
1287 arguments to be passed in the call without error.
1288 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1289 option is consistent with the required arguments of the generic
1290 function lambda list.
1291 * bug fix: REQUIRE accepts a string designator. (Thanks to
1293 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1294 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1295 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1296 * bug fix: The compiler now checks for duplicated variables in macro
1298 * bug fix: SETQ on globals returns the correct value.
1299 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1300 not a COMPILER-ERROR (followed by some other strange error on
1301 choosing the CONTINUE restart).
1302 * bug fix: make.sh and friends are now more consistent in the way that
1303 they look for GNU "make".
1305 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1306 * experimental native threads support (on x86 Linux >=2.4 only).
1307 This is not compiled in by default: you need to add :SB-THREAD to
1308 the target features. See the "Beyond ANSI" chapter of the manual
1310 * fix for longstanding nonANSIism: The old distinction between
1311 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1312 The return value from CL:FIND-CLASS is now a CLOS class, and
1313 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1314 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1315 * An interface to the MetaObject Protocol, as described in Kiczales,
1316 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1317 MIT Press, 1991, is available from the SB-MOP package.
1318 * incompatible change: the SB-PCL package should now be considered
1319 a private implementation detail, and no longer a semi-private MOP
1321 * minor incompatible change: due to rearrangement for threads, the
1322 control stack and binding stack are now allocated at arbitrary
1323 addresses instead of being hardcoded per-port. Users affected by
1324 this probably have to be doing advanced things with shared
1325 libraries, and will know who they are.
1326 * minor incompatible change: Previously, all --eval forms used were
1327 processed with READ before any of them were processed with EVAL.
1328 Now each --eval form is processed with both READ and EVAL before
1329 the next --eval form is processed. (Thus package operations like
1330 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1331 work as the user might reasonably expect.)
1332 * minor incompatible change: *STANDARD-INPUT* is now only an
1333 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1335 * minor incompatible change: Y-OR-N-P is now character-oriented,
1336 not line oriented. Also, YES-OR-NO-P now works without errors.
1337 (thanks to Antonio Martinez)
1338 * sb-aclrepl module improvements: an integrated inspector, added
1339 repl features, and a bug fix to :trace command.
1340 * Known functions, which cannot be open coded by the backend, are
1341 considered to be able to check types of their arguments. (fixing
1342 a bug report by Nathan J. Froyd)
1343 * fixed a bug in computing method discriminating functions: It is
1344 now possible to define methods specialized on classes which have
1345 forward-referenced superclasses. (thanks to Gerd Moellmann)
1346 * fixed evaluation order in optional entries (reported by Gilbert
1348 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1349 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1350 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1351 required-by-AMOP NAME argument, as well as CLASS and
1352 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1353 * fixed bug 20: DEFMETHOD can define methods using names that are
1354 not the proper names of classes to designate class specializers.
1355 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1356 Unix file descriptors, instead of blowing up. (thanks to Antonio
1358 * Garbage collection refactoring: user-visible change is that a
1359 call to the GC function during WITHOUT-GCING will not do garbage
1360 collection until the end of the WITHOUT-GCING. If you were doing
1361 this you were probably losing anyway.
1362 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1363 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1364 TYPEP the latter but not the former.
1365 * The compiler issues a full WARNING for calls to undefined functions
1366 with names from the CL package.
1367 * MAP-INTO for a vector destination is open coded. (reported by
1368 Brian Downing on c.l.l)
1369 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1370 documentation string.
1371 * fixed some bugs revealed by Paul Dietz' test suite:
1372 ** COPY-ALIST now signals an error if its argument is a dotted
1374 ** Condition slots are now accessed more correctly in the presence
1375 of multiple initargs for a given slot.
1376 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1377 exclude restarts of the same name associated with a different
1379 ** DEFCLASS of forward-referenced classes with another
1380 forward-referenced class in the superclasses list no longer
1382 ** Condition slots are now initialized once each, not multiple
1383 times. (thanks to Gerd Moellmann)
1384 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1386 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1387 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1388 arguments contain duplicated elements.
1389 ** RESTART-CASE understands local macros.
1390 ** RESTART-CASE associates exactly its own restarts with a condition.
1391 ** ENDP in safe mode checks its argument to be of type LIST.
1392 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1393 in question is unbound.
1394 ** Optimized MAKE-INSTANCE functions no longer cause internal
1395 assertion failures in the presence of duplicate initargs.
1396 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1397 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1398 propagated even in the ctor optimized implementation of
1400 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1402 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1403 * a better implementation of SXHASH on (simple) bit vectors,
1404 measured both in execution speed and in distribution of results
1405 over the positive fixnums, has been installed. Likewise, a better
1406 implementation of EQUAL for simple bit vectors is now available.
1407 * fixed CEILING optimization for a divisor of form 2^k.
1408 * fixed bug 240 (emitting extra style warnings "using the lexical
1409 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1410 by Antonio Martinez)
1411 * fixed SXHASH, giving different results for NIL depending on type
1412 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1413 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1414 variable. (found by Rolf Wester)
1415 * SBCL does not ignore type declarations for special
1416 variables. (reported by rif on c.l.l 2003-03-05)
1417 * some bug fixes in contrib/sb-aclrepl/
1418 * fixed some bugs revealed by Paul Dietz' test suite:
1419 ** a bug in the CONS type specifier, whereby the CAR and CDR
1420 types got intertwined, has been fixed;
1421 ** the type system is now able to reason about the interaction
1422 between INTEGER and RATIO types more completely;
1423 ** APPEND, [N]REVERSE and NRECONC check that those their
1424 arguments, which must be proper lists, are really so;
1425 ** An array specialized to be unable to hold elements has been
1426 implemented, as required -- yes, really -- by ANSI;
1427 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1428 on malformed property lists;
1430 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1431 * incompatible packaging change: in line with Unix convention,
1432 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1433 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1434 * REQUIRE and PROVIDE are now optionally capable of doing something
1435 useful. See the documentation string for REQUIRE.
1436 * infrastructure for a managed SBCL contrib system: contributed
1437 modules in this release include:
1438 ** the ASDF system definition facility;
1439 ** an interface to the BSD Sockets API;
1440 ** an ACL-like convenience interface to the repl;
1441 (thanks to Kevin Rosenberg)
1442 ** an implementation of ROTATE-BYTE, with efficient implementation
1444 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1445 gives a closer approximation to the right answer than previously.
1446 (thanks to Raymond Toy)
1447 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1448 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1449 optional environment arguments, as required by ANSI.
1450 * fixed bugs in other functions taking environment objects, allowing
1451 calls with an explicit NIL environment argument to be compiled
1453 * fixed bug 228: primary return values from
1454 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1455 COMPILE or FUNCTION.
1456 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1457 :TYPE will no longer signal errors on innocuous objects.
1458 * fixed bug 231b: SETQ is better at respecting type declarations in
1459 the lexical environment.
1460 * fixed a bug in DEFCLASS: classes named by symbols with no or
1461 unprintable packages can now be defined.
1462 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1463 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1464 * fixed compiler failure related to checking types of functions.
1465 (reported by Robert E. Brown)
1466 * the compiler is now much more consistent in its error-checking
1467 treatment of bounding index arguments to sequence functions: in
1468 (SAFETY 3) code, errors will be signalled in almost all cases if
1469 invalid sequence bounding indices are passed to functions defined
1470 by ANSI to operate on sequences.
1471 * fixed a bug in the build procedure: documentation of SBCL-specific
1472 packages is now preserved and available in the final Lisp image.
1473 * lifted FDEFINITION lookup out of loops in the implementation of
1474 many list operations. (thanks to Robert E. Brown)
1475 * fixed a bug in the reader: the #n# reader macro now works for
1476 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1477 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1478 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1479 E. Brown and Tony Martinez respectively)
1480 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1481 SLOT-BOUNDP: in particular, it is now able to optimize them much
1482 better, and is now not vulnerable to having packages renamed.
1483 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1484 always called when a slot is not present in an instance. (thanks
1486 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1487 longer signal an error on unreading a character following EOF on
1488 the previous constituent stream. (thanks to Tony Martinez)
1489 * fixed some bugs revealed by Paul Dietz' test suite:
1490 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1491 not just nonnegative fixnums;
1492 ** the logical bit-array operators such as BIT-AND now accept an
1493 explicit NIL for their "opt-arg" argument (to indicate a
1494 freshly-consed result bit-array);
1495 ** ELT now signals an error on an invalid sequence index in safe
1497 ** the type system is now cleverer about negations of numeric
1498 types, and consequently understands the BIGNUM and RATIO types
1500 ** the type system is now cleverer about the interaction between
1501 INTEGER and RATIO types: while bugs still remain, many more
1502 cases are accurately computed;
1503 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1504 if it is in the last clause;
1505 ** CONSTANTLY now correctly returns a side-effect-free function in
1507 ** DECLARE is no longer treated as a special-operator; in
1508 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1509 * incremented fasl file version number due to the change in the
1510 DEFSTRUCT-SLOT-DESCRIPTION structure.
1512 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1513 * minor incompatible change: code processed by the "interpreter" or
1514 EVAL now has a compilation optimization policy of (DEBUG 2)
1515 (changed from (DEBUG 1)) to improve debuggability of interactive
1516 development, and to allow the use of the debug RETURN command in
1518 * an experimental implementation of the RETURN command for the
1519 debugger has been included. (thanks to Frederik Kuivinen)
1520 * fixed bug 62: constraints were not propagated into a loop.
