1 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
4 * fixed bugs 19 and 317: fixed-format floating point printing is
5 more accurate. This also fixes a bug reported by Adam Warner
6 related to the ~@F format directive.
7 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
9 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
10 dispatch table if the from-char is a dispatch macro character.
11 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
12 by the wordsize no longer ignore the last word. (reported by Lutz
14 * fixed bug in type checking of dynamic-extent variables. (reported
16 * optimization: sequence traversal functions use their freedom to
17 coerce function designators to functions.
18 * optimization: code with many calls to CLOS methods specialized on
19 CLOS classes has had redundant type checks removed.
20 * optimization: added declarations to speed up operations that access
21 the internal character database (for example STRING-UPCASE)
22 * optimization: comparison operations between floats and sufficiently small
23 fixnums no longer create extra rationals
24 * fixed some bugs related to Unicode integration:
25 ** portions of multibyte characters at the end of buffers for
26 character-based file input are correctly transferred to the
27 start of the buffer at the next read.
28 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
29 passing it through to OPEN.
30 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
31 argument given to internal calls to OPEN.
32 ** the built-in comment readers (introduced by character sequences
33 ";" and "#|") are more forgiving to encoding errors; they will
34 STYLE-WARN and then attempt to resync the stream at a character
35 boundary. (thanks to Teemu Kalvas)
36 * fixed some bugs in the x86-64 port:
37 ** Negative short int return values from c-calls are sign-extended
39 ** The stack is aligned to 16-bytes for c-calls, as required by
40 the ABI. (thanks to Cheuksan Wang)
41 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
42 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
43 for enums. (thanks to Vincent Arkesteijn)
44 * fixed some bugs revealed by Paul Dietz' test suite:
45 ** Space, Tab, Linefeed, Return and Page have the invalid
46 secondary constituent character trait.
47 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
49 ** WITH-INPUT-FROM-STRING should only update the index place on
51 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
54 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
55 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
56 platform. The port supports 61-bit fixnums, large memory spaces
57 and reloading shared object files.
58 * enhancement: saving cores with foreign code loaded is now
59 supported on ppc/Darwin in addition to the previously supported
61 * enhancement: the statistical profiler now walks deeper into the
62 call stack for more meaningful call-graphs and accrued time
63 reports (x86/x86-64 only). It also now reports time spent in
65 * enhancement: it is now possible to trace most individual methods
66 of a generic function in addition to tracing the generic function
68 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
70 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
71 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
72 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
73 produces an error. (thanks to Vincent Arkesteijn)
74 * bug fix: NAMESTRING on pathnames with :WILD components in their
75 directories works correctly. (thanks to Artem V. Andreev)
76 * fixed bug 125: compiler preserves identity of closures. (reported
78 * bug fixed: functions with &REST arguments sometimes failed with
79 "Undefined function" when compiled with (DEBUG 3). (reported by
81 * bug fix: overflow during compiling of setting element of a bit
82 vector with constant index and value. (reported by Timmy Douglas)
83 * build fix: fixed the dependence on *LOAD-PATHNAME* and
84 *COMPILE-FILE-PATHNAME* being absolute pathnames.
85 * on x86 compiler partially supports stack allocation of dynamic-extent
87 * GO and RETURN-FROM do not check the extent of their exit points
88 when compiled with SAFETY 0.
89 * fixed some bugs related to Unicode integration:
90 ** encoding and decoding errors are now much more robustly
91 handled; it should now be possible to recover even from invalid
92 input or output to the terminal. (thanks to Teemu Kalvas)
93 ** provided a first cut at implementing STRING-TO-OCTETS and
94 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
95 ** altered the SB-MD5 contributed module slightly, changing the
96 interface just enough for it to be supportable for builds where
97 lisp characters are not eight bits.
98 * fixed some bugs revealed by Paul Dietz' test suite:
99 ** the FORMATTER-generated functions for ~V[ conditionals require
100 the correct number of arguments.
101 ** READ-FROM-STRING returns the mandated second value when applied
102 to displaced strings.
103 ** the #\Rubout and #\Backspace characters are treated as invalid
104 constituent characters by the tokenizer.
106 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
107 * new feature: reloading changed shared object files with
108 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
109 * new feature: references to foreign variables and functions
110 can now be compiled and loaded before the corresponding shared
111 object file is loaded, as long as the foreign definitions are
112 available at runtime.
113 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
114 supported. (thanks to Dan Debertin)
115 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
116 just keywords. (thanks to Vincent Arkesteijn)
117 * fixed bug #331: structure-class instances corresponding to
118 DEFSTRUCT forms are now created eagerly.
119 * fixed bug #345: backtraces from calls to undefined functions work
120 on x86 as well. Related bug #61 is now also partially fixed on x86
121 (backtraces from throws to unknown catch tags.)
122 * bug fix: lambda-list parsing is now stricter vrt. order and number
123 of lambda-list keywords.
124 * bug fix: as specified by AMOP, an error is signalled if a
125 class-option appears multiple times in a DEFCLASS form. (reported
127 * bug fix: RANDOM can be compiled when the compiler derives the type
128 of its numeric argument as a disjoint set of small integers.
129 (reported by Paul Dietz)
130 * bug fix: starting a core saved with shared objects loaded when
131 those objects are not available no longer causes threaded SBCL to
132 hang. (reported by Sean Ross)
133 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
135 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
136 argument, which must always be NIL. (reported by Kalle Niemitalo)
137 * bug fix: printing 1.0d+23 no longer results in an error.
138 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
139 * bug fix: structure slot setters preserve evaluation order. (thanks
141 * fixed some bugs related to Unicode integration:
142 ** RUN-PROGRAM can allow its child to take input from a Lisp
143 stream. (reported by Stefan Scholl)
144 ** ASDF-INSTALL successfully downloads .tar.gz files in all
145 locales. (reported by Ken Causey)
146 * fixed some bugs revealed by Paul Dietz' test suite:
147 ** INCF, DECF and REMF evaluate their place form as specified in
149 ** FORMATTER expands ~{ ~} iteration directives with V or #
150 parameters correctly.
151 ** FORMATTER deals with the ~@[ ~] conditional directive where the
152 consequent uses no arguments correctly.
153 ** the system has a partial understanding of the (COMPLEX RATIO)
155 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
158 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
159 * new feature: a build-time option (controlled by the :SB-UNICODE
160 keyword feature, enabled by default) for building the system with
161 support for the entire 21-bit character space defined by the
163 * new feature: the system now has rudimentary external-format
164 support; the primary user-visible change at this time is that
165 characters with the high bit set (such as lower-case-e-acute) will
166 print correctly to a terminal in a UTF-8 environment.
167 * minor incompatible change: BASE-CHAR no longer names a class;
168 however, CHARACTER continues to do so, as required by ANSI.
169 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
170 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
171 has been moved to the SB-EXT package (temporarily re-exported from
173 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
174 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
176 * fixed bug #351: better error-handlind and reporting for malformed
178 * fixed bug #350: bignum-printing is now more memory-efficient,
179 allowing printing of very large bignums, eg. (expt 2 10000000).
180 (reported by Bruno Haible)
181 * fixed bug #302: better primitive-type selection for intersection
183 * fixed bug #308: non-graphic characters now all have names, as
184 required. (reported by Bruno Haible)
185 * bug fix: redefining a class with different superclasses now correctly
186 removes it from the direct-subclasses of its previous superclasses.
187 (reported by David Morse)
188 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
189 new value now works. (reported by Bruno Haible)
190 * bug fix: correct canonicalization of multiple non-standard slot
191 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
192 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
193 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
194 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
195 now exists, an signals an error.
196 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
197 during macroexpansion for non-iteration variables as well. (reported
198 by Bruno Haible for CMUCL)
199 * bug fix: Cyclic structures and unprintable objects in compiler
200 messages no longer cause errors. (reported by Bruno Haible)
201 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
202 and READ-FROM-STRING all now return a primary value of NIL if
203 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
204 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
205 (reported by Bruno Haible for CMUCL)
206 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
207 do not designate a condition. (reported by Bruno Haible for
209 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
210 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
211 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
212 permitting resolution in favour of any of the conflicting symbols.
213 (reported by Bruno Haible for CMUCL)
214 * bug fix: EQUAL compiler optimizations is less aggressive on
215 strings which can potentially compare true despite having distinct
216 specialized array element types.
217 * bug fix: unit enumerations can be defined without dividing by
218 zero. (thanks to Vincent Arkesteijn)
219 * FORMAT compile-time argument count checking has been enhanced.
220 (report from Bruno Haible for CMUCL)
221 * a partial workaround for the bug 262: the compiler does not try to
222 inline-expand a local function doing RETURN-FROM from a deleted
223 BLOCK. (thanks to Peter Denno for the bug report and to David
224 Wragg for the simple test case)
225 * fixed some bugs revealed by Paul Dietz' test suite:
226 ** RENAME-PACKAGE allows all package designators as new package
228 ** constraint propagation and lambda variable substitution are
229 more cautious in dealing with partially deleted code.
230 ** compiler sometimes lost reoptimization passes.
231 ** CERROR, given a condition as condition designator, passes its
232 remaining arguments to the continue format control without
234 ** Case-altering FORMAT directives work correctly on non-ASCII
236 ** The REAL type specifier handles bounds outside the floating
237 point ranges without signalling FLOATING-POINT-OVERFLOW.
238 ** Functions with IR1-transformations can create intercomponent
239 references to global functions.
240 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
242 ** FORMAT ~R treats a nil value for its first parameter correctly.
244 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
245 * enhancement: saving cores with foreign code loaded is now
246 supported on x86/NetBSD and sparc/Linux in addition to the previously
248 * bug fix: on some platforms repeated installations caused multiple
249 copies of HTML documentation to be installed -- should not happen
250 any more. (reported by Stefan Scholl)
251 * bug fix: parsing self-recursive alien record types multiple times
252 no longer causes infinite recursion. (reported by Thomas F. Burdick,
253 original patch by Helmut Eller for CMUCL)
254 * bug fix: stack-exhaustion detection works now on NetBSD as well.
255 (thanks to Richard Kreuter)
256 * bug fix: defining classes whose accessors are methods on existing
257 generic functions in other (locked) packages no longer signals
258 bogus package lock violations. (reported by François-René Rideau)
259 * bug fix: special variables as DEFMETHOD parameters no longer have
260 associated bogus type declarations. (reported by David Wragg and
262 * bug fix: read-write consistency on streams of element-type
263 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
264 * bug fix: redefiniton of the only method of a generic function with
265 no DEFGENERIC no longer emits a full WARNING. In addition,
266 redefinition of generic functions with no DEFGENERIC to an
267 incompatible lambda list now signals an error. (thanks to Zach
269 * bug fix: DEFGENERIC now works even when there's a function of the
270 same name in an enclosing lexical environment. (thanks to Zach
272 * fixed compiler failure, caused by instrumenting code during
273 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
274 * optimization: added loop analysis and improved register allocation
275 * fixed some bugs revealed by Paul Dietz' test suite:
276 ** POSITION on displaced vectors with non-zero displacement
277 returns the right answer.
278 ** (SIMPLE-STRING) is a valid type specifier for sequence
280 ** *PRINT-LEVEL* handling for slotless structures is pedantically
282 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
283 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
285 ** The justification version of the FORMAT ~< directive treats
286 non-zero minpad parameter correctly.
288 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
289 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
290 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
291 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
292 the supported interface.
293 * new feature: Single-stepping of code compiled with DEBUG 2 or
294 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
295 * new feature: saving cores with foreign code loaded is now
296 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
297 Timothy Moore's work for CMUCL)
298 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
299 parameters to * instead of NIL if no initform is supplied.