1521 * fixed bug in embedded calls of SORT (reported and investigated by
1523 * fixed some bugs revealed by Paul Dietz' test suite:
1524 ** printing and reading of arrays with some dimensions having
1525 length 0 (thanks to Gerd Moellmann);
1526 ** BOA constructor with &AUX argument without a default value does
1527 not cause a type error;
1528 ** CONSTANTP now returns true for all self-evaluating objects.
1530 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1531 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1532 accessors that are related by inheritance, as specified in the
1533 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1534 Valtteri Vuorikoski)
1535 * The compiler is now able to inline functions that were defined in
1536 a complex lexical environment (e.g. inside a MACROLET).
1537 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1539 * Support for the upcoming FreeBSD-5.0 release has been included.
1540 (thanks to Dag-Erling Smorgrav)
1541 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1542 effect when it is not in a toplevel context.
1543 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1544 stand a better chance of being correct. (thanks to Gerd
1546 * fixed bug in COERCE, which now signals an error on coercing a
1547 rational to a bounded real type which excludes the expected
1549 * The compiler is now able to derive types more accurately from the
1550 COERCE and COMPILE functions.
1551 * fixed bug 223: functional binding is considered to be constant
1552 only for symbols in the CL package.
1553 * fixed bug 231: SETQ did not check the type of a variable being set
1554 (reported by Robert E. Brown)
1555 * A new optimization for MAKE-INSTANCE has been included, fixing
1556 various bugs (including relating to :ALLOCATION :CLASS slots and
1557 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1559 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1560 ** As required by ANSI, LOOP now disallows anonymous collection
1561 clauses such as COLLECT I in conjunction with aggregate boolean
1562 clauses such as THEREIS (= I 1);
1563 ** LOOP now signals an error when any variable is reused in the
1564 same loop (including the potentially useful construct analogous
1565 to WITH A = 1 WITH A = (1+ A);
1566 ** IT is only a special loop symbol within the first clause of a
1567 conditional loop clause;
1568 ** LOOP with a typed iteration variable over a hashtable now
1569 signals a type error iff it should.
1570 * fixed some other bugs revealed by Paul Dietz' test suite:
1571 ** FILE-STREAM now names the class previously known as FD-STREAM;
1572 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1573 argument) no longer signals an error;
1574 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1575 of the slot symbol, rather than using the current package
1576 ((:CONC-NAME "") continues to intern the slot's name in the
1578 * incremented fasl file version number, because of the incompatible
1579 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1580 of the new implementation of DEFINE-COMPILER-MACRO.
1582 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1583 * Support for building SBCL for MIPS platforms running in
1584 little-endian mode has now been checked in, and basic
1585 functionality on said platforms verified.
1586 * minor incompatible change: PCL now records the pathname of a file
1587 in which methods and the like are defined, rather than its
1589 * minor incompatible change: TRUENAME now considers the truename of
1590 a file naming a directory to be the pathname with :DIRECTORY
1591 component indicating that directory.
1592 * minor incompatible change: a NAMED clause in the extended form of
1593 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1594 reason for the previous behaviour is unclear.
1595 * more systematization and improvement of CLOS and MOP conformance
1596 in PCL (thanks to Gerd Moellman and Pierre Mai):
1597 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1599 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1600 primary methods with no specializers;
1601 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1603 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1604 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1605 FORWARD-REFERENCED-CLASSes; error reporting on
1606 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1608 ** SXHASH on CLOS instances now uses a slot internal to the
1609 instance to return different numbers on distinct instances,
1610 while preserving the same return value through invocations of
1612 ** DEFMETHOD signals errors when methods with longer incongruent
1613 lambda lists are added to generic functions;
1614 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1615 CLASS, as specified in AMOP;
1616 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1617 on the order returned by the primary method for classes of
1618 class STANDARD-CLASS;
1619 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1620 * fixed some bugs shown by Paul Dietz' test suite:
1621 ** DOLIST puts its body in TAGBODY;
1622 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1624 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1625 value producing form;
1626 ** if more variables are given to PROGV than values, extra
1627 variables are bound and made to have no value;
1628 ** NSUBSTITUTE on list arguments gets the right answer with
1630 ** ELT signals an error of type TYPE-ERROR when the index argument
1631 is not a valid sequence index;
1632 ** LOOP signals (at macroexpansion time) an error of type
1633 PROGRAM-ERROR when duplicate variable names are found;
1634 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1635 ** FUNCALL of special-operators now cause an error of type
1637 ** PSETQ now works as required in the presence of side-effecting
1638 symbol-macro places;
1639 ** NCONC accepts any object as its last argument;
1640 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1642 ** loop-for-as-package does not require a package to be explicitely
1644 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1646 * fixed bug 166: compiler preserves "there is a way to go"
1647 invariant when deleting code.
1648 * fixed bug 172: macro lambda lists with required arguments after
1649 &REST arguments now cause an error to be signalled. (thanks to
1651 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1652 bug. (thanks to Matthew Danish)
1653 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1655 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1656 arguments are lexically rebound. (thanks to Gerd Moellmann and
1658 * fixed bug 194: error messages are now more informative when there
1659 is no primary method applicable in a call to a generic
1660 function. (thanks to Gerd Moellmann)
1661 * fixed bug in command line argument checking (thanks to Julian
1663 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1664 (thanks to Matthew Danish)
1665 * incremented fasl file version number, because of the
1666 SXHASH-related changes in the layout of CLOS data structures
1668 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1669 * minor incompatible change: The runtime (the Unix executable named
1670 "sbcl") is now much pickier about the .core files it will load.
1671 Essentially it now requires .core files to descend from the same
1672 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1673 as the runtime does. (The intent is to prevent the crashes which
1674 can occur, and which can even be reported as mysterious failures,
1675 when people patch the sources or change the build parameters
1676 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1677 sbcl and .core files.)
1678 * fixed bug: VALUES-LIST is no longer optimized away.
1679 * fixed bug 142: The FFI conversion of C string values to Lisp
1680 string values no longer conses excessively. (thanks to Nathan
1681 Froyd porting Raymond Toy's fix to CMU CL)
1682 * began to systematize and improve MOP conformance in PCL (thanks to
1683 Nathan Froyd, Gerd Moellman and Pierre Mai):
1684 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1686 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1687 ** FINALIZE-INHERITANCE is now called on class finalization;
1688 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1689 argument precedence order.
1690 * fixed bug 202: The compiler no longer fails on functions whose
1691 derived types contradict their declared type.
1692 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1693 so it can be non-toplevel.
1694 * The fasl file version number has changed (because of the new
1695 implementation of DEFMACRO).
1696 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1697 safe code, that any length requirement by their type-specifier
1698 argument is valid. The exceptions to this are described in bug
1700 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1701 functions now check the types of their inputs as required by ANSI.
1702 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1703 introduced symbol is DECLAREd to be SPECIAL.
1704 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1705 * fixed bug: :COUNT argument to sequence functions may be negative
1706 * fixed bug: body of DO-SYMBOLS may contain declarations
1707 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1708 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1709 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1710 symbol macro only once
1711 * fixed printing of call frame when argument list is unavailable
1712 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1713 * compiler no longer signals WARNING on unknown keyword
1716 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1717 * A beta-quality port to the mips architecture running Linux,
1718 based on the old CMUCL backend, has been made. It has been tested
1719 on a big-endian kernel, and works sufficiently well to be able to
1720 rebuild itself; it has not been tested in little-endian mode.
1721 * fixed an inconsistency between gencgc.c and purify.c which made
1722 dumping/loading .core files unreliable
1723 * fixed bug 120a: The compiler now deals correctly with IFs where
1724 the consequent is the same as the alternative, instead of
1725 misderiving the return type. (thanks to Alexey Dejneka)
1726 * fixed bug 113: Logical pathnames are now dumpable (the logical
1727 host is resolved at load-time, throwing an error if it is not
1729 * fixed bug 174: FORMAT's error message is slightly clearer when a
1730 non-printing character is used in a format directive.
1731 * fixed several bugs in compiler checking of type declarations, i.e.