300 (reported by Johan Bockgaard)
301 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
302 correctly when FUNCALL appears as the car of the form. Note:
303 despite this FUNCALL forms are not currently subject to
304 compiler-macro expansion. (port of Raymond Toy's fix for the
305 same from CMUCL, reported by Johan Bockgaard)
306 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
307 (thanks for Teemu Kalvas)
308 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
309 case instead of terminating the iteration (thanks for Julian
310 Squires, Sean Champ and Raymond Toy)
311 * bug fix: incorrect expansion of defgeneric that caused a style
312 warning. (thanks for Zach Beane)
313 * bug fix: slot accessor effective method computation works properly
314 for classes with multiple non-standard applicable methods on
315 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
316 * on x86 compiler supports stack allocation of results of LIST and
317 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
318 CMUCL implementation by Gerd Moellmann)
319 * fixed some bugs revealed by Paul Dietz' test suite:
320 ** FORMAT strings with both the ~<~:;~> form of the justification
321 directive and pretty-printing directives cause an error.
323 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
324 * incompatible change: the internal functions
325 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
326 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
327 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
328 instead of the old functions.
329 * new feature: on platforms where "dladdr" is available foreign
330 function names now appear in backtraces. (based on Helmut Eller's
332 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
333 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
335 * documentation: ASDF manual is now distributed with SBCL.
336 * bug fix: SBCL can now load its contributed modules with REQUIRE
337 even if the system-provided entries have been removed from
338 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
340 * bug fix: backtraces involving undefined functions or assembly
341 routines are more informative. (thanks to Brian Downing)
342 * bug fix: mutually referent alien structures now work correctly.
343 (reported by Rick Taube)
344 * bug fix: structures defined by WITH-ALIEN can be referred to
345 within other definitions in the same WITH-ALIEN.
346 * bug fix: division operators (MOD, TRUNCATE and the like) with
347 constant zero divisors and integer dividends no longer generate
349 * bug fix: provide default methods for INPUT-STREAM-P and
350 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
351 * bug fix: improve the detection and resolution of MOP metacycles
352 (where effective method computation depends on the generic function
353 having its effective method computed). (reported by Bruno Haible)
354 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
355 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
356 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
357 * bug fix: pathnames with relative directory components can be
358 represented relative to default pathnames.
359 * optimization: in taking the GCD of bignums, reduce the two bignums
360 to approximately the same size (using Euclid's algorithm) before
361 applying the more sophisticated binary GCD. (thanks to Juho
363 * optimization: COUNT on bitvectors now operates word-at-a-time.
364 * optimization: ASH with a positive, but not necessarily constant,
365 (leftwards) shift, when in a modular context, is compiled to a
367 * fixed some bugs revealed by Paul Dietz' test suite:
368 ** FORMAT variable parameters ("~V<char>") are defaulted properly
369 if the corresponding argument is NIL.
370 ** FORMAT directives accepting a minpad parameter treat negative
372 ** literal commas as character parameters to FORMAT directives are
374 ** literal spaces directly after ~<Newline> directives within a
375 format-logical-block (~:< ~@:>) do not induce :FILL-style
376 conditional newlines.
377 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
378 as specified: it no longer includes conditional newlines).
379 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
381 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
382 :PER-LINE-PREFIX argument does not evaluate to a string.
384 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
385 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
386 the manual for details; for now, package locks can be disabled by
387 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
388 if no major problems are found then it is likely that they will be
389 compiled in unconditionally.
390 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
391 unsupported operators on all platforms. To load a shared library
392 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
393 object file, link it into a shared library outside of SBCL and
394 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
396 * bug workaround: ROOM T and the GC were not working together
397 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
398 reporting facility aren't preserved. That reporting has been
399 disabled, so now until and unless someone figures out how to make it
400 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
401 * minor incompatible change: as threatened around sbcl-0.8.0, the
402 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
403 The symbols are also exported from SB-PCL for backwards
404 compatibility, but more so than before SB-PCL should be treated as
405 an implementation-internal package.
406 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
408 * the SB-SPROF contrib now works on (most) non-x86 architectures.
409 It is known as of this release not to work on the Alpha, however.
410 * fixed bug #167: errors signalled due to illegal syntax in method
411 bodies are now more legible.
412 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
413 designators and can hence be used with TYPEP.
414 * fixed bug #333: CHECK-TYPE now ensures that the type error
415 signalled, if any, has the right object to be accessed by
416 TYPE-ERROR-DATUM. (reported by Tony Martinez)
417 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
418 5.1.2.3 for multiple-value place subforms. (reported by Kalle
420 * fixed bug #334: programmatic addition of slots using specialized
421 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
422 and :ALLOCATION :CLASS slots.
423 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
424 float exponents. (rereported by Peter Seibel)
425 * fixed a bug: #\Space (and other whitespace characters) are no
426 longer considered to be macro characters in standard syntax by
428 * fixed bug: initialization of condition class metaobjects no longer
429 causes an instance of the condition to be created. (reported by
431 * fixed bug: it is now possible to have slots such that
432 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
433 description is neither :INSTANCE nor :CLASS.
434 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
435 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
436 SLOT-VALUE-USING-CLASS), and no longer causes errors with
437 non-standard SLOT-DEFINITION-ALLOCATION values.
438 * fixed bugs: various race conditions handling exiting threads.
439 CL-PPCRE's thread torture test now passes.
440 * fixed arguably-a-bug: GC time proportional to number of threads in
441 system even when most of them are idle
442 * optimization: improved performance of BIT and SBIT on bit-vectors.
443 * .fasl file incompatibility: The fasl file version number has
444 been incremented because of changes associated with package locks.
446 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
447 * minor incompatible change: the system no longer provides
448 optimization hints (conditions which are TYPEP
449 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
450 mode; these hints are still emitted when higher SPEED optimization
452 * new contrib module: a sampling profiler (profiling by statistical
453 sampling, rather than by instrumenting functions) is available as
454 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
455 * the behaviour of the standard function ED is now customizeable by
456 third parties through a hook variable: see ED's documentation
457 string for information on the protocol.
458 * the compiler no longer emits efficiency notes for (FUNCALL X)
459 when the type of X is uncertain under default optimization
461 * fixed bug 276: mutating a binding of a specialized parameter to a
462 method to something that is not TYPEP the specializer is now
464 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
465 exhibit the required behaviour on the x86 platform. (thanks to
466 Peter van Eynde, Eric Marsden and Bruno Haible)
467 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
468 even for difficult arguments. (reported by Peter Graves)
469 * fixed bug 141a: the backquote printer now descends quoted
471 * fixed another bug in backquote printing: no more destructive
472 modification of the form's list structure. (reported by Brian
474 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
475 move between its address being taken and the call to
476 interrupt_thread, fixing a crashing race condition.
477 * the SB-POSIX contrib implementation has been adjusted so that it
478 no longer exhibits ridiculously poor performance when constructing
479 instances corresponding to C structs.
481 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
482 * minor incompatible change: the sb-grovel contrib now treats C
483 structures as alien (in the sense of SB-ALIEN) objects rather than
484 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
485 has implications for memory management of client code
486 (sb-grovel-returned objects must now be manually managed) and for
487 type safety (alien objects now have full types).
488 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
489 used to control emission of compiler diagnostics, rather than the
490 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
491 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
492 quality should be considered deprecated.
493 * install.sh now installs the user manual as well
494 * (not quite a new documentable feature, but worth considering in
495 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
496 beginnings of a semantically meaningful condition hierarchy is
497 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
499 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
500 designator as the defaults argument.
501 * fixed bug: Displaced arrays whose displaced-to array has become
502 too small now cause ARRAY-DIMENSION to signal an error, providing
503 for safer bounds-checking. (reported by Bruno Haible)
504 * fixed bug: DEFCLASS slot definitions with identical :READER and
505 :WRITER names now signal a reasonable error. (reported by Thomas
507 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
509 * fixed bug: Invalid slot specification errors now print correctly.
510 (thanks to Zach Beane)
511 * fixed bug 320: Shared to local slot value transfers in class
512 redefinitions now happen corrently from superclasses as
513 well. (reported by Bruno Haible)
514 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
516 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
517 improper lists correctly. (reported by Bruno Haible)
518 * fixed bug 313: source-transform for <fun-name> was erroneously
519 applied to a call of a value of a variable with name <fun-name>.
520 (reported by Antonio Menezes Leitao)
521 * fixed bug 307: The obsolete instance protocol ensures that
522 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
523 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
524 remove dead unknown-values globs from the middle of the stack.
525 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
526 easier to resume long computations after using *BREAK-ON-SIGNALS*
527 to diagnose and fix failures. (thanks to Nikodemus Siivola)
528 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
529 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
530 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
531 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
532 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
534 * fixed a (fairly theoretical) bug in string printing: if
535 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
536 does not have array-element-type equal to the most general string
538 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
539 function name. (thanks to Nikodemus Siivola)
540 * fixed bug: compile-time format string checker failed on
541 non-closed ~{. (reported by Thomas F Burdick)
542 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
543 don't warn on legal loop code involving more than one aggregate
544 boolean. (thanks to Nikodemus Siivola)
545 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
546 in keyword position to MAKE-PACKAGE was not being reported as
548 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
549 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
551 * fixed bugs 280 and 312: the checking for multiple definitions in a
552 file is less likely to become confused by uses of inline
554 * fixed bug: the #S reader macro performs the keyword coercion
555 specified for slot names. (reported by Kalle Niemitalo)
556 * fixed bug: lambda lists may contain symbols whose names start with
557 & but are not lambda-list-keywords; their occurrence triggers a
559 * fixed bug 321: define-method-combination argument lambda lists do
560 not cause invalid code to be generated when &optional or &aux
561 variables have default values. (reported by Bruno Haible)
562 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
563 have CLOS classes; this appears also to have fixed reported
564 instability in stack exhaustion detection.
565 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
566 MUFFLE-WARNING when no associated restart is present is now
568 * optimization: rearranged the expansion of various defining macros
569 so that each expands into only one top-level form in a
570 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
572 * optimization: used a previously little-used slot in symbols to
573 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
575 * fixed some bugs revealed by Paul Dietz' test suite:
576 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
578 ** Fixed an optimization bug related to inheritance of initforms
579 from local to shared slots.
580 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
581 ** CAST optimizer forgot to flush argument derived type.
582 ** print/read consistency on floats is now orders of magnitude
583 more likely. (thanks also to Bruno Haible for a similar report
585 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
586 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
587 for objects of type REAL. Make it so.
588 ** SXHASH is specified (infelicitously) to respect similarity,
589 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
590 it so. (thanks to Markus Ziegler)
591 ** on the Alpha, the compiler succeeds in compiling functions
592 returning a known number of arguments greater than 63.
593 ** fixed handling of invalid NIL arguments in keyword position
595 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
597 ** provided a readably-printable representation for RANDOM-STATE
599 ** ensured that pathnames, where they have a namestring, always
600 print using #P"..." syntax.
602 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
603 * Support for the forthcoming 2.0 version of the NetBSD kernel
604 running on x86 hardware has been added. (thanks to Perry
605 E. Metzger most immediately, and others for their past work)
606 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
607 current release version; SBCL's OpenBSD support had been broken
608 since about the time of OpenBSD's switch to ELF binary format.)
609 * [placeholder for DX summary]
610 ** user code with &REST lists declared dynamic-extent, under high
611 speed or space and low safety and debug optimization policy.
612 * The manual has been converted to Texinfo format and the debugger
613 chapter from the cmucl manual has been added.