1732 violations of the Python "declarations are assertions" principle
1733 (thanks to Alexey Dejneka)
1734 * fixed several bugs in PCL's error checking (thanks to Gerd
1736 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1738 * fixed bug in compilation of functions as first class values
1739 (thanks to Antonio Martinez-Shotton)
1740 * The compiler's handling TYPE-ERRORs which it can prove will
1741 inevitably happen at runtime has been cleaned up and corrected
1742 in several ways. (thanks to Alexey Dejneka)
1743 * improved argument type checking for various basic arithmetic
1744 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1745 lavished on them in the past that they can be compiled in many
1746 ways in different special cases
1747 * fixed bug 181: compiler checks validity of user supplied type
1749 * cleaned up code flushing in optimization: Function calls which
1750 should signal errors for safety purposes (e.g. which ANSI says
1751 should signal errors when their arguments are of incorrect type)
1752 are no longer optimized away.
1753 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1754 * incremented fasl file version number, because changes in the
1755 implementation of sequence functions like COERCE caused
1756 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1757 in old inline expansions) to become undefined. (Actually these
1758 changes were later undone, so we might very well be binary
1759 compatible with 0.7.7 after all, but leaving the version number
1760 incremented seemed like the simplest and most conservative
1763 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1764 * An alpha-quality port to the parisc architecture running Linux,
1765 based on the old CMUCL backend, has been made. This, even more so
1766 than the other backends, should be considered still a work in
1767 progress; known problems include that the Linux kernel in 64-bit
1768 mode does not propagate the correct sigcontext structure to
1769 userspace, and consequently SBCL on a parisc64 kernel will not
1771 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1772 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1773 suppressed.) Also now that the compiler is looking at declarations
1774 in the environment, it checks optimization declarations as well,
1775 and suppresses inlining when (> DEBUG SPEED).
1776 * More fixes have been made to treatment of floating point exception
1777 treatment and other Unix signals. In particular, floating point
1778 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1779 * The detection and handling of control stack exhaustion (infinite
1780 or very deeply nested recursion) has changed. Stack exhaustion
1781 detection is now done by write-protecting pages at the OS level
1782 and applies at all optimization settings; when found, a
1783 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1784 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1785 be caught using IGNORE-ERRORS.
1786 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1787 correctly when transforming recursive LABELS functions to LETs.
1788 (thanks to Alexey Dejneka)
1789 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1790 that are names of constants or global variables.
1791 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1792 alien routines with docstrings.
1793 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1794 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1796 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1797 (thanks to Eric Marsden)
1798 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1799 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1800 * Bug fix: The SPARC backend can now compile functions involving
1801 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1802 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1803 to a functions expecting a PATHNAME-DESIGNATOR.
1804 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1805 lambda lists. (thanks to Alexey Dejneka)
1806 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1807 functions so that EXPORTing the name of the class doesn't cause
1808 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1809 to Antonio Martinez for reporting this)
1810 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1811 constant-folded without failing an assertion. (thanks to Einar
1812 Floystad Dorum for reporting this)
1813 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1814 types are now dealt with more correctly by the compiler.
1815 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1816 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1817 behaviour analogous to RENAME-FILE. This puts its behaviour more
1818 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1820 * The fasl file version number has changed again. (because of the
1821 bug fix involving the names of PCL MAKE-INSTANCE functions)
1823 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1824 * bug fix: Floating point exceptions are treated much more
1825 consistently on the x86/Linux and PPC/Linux platforms.
1826 * Array initialization with :INITIAL-ELEMENT is now much faster for
1827 cases when the compiler cannot open code the array creation, but
1828 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1829 array accesses have also seen a speed increase.
1830 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1831 specified. (This was at the root of some bad interactions between
1832 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1834 * bug fix: Internal error arguments for undefined functions are now
1835 computed correctly on the PPC/Linux platform.
1836 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1837 Raymond Toy's patch for CMU CL)
1838 * Support for the Solaris 9 operating environment has been included
1839 (thanks to Daniel Merritt)
1840 * A very ugly but hopefully complete draft of the missing FFI chapter
1841 of the manual has been created by reformatting the corresponding
1842 CMU CL manual chapter into (currently very ugly and incoherent)
1843 DocBook and bringing it up to date for SBCL behavior. Thus, the
1844 manual is now essentially complete, at least by my extreme
1845 once-and-only-once standards, whereby it's acceptable to refer to
1846 the doc strings of SB-EXT functions as the primary documentation.
1847 * The fasl file version number has changed again, due to cleanup of
1848 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1849 is no longer a static symbol.)
1851 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1852 * SBCL now builds with OpenMCL (version 0.12) as the
1853 cross-compilation host; also, more progress has been made toward
1854 bootstrapping under CLISP.
1855 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1857 * bug 158 fixed: The compiler can now deal with integer loop
1858 increments different from 1; fixing this turned out also to fix
1860 * bug 169 fixed: no more bogus warnings about using lexical bindings
1861 despite the presence of perfectly good SPECIAL declarations (thanks
1862 to David Lichteblau)
1863 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1864 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1865 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1867 * bug 180 fixed: Method combination specifications no longer ignore
1868 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1869 * bug fix: Structure type predicate functions now check their argument
1870 count as they should.
1871 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1872 correctly. (thanks to Pierre Mai)
1873 * minor incompatible change: The --noprogrammer option is deprecated
1874 in favor of the new --disable-debugger option, which is very similar.
1875 (The major difference is that it takes effect at a slightly different
1876 time at startup, causing handling of errors in --sysinit and
1877 --userinit files will be affected differently.) The
1878 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1879 been added to allow this functionality to be controlled from ordinary
1880 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1881 maintainers, who might want to run non-interactive scripts to
1882 build SBCL cores which will later be used interactively.)
1883 * minor incompatible change: The LOAD function no longer, when given
1884 a wild pathname to load, loads all files matching that pathname.
1885 Instead, an error of type FILE-ERROR is signalled.
1887 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1888 * bug 147 fixed: The compiler preserves its block link/count
1889 invariants more correctly now so that it doesn't crash. (thanks
1891 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1893 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1894 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1895 no-value-for-this-character as NIL (as specified by ANSI).
1896 * HOST-NAMESTRING on physical pathnames now returns a string that is
1897 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1898 (thanks to Christophe Rhodes)
1899 * The Alpha port handles icache flushing more correctly. (thanks to
1901 * More progress has been made toward bootstrapping under CLISP. (thanks
1902 to Christophe Rhodes)
1903 * The fasl file format has changed again, because dynamic loading
1904 on OpenBSD (which has non-ELF object files) motivated some cleanups
1905 in the way that foreign symbols are transformed and passed around.
1906 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1907 is no longer treated as whitespace by the reader, but instead as
1908 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1909 |A
\7fB|, instead of A as it used to.
1911 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1912 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1913 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1914 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1915 this as of 0.7.1.45, but was left out of the previous news section
1916 (thanks to Dan Barlow)
1917 * SBCL now runs on the Solaris operating system on SPARC architectures
1918 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1919 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1920 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1921 * SPARC backend cleanups, allowing builds of cores optimized for V8
1922 and V9 SPARCS, and also emission of code targeted to a particular
1923 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1925 * SBCL is closer to bootstrapping under CLISP, thanks to various
1926 fixes by Christophe Rhodes.
1927 * The fasl file format has changed again, to allow the compiler's
1928 INFO database to support symbol macros.
1929 * The user manual (in doc/) is formatted into HTML more nicely.
1930 (thanks to coreythomas)
1931 * The system is smarter about SUBTYPEP relationships, especially
1932 those involving NOT types (including types such as ATOM which are
1933 represented internally using NOT types). Thus SUBTYPEP is less
1934 likely to return (VALUES NIL NIL) in general, and in particular
1935 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1937 * The fasl file format has changed again, because the internal
1938 representation of types now includes a new slot to support the new
1939 SUBTYPEP-of-NOT-types logic.
1940 * (not a change in the main branch of SBCL, but a related prototype
1941 which can hopefully be merged into the main branch of SBCL in the
1942 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1943 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1944 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1945 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1946 dispatch macro character combinations. (thanks to Alexey Dejneka)
1947 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1948 that superficially look like logical namestrings correctly.
1949 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1951 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1952 default supplied-p) for &optional and &key arguments. (thanks to
1955 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1956 * incompatible change: The compiler is now less aggressive about
1957 tail call optimization, doing it only when (> SPACE DEBUG) or
1958 (> SPEED DEBUG). (This is an incompatible change because there are
1959 programs which relied on the old CMU-CL-style behavior to optimize
1960 away their unbounded recursion which will now die of stack overflow.)