614 * A facility has been added to extract documentation strings from
615 sbcl and store them as Texinfo-formatted snippets for inclusion in
616 the manual (via Texinfo's @include directive)
617 * bug fix: compiler emitted division in optimized DEREF. (thanks for
618 the test case to Dave Roberts)
619 * bug fix: multidimensional simple arrays loaded from FASLs had fill
620 pointers. (reported by Sean Ross)
621 * bug fix: PROFILE output is printed nicely even for large numerical
622 values. (thanks to Zach Beane)
623 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
624 greater than 32 handle EOF correctly.
625 * bug fix: on X86 an immediate argument of the IMUL instruction is
626 correctly printed in disassembly. (thanks to Lutz Euler)
627 * bug fix: class slots in redefined classes preserve their old
628 values. (thanks to Bruno Haible and Nikodemus Siivola)
629 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
630 succeeds. (reported by Marco Baringer)
631 * fixed some bugs revealed by Paul Dietz' test suite:
632 ** READ-SEQUENCE now works on ECHO-STREAMs.
633 ** RATIONALIZE works more according to its specification. (thanks
636 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
637 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
638 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
639 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
640 should matter to you only if you rebind the printer control
641 variables and then find you want different bindings in the
642 debugger than in the ordinary execution of your program.)
643 * The runtime build system has been tweaked to support building
644 (on SPARC/SunOS) using a C compiler which invokes Sun's own
645 assembler and linker. (thanks to Nikodemus Siivola)
646 * Unbound, undefined, undeclared variables now trigger full
647 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
648 more often programmer error than deliberate exploitation of undefined
650 * optimization: the hash algorithm for strings has changed to one
651 that is less vulnerable to spurious collisions. (thanks to Juho
653 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
654 needless bounds checking. (thanks to Juho Snellman)
655 * optimization: implemented multiplication as a modular
656 (UNSIGNED-BYTE 32) operation on the PPC backend.
657 * fixed some bugs revealed by Paul Dietz' test suite:
658 ** ADJUST-ARRAY now copies the datum in a zero rank array if
660 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
661 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
662 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
664 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
666 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
668 ** ECHO-STREAMs no longer attempt to echo the end of file value to
669 their output stream on EOF from read.
670 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
671 have been read to end-of-file.
672 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
674 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
675 description of determination of which consecutive characters
677 ** Printing the "Space" character with escaping on now yields "#\\ ",
678 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
679 ** Reading floating-point numbers with *READ-BASE* set to a number
680 less than 10 works correctly.
681 ** Reading floating-point numbers with *READ-BASE* set to a number
682 more than 10 works correctly.
683 ** Printing with *PRINT-READABLY* targets the standard readtable, not
684 the readtable currently in effect.
686 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
687 * minor incompatible change: parsing of namestrings on a physical
688 (Unix) host has changed; numbers after the final #\. in a
689 namestring are no longer interpreted as a version field. This is
690 intented to be largely invisible to the user, except that the
691 meaning of the namestring "*.*.*" has changed: it now refers to a
692 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
693 should usually be replaced by
694 (make-pathname :name :wild :type :wild :version :wild)
695 with the added benefit that this is more likely to be portable.
696 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
697 option to OPEN now signals an error if the file being opened
698 exists; this may have an impact on existing code.
699 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
700 signals received, on the PowerPC platforms (both Linux and
701 Darwin). (thanks to Pierre Mai for pointing out the location of
703 * several fixes on OS X: The system now builds and runs cleanly on
704 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
705 (thanks to Brian Mastenbrook)
706 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
707 second-resolution: integer multiples of 1/3600 between -24 and 24.
708 (thanks to Vincent Arkesteijn)
709 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
710 type when called with 1 argument; PEEK-CHAR checked type of
711 PEEK-TYPE only after having read first character from a
712 stream. (reported by Peter Graves)
713 * bug fix: the garbage collector now has much better locality
714 behaviour, and in particular no longer treats all memory as being
715 exhausted when many small objects point to each other in a deeply
717 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
718 recognized as being TYPEP their class.
719 * bug fix: the PUSHNEW documentation string has been corrected.
720 (thanks to Vincent Arkesteijn)
721 * bug fix: defaulting of the value for the last of an atypically
722 large number of multiple values being bound was not being
723 performed correctly on the Alpha or PPC platforms
724 * optimization: implemented multiplication as a modular
725 (UNSIGNED-BYTE 32) operation on the x86 backend.
726 * optimization: SEARCH on simple-base-strings can now be open-coded.
727 (see also contrib/compiler-extras.lisp for inspiration for
728 teaching the compiler about the Boyer-Moore algorithm).
729 * value, returned by MAX (and MIN) called with several EQUALP, but
730 not EQL, arguments now does not depend on compiler settings.
731 * fixed some bugs revealed by Paul Dietz' test suite:
732 ** in stack analysis liveness information is propagated from
733 non-local entry points.
734 ** pathwise CAST removing failed when the CAST node did not start
736 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
737 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
739 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
740 host is already defined.
741 ** RENAME-FILE works on streams instead of signalling an internal
743 ** PEEK-CHAR uses the current readtable when determining whether
744 or not a character is whitespace.
745 ** MERGE-PATHNAMES handles the case when the pathname does not
746 specify a name while the default-pathname specifies a version.
747 ** Pathnames now stand a better chance of respecting print/read
749 ** Attempting to use standardized file system operators with a
750 pathname with invalid :DIRECTORY components signals a
752 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
754 ** DIRECTORY on logical pathnames is more correct.
755 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
756 signal a TYPE-ERROR if their argument is not a stream.
757 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
758 designator argument does not designate a stream.
759 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
760 examining the synonym.
761 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
763 ** OPEN and WITH-OPEN-STREAM allow opening streams with
764 element-type larger than ([UN]SIGNED-BYTE 32).
766 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
767 * When built with the :SB-FUTEX feature, threaded builds now take
768 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
769 for faster/more reliable mutex and condition variable support.
770 * Incompatible change (but one you probably shouldn't have been using
771 anyway): the interface and code for arbitrating between multiple
772 threads in the same user session has been redesigned.
773 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
774 error when called without an explicit environment argument.
775 (thanks to Nikodemus Siivola)
776 * bug fix: buffered :DIRECTION :IO streams are less likely to become
777 confused about their position. (thanks to Adam Warner and Gerd
779 * bug fix: Pretty printing backquoted forms with unquotations in the
780 argument list position of various code constructs such as LAMBDA
781 now works correctly. (reported by Paul Dietz)
782 * bug fix: Pretty printing unquotations no longer loses all
783 stream position information.
784 * optimization: performance of string output streams is now less
785 poor for multiple small sequence writes.
786 * optimization: performance of CSUBTYPEP in the presence of complex
787 expressions involving CONS and NOT many times has been improved.
788 (reported by Paul Dietz)
789 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
791 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
792 streams for the REPL, for the debugger, and so on. (thanks to
794 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
795 * fixed some bugs revealed by Paul Dietz' test suite:
796 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
797 is ignored if it is too small, rather than propagating through
798 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
799 ** extremely complex negations of CONS types were not being
800 sufficiently canonicalized, leading to inconsistencies in
802 ** VALUES tranformer lost derived type.
804 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
805 * fixed a bootstrapping bug: the build process no longer assumes
806 that the various BOOLE-related constants have the same value in
807 host and target lisps. (noted by Paul Dietz' test suite on an
808 SBCL binary built from CLISP)
809 * The system can now be dynamically linked on the MIPS platform,
810 which enables dynamic loading of foreign code from Lisp. (thanks
811 to Ralf Baechle for discussions on the MIPS ABI)
812 * The system now records debugging information for its own source
813 files in a filesystem-position-independent manner, relative to
814 the "SYS" logical host.
815 * fixed a compiler bug: MV-LET convertion did not check references
816 to the "max args" entry point. (reported by Brian Downing)
817 * tweaked disassembly notes to be less confident about proclaiming
818 some instruction as an LRA. (thanks to Brian Downing)
819 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
820 now each have their own history, command character, and other
821 characteristics. (thanks to David Lichteblau)
822 * fixed some bugs revealed by Paul Dietz' test suite:
823 ** compiler failure in compiling LOGAND expressions including a
825 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
826 shift greater than 32.
827 ** FLUSH-DEST did not mark blocks for type check regeneration.
828 ** HANDLER-CASE failed to accept declarations in handler clauses
829 in some circumstances.
831 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
832 * New code in contrib/sb-introspect (still probably not entirely
833 stable yet) provides some support for smart Lisp development
834 environments like SLIME.
835 * The conditions signalled for errors occurring when loading .fasl
836 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
837 in a way which should help ASDF recover gracefully.
838 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
839 stuff has been cleaned up. If you code contrib/ stuff, this might
840 affect you, and you can look at contrib/README, contrib/STANDARDS,
841 and/or the 0.8.4.27 diff to check.
842 * In full calls the compiler now does not generate checks for declared
843 argument types for all arguments.
844 * various threading fixes
845 ** and some experimental patches which didn't make it into
846 the main tree for this release, but which are shipped in
847 contrib/experimental-thread.patch as a possible fix for some
848 failures (deadlock, spinning...) in GC-intensive multithreaded
850 * fixed PPC build problem (source code incompatibility of different
851 library versions): added offsetof() hackery which attempts to divine
852 where glibc maintainers put uc_mcontext today
853 * fixed bug 282: compiler does not trust type assertions while passing
854 arguments to a full call.
855 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
856 &OPTIONAL VALUES type parameter.
857 * fix bug 214: algorithm for noting rejected templates is now more
858 similar to that of template seletion. (also reported by rydis on
860 * fixed bug 141b: printing backquoted information readably and prettily
861 inserts a space where necessary.
862 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
863 (CHARACTER 10) are now reported as errors, rather than propagated
864 as unknown types. (reported by piso on #lisp)
865 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
866 with values NIL and :ERROR. (thanks to Milan Zamazal)
867 * fixed bug 191c: CLOS now does proper keyword argument checking as
868 described in CLHS 7.6.5 and 7.6.5.1.
869 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
870 longer raise an error; further, using a list as a for-as-arithmetic
871 counter now raises a meaningful error.
872 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
873 now understood by sequence creation functions such as MAKE-SEQUENCE
875 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
876 a STRING-INPUT-STREAM.
877 * compiler enhancement: SIGNUM is now better able to derive the type
879 * type declarations inside WITH-SLOTS are checked. (reported by
881 * fixed some bugs revealed by Paul Dietz' test suite:
882 ** incorrect optimization of TRUNCATE for a positive first
883 argument and negative second.
884 ** compiler failure in let-convertion during flushing dead code.
885 ** compiler failure while deriving type of TRUNCATE on an
886 interval, containing 0.
887 ** ASH of a negative bignum by a negative bignum count now returns
889 ** intersection of CONS types now canonicalizes properly, fixing
890 inconsistencies in SUBTYPEP.
892 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
893 * incompatible change: The --disable-debugger command line
894 option now clobbers the debugger at a more fundamental
895 level, by redefining #'INVOKE-DEBUGGER instead of by
896 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
897 is specified by ANSI to ignore *DEBUGGER-HOOK* and
898 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
899 enter the debugger REPL and then suffer recursive errors
900 because *DEBUG-IO* is also messed up in --disable-debugger mode;
901 while under the new system, BREAK in --disable-debugger mode
902 terminates the system just as an unhandled error would.
903 * fixed compiler performance when processing loops with a step >1;
904 * bug fix: DOCUMENTATION now retrieves generic function
905 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
906 support has been systematized, and now supports the methods
907 specified by ANSI, along with a default method and a method for
908 slot documentation. (reported by Nathan Froyd)
909 * bug fix: effective methods associated with a generic function are
910 no longer cached over a change of that generic function's method
911 combination. (reported by Andreas Fuchs)
912 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
913 names a non-existent directory. (thanks to Andreas Fuchs)
914 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
915 the PPC platform, lead to stack corruption; this has been fixed.
916 (reported by Rainer Joswig)
917 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
918 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
919 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
920 arithmetic in general, is now much more reliable on the PPC
922 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
923 platform now returns the right answer.
924 * optimization: restored some effective method precomputation in
925 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
926 precomputation is now tunable.
927 * optimization: compiler-internal data structure use has been
928 reviewed, and changes have been made that should improve the
929 performance of the compiler by about 20%.
930 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
931 simple-base-strings and simple-bit-vectors is improved.
932 * optimization: the optimization of 32-bit logical and arithmetic
933 functions introduced in version 0.8.3 on the x86 has been
934 implemented on the mips, ppc and sparc platforms; an
935 implementation of the same facility, but for 64-bit arithmetic,
936 has been added for the alpha.
937 * microoptimization: the compiler is better able to make use of the
938 x86 LEA instruction for multiplication by constants.
939 * bug fix: in some situations compiler did not report usage of
940 generic arithmetic in (SPEED 3) policy.
941 * bug 145b fix: compiler used wrong type specifier while converting
942 MEMBER-types to numeric.
943 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
945 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
946 index argument now works properly on the Alpha platform.
947 * bug fix: floating point exception treatment on the Alpha platform
949 * bug fix: FILE-POSITION works much better on string input and
950 output streams. (thanks to Nikodemus Siivola)
951 * bug fix: many threading/garbage collection symptoms sorted.
952 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
953 might be pseudo-atomic.
954 * internal change: Stopping for GC is now done with signals not
955 ptrace. GC is now done in whichever thread wanted it, instead of
957 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
959 * bug fix: result form in DO is not contained in the implicit
961 * incompatible change: ICR structure is changed; the value part of
962 CONTINUATION is now called LVAR; corresponding functions are
963 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
964 * added type deriver for ISQRT (thanks to Robert E. Brown).
965 * bug fix: better support for loading from the command line when an
966 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
968 * fixed some bugs revealed by Paul Dietz' test suite:
969 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
970 ** ROUND and FROUND now give the right answer when given very
971 small float arguments.
972 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
974 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
975 ** (GCD 0 <negative-integer>) returned <negative-integer>.
976 ** LCM should return a non-negative integer.
977 ** PARSE-INTEGER returned the index of a terminator instead of the
978 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
979 ** PARSE-INTEGER returned an incorrect index being applied to a
981 ** LCM with two arguments of 0 returns 0 rather than signalling
983 ** unsigned addition of a 32-bit constant with the high bit set no
984 longer causes an internal compiler error on the x86.
985 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
986 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
987 with negative last argument.
988 ** byte specifiers with zero size and position no longer cause
989 an error during type derivation.
990 ** bignum multiplication on the Alpha platform now returns the
992 * porting: The system now builds on SuSE AMD64, although it still
993 generates a 32-bit binary.
994 * .fasl file incompatibility: The fasl file version number has
995 been incremented (because of the changes to internal compiler
996 data structures referred to above).
998 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
999 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1000 more accurately, on the Darwin kernel running on PowerPC hardware.
1001 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1002 * Compiler code deletion notes now signal a condition of type
1003 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1004 an associated MUFFLE-WARNING restart.
1005 * The compiler now performs limited argument count validation of
1006 constant format strings in FORMAT, and where appropriate in ERROR,
1007 CERROR and WARN. (thanks to Gerd Moellmann)
1008 * New ASDF-INSTALL contrib can be used for automatic download and
1009 installation of third-party Lisp code from CCLAN or other sites
1011 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1012 which forces another thread to execute a function supplied by the
1014 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1015 now accept and act upon their :ELEMENT-TYPE keyword argument.
1016 (reported by Martin Atzmueller, Edi Weitz)
1017 * bug fix: FILE-POSITION now accepts position designators up to
1018 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1019 is the greater. (thanks to Patrik Nordebo)
1020 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1021 Kalle Olavi Niemitalo)
1022 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1023 printable. (reported by Eric Marsden)
1024 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1025 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1026 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1027 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1028 (thanks to Stig Sandoe)
1029 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1030 UNREAD-CHAR now correctly includes the unread character in the
1031 target sequence. (thanks to Gerd Moellmann)
1032 * bug fix in threaded builds: the system can now be suspended and
1033 resumed by shell job control with minimal disruption.
1034 * bug fixes in times and timezones >2038AD
1035 * better handling of "where is GNU make?" problem in build scripts
1036 (thanks to Nikodemus Siivola)
1037 * new optimization: inside a named function any reference to a
1038 function with the same name is considered to be a self-reference;
1039 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1040 optimization quality.
1041 * new optimization on x86: logical functions and + now have
1042 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1043 used when the result is truncated to 32 bits.
1044 * VALUES declaration is partially enabled.
1045 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1046 * fixed some bugs revealed by Paul Dietz' test suite:
1047 ** The system now obeys the constraint imposed by
1048 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1049 types form a lattice under type intersection.
1050 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1051 ** ASSOC now ignores NIL elements in an alist.
1052 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1053 and (1+ MOST-POSITIVE-FIXNUM) answers.
1054 ** The addition of a method with invalid qualifiers to a generic
1055 function does not cause an error to be signalled immediately;
1056 a warning is signalled, and the error is generated only on
1057 calling the generic function.
1058 * changed .fasl file version number, in part to add type codes for
1059 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1060 obscure ANSI requirements
1062 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1063 * fixed bug 148: failure to inline-expand a local function left
1064 garbage, confusing the compiler.
1065 * fixed bugs 3cd: structure slot readers perform type check if the
1066 slot can have an invalid value (i.e. it is either not initialized
1067 or can be written with a less specific slot writer).
1068 * bug fix: the compiler now traps array references to elements off
1069 the end of an array; previously, the bounds checking in some
1070 circumstances could go off-by-one.
1071 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1073 * type declarations for array element types now obey the description
1074 on the CLHS page "Declaration TYPE", as per discussions on
1075 sbcl-help around 2003-05-08. This means that a declaration
1076 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1077 declaration, all references to BAR will be asserted or assumed
1078 (with THE, so dependent on compiler policy) to involve objects of
1079 type FOO. Note that no such declaration is implied in
1080 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1081 * declared types of functions from the "Conditions"
1082 chapter. (reported by Paul Dietz)
1083 * bug fix: CERROR accepts a function as its first argument.
1084 * bug fix: NTH an NTHCDR accept a bignum as index
1085 arguments. (reported by Adam Warner)
1086 * optimization: character compare routines now optimize comparing
1087 against a constant character. (reported by Gilbert Baumann)
1088 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1089 argument now works properly on the MIPS platform.
1090 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1091 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1092 declared to have a functional type, does not cause an error
1094 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1095 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1096 * fixed bug: if last continuation of a deleted block has a
1097 destination, this destination should be deleted too. (reported by
1099 * fixed a bug in the bootstrap process: the host compiler's values
1100 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1101 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1102 test case from Patrik Nordebo)
1103 * improved the ability of the disassembler on the PPC platform to
1104 provide helpful disassembly notes.
1105 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1106 the class in more cases than previously.
1107 * bug fix: FILE-POSITION now understands :START and :END for
1108 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1109 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1110 8) data. (thanks to David Lichteblau)
1111 * bug fix: it is possible to add a method to a generic function
1112 without lambda list.
1113 * bug fix: reader failed to signal END-OF-FILE inside an
1114 object representation. (reported by Nikodemus Siivola)
1115 * fixed some bugs revealed by Paul Dietz' test suite:
1116 ** LAST and [N]BUTLAST should accept a bignum.
1117 ** condition slot accessors are methods.
1118 ** (VECTOR NIL) is a subtype of STRING.
1120 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1121 * minor incompatible change: some nonsensical specialized lambda
1122 lists (used in DEFMETHOD) which were previously ignored now signal
1124 * minor incompatible change: the system is now aware of the types of
1125 variables in the COMMON-LISP package, and will signal errors for
1126 most violations of these type constraints (where previously they
1127 were silently accepted).
1128 * minor incompatible change: COMPILE-FILE now uses the freedom
1129 afforded (ANSI 3.2.2.3) to use derived function types for
1130 functions defined in the same file. This also permits the system
1131 to warn on static type mismatches and function
1132 redefinition. (Currently it does not work with high DEBUG level.)
1133 * minor incompatible change: VALUES declaration is disabled.
1134 * When issuing notes, the compiler now signals a condition of type
1135 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1136 restart for use in user handlers. It is expected that the
1137 COMPILER-NOTE condition will eventually become a condition
1138 supertype to a hierarchy of note types, which will then be
1139 handleable in a similar fashion. However, at the moment, no such
1140 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1141 but it's an implementation detail, not a classification for the
1143 * Changes in type checking closed the following bugs:
1144 ** type checking of unused values (192b, 194d, 203);
1145 ** template selection based on unsafe type assertions (192c, 236);
1146 ** type checking in branches (194bc).
1147 * A short form of VALUES type specifier has ANSI meaning (it has
1148 increased the number of situations when SBCL cannot perform type
1150 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1151 keywords or constants is permissible.
1152 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1153 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1154 classes. (thanks to Antonio Martinez)
1155 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1156 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1157 argument to a pretty-print stream. Instead, it leaves any such
1158 operations to DESCRIBE-OBJECT methods.
1159 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1161 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1162 lambda lists. (thanks to David Lichteblau)
1163 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1164 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1165 (thanks to Gerd Moellmann)
1166 * fixed bug 246: increased compilation speed of long
1167 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1169 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1171 * DEFCONSTANT now throws a condition of type
1172 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1173 constant to a non-EQL value; CONTINUE and ABORT restarts
1174 respectively change and preserve the value.
1175 * fixed bug 63: The code walker, part of the implementation of CLOS,
1176 is now better at handling symbol macros.
1177 * bug fix: There is no longer an internal implementation type named
1178 CL:LENGTH. (reported by Raymond Toy)
1179 * bug fix: In macro-like defining macros/special operators the
1180 implicit block does not enclose the lambda list.
1181 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1182 enclosed in parentheses) are not suitable as type specifiers, and
1183 their use properly signals an error now.
1184 * bug fix: An argument count mismatch for a type specifier in code
1185 being compiled no longer causes an unhandled error at compile
1186 time, but signals a compile-time warning.
1187 * fixed simple vector readable printing
1188 * bug fix: DESCRIBE takes more care over whether the class
1189 precedence list slot of a class is bound before accessing it.
1190 (reported by Markus Krummenacker)
1191 * bug fix: FORMATTER can successfully compile pretty-printer format
1192 strings which use variants of the ~* directive inside.
1193 * bug fix: SEARCH now applies its TEST predicate to the elements of
1194 the arguments in the correct order. (thanks to Wolfhard Buss)
1195 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1196 in type inference. (thanks to Robert E. Brown)
1197 * bug fix: Reading in symbols with an explicit package name of ""
1198 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1199 * fixed some bugs revealed by Paul Dietz' test suite:
1200 ** NIL is now allowed as a structure slot name.
1201 ** Arbitrary numbers, not just REALs, are allowed in certain
1202 circumstances in LOOP for-as-arithmetic clauses.
1203 ** Multiple class redefinitions before slot access no longer
1204 causes a type error.
1205 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1206 association between the name and a class.