1961 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1962 for non-GENCGC systems has been increased to 20M (since that
1963 seems much closer to the likely performance optimum for modern
1964 systems than the old 4M value was)
1965 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1966 and *DEBUG-PRINT-LEVEL*
1967 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1968 of CMU CL's support for SPARC, and various endianness and other
1969 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1970 * new syntactic sugar for the Unix command line: --load foo.bar is now
1971 an alternate notation for --eval '(load "foo.bar")'.
1973 ** The system now detects stack overflow and handles it gracefully,
1974 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1975 optimization settings. (This is a good thing in general, and
1976 its introduction in this version should be particularly timely
1977 for anyone whose code fails because of suppression of tail
1979 ** The system now hunts for the C variable "environ" in a more
1980 devious way, to avoid segfaults when the C library version differs
1981 between compile time and run time. (thanks to Christophe Rhodes)
1982 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1983 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1984 ** The compiler no longer issues bogus style warnings for undefined
1985 classes in the same source file as the DEFCLASSes which defined
1986 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1988 ** fixes in CONDITION class precedence list for undefined function
1989 errors (thanks to Alexei Dejneka)
1990 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1991 correctly. (thanks to Dan Barlow)
1992 ** portability fixes aiming at bootstrapping under CLISP (thanks
1993 to Dave McDonald and Christophe Rhodes)
1994 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1995 ** fixes in type translation and and type inference (thanks to
1997 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1998 ** various fixes in the new ports (thanks to Dan Barlow)
1999 * several changes related to debugging:
2000 ** suppression of tail recursion, as noted above
2001 ** stack overflow detection, as noted above
2002 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2003 is now the default. (For some time encapsulation has been more
2004 reliable than the breakpoint-based :ENCAPSULATE NIL
2005 implementation, at least on X86 systems; and I just noticed that
2006 encapsulation also seems closer to the spirit of the ANSI
2009 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2011 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2012 up properly again. (There was a packaging bug in 0.7.0 which
2013 left their definitions in SB-SYS::LOAD-FOREIGN and
2014 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2015 vital for most things which interface to C-level interfaces,
2016 like extensions working with sockets or databases or
2017 Perl-compatible regexes or whatever, and the need to fix
2018 this bug is the main reason that 0.7.1 was released so
2020 ** DEFGENERIC is now choosier about the methods it redefines, so that
2021 reLOADing a previously-LOADed file containing DEFGENERICs does
2022 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2023 cycle now works as it should. (thanks to Alexey Dejneka)
2024 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2025 fixed by Christophe Rhodes. (He actually submitted this patch
2026 months ago, and I delayed until after 0.7.0.)
2027 ** Bug 111 (internal compiler confusion about runtime checks on
2028 FUNCTION types) was fixed by Alexey Dejneka.
2029 * Some internal cleanups (getting rid of variables which aren't
2030 needed now that the byte interpreter is gone) caused the fasl
2031 file format number to change again.
2033 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2034 * major incompatible change: The default fasl file extension, i.e. the
2035 default extension for files produced by COMPILE-FILE, has changed
2036 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2038 ** There are many changes in the implementation of the compiler.
2039 SBCL is now essentially a compiler-only implementation of ANSI
2040 Common Lisp. EVAL still "interprets" a few special cases, but
2041 almost all the interesting cases are handled by creating
2042 a LAMBDA expression, calling COMPILE on it, then calling
2043 FUNCALL on the result.
2044 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2045 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2046 Since the code is newer, there might still be some new bugs
2047 (though not as many as before Martin Atzmueller's fixes:-). But
2048 the new code is substantially simpler and clearer, and hopefully
2049 any remaining bugs will be simpler, less fundamental, and more
2050 fixable then the bugs in the old code.
2051 ** The revised compiler is still a little unsteady on its feet.
2053 *** The debugging information it produces (particularly the names
2054 of FUNCTION objects) is sometimes much less useful than what
2055 the old compiler produced.
2056 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2057 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2058 within a MACROLET) has been temporarily weakened.
2059 ** There are new compiler optimizations for various functions:
2060 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2061 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2062 *** the math functions TRUNCATE, FLOOR, and CEILING
2063 *** the function-of-all-trades COERCE
2064 Mostly these should be transparent, but there's one
2065 potentially-annoying problem (bug 117): when the compiler
2066 inline-expands a function and does type analysis on the result,
2067 it can create control paths which have type mismatches, and
2068 when it can't prove that those control paths aren't taken,
2069 it will issue WARNINGs about the type mismatches. This is
2070 a particular problem in practice for the new sequence functions.
2071 It's not clear how this should be fixed, and for now, a
2072 workaround is given in the entry for 117 in the BUGS file.
2073 ** (Because of the interaction between the two previous items --
2074 occasional inlining problems and new inline expansions -- some
2075 of the new sequence function optimizations won't really kick in
2076 completely until debugging information, and then inlining, are
2077 straightened out in some future version.)
2078 * minor incompatible changes:
2079 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2080 DIRECTORY no longer implicitly promotes NIL slots of its
2081 pathname argument to :WILD. In particular, when you ask for the
2082 contents of a directory (which you used to be able to do without
2083 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2084 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2085 ** changes in behavior that ANSI explicitly defines to be
2086 implementation dependent:
2087 *** The new compiler-only implementation still conforms with ANSI,
2088 but acts a little different than before. Besides the obvious
2089 changes in performance tradeoffs (that the cost per form passed
2090 to EVAL has gone up, and the cost per form executed by EVAL
2091 has gone down), the behavior of the system changes a little
2092 because there are no longer any interpreted function objects.
2093 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2094 e.g. doing COMPILE on the output of interactive DEFUN is
2096 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2098 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2099 has changed from (:CL) to NIL.
2100 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2101 e.g. #\Newline and #\Tab, don't have names specified in the
2102 ANSI Common Lisp standard, is now based on their ASCII symbolic
2103 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2104 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2105 are no longer used for output.
2106 ** changes in internal implementation constants:
2107 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2108 4 million. (If your application spends a lot of time GCing and
2109 you have a lot of RAM, you might want to experiment with
2110 increasing it even more.)
2111 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2112 However, almost all old code should still continue to work without
2113 immediate update, as SB-C-CALL is now a (deprecated) nickname
2115 ** Old operator names in the style DEF-FOO are now deprecated in
2116 favor of new corresponding names DEFINE-FOO, for consistency with
2117 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2118 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2119 internal symbols, but a few supported extensions like
2120 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2121 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2122 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2123 etc. as you get deeper into recursive calls to the debugger
2124 command loop, instead of the old "5]", "5]]", "5]]]"
2125 sequence. (I was motivated to do this when squabbles between
2126 ILISP and SBCL left me very deeply nested in the debugger. In the
2127 short term, this change will probably provoke more ILISP/SBCL
2128 squabbles, but hopefully it will be an improvement in the long run.)
2129 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2130 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2131 function, since declaiming return types involving aliens is
2132 (1) annoyingly messy to do by hand and (2) vital to efficient
2133 compilation of code which calls such functions.
2134 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2135 longer reexported by the SB-EXT package. They're solely useful
2136 for alien code, so it seems more logical that you should get
2137 them from the SB-ALIEN package, not in SB-EXT.
2138 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2139 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2140 features. Instead, the code that they used to control is always
2141 built into the system.
2142 * many other bug fixes
2143 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2144 advantage of the new EVAL-WHEN stuff and to clean them up in
2145 general, and they are now more ANSI-compliant in a number of
2146 ways. Martin Atzmueller is responsible for a lot of this.
2147 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2149 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2150 *** correct ERROR type for various file operations
2151 *** some fixes for Lisp streams
2152 *** DEFMETHOD syntax checking
2153 *** changing old weird representation of debug information as
2154 strings (which, among their other deficiencies, don't transform
2155 correctly when you rename packages, and don't change their
2156 print representation when you change things like *PACKAGE*
2157 and *PRINT-LENGTH*) to symbols and lists of symbols
2158 He also made several improvements and fixed several bugs in DESCRIBE.
2159 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2161 *** misbehavior of WRITE-STRING/WRITE-LINE
2162 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2163 and several other LOOP problems as well
2164 *** DIRECTORY when similar filenames are present
2165 *** DEFGENERIC with :METHOD options
2166 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2167 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2168 *** argument ordering in FIND with :TEST option
2169 *** mishandled package designator argument in APROPOS-LIST
2170 *** various problems in the backquote readmacro
2171 *** a bug in APROPOS
2172 *** probably some others that I'm not describing very well here,
2173 since the CVS log documents them by reference to sbcl-devel
2174 messages, and the SourceForge archives aren't working well.:-(
2175 ** Dan Barlow improved the Alpha port (and is making progress on the
2176 PPC port, for those of you who think different).
2177 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2178 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2179 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2180 continues to work on the SPARC port (for those of you in a position
2181 to look down upon our little PC-compatible boxes from a great height).
2182 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2183 and uses that copy, rather than the current dynamic value, when
2184 it's trying to decide whether to truncate output. Thus e.g.