1207 ** Generic functions with non-standard method-combination and over
1208 five methods all of which return constants no longer return NIL
1209 after the first few invocations. (thanks to Gerd Moellmann)
1210 ** CALL-NEXT-METHOD with no arguments now passes the original
1211 values of the arguments, even in the presence of assignment.
1212 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1214 ** DISASSEMBLE works with closures and funcallable instances.
1215 ** ADD-METHOD now returns the generic function, not the new method.
1216 ** FIND-METHOD signals an error if the lengths of the specializers
1217 is incompatible with the generic function, even if the ERRORP
1219 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1220 which its argument is a member.
1221 ** DEFCLASS only redefines the class named by its class-name
1222 argument if that name is the proper name of the class;
1223 otherwise, it creates a new class.
1224 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1225 of the UNBOUND-SLOT condition to the name of the slot.
1226 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1227 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1228 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1229 SLOT-MAKUNBOUND in the specified fashion.
1231 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1232 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1233 cross-compilation host. As a consequence, we can now bootstrap our
1234 way up to SBCL starting with a bare gcc toolchain and human-readable
1235 source code (first the source to CLISP, then the source to SBCL).
1236 * A contributed module containing a partial implementation of the
1237 simple-streams interface has been included. (thanks to Rudi
1239 * A contributed module implementing the RFC1321 Message Digest
1240 Algorithm, known as MD5, has been included.
1241 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1242 no longer has any effect, as the code controlled by this feature
1243 has been deleted. (As far as we know, no-one has ever built using
1244 this feature, and its semantics were confused in any case).
1245 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1246 work on conditions (as required by the ANSI specification),
1247 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1248 expected behaviour on conditions. Users should note, however,
1249 that such behaviour is not required by the ANSI specification,
1250 and so use of this behaviour may render their code unportable.
1251 * fixed some bugs revealed by Paul Dietz' test suite:
1252 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1254 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1255 ** SXHASH on condition objects no longer returns NIL.
1256 ** :ALLOCATION :CLASS slots are better treated; their values are
1257 updated on class redefinition, and initforms inherited from
1258 superclasses are applied.
1259 ** REMOVE-METHOD returns its generic function argument even when
1260 no method was removed.
1261 ** SHARED-INITIALIZE now initializes the values of the requested
1262 slots, including those with :ALLOCATION :CLASS.
1263 ** ALLOCATE-INSTANCE now works on structure classes defined via
1264 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1266 ** SLOT-EXISTS-P now works on conditions, as well as structures
1268 ** MAKE-LOAD-FORM now has the required methods on
1269 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1270 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1271 keyword, and now implements the SLOT-NAMES argument.
1272 ** methods with &OPTIONAL arguments no longer allow too many
1273 arguments to be passed in the call without error.
1274 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1275 option is consistent with the required arguments of the generic
1276 function lambda list.
1277 * bug fix: REQUIRE accepts a string designator. (Thanks to
1279 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1280 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1281 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1282 * bug fix: The compiler now checks for duplicated variables in macro
1284 * bug fix: SETQ on globals returns the correct value.
1285 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1286 not a COMPILER-ERROR (followed by some other strange error on
1287 choosing the CONTINUE restart).
1288 * bug fix: make.sh and friends are now more consistent in the way that
1289 they look for GNU "make".
1291 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1292 * experimental native threads support (on x86 Linux >=2.4 only).
1293 This is not compiled in by default: you need to add :SB-THREAD to
1294 the target features. See the "Beyond ANSI" chapter of the manual
1296 * fix for longstanding nonANSIism: The old distinction between
1297 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1298 The return value from CL:FIND-CLASS is now a CLOS class, and
1299 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1300 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1301 * An interface to the MetaObject Protocol, as described in Kiczales,
1302 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1303 MIT Press, 1991, is available from the SB-MOP package.
1304 * incompatible change: the SB-PCL package should now be considered
1305 a private implementation detail, and no longer a semi-private MOP
1307 * minor incompatible change: due to rearrangement for threads, the
1308 control stack and binding stack are now allocated at arbitrary
1309 addresses instead of being hardcoded per-port. Users affected by
1310 this probably have to be doing advanced things with shared
1311 libraries, and will know who they are.
1312 * minor incompatible change: Previously, all --eval forms used were
1313 processed with READ before any of them were processed with EVAL.
1314 Now each --eval form is processed with both READ and EVAL before
1315 the next --eval form is processed. (Thus package operations like
1316 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1317 work as the user might reasonably expect.)
1318 * minor incompatible change: *STANDARD-INPUT* is now only an
1319 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1321 * minor incompatible change: Y-OR-N-P is now character-oriented,
1322 not line oriented. Also, YES-OR-NO-P now works without errors.
1323 (thanks to Antonio Martinez)
1324 * sb-aclrepl module improvements: an integrated inspector, added
1325 repl features, and a bug fix to :trace command.
1326 * Known functions, which cannot be open coded by the backend, are
1327 considered to be able to check types of their arguments. (fixing
1328 a bug report by Nathan J. Froyd)
1329 * fixed a bug in computing method discriminating functions: It is
1330 now possible to define methods specialized on classes which have
1331 forward-referenced superclasses. (thanks to Gerd Moellmann)
1332 * fixed evaluation order in optional entries (reported by Gilbert
1334 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1335 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1336 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1337 required-by-AMOP NAME argument, as well as CLASS and
1338 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1339 * fixed bug 20: DEFMETHOD can define methods using names that are
1340 not the proper names of classes to designate class specializers.
1341 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1342 Unix file descriptors, instead of blowing up. (thanks to Antonio
1344 * Garbage collection refactoring: user-visible change is that a
1345 call to the GC function during WITHOUT-GCING will not do garbage
1346 collection until the end of the WITHOUT-GCING. If you were doing
1347 this you were probably losing anyway.
1348 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1349 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1350 TYPEP the latter but not the former.
1351 * The compiler issues a full WARNING for calls to undefined functions
1352 with names from the CL package.
1353 * MAP-INTO for a vector destination is open coded. (reported by
1354 Brian Downing on c.l.l)
1355 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1356 documentation string.
1357 * fixed some bugs revealed by Paul Dietz' test suite:
1358 ** COPY-ALIST now signals an error if its argument is a dotted
1360 ** Condition slots are now accessed more correctly in the presence
1361 of multiple initargs for a given slot.
1362 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1363 exclude restarts of the same name associated with a different
1365 ** DEFCLASS of forward-referenced classes with another
1366 forward-referenced class in the superclasses list no longer
1368 ** Condition slots are now initialized once each, not multiple
1369 times. (thanks to Gerd Moellmann)
1370 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1372 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1373 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1374 arguments contain duplicated elements.
1375 ** RESTART-CASE understands local macros.
1376 ** RESTART-CASE associates exactly its own restarts with a condition.
1377 ** ENDP in safe mode checks its argument to be of type LIST.
1378 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1379 in question is unbound.
1380 ** Optimized MAKE-INSTANCE functions no longer cause internal
1381 assertion failures in the presence of duplicate initargs.
1382 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1383 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1384 propagated even in the ctor optimized implementation of
1386 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1388 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1389 * a better implementation of SXHASH on (simple) bit vectors,
1390 measured both in execution speed and in distribution of results
1391 over the positive fixnums, has been installed. Likewise, a better
1392 implementation of EQUAL for simple bit vectors is now available.
1393 * fixed CEILING optimization for a divisor of form 2^k.
1394 * fixed bug 240 (emitting extra style warnings "using the lexical
1395 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1396 by Antonio Martinez)
1397 * fixed SXHASH, giving different results for NIL depending on type
1398 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1399 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1400 variable. (found by Rolf Wester)
1401 * SBCL does not ignore type declarations for special
1402 variables. (reported by rif on c.l.l 2003-03-05)
1403 * some bug fixes in contrib/sb-aclrepl/
1404 * fixed some bugs revealed by Paul Dietz' test suite:
1405 ** a bug in the CONS type specifier, whereby the CAR and CDR
1406 types got intertwined, has been fixed;
1407 ** the type system is now able to reason about the interaction
1408 between INTEGER and RATIO types more completely;
1409 ** APPEND, [N]REVERSE and NRECONC check that those their
1410 arguments, which must be proper lists, are really so;
1411 ** An array specialized to be unable to hold elements has been
1412 implemented, as required -- yes, really -- by ANSI;
1413 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1414 on malformed property lists;
1416 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1417 * incompatible packaging change: in line with Unix convention,
1418 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1419 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1420 * REQUIRE and PROVIDE are now optionally capable of doing something
1421 useful. See the documentation string for REQUIRE.
1422 * infrastructure for a managed SBCL contrib system: contributed
1423 modules in this release include:
1424 ** the ASDF system definition facility;
1425 ** an interface to the BSD Sockets API;
1426 ** an ACL-like convenience interface to the repl;
1427 (thanks to Kevin Rosenberg)
1428 ** an implementation of ROTATE-BYTE, with efficient implementation
1430 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1431 gives a closer approximation to the right answer than previously.
1432 (thanks to Raymond Toy)
1433 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1434 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1435 optional environment arguments, as required by ANSI.
1436 * fixed bugs in other functions taking environment objects, allowing
1437 calls with an explicit NIL environment argument to be compiled
1439 * fixed bug 228: primary return values from
1440 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1441 COMPILE or FUNCTION.
1442 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1443 :TYPE will no longer signal errors on innocuous objects.
1444 * fixed bug 231b: SETQ is better at respecting type declarations in
1445 the lexical environment.
1446 * fixed a bug in DEFCLASS: classes named by symbols with no or
1447 unprintable packages can now be defined.
1448 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1449 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1450 * fixed compiler failure related to checking types of functions.
1451 (reported by Robert E. Brown)
1452 * the compiler is now much more consistent in its error-checking
1453 treatment of bounding index arguments to sequence functions: in
1454 (SAFETY 3) code, errors will be signalled in almost all cases if
1455 invalid sequence bounding indices are passed to functions defined
1456 by ANSI to operate on sequences.
1457 * fixed a bug in the build procedure: documentation of SBCL-specific
1458 packages is now preserved and available in the final Lisp image.
1459 * lifted FDEFINITION lookup out of loops in the implementation of
1460 many list operations. (thanks to Robert E. Brown)
1461 * fixed a bug in the reader: the #n# reader macro now works for
1462 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1463 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1464 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1465 E. Brown and Tony Martinez respectively)
1466 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1467 SLOT-BOUNDP: in particular, it is now able to optimize them much
1468 better, and is now not vulnerable to having packages renamed.
1469 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1470 always called when a slot is not present in an instance. (thanks
1472 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1473 longer signal an error on unreading a character following EOF on
1474 the previous constituent stream. (thanks to Tony Martinez)
1475 * fixed some bugs revealed by Paul Dietz' test suite:
1476 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1477 not just nonnegative fixnums;
1478 ** the logical bit-array operators such as BIT-AND now accept an
1479 explicit NIL for their "opt-arg" argument (to indicate a
1480 freshly-consed result bit-array);
1481 ** ELT now signals an error on an invalid sequence index in safe
1483 ** the type system is now cleverer about negations of numeric
1484 types, and consequently understands the BIGNUM and RATIO types
1486 ** the type system is now cleverer about the interaction between
1487 INTEGER and RATIO types: while bugs still remain, many more
1488 cases are accurately computed;
1489 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1490 if it is in the last clause;
1491 ** CONSTANTLY now correctly returns a side-effect-free function in
1493 ** DECLARE is no longer treated as a special-operator; in
1494 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1495 * incremented fasl file version number due to the change in the
1496 DEFSTRUCT-SLOT-DESCRIPTION structure.
1498 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1499 * minor incompatible change: code processed by the "interpreter" or
1500 EVAL now has a compilation optimization policy of (DEBUG 2)
1501 (changed from (DEBUG 1)) to improve debuggability of interactive
1502 development, and to allow the use of the debug RETURN command in
1504 * an experimental implementation of the RETURN command for the
1505 debugger has been included. (thanks to Frederik Kuivinen)
1506 * fixed bug 62: constraints were not propagated into a loop.