2185 (let ((*print-lines* 50))
2186 (pprint-logical-block (stream nil)
2188 (let ((*print-lines* 8))
2189 (print (aref possiblybigthings i) stream)))))
2190 should now truncate the logical block only at 50 lines, instead of
2191 often truncating it at 8 lines, as it did before.
2192 * The doc/cmucl/ directory, containing old CMU CL documentation
2193 from the time of the fork, is no longer part of the base system.
2194 SourceForge has shut down its anonymous FTP service, and with it
2195 my original plan for distributing the old CMU CL documentation
2196 there. For now, if you need these files you can download an old
2197 SBCL source release and extract them from it.
2198 * The fasl file version number changed again, for dozens of reasons,
2199 some of which are apparent above.
2201 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2202 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2203 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2204 patch, so that DISASSEMBLE output is much nicer.
2205 * The code in the SB-PROFILE package now seems reasonably stable.
2206 I still haven't decided what the final interface should look like
2207 (I'd like PROFILE to interact cleanly with TRACE, since both
2208 facilities use function encapsulation) but if you have a need
2209 for profiling now, you can probably use it successfully with
2210 the current CMU-CL-style interface.
2211 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2212 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2213 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2214 link now returns the dangling link itself, and for similar
2215 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2216 link itself. (In these cases the old code signalled an error and
2217 looped endlessly, respectively.) Thus, DIRECTORY now works even
2218 in the presence of dangling and cyclic symbolic links.
2219 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2220 is now a supported extension again, since the consensus on
2221 sbcl-devel was that it can be useful for ordinary development
2222 work, not just for debugging SBCL itself.
2223 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2224 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2225 that currently-defined functions might be redefined later with
2226 different return types.
2227 * Hash tables can be printed readably, as inspired by CMU CL code
2228 of Eric Marsden and SBCL code of Martin Atzmueller.
2229 * better error handling in CLOS method combination, thanks to
2230 Martin Atzmueller porting Pierre Mai's CMU CL patches
2231 * more overflow fixes for >16Mbyte I/O buffers
2232 * A bug in READ has been fixed, so that now a single Ctrl-D
2233 character suffices to cause end-of-file on character streams.
2234 In particular, now you only need one Ctrl-D at the command
2235 line (not two) to exit SBCL.
2236 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2238 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2239 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2240 does the right thing.
2241 * The implementation of some type tests, especially for CONDITION
2242 types, is now tidier and maybe faster, due to CMU CL code
2243 originally by Douglas Crosher, ported by Martin Atzmueller.
2244 * Some math functions have been fixed, and there are new
2245 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2246 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2247 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2248 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2249 likely to be built into sbcl-0.7.0.)
2250 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2251 CL, ported to SBCL by Martin Atzmueller)
2252 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2253 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2254 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2255 * There's a new slam.sh hack to shorten the edit/compile/debug
2256 cycle for low-level changes to SBCL itself, and a new
2257 :SB-AFTER-XC-CORE target feature to control the generation of
2258 the after-xc.core file needed by slam.sh.
2259 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2260 COMPILE-FILE is no longer supported, so that now every function
2261 gets an entry point, so that block compilation looks a little
2262 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2263 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2264 returns the number of bytes consed since the system started,
2265 rather than the number consed since the first time the function
2266 was called. (The new definition parallels ANSI functions like
2267 CL:GET-INTERNAL-RUN-TIME.)
2268 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2269 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2270 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2271 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2272 * Fasl file version numbers are now independent of the target CPU,
2273 since historically most system changes which required version
2274 number changes have affected all CPUs equally. Similarly,
2275 the byte fasl file version is now equal to the ordinary
2278 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2279 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2280 is no longer recognized. I apologize for this, because it was
2281 listed in SB-EXT as a supported extension, but I found that
2282 its existing behavior was poorly specified, as well as incorrectly
2283 specified, and it looked like too much of a mess to straighten it
2284 out. I have enough on my hands trying to get ANSI stuff to work..
2285 * many patches ported from CMU CL by Martin Atzmueller, with
2286 half a dozen bug fixes in pretty-printing and the debugger, and
2287 half a dozen others elsewhere
2288 * fixed bug 13: Floating point infinities are now supported again.
2289 They might still be a little bit flaky, but thanks to bug reports
2290 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2291 as flaky as they were.
2292 * The --noprogrammer command line option is now supported. (Its
2293 behavior is slightly different in detail from what the old man
2294 page claimed it would do, but it's still appropriate under the
2295 same circumstances that the man page talks about.)
2296 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2297 are now supported, and enabled by default. Thus, the compiler can
2298 handle many floating point and complex operations much less
2299 inefficiently. (Thus e.g. you can implement a complex FFT
2301 * The compiler now detects type mismatches between DECLAIM FTYPE
2302 and DEFUN better, and implements CHECK-TYPE more correctly, and
2303 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2304 thanks to patches from Martin Atzmueller.
2305 * various fixes to make the cross-compiler more portable to
2306 ANSI-conforming-but-different cross-compilation hosts (notably
2307 Lispworks for Windows, following bug reports from Arthur Lemmens)
2308 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2309 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2310 fixed thanks to Pierre Mai's CMU CL patches.
2311 * fixes to keep the system from overflowing internal counters when
2312 it tries to use i/o buffers larger than 16M bytes
2313 * fixed bug 45a: Various internal functions required to support
2314 complex special functions have been merged from CMU CL sources.
2315 (When I was first setting up SBCL, I misunderstood a compile-time
2316 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2317 * improved support for type intersection and union, fixing bug 12
2318 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2319 more obscure bugs as well
2320 * some steps toward byte-compiling non-performance-critical
2321 parts of the system, courtesy of patches from Martin Atzmueller
2322 * Christophe Rhodes has made some debian packages of sbcl at
2323 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2324 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2325 stable, but are nonetheless usable. When he's ready, I'd be happy
2326 to add them to the SourceForge "File Releases" section. (And if
2327 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2328 * new fasl file format version number (because of changes in
2329 internal representation of (OR ..) types to accommodate the new
2330 support for (AND ..) types, among other things)
2332 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2333 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2335 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2337 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2338 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2339 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2340 * DESCRIBE now works on structure objects again.
2341 * Most function call argument type mismatches are now handled as
2342 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2343 know whether the function will be redefined before the call is
2344 executed. (The compiler could flag local calls with full WARNINGs,
2345 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2346 it doesn't keep track of enough information to know whether calls
2347 are local in this sense.)
2348 * Compiler output is now more verbose, with messages truncated
2349 later than before. (There should be some supported way for users
2350 to override the default verbosity, but I haven't decided how to
2351 provide it yet, so this behavior is still controlled by the internal
2352 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2353 src/compiler/ir1util.lisp.)
2354 * Fasl file format version numbers have increased again, because
2355 support for the Gray streams extension changes the layout of the
2356 system's STREAM objects.
2357 * The Gray subclassable streams extension now works, thanks to a
2358 patch from Martin Atzmueller.
2359 * The full LOAD-FOREIGN extension (not just the primitive
2360 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2361 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2362 but like most other programs, it defaults to copying the Unix
2363 environment from the original process instead of starting the
2364 new process in an empty environment.
2365 * Extensions which manipulate the Unix environment now support
2366 an :ENVIRONMENT keyword option which doesn't smash case or
2367 do other bad things. The CMU-CL-style :ENV option is retained
2368 for porting convenience.
2369 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2370 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2372 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2374 * A patch from Martin Atzmueller seems to have solved the SIGINT
2375 problem, and as far as we know, signal-handling now works cleanly.
2376 (If you find any new bugs, please report them!)
2377 * The system no longer defaults Lisp source file names to types
2378 ".l", ".cl", or ".lsp", but only to ".lisp".
2379 * The compiler no longer uses special default file extensions for
2380 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2381 seems to expect a single default extension for all compiled code,
2382 and there's no compelling reason to try to stretch the standard
2383 to allow two different extensions.) Instead, byte-compiled files
2384 default to the same extension as native-compiled files.
2385 * Fasl file format version numbers have increased again, because
2386 a rearrangement of internal implementation packages made some
2387 dumped symbols in old fasl files unreadable in new cores.
2388 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2389 many fewer weird special cases.
2390 * Bug #17 (differing COMPILE-FILE behavior between logical and
2391 physical pathnames) has been fixed, and some related misbehavior too,
2392 thanks to a patch from Martin Atzmueller.
2393 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2394 by Tim Moore, ported to SBCL by Martin Atzmueller.
2395 * Martin Atzmueller fixed several filesystem-related problems,
2396 including bug #36, in part by porting CMU CL patches, which were
2397 written in part by Paul Werkowski.
2398 * More compiler warnings in src/runtime/ are gone, thanks to
2399 more patches from Martin Atzmueller.
2400 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2403 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2405 * DESCRIBE now works on CONDITION objects.
2406 * The debugger now handles errors which arise when trying to print
2407 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2409 * The build system now uses an additional file, customize-target-features.lisp,
2410 to allow local modifications to the target *FEATURES* list. (The point of
2411 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2412 debugging features enabled, without having to worry about propagating it
2413 into everyone's system when I do a "cvs update".) When no
2414 customize-target-features.lisp file exists, the target *FEATURES* list
2415 should be constructed the same way as before.
2416 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2417 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2418 warn about behavior which is undefined under the ANSI spec. Note: This
2419 is specified by ANSI, but it's not very popular with programmers.
2420 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2421 macro in the SBCL sources for an example of a workaround which you
2422 might use to make portable ANSI-standard code which does what you want.
2423 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2424 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2425 when it's used as a non-toplevel form. (This is required in order
2426 to implement the DEFCONSTANT-EQX macro.)
2427 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2428 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2429 funny things, due to bugs in the implementation of EVAL-WHEN
2430 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2431 non-toplevel DEFCONSTANTs are uncommon.)
2432 * The core file version number and fasl file version number have been
2433 incremented, because the old noncompliant DEFCONSTANT behavior involved
2434 calling functions which no longer exist, and because I also took the
2435 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2436 * fixed bug 1 (error handling before read-eval-print loop starts), and
2437 redid debugger restarts and related debugger commands somewhat while
2439 ** The QUIT debugger command is gone, since it did something
2440 rather different than the SB-EXT:QUIT command, and since it never
2441 worked properly outside the main toplevel read/eval/print loop.
2442 Invoking the new TOPLEVEL restart provides the same functionality.
2443 ** The GO debugger command is also gone, since you can just invoke
2444 the CONTINUE restart directly instead.
2445 ** The TOP debugger command is also gone, since it's redundant with the
2446 FRAME 0 command, and since it interfered with abbreviations for the
2448 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2450 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2451 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2452 some porting work by Martin Atzmueller)
2453 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2454 with a boatload of patches which he ported to SBCL. Now that those
2456 ** The system tries to make sure that its low-priority messages
2457 are prefixed by semicolons, to help people who like to use
2458 syntax highlighting in their ILISP buffer. (This patch
2459 was originally due to Raymond Toy.)
2460 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2461 patches originally written by Raymond Toy.
2462 ** The compiler understands coercion between single-value and
2463 multiple-VALUES type expressions better, getting rid of some very
2464 weird behavior, thanks to patches originally by Robert MacLachlan
2465 and Douglas Crosher.
2466 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2467 lambda lists, thanks to a patch originally by Pierre Mai.
2468 ** The system no longer bogusly warns about "abbreviated type
2470 ** The compiler gets less confused by inlining and RETURN-FROM,
2471 thanks to some patches originally by Tim Moore.
2472 ** The system no longer hangs when dumping circular lists to fasl
2473 files, thanks to a patch originally from Douglas Crosher.
2474 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2475 undefined function error.
2476 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2477 in structure slot initforms) for now, documented workaround instead:-|
2478 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2479 * fixed bug 5: added stubs for various Gray stream functions called
2480 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2481 installed, at least appropriate type errors are generated
2482 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2483 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2484 consistently in DEFMETHOD forms.
2485 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2486 it doesn't seem to affect SBCL after all
2487 * The C runtime system now builds with better optimization and many
2488 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2490 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2492 * The system is now under CVS at SourceForge (instead of the
2493 CVS repository on my home machine).
2494 * The new signal handling code has been tweaked to treat register
2495 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2496 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2497 downstream, e.g. in debug-int.lisp, has implicit dependencies
2498 on the unsignedness of integer representation of machine words,
2499 and that caused the system to bomb out with infinite regress
2500 when trying to recover from type errors involving signed values,
2501 e.g. (BUTLAST '(1 2 3) -1).)
2502 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2503 (This was one of the bugs Peter Van Eynde reported back in July.)
2504 * The system now uses code inspired by Colin Walters' O(N)
2505 implementation of MAP (from the cmucl-imp@cons.org mailing
2506 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2507 inline the MAP operation, and there is more than one
2508 sequence argument to the MAP call (so that it can't just
2509 do ETYPECASE once and for all based on the type of the
2510 single sequence argument). (The old non-inline implementation
2511 of the general M-argument sequence-of-length-N case required
2512 O(M*N*N) time when any of the sequence arguments were LISTs.)
2513 * The QUIT :UNIX-CODE keyword argument has been renamed to
2514 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2515 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2517 * Raymond Wiker's patches to port dynamic loading from Linux to
2518 FreeBSD have been added.
2519 * The BUGS file is now more nearly up to date, thanks in large part
2520 to Martin Atzmueller's review of it.
2521 * The debugger now flushes standard output streams before it begins
2522 its output ("debugger invoked" and so forth).
2523 * The core version number and fasl file version number have both
2524 been incremented, because of incompatible changes in the layout
2526 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2527 used to not be called for a saved Lisp image.)
2528 * Martin Atzmueller's version of a patch to fix a compiler crash,
2529 as posted on sbcl-devel 13 September 2000, has been installed.
2530 * Instead of installing Martin Atzmueller's patch for the
2531 compiler transform for SUBSEQ, I deleted the compiler transform,
2532 and transforms for some similar consing operations.
2533 * A bug in signal handling which kept TRACE from working on OpenBSD
2535 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2536 away by constant folding
2537 * The system now defines its address space constants in one place
2538 (in the Lisp sources), and propagates them automatically elsewhere
2539 (through GENESIS and the sbcl.h file). Therefore, patching the
2540 address map is less unnecessarily tedious and error-prone. The
2541 Lisp names of address space constants have also been systematized.
2542 * CVS tags like dollar-Header-dollar have been removed from
2543 the sources, because they have never saved me trouble and
2544 they've been source of trouble working with patches and other
2545 diff-related operations.
2546 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2547 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2549 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2551 * The system has been ported to OpenBSD.
2552 * The system now compiles with a simple "sh make.sh" on the systems
2553 that it's supported on. I.e., now you no longer need to tweak
2554 text in the target-features.lisp-expr and symlinks in src/runtime/
2555 by hand, the make.sh takes care of it for you.
2556 * The system is no longer so grossly inefficient when compiling code
2557 involving vectors implemented as general (not simple) vectors (VECTOR T),
2558 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2559 which dares to use the various sequence functions on non-simple
2560 vectors, takes less of a performance hit.
2561 * There is now a primitive type predicate VECTOR-T-P
2562 to test for the (VECTOR T) type, so that e.g.
2563 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2564 can now be compiled with some semblance of efficiency. (The old code
2565 turned the type declaration into a full call to %TYPEP at runtime!)
2566 * AREF on (VECTOR T) is still not fast, since it's still compiled
2567 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2568 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2569 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2570 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2571 without making full calls to SUBTYPEP at runtime.
2572 (Some analogous efficiency issues for non-simple vectors specialized to
2573 element types other than T, or for non-simple multidimensional arrays,
2574 have not been addressed. They could almost certainly be handled the
2575 same way if anyone is motivated to do so.)
2576 * The changes in array handling break binary compatibility, so
2577 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2578 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2579 instead of (VALUES T T).
2580 * By following the instructions that Dan Barlow posted to sbcl-devel
2581 on 2 July 2000, I was able to enable primitive dynamic object
2582 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2583 functionality is not implemented (since it calls ld to resolve
2584 library references automatically, requiring RUN-PROGRAM for its
2585 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2586 try to resolve library references) is now supported.
2587 * The system now flushes the standard output streams when it terminates,
2588 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2589 them at several other probably-convenient times, e.g. in each pass of
2590 the toplevel read-eval-print loop, and after evaluating a form given
2591 as an "--eval" command-line option. (These changes were motivated by a
2592 discussion of stream flushing issues on cmucl-imp in August 2000.)
2593 * The source transform for TYPEP of array types no longer assumes
2594 that an array whose element type is a not-yet-defined type
2595 is implemented as an array of T, but instead punts, so that the
2596 type will be interpreted at runtime.
2597 * There is now some support for cross-compiling in make.sh: each of
2598 the phases of make.sh has its own script. (This should be transparent
2599 to people doing ordinary, non-cross-compile builds.)
2600 * Since my laptop doesn't have hundreds of megabytes of memory like
2601 my desktop machine, I became more motivated to do some items on
2602 my to-do list in order to reduce the size of the system a little:
2603 ** Arrange for various needed-only-at-cold-init things to be
2604 uninterned after cold init. To support this, those things have
2605 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2606 symbols with such names are now uninterned after cold init).
2607 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2608 fasl files for cold load.
2609 ** Remove the old compiler structure pooling code (which used to
2610 be conditional on the target feature :SB-ALLOC) completely.