1507 * fixed bug in embedded calls of SORT (reported and investigated by
1509 * fixed some bugs revealed by Paul Dietz' test suite:
1510 ** printing and reading of arrays with some dimensions having
1511 length 0 (thanks to Gerd Moellmann);
1512 ** BOA constructor with &AUX argument without a default value does
1513 not cause a type error;
1514 ** CONSTANTP now returns true for all self-evaluating objects.
1516 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1517 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1518 accessors that are related by inheritance, as specified in the
1519 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1520 Valtteri Vuorikoski)
1521 * The compiler is now able to inline functions that were defined in
1522 a complex lexical environment (e.g. inside a MACROLET).
1523 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1525 * Support for the upcoming FreeBSD-5.0 release has been included.
1526 (thanks to Dag-Erling Smorgrav)
1527 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1528 effect when it is not in a toplevel context.
1529 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1530 stand a better chance of being correct. (thanks to Gerd
1532 * fixed bug in COERCE, which now signals an error on coercing a
1533 rational to a bounded real type which excludes the expected
1535 * The compiler is now able to derive types more accurately from the
1536 COERCE and COMPILE functions.
1537 * fixed bug 223: functional binding is considered to be constant
1538 only for symbols in the CL package.
1539 * fixed bug 231: SETQ did not check the type of a variable being set
1540 (reported by Robert E. Brown)
1541 * A new optimization for MAKE-INSTANCE has been included, fixing
1542 various bugs (including relating to :ALLOCATION :CLASS slots and
1543 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1545 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1546 ** As required by ANSI, LOOP now disallows anonymous collection
1547 clauses such as COLLECT I in conjunction with aggregate boolean
1548 clauses such as THEREIS (= I 1);
1549 ** LOOP now signals an error when any variable is reused in the
1550 same loop (including the potentially useful construct analogous
1551 to WITH A = 1 WITH A = (1+ A);
1552 ** IT is only a special loop symbol within the first clause of a
1553 conditional loop clause;
1554 ** LOOP with a typed iteration variable over a hashtable now
1555 signals a type error iff it should.
1556 * fixed some other bugs revealed by Paul Dietz' test suite:
1557 ** FILE-STREAM now names the class previously known as FD-STREAM;
1558 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1559 argument) no longer signals an error;
1560 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1561 of the slot symbol, rather than using the current package
1562 ((:CONC-NAME "") continues to intern the slot's name in the
1564 * incremented fasl file version number, because of the incompatible
1565 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1566 of the new implementation of DEFINE-COMPILER-MACRO.
1568 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1569 * Support for building SBCL for MIPS platforms running in
1570 little-endian mode has now been checked in, and basic
1571 functionality on said platforms verified.
1572 * minor incompatible change: PCL now records the pathname of a file
1573 in which methods and the like are defined, rather than its
1575 * minor incompatible change: TRUENAME now considers the truename of
1576 a file naming a directory to be the pathname with :DIRECTORY
1577 component indicating that directory.
1578 * minor incompatible change: a NAMED clause in the extended form of
1579 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1580 reason for the previous behaviour is unclear.
1581 * more systematization and improvement of CLOS and MOP conformance
1582 in PCL (thanks to Gerd Moellman and Pierre Mai):
1583 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1585 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1586 primary methods with no specializers;
1587 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1589 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1590 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1591 FORWARD-REFERENCED-CLASSes; error reporting on
1592 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1594 ** SXHASH on CLOS instances now uses a slot internal to the
1595 instance to return different numbers on distinct instances,
1596 while preserving the same return value through invocations of
1598 ** DEFMETHOD signals errors when methods with longer incongruent
1599 lambda lists are added to generic functions;
1600 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1601 CLASS, as specified in AMOP;
1602 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1603 on the order returned by the primary method for classes of
1604 class STANDARD-CLASS;
1605 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1606 * fixed some bugs shown by Paul Dietz' test suite:
1607 ** DOLIST puts its body in TAGBODY;
1608 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1610 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1611 value producing form;
1612 ** if more variables are given to PROGV than values, extra
1613 variables are bound and made to have no value;
1614 ** NSUBSTITUTE on list arguments gets the right answer with
1616 ** ELT signals an error of type TYPE-ERROR when the index argument
1617 is not a valid sequence index;
1618 ** LOOP signals (at macroexpansion time) an error of type
1619 PROGRAM-ERROR when duplicate variable names are found;
1620 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1621 ** FUNCALL of special-operators now cause an error of type
1623 ** PSETQ now works as required in the presence of side-effecting
1624 symbol-macro places;
1625 ** NCONC accepts any object as its last argument;
1626 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1628 ** loop-for-as-package does not require a package to be explicitely
1630 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1632 * fixed bug 166: compiler preserves "there is a way to go"
1633 invariant when deleting code.
1634 * fixed bug 172: macro lambda lists with required arguments after
1635 &REST arguments now cause an error to be signalled. (thanks to
1637 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1638 bug. (thanks to Matthew Danish)
1639 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1641 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1642 arguments are lexically rebound. (thanks to Gerd Moellmann and
1644 * fixed bug 194: error messages are now more informative when there
1645 is no primary method applicable in a call to a generic
1646 function. (thanks to Gerd Moellmann)
1647 * fixed bug in command line argument checking (thanks to Julian
1649 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1650 (thanks to Matthew Danish)
1651 * incremented fasl file version number, because of the
1652 SXHASH-related changes in the layout of CLOS data structures
1654 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1655 * minor incompatible change: The runtime (the Unix executable named
1656 "sbcl") is now much pickier about the .core files it will load.
1657 Essentially it now requires .core files to descend from the same
1658 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1659 as the runtime does. (The intent is to prevent the crashes which
1660 can occur, and which can even be reported as mysterious failures,
1661 when people patch the sources or change the build parameters
1662 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1663 sbcl and .core files.)
1664 * fixed bug: VALUES-LIST is no longer optimized away.
1665 * fixed bug 142: The FFI conversion of C string values to Lisp
1666 string values no longer conses excessively. (thanks to Nathan
1667 Froyd porting Raymond Toy's fix to CMU CL)
1668 * began to systematize and improve MOP conformance in PCL (thanks to
1669 Nathan Froyd, Gerd Moellman and Pierre Mai):
1670 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1672 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1673 ** FINALIZE-INHERITANCE is now called on class finalization;
1674 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1675 argument precedence order.
1676 * fixed bug 202: The compiler no longer fails on functions whose
1677 derived types contradict their declared type.
1678 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1679 so it can be non-toplevel.
1680 * The fasl file version number has changed (because of the new
1681 implementation of DEFMACRO).
1682 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1683 safe code, that any length requirement by their type-specifier
1684 argument is valid. The exceptions to this are described in bug
1686 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1687 functions now check the types of their inputs as required by ANSI.
1688 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1689 introduced symbol is DECLAREd to be SPECIAL.
1690 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1691 * fixed bug: :COUNT argument to sequence functions may be negative
1692 * fixed bug: body of DO-SYMBOLS may contain declarations
1693 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1694 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1695 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1696 symbol macro only once
1697 * fixed printing of call frame when argument list is unavailable
1698 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1699 * compiler no longer signals WARNING on unknown keyword
1702 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1703 * A beta-quality port to the mips architecture running Linux,
1704 based on the old CMUCL backend, has been made. It has been tested
1705 on a big-endian kernel, and works sufficiently well to be able to
1706 rebuild itself; it has not been tested in little-endian mode.
1707 * fixed an inconsistency between gencgc.c and purify.c which made
1708 dumping/loading .core files unreliable
1709 * fixed bug 120a: The compiler now deals correctly with IFs where
1710 the consequent is the same as the alternative, instead of
1711 misderiving the return type. (thanks to Alexey Dejneka)
1712 * fixed bug 113: Logical pathnames are now dumpable (the logical
1713 host is resolved at load-time, throwing an error if it is not
1715 * fixed bug 174: FORMAT's error message is slightly clearer when a
1716 non-printing character is used in a format directive.
1717 * fixed several bugs in compiler checking of type declarations, i.e.
1718 violations of the Python "declarations are assertions" principle
1719 (thanks to Alexey Dejneka)
1720 * fixed several bugs in PCL's error checking (thanks to Gerd
1722 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1724 * fixed bug in compilation of functions as first class values
1725 (thanks to Antonio Martinez-Shotton)
1726 * The compiler's handling TYPE-ERRORs which it can prove will
1727 inevitably happen at runtime has been cleaned up and corrected
1728 in several ways. (thanks to Alexey Dejneka)
1729 * improved argument type checking for various basic arithmetic
1730 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1731 lavished on them in the past that they can be compiled in many
1732 ways in different special cases
1733 * fixed bug 181: compiler checks validity of user supplied type
1735 * cleaned up code flushing in optimization: Function calls which
1736 should signal errors for safety purposes (e.g. which ANSI says
1737 should signal errors when their arguments are of incorrect type)
1738 are no longer optimized away.
1739 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1740 * incremented fasl file version number, because changes in the
1741 implementation of sequence functions like COERCE caused
1742 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1743 in old inline expansions) to become undefined. (Actually these
1744 changes were later undone, so we might very well be binary
1745 compatible with 0.7.7 after all, but leaving the version number
1746 incremented seemed like the simplest and most conservative
1749 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1750 * An alpha-quality port to the parisc architecture running Linux,
1751 based on the old CMUCL backend, has been made. This, even more so
1752 than the other backends, should be considered still a work in
1753 progress; known problems include that the Linux kernel in 64-bit
1754 mode does not propagate the correct sigcontext structure to
1755 userspace, and consequently SBCL on a parisc64 kernel will not
1757 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1758 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1759 suppressed.) Also now that the compiler is looking at declarations
1760 in the environment, it checks optimization declarations as well,
1761 and suppresses inlining when (> DEBUG SPEED).
1762 * More fixes have been made to treatment of floating point exception
1763 treatment and other Unix signals. In particular, floating point
1764 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1765 * The detection and handling of control stack exhaustion (infinite
1766 or very deeply nested recursion) has changed. Stack exhaustion
1767 detection is now done by write-protecting pages at the OS level
1768 and applies at all optimization settings; when found, a
1769 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1770 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1771 be caught using IGNORE-ERRORS.
1772 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1773 correctly when transforming recursive LABELS functions to LETs.
1774 (thanks to Alexey Dejneka)
1775 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1776 that are names of constants or global variables.
1777 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1778 alien routines with docstrings.
1779 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1780 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1782 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1783 (thanks to Eric Marsden)
1784 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1785 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1786 * Bug fix: The SPARC backend can now compile functions involving
1787 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1788 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1789 to a functions expecting a PATHNAME-DESIGNATOR.
1790 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1791 lambda lists. (thanks to Alexey Dejneka)
1792 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1793 functions so that EXPORTing the name of the class doesn't cause
1794 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1795 to Antonio Martinez for reporting this)
1796 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1797 constant-folded without failing an assertion. (thanks to Einar
1798 Floystad Dorum for reporting this)
1799 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1800 types are now dealt with more correctly by the compiler.
1801 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1802 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1803 behaviour analogous to RENAME-FILE. This puts its behaviour more
1804 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1806 * The fasl file version number has changed again. (because of the
1807 bug fix involving the names of PCL MAKE-INSTANCE functions)
1809 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1810 * bug fix: Floating point exceptions are treated much more
1811 consistently on the x86/Linux and PPC/Linux platforms.