2611 ** Redo the representation of some data in cold init to be more compact.
2612 (I also looked into supporting byte compiled code at bootstrap time,
2613 which would probably reduce the size of the system a lot, but that
2614 looked too complicated, so I punted for now.)
2615 * The maximum signal nesting depth in the src/runtime/ support code has
2616 been reduced from 4096 to 256. (I don't know any reason for the very
2617 large old value. If the new smaller value turns out to break something,
2618 I'll probably just bump it back up.)
2619 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2621 * Many, many bugs reported by Peter Van Eynde have been added to
2622 the BUGS list; some have even been fixed.
2623 * While enabling dynamic object file loading, I tried to make the
2624 code easier to understand, renaming various functions and variables
2625 with less ambiguous names, and changing some function calling
2626 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2627 * While trying to figure out how to do the OpenBSD port, I tried to
2628 clean up some of the code in src/runtime/. In particular, I dropped
2629 support for non-POSIX signal handling, added various comments,
2630 tweaked the code to reduce the number of compilation warnings, and
2631 renamed some files to increase consistency.
2632 * To support the new automatic configuration functionality in make.sh,
2633 the source file target-features.lisp-expr has been replaced with the
2634 source file base-target-features.lisp-expr and the machine-generated
2635 file local-target-features.lisp-expr.
2636 * fixed a stupid quoting error in make.sh so that using CMU CL
2637 "lisp -batch" as cross-compilation host works again
2639 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2641 * DESCRIBE no longer tries to call itself recursively to describe
2642 bound/fbound values, so that it no longer fails on symbols which are
2643 bound to themselves (like keywords, T, and NIL).
2644 * DESCRIBE now works on generic functions.
2645 * The printer now prints less-screwed-up representations of closures
2646 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2647 * A private symbol is used instead of the :EMPTY keyword previously
2648 used to mark empty slots in hash tables. Thus
2649 (DEFVAR *HT* (MAKE-HASH-TABLE))
2650 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2651 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2652 now does what ANSI says that it should. (You can still get
2653 similar noncompliant behavior if bang on the hash table
2654 implementation with all the symbols you get back from
2655 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2656 This breaks binary compatibility, since tests for equality to
2657 :EMPTY are wired into things like the macroexpansion of
2658 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2660 * There's now a minimal placeholder implementation for CL:STEP,
2661 as required by ANSI.
2662 * An obscure bug in the interaction of the normal compiler, the byte
2663 compiler, inlining, and structure predicates has been patched
2664 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2665 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2667 * Missing ordinary arguments in a macro call are now detected even
2668 when the macro lambda list contains &KEY or &REST.
2669 * The debugger no longer complains about encountering the top of the
2670 stack when you type "FRAME 0" to explicitly instruct it to go to
2671 the top of the stack. And it now prints the frame you request even
2672 if it's the current frame (instead of saying "You are here.").
2673 * As specified by ANSI, the system now always prints keywords
2674 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2676 * The default initial SIZE of HASH-TABLEs is now smaller.
2677 * Type information from CLOS class dispatch is now propagated
2678 into DEFMETHOD bodies, so that e.g.
2679 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2681 is now basically equivalent to
2682 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2683 (DECLARE (TYPE SINGLE-FLOAT X))
2685 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2686 operation, without having to do run-time type dispatch.
2687 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2688 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2689 * Since I don't seem to be making big file reorganizations very often
2690 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2691 is overflowing my ability to conveniently back them up), I've finally
2692 checked the system into CVS. (The CVS repository is on my home system,
2693 not at SourceForge -- putting it on SourceForge might come later.)
2694 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2695 high-level GC-NOTIFY-FOO functions send their output. (There's
2696 still very little control of where low-level verbose GC functions
2697 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2698 less than it used to -- the GC-NOTIFY-FOO functions are now under
2699 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2700 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2701 in only one place in the source code, and propagates it automatically
2702 everywhere that it's needed. Thus e.g. when I bump the version from
2703 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2704 * The C source files now include boilerplate legalese and documentation
2705 at the head of each file (just as the Lisp source files already did).
2706 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2707 to his page will be replaced with a link to his new CLiki service.
2709 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2711 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2712 * The build process now looks for GNU make under the default name "gmake",
2713 instead of "make" as it used to. If GNU make is not available as "gmake"
2714 on your system, you can change this default behavior by setting the
2715 GNUMAKE environment variable.
2716 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2717 can build without error under CMU CL.
2719 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2721 * There is now a partial SBCL user manual (with some new text and some
2722 text cribbed from the CMU CL manual).
2723 * The beginnings of a profiler have been added (starting with the
2724 CMU CL profiler and simplifying and cleaning up). Eventually the
2725 main interface should be through the TRACE macro, but for now,
2726 it's still accessed through vaguely CMU-CL-style functions and macros
2727 exported from the package SB-PROFILE.
2728 * Some problems left over from porting CMU CL to the new
2729 cross-compilation bootstrap process have been cleaned up:
2730 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2731 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2732 part when getting rid of the compiler *BACKEND* stuff.)
2733 ** The value of *NULL-TYPE* was screwed up, because it was
2734 being initialized before the type system knew the final
2735 definition of the 'NULL type. This screwed up several key
2736 optimizations in the compiler, causing inefficiency in all sorts
2737 of places. (I found it because I wanted to understand why
2738 GET-INTERNAL-RUN-TIME was consing.)
2739 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2740 PROCLAIM FTYPE information. Unfortunately this broke binary
2741 compatibility again, since now the forms output by DEFGENERIC
2742 to refer to functions which didn't exist in 0.6.3.
2743 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2744 can use the new (as of 0.6.3) transform for SOME into MAP into
2746 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2747 values to (INTEGER 0 1000000), so that the time code will no longer
2748 occasionally get blown up by Linux returning 1000000 microseconds
2749 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2750 its output conform to the ANSI spec. (Alas, this makes its output
2751 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2752 * A full call to MAP NIL with a single sequence argument no longer conses.
2753 * fixes to problems pointed out by Martin Atzmueller:
2754 * The manual page no longer talks about multiprocessing as though
2755 it were currently supported.
2756 * The ILISP support patches have been removed from the distribution,
2757 because as of version 5.10.1, ILISP now supports SBCL without us
2758 having to maintain patches.
2759 * added a modified version of Raymond Toy's recent CMU CL patch for
2760 EQUALP comparison of HASH-TABLE
2762 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2764 * The system still can't cross-compile itself with
2765 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2766 entails), but at least it can compile more of itself that way
2767 than it used to be able to, and various buglets which were uncovered
2768 by trying to cross-compile itself that way have now been fixed.
2769 * This release breaks binary compatibility again. This time
2770 at least I've incremented the FASL file format version to 2, so that the
2771 problem can be detected reliably instead of just causing weird errors.
2772 * various new style warnings:
2773 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2774 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2775 ** using the lexical binding of a variable named in the *FOO* style
2776 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2777 as specified by ANSI.
2778 * *RANDOM-STATE* is no longer automatically initialized from
2779 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2780 default behavior of the system is to repeat its behavior every time
2781 it's run. If you'd like to change this behavior, you can always
2782 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2783 old convention there was no comparably easy way to get the system to
2784 repeat its behavior every time it was run.
2785 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2786 FTYPE declarations has been removed, in favor of their ANSI
2787 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2788 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2789 the types of their sequence arguments can be determined at compile time.
2790 This is done through a new open code expansion for MAP which eliminates
2791 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2792 argument types can be determined at compile time.
2793 * The optimizer now transforms COERCE into an identity operation when it
2794 can prove that the coerced object is already of the correct type. (This
2795 can be a win for machine generated code, including the output of other
2796 optimization transforms, such as the MAP transform above.)
2797 * Credit information has been moved from source file headers into CREDITS.
2798 * Source file headers have been made more standard.
2799 * The CASE macro now compiles without complaining even when it has
2802 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2804 * (Note that the way that the PCL macroexpansions were rewritten
2805 to accommodate the change in DEFGENERIC below breaks binary
2806 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2807 not run under sbcl-0.6.2. Once we get out of alpha releases,
2808 i.e. hit release 1.0.0, we'll probably try to maintain binary
2809 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2810 and sbcl-1.4.4. Until then, however, it might be fairly common
2811 for maintenance releases to break binary compatibility.)
2812 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2814 * The reporting of unhandled signals has been changed to print some
2815 explanatory text as well as the report form. (Previously only
2816 the report form was printed.)
2817 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2818 it defines, so that the compiler no longer issues undefined function
2819 warnings for compiled-but-not-yet-loaded generic functions.
2820 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2821 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2822 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2823 section of the ANSI spec are supported.
2824 * The "" nickname for the "KEYWORD" package has been removed.