1812 * Array initialization with :INITIAL-ELEMENT is now much faster for
1813 cases when the compiler cannot open code the array creation, but
1814 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1815 array accesses have also seen a speed increase.
1816 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1817 specified. (This was at the root of some bad interactions between
1818 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1820 * bug fix: Internal error arguments for undefined functions are now
1821 computed correctly on the PPC/Linux platform.
1822 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1823 Raymond Toy's patch for CMU CL)
1824 * Support for the Solaris 9 operating environment has been included
1825 (thanks to Daniel Merritt)
1826 * A very ugly but hopefully complete draft of the missing FFI chapter
1827 of the manual has been created by reformatting the corresponding
1828 CMU CL manual chapter into (currently very ugly and incoherent)
1829 DocBook and bringing it up to date for SBCL behavior. Thus, the
1830 manual is now essentially complete, at least by my extreme
1831 once-and-only-once standards, whereby it's acceptable to refer to
1832 the doc strings of SB-EXT functions as the primary documentation.
1833 * The fasl file version number has changed again, due to cleanup of
1834 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1835 is no longer a static symbol.)
1837 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1838 * SBCL now builds with OpenMCL (version 0.12) as the
1839 cross-compilation host; also, more progress has been made toward
1840 bootstrapping under CLISP.
1841 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1843 * bug 158 fixed: The compiler can now deal with integer loop
1844 increments different from 1; fixing this turned out also to fix
1846 * bug 169 fixed: no more bogus warnings about using lexical bindings
1847 despite the presence of perfectly good SPECIAL declarations (thanks
1848 to David Lichteblau)
1849 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1850 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1851 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1853 * bug 180 fixed: Method combination specifications no longer ignore
1854 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1855 * bug fix: Structure type predicate functions now check their argument
1856 count as they should.
1857 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1858 correctly. (thanks to Pierre Mai)
1859 * minor incompatible change: The --noprogrammer option is deprecated
1860 in favor of the new --disable-debugger option, which is very similar.
1861 (The major difference is that it takes effect at a slightly different
1862 time at startup, causing handling of errors in --sysinit and
1863 --userinit files will be affected differently.) The
1864 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1865 been added to allow this functionality to be controlled from ordinary
1866 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1867 maintainers, who might want to run non-interactive scripts to
1868 build SBCL cores which will later be used interactively.)
1869 * minor incompatible change: The LOAD function no longer, when given
1870 a wild pathname to load, loads all files matching that pathname.
1871 Instead, an error of type FILE-ERROR is signalled.
1873 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1874 * bug 147 fixed: The compiler preserves its block link/count
1875 invariants more correctly now so that it doesn't crash. (thanks
1877 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1879 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1880 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1881 no-value-for-this-character as NIL (as specified by ANSI).
1882 * HOST-NAMESTRING on physical pathnames now returns a string that is
1883 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1884 (thanks to Christophe Rhodes)
1885 * The Alpha port handles icache flushing more correctly. (thanks to
1887 * More progress has been made toward bootstrapping under CLISP. (thanks
1888 to Christophe Rhodes)
1889 * The fasl file format has changed again, because dynamic loading
1890 on OpenBSD (which has non-ELF object files) motivated some cleanups
1891 in the way that foreign symbols are transformed and passed around.
1892 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1893 is no longer treated as whitespace by the reader, but instead as
1894 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1895 |A
\7fB|, instead of A as it used to.
1897 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1898 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1899 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1900 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1901 this as of 0.7.1.45, but was left out of the previous news section
1902 (thanks to Dan Barlow)
1903 * SBCL now runs on the Solaris operating system on SPARC architectures
1904 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1905 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1906 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1907 * SPARC backend cleanups, allowing builds of cores optimized for V8
1908 and V9 SPARCS, and also emission of code targeted to a particular
1909 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1911 * SBCL is closer to bootstrapping under CLISP, thanks to various
1912 fixes by Christophe Rhodes.
1913 * The fasl file format has changed again, to allow the compiler's
1914 INFO database to support symbol macros.
1915 * The user manual (in doc/) is formatted into HTML more nicely.
1916 (thanks to coreythomas)
1917 * The system is smarter about SUBTYPEP relationships, especially
1918 those involving NOT types (including types such as ATOM which are
1919 represented internally using NOT types). Thus SUBTYPEP is less
1920 likely to return (VALUES NIL NIL) in general, and in particular
1921 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1923 * The fasl file format has changed again, because the internal
1924 representation of types now includes a new slot to support the new
1925 SUBTYPEP-of-NOT-types logic.
1926 * (not a change in the main branch of SBCL, but a related prototype
1927 which can hopefully be merged into the main branch of SBCL in the
1928 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1929 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1930 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1931 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1932 dispatch macro character combinations. (thanks to Alexey Dejneka)
1933 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1934 that superficially look like logical namestrings correctly.
1935 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1937 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1938 default supplied-p) for &optional and &key arguments. (thanks to
1941 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1942 * incompatible change: The compiler is now less aggressive about
1943 tail call optimization, doing it only when (> SPACE DEBUG) or
1944 (> SPEED DEBUG). (This is an incompatible change because there are
1945 programs which relied on the old CMU-CL-style behavior to optimize
1946 away their unbounded recursion which will now die of stack overflow.)
1947 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1948 for non-GENCGC systems has been increased to 20M (since that
1949 seems much closer to the likely performance optimum for modern
1950 systems than the old 4M value was)
1951 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1952 and *DEBUG-PRINT-LEVEL*
1953 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1954 of CMU CL's support for SPARC, and various endianness and other
1955 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1956 * new syntactic sugar for the Unix command line: --load foo.bar is now
1957 an alternate notation for --eval '(load "foo.bar")'.
1959 ** The system now detects stack overflow and handles it gracefully,
1960 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1961 optimization settings. (This is a good thing in general, and
1962 its introduction in this version should be particularly timely
1963 for anyone whose code fails because of suppression of tail
1965 ** The system now hunts for the C variable "environ" in a more
1966 devious way, to avoid segfaults when the C library version differs
1967 between compile time and run time. (thanks to Christophe Rhodes)
1968 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1969 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1970 ** The compiler no longer issues bogus style warnings for undefined
1971 classes in the same source file as the DEFCLASSes which defined
1972 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1974 ** fixes in CONDITION class precedence list for undefined function
1975 errors (thanks to Alexei Dejneka)
1976 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1977 correctly. (thanks to Dan Barlow)
1978 ** portability fixes aiming at bootstrapping under CLISP (thanks
1979 to Dave McDonald and Christophe Rhodes)
1980 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1981 ** fixes in type translation and and type inference (thanks to
1983 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1984 ** various fixes in the new ports (thanks to Dan Barlow)
1985 * several changes related to debugging:
1986 ** suppression of tail recursion, as noted above
1987 ** stack overflow detection, as noted above
1988 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1989 is now the default. (For some time encapsulation has been more
1990 reliable than the breakpoint-based :ENCAPSULATE NIL
1991 implementation, at least on X86 systems; and I just noticed that
1992 encapsulation also seems closer to the spirit of the ANSI
1995 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1997 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1998 up properly again. (There was a packaging bug in 0.7.0 which
1999 left their definitions in SB-SYS::LOAD-FOREIGN and
2000 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2001 vital for most things which interface to C-level interfaces,
2002 like extensions working with sockets or databases or
2003 Perl-compatible regexes or whatever, and the need to fix
2004 this bug is the main reason that 0.7.1 was released so
2006 ** DEFGENERIC is now choosier about the methods it redefines, so that
2007 reLOADing a previously-LOADed file containing DEFGENERICs does
2008 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2009 cycle now works as it should. (thanks to Alexey Dejneka)
2010 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2011 fixed by Christophe Rhodes. (He actually submitted this patch
2012 months ago, and I delayed until after 0.7.0.)
2013 ** Bug 111 (internal compiler confusion about runtime checks on
2014 FUNCTION types) was fixed by Alexey Dejneka.
2015 * Some internal cleanups (getting rid of variables which aren't
2016 needed now that the byte interpreter is gone) caused the fasl
2017 file format number to change again.
2019 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2020 * major incompatible change: The default fasl file extension, i.e. the
2021 default extension for files produced by COMPILE-FILE, has changed
2022 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2024 ** There are many changes in the implementation of the compiler.
2025 SBCL is now essentially a compiler-only implementation of ANSI
2026 Common Lisp. EVAL still "interprets" a few special cases, but
2027 almost all the interesting cases are handled by creating
2028 a LAMBDA expression, calling COMPILE on it, then calling
2029 FUNCALL on the result.
2030 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2031 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2032 Since the code is newer, there might still be some new bugs
2033 (though not as many as before Martin Atzmueller's fixes:-). But
2034 the new code is substantially simpler and clearer, and hopefully
2035 any remaining bugs will be simpler, less fundamental, and more
2036 fixable then the bugs in the old code.
2037 ** The revised compiler is still a little unsteady on its feet.
2039 *** The debugging information it produces (particularly the names
2040 of FUNCTION objects) is sometimes much less useful than what
2041 the old compiler produced.
2042 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2043 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2044 within a MACROLET) has been temporarily weakened.
2045 ** There are new compiler optimizations for various functions:
2046 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2047 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2048 *** the math functions TRUNCATE, FLOOR, and CEILING
2049 *** the function-of-all-trades COERCE
2050 Mostly these should be transparent, but there's one
2051 potentially-annoying problem (bug 117): when the compiler
2052 inline-expands a function and does type analysis on the result,
2053 it can create control paths which have type mismatches, and
2054 when it can't prove that those control paths aren't taken,
2055 it will issue WARNINGs about the type mismatches. This is
2056 a particular problem in practice for the new sequence functions.
2057 It's not clear how this should be fixed, and for now, a
2058 workaround is given in the entry for 117 in the BUGS file.
2059 ** (Because of the interaction between the two previous items --
2060 occasional inlining problems and new inline expansions -- some
2061 of the new sequence function optimizations won't really kick in
2062 completely until debugging information, and then inlining, are
2063 straightened out in some future version.)
2064 * minor incompatible changes:
2065 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2066 DIRECTORY no longer implicitly promotes NIL slots of its
2067 pathname argument to :WILD. In particular, when you ask for the
2068 contents of a directory (which you used to be able to do without
2069 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2070 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2071 ** changes in behavior that ANSI explicitly defines to be
2072 implementation dependent:
2073 *** The new compiler-only implementation still conforms with ANSI,
2074 but acts a little different than before. Besides the obvious
2075 changes in performance tradeoffs (that the cost per form passed
2076 to EVAL has gone up, and the cost per form executed by EVAL
2077 has gone down), the behavior of the system changes a little
2078 because there are no longer any interpreted function objects.
2079 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2080 e.g. doing COMPILE on the output of interactive DEFUN is
2082 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2084 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2085 has changed from (:CL) to NIL.
2086 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2087 e.g. #\Newline and #\Tab, don't have names specified in the
2088 ANSI Common Lisp standard, is now based on their ASCII symbolic
2089 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2090 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2091 are no longer used for output.
2092 ** changes in internal implementation constants:
2093 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2094 4 million. (If your application spends a lot of time GCing and
2095 you have a lot of RAM, you might want to experiment with
2096 increasing it even more.)