2825 The reader still handles symbol tokens which begin with a package marker
2826 as keywords, but it doesn't expose its mechanism for doing so in the
2827 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2828 * The system now issues STYLE-WARNINGs for contradictory TYPE
2829 proclamations. (Warnings for contradictory FTYPE proclamations would
2830 be nice too, but those can't be done usefully unless the type system
2831 is made smarter about FUNCTION types.)
2832 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2833 systematized, so that "*target-*.lisp is supposed to exist only on the
2834 target and imply that there's a related file which exists on the
2835 host, and *host-*.lisp is supposed to exist only on the host and imply
2836 that there's a related file which exists on the target. This involves a
2837 lot of renaming. Hopefully the acute confusion caused by the renaming
2838 will be justified by the reduction in chronic confusion..
2839 ** runtime-type.lisp -> early-target-type.lisp
2840 ** target-type.lisp -> late-target-type.lisp
2841 ** early-host-format.lisp -> early-format.lisp
2842 ** late-host-format.lisp -> late-format.lisp
2843 ** host-error.lisp -> misc-error.lisp
2844 ** early-error.lisp -> early-target-error.lisp
2845 ** late-error.lisp -> late-target-error.lisp
2846 ** host-defboot.lisp -> early-defboot.lisp
2847 ** code/misc.lisp -> code/target-misc.lisp
2848 ** code/host-misc.lisp -> code/misc.lisp
2849 ** code/numbers.lisp -> code/target-numbers.lisp
2850 ** code/early-numbers.lisp -> numbers.lisp
2851 ** early-host-type.lisp -> early-type.lisp
2852 ** late-host-type.lisp -> late-type.lisp
2853 ** host-typep.lisp -> typep.lisp
2854 ** load.lisp -> target-load.lisp
2855 ** host-load.lisp -> load.lisp
2856 ** host-disassem.lisp -> disassem.lisp
2857 ** host-insts.lisp -> insts.lisp
2858 ** byte-comp.lisp -> target-byte-comp.lisp
2859 ** host-byte-comp.lisp -> byte-comp.lisp
2860 ** host-signal.lisp -> signal.lisp
2861 ** host-defstruct.lisp -> defstruct.lisp
2862 ** late-target-type.lisp -> deftypes-for-target.lisp
2863 Furthermore, several other previously target-only files foo.lisp (e.g.
2864 hash-table.lisp and random.lisp) have been split into a target-and-host
2865 foo.lisp file and a target-only target-foo.lisp file, with their key type
2866 definitions in the target-and-host part, so that the cross-compiler will
2867 know more about target types.
2868 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2869 gone away. In their place are various *BACKEND-FOO* variables
2870 corresponding to the slots of the old structure.
2871 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2872 into the target SBCL has been fixed.
2873 * The chill.lisp system for loading cold code into a running SBCL
2875 * Support for the CMU CL "scavenger hook" extension has been removed.
2876 (It was undocumented and unused in the CMU CL sources that SBCL was
2877 derived from, and stale in sbcl-0.6.1.)
2878 * Various errors in the cross-compiler type system were detected
2879 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2880 (enabling various consistency checks). Many of them were fixed,
2881 but some hard problems remain, so the compiler is back to
2882 running without *TYPE-SYSTEM-INITIALIZED* for now.
2883 * As part of the cross-compiler type system cleanup, I implemented
2884 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2885 * I have started adding UNCROSS calls throughout the type system
2886 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2887 on cross-compiler input files will be able to go away, and various
2889 * CONSTANTP now returns true for quoted forms (as explicitly required
2892 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2894 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2895 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2896 (MAKE-LIST -1). In the longer term, it ought to become true
2897 that declarations are assertions even at SAFETY 1. For now, it's not
2898 quite true even at SAFETY 3, but it's at least more nearly true..
2899 (Note that this change seems to increases the size of the system by
2900 O(5%) and to decrease the speed of the compiler by 20% or more.)
2901 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2902 for the problem of printing dozens of lines of distracting information
2903 about low-level system machinery as part of the top stack frame
2904 on entry to the debugger when an undefined function was called.
2905 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2906 is not reset to COMMON-LISP-USER.
2907 * Compilation of stuff related to dyncount.lisp has been made conditional
2908 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2909 smaller. The various dyncount-related symbols have been moved into
2910 a new "SB-DYNCOUNT" package.
2911 * tty-inspect.lisp has been renamed to inspect.lisp.
2912 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2913 feature has gone away. (When we eventually port to other flavors of
2914 libc and/or Unix, we'll try to make the differences between flavors
2915 invisible at the user level.)
2916 * Various other *FEATURES* tags, and/or their associated conditionals,
2917 have been removed if obsolescent, or given better documentation, or
2918 sometimes given more-mnemonic names.
2920 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2922 * tidied up "make.sh" script
2923 * tidied up system directory structure
2924 * better "clean.sh" behavior
2925 * added doc/FOR-CMUCL-DEVELOPERS
2926 * many many small tweaks to output format, e.g. removing possibly-confusing
2927 trailing #\. character in DESCRIBE-INSTANCE
2928 * (EQUALP #\A 'A) no longer signals an error.
2929 * new hashing code, including EQUALP hashing
2930 * tidied up Lisp initialization and toplevel
2931 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2932 * command line argument processing
2933 * added POSIX-GETENV function to deal with Unix-ish environment variables
2934 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2935 terminating SBCL on EOF
2936 * non-verbose GC by default
2937 * There is no more "sbcl" shell script; the sbcl file is now the C
2938 runtime executable (just like CMU CL).
2939 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2941 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2942 symbol and package information as Lisp native symbol and package objects
2943 instead of strings naming symbols and strings naming packages. This way,
2944 whenever packages are renamed (as in warm init), debug information is
2945 transformed along with everything else.
2946 * tweaked the optimization policy declarations which control the building
2947 of SBCL itself. Now, among other things, the system no longer saves
2948 source location debugging information. (This helps two problems at once
2949 by reducing SBCL size and by keeping SBCL from trying to look for its
2950 sources -- which may not exist -- when reporting errors.)
2951 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2952 debugging and testing purposes
2953 * cleaned up printing, making the printer call PRINT-OBJECT for
2954 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2955 methods, giving nearly-ANSI behavior
2956 * converted almost all special variables to use *FOO* naming convention
2957 * deleted PARSE-TIME functionality, since it can be done portably
2958 * moved some files out of cold init into warm init
2959 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2961 * regularized formatting of source files
2962 * added an install.sh script
2963 * fixed ridiculous memory usage of cross-compiler by making
2964 compiler/alloc.lisp not try to do pooling unless it can hook
2965 itself into the GC of the cross-compilation host. Now the system
2966 builds nicely on my old laptop.
2967 * added :SB-ALLOC in target-features.lisp-expr
2968 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2969 was not implemented)
2970 * re-did condition handling and note reporting in the compiler. Notes
2971 are no longer handled by signalling conditions. Style warnings
2972 and warnings are handled more correctly and reported in such a way
2973 that it's easy to find one or the other in your output (so that you
2974 can e.g. figure out which of many problems caused COMPILE-FILE to
2976 * changed the severity of several compiler warnings from full WARNING
2977 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2978 compiler note reporting so that it doesn't use the condition system
2979 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2980 COMPILE-FILE command)
2981 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2982 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2983 no longer supported, since their implementation was deeply intertwingled
2984 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2985 * removed bogus "support" for compiler macros named (SETF FOO), and
2986 removed the compiler macro for SETF INFO (but only after making a fool
2987 of myself on the cmucl-imp mailing list by posting a bogus patch for
2988 DEFINE-COMPILER-MACRO..)
2989 * Compiled files containing forms which have side effects on the Lisp
2990 reader (such as DEFPACKAGE forms) are now handled more correctly.
2991 (Compiler queuing of top level lambdas has been suppressed by setting
2992 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2993 * deleted various currently-unused source files, e.g. gengc.lisp. They
2994 may be added back at some point e.g. when porting to other architectures,
2995 but until they are it's distracting to distribute them and to try to
2997 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2998 there were so many of them they're just distractions, and UNCROSS is
2999 known to be able to handle the current sources
3000 * moved PROFILE functionality into TRACE, so that it will be clear
3001 how the wrapping and unwrapping of functions when you profile them
3002 interacts with the wrapping and unwrapping of functions when you
3003 trace them. (Actually, the functionality isn't there yet, but at least
3004 the interface specification is there. Hopefully, the functionality will
3005 arrive with some future maintenance release.)
3006 * removed host-oops.lisp
3007 * changed signature of QUIT function to allow UNIX-CODE argument
3008 * fixed READ-SEQUENCE bug
3009 * tweaked verbose GC output so that it looks more like the progress
3010 output that ANSI specifies for functions like LOAD
3011 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3012 * added <http://sbcl.sourceforge.com> to the banner information printed by