2097 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2098 However, almost all old code should still continue to work without
2099 immediate update, as SB-C-CALL is now a (deprecated) nickname
2101 ** Old operator names in the style DEF-FOO are now deprecated in
2102 favor of new corresponding names DEFINE-FOO, for consistency with
2103 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2104 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2105 internal symbols, but a few supported extensions like
2106 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2107 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2108 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2109 etc. as you get deeper into recursive calls to the debugger
2110 command loop, instead of the old "5]", "5]]", "5]]]"
2111 sequence. (I was motivated to do this when squabbles between
2112 ILISP and SBCL left me very deeply nested in the debugger. In the
2113 short term, this change will probably provoke more ILISP/SBCL
2114 squabbles, but hopefully it will be an improvement in the long run.)
2115 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2116 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2117 function, since declaiming return types involving aliens is
2118 (1) annoyingly messy to do by hand and (2) vital to efficient
2119 compilation of code which calls such functions.
2120 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2121 longer reexported by the SB-EXT package. They're solely useful
2122 for alien code, so it seems more logical that you should get
2123 them from the SB-ALIEN package, not in SB-EXT.
2124 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2125 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2126 features. Instead, the code that they used to control is always
2127 built into the system.
2128 * many other bug fixes
2129 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2130 advantage of the new EVAL-WHEN stuff and to clean them up in
2131 general, and they are now more ANSI-compliant in a number of
2132 ways. Martin Atzmueller is responsible for a lot of this.
2133 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2135 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2136 *** correct ERROR type for various file operations
2137 *** some fixes for Lisp streams
2138 *** DEFMETHOD syntax checking
2139 *** changing old weird representation of debug information as
2140 strings (which, among their other deficiencies, don't transform
2141 correctly when you rename packages, and don't change their
2142 print representation when you change things like *PACKAGE*
2143 and *PRINT-LENGTH*) to symbols and lists of symbols
2144 He also made several improvements and fixed several bugs in DESCRIBE.
2145 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2147 *** misbehavior of WRITE-STRING/WRITE-LINE
2148 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2149 and several other LOOP problems as well
2150 *** DIRECTORY when similar filenames are present
2151 *** DEFGENERIC with :METHOD options
2152 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2153 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2154 *** argument ordering in FIND with :TEST option
2155 *** mishandled package designator argument in APROPOS-LIST
2156 *** various problems in the backquote readmacro
2157 *** a bug in APROPOS
2158 *** probably some others that I'm not describing very well here,
2159 since the CVS log documents them by reference to sbcl-devel
2160 messages, and the SourceForge archives aren't working well.:-(
2161 ** Dan Barlow improved the Alpha port (and is making progress on the
2162 PPC port, for those of you who think different).
2163 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2164 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2165 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2166 continues to work on the SPARC port (for those of you in a position
2167 to look down upon our little PC-compatible boxes from a great height).
2168 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2169 and uses that copy, rather than the current dynamic value, when
2170 it's trying to decide whether to truncate output. Thus e.g.
2171 (let ((*print-lines* 50))
2172 (pprint-logical-block (stream nil)
2174 (let ((*print-lines* 8))
2175 (print (aref possiblybigthings i) stream)))))
2176 should now truncate the logical block only at 50 lines, instead of
2177 often truncating it at 8 lines, as it did before.
2178 * The doc/cmucl/ directory, containing old CMU CL documentation
2179 from the time of the fork, is no longer part of the base system.
2180 SourceForge has shut down its anonymous FTP service, and with it
2181 my original plan for distributing the old CMU CL documentation
2182 there. For now, if you need these files you can download an old
2183 SBCL source release and extract them from it.
2184 * The fasl file version number changed again, for dozens of reasons,
2185 some of which are apparent above.
2187 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2188 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2189 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2190 patch, so that DISASSEMBLE output is much nicer.
2191 * The code in the SB-PROFILE package now seems reasonably stable.
2192 I still haven't decided what the final interface should look like
2193 (I'd like PROFILE to interact cleanly with TRACE, since both
2194 facilities use function encapsulation) but if you have a need
2195 for profiling now, you can probably use it successfully with
2196 the current CMU-CL-style interface.
2197 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2198 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2199 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2200 link now returns the dangling link itself, and for similar
2201 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2202 link itself. (In these cases the old code signalled an error and
2203 looped endlessly, respectively.) Thus, DIRECTORY now works even
2204 in the presence of dangling and cyclic symbolic links.
2205 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2206 is now a supported extension again, since the consensus on
2207 sbcl-devel was that it can be useful for ordinary development
2208 work, not just for debugging SBCL itself.
2209 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2210 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2211 that currently-defined functions might be redefined later with
2212 different return types.
2213 * Hash tables can be printed readably, as inspired by CMU CL code
2214 of Eric Marsden and SBCL code of Martin Atzmueller.
2215 * better error handling in CLOS method combination, thanks to
2216 Martin Atzmueller porting Pierre Mai's CMU CL patches
2217 * more overflow fixes for >16Mbyte I/O buffers
2218 * A bug in READ has been fixed, so that now a single Ctrl-D
2219 character suffices to cause end-of-file on character streams.
2220 In particular, now you only need one Ctrl-D at the command
2221 line (not two) to exit SBCL.
2222 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2224 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2225 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2226 does the right thing.
2227 * The implementation of some type tests, especially for CONDITION
2228 types, is now tidier and maybe faster, due to CMU CL code
2229 originally by Douglas Crosher, ported by Martin Atzmueller.
2230 * Some math functions have been fixed, and there are new
2231 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2232 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2233 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2234 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2235 likely to be built into sbcl-0.7.0.)
2236 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2237 CL, ported to SBCL by Martin Atzmueller)
2238 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2239 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2240 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2241 * There's a new slam.sh hack to shorten the edit/compile/debug
2242 cycle for low-level changes to SBCL itself, and a new
2243 :SB-AFTER-XC-CORE target feature to control the generation of
2244 the after-xc.core file needed by slam.sh.
2245 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2246 COMPILE-FILE is no longer supported, so that now every function
2247 gets an entry point, so that block compilation looks a little
2248 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2249 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2250 returns the number of bytes consed since the system started,
2251 rather than the number consed since the first time the function
2252 was called. (The new definition parallels ANSI functions like
2253 CL:GET-INTERNAL-RUN-TIME.)
2254 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2255 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2256 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2257 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2258 * Fasl file version numbers are now independent of the target CPU,
2259 since historically most system changes which required version
2260 number changes have affected all CPUs equally. Similarly,
2261 the byte fasl file version is now equal to the ordinary
2264 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2265 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2266 is no longer recognized. I apologize for this, because it was
2267 listed in SB-EXT as a supported extension, but I found that
2268 its existing behavior was poorly specified, as well as incorrectly
2269 specified, and it looked like too much of a mess to straighten it
2270 out. I have enough on my hands trying to get ANSI stuff to work..
2271 * many patches ported from CMU CL by Martin Atzmueller, with
2272 half a dozen bug fixes in pretty-printing and the debugger, and
2273 half a dozen others elsewhere
2274 * fixed bug 13: Floating point infinities are now supported again.
2275 They might still be a little bit flaky, but thanks to bug reports
2276 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2277 as flaky as they were.
2278 * The --noprogrammer command line option is now supported. (Its
2279 behavior is slightly different in detail from what the old man
2280 page claimed it would do, but it's still appropriate under the
2281 same circumstances that the man page talks about.)
2282 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2283 are now supported, and enabled by default. Thus, the compiler can
2284 handle many floating point and complex operations much less
2285 inefficiently. (Thus e.g. you can implement a complex FFT
2287 * The compiler now detects type mismatches between DECLAIM FTYPE
2288 and DEFUN better, and implements CHECK-TYPE more correctly, and
2289 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2290 thanks to patches from Martin Atzmueller.
2291 * various fixes to make the cross-compiler more portable to
2292 ANSI-conforming-but-different cross-compilation hosts (notably
2293 Lispworks for Windows, following bug reports from Arthur Lemmens)
2294 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2295 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2296 fixed thanks to Pierre Mai's CMU CL patches.
2297 * fixes to keep the system from overflowing internal counters when
2298 it tries to use i/o buffers larger than 16M bytes
2299 * fixed bug 45a: Various internal functions required to support
2300 complex special functions have been merged from CMU CL sources.
2301 (When I was first setting up SBCL, I misunderstood a compile-time
2302 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2303 * improved support for type intersection and union, fixing bug 12
2304 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2305 more obscure bugs as well
2306 * some steps toward byte-compiling non-performance-critical
2307 parts of the system, courtesy of patches from Martin Atzmueller
2308 * Christophe Rhodes has made some debian packages of sbcl at
2309 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2310 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2311 stable, but are nonetheless usable. When he's ready, I'd be happy
2312 to add them to the SourceForge "File Releases" section. (And if
2313 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2314 * new fasl file format version number (because of changes in
2315 internal representation of (OR ..) types to accommodate the new
2316 support for (AND ..) types, among other things)
2318 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2319 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2321 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2323 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2324 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2325 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2326 * DESCRIBE now works on structure objects again.
2327 * Most function call argument type mismatches are now handled as
2328 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2329 know whether the function will be redefined before the call is
2330 executed. (The compiler could flag local calls with full WARNINGs,
2331 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2332 it doesn't keep track of enough information to know whether calls
2333 are local in this sense.)
2334 * Compiler output is now more verbose, with messages truncated
2335 later than before. (There should be some supported way for users
2336 to override the default verbosity, but I haven't decided how to
2337 provide it yet, so this behavior is still controlled by the internal
2338 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2339 src/compiler/ir1util.lisp.)
2340 * Fasl file format version numbers have increased again, because
2341 support for the Gray streams extension changes the layout of the
2342 system's STREAM objects.
2343 * The Gray subclassable streams extension now works, thanks to a
2344 patch from Martin Atzmueller.
2345 * The full LOAD-FOREIGN extension (not just the primitive
2346 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2347 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2348 but like most other programs, it defaults to copying the Unix
2349 environment from the original process instead of starting the
2350 new process in an empty environment.
2351 * Extensions which manipulate the Unix environment now support
2352 an :ENVIRONMENT keyword option which doesn't smash case or
2353 do other bad things. The CMU-CL-style :ENV option is retained
2354 for porting convenience.
2355 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2356 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2358 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2360 * A patch from Martin Atzmueller seems to have solved the SIGINT
2361 problem, and as far as we know, signal-handling now works cleanly.
2362 (If you find any new bugs, please report them!)
2363 * The system no longer defaults Lisp source file names to types
2364 ".l", ".cl", or ".lsp", but only to ".lisp".
2365 * The compiler no longer uses special default file extensions for
2366 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2367 seems to expect a single default extension for all compiled code,
2368 and there's no compelling reason to try to stretch the standard
2369 to allow two different extensions.) Instead, byte-compiled files
2370 default to the same extension as native-compiled files.
2371 * Fasl file format version numbers have increased again, because
2372 a rearrangement of internal implementation packages made some
2373 dumped symbols in old fasl files unreadable in new cores.
2374 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2375 many fewer weird special cases.
2376 * Bug #17 (differing COMPILE-FILE behavior between logical and
2377 physical pathnames) has been fixed, and some related misbehavior too,
2378 thanks to a patch from Martin Atzmueller.
2379 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2380 by Tim Moore, ported to SBCL by Martin Atzmueller.
2381 * Martin Atzmueller fixed several filesystem-related problems,
2382 including bug #36, in part by porting CMU CL patches, which were
2383 written in part by Paul Werkowski.
2384 * More compiler warnings in src/runtime/ are gone, thanks to
2385 more patches from Martin Atzmueller.
2386 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2389 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2391 * DESCRIBE now works on CONDITION objects.
2392 * The debugger now handles errors which arise when trying to print
2393 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2395 * The build system now uses an additional file, customize-target-features.lisp,
2396 to allow local modifications to the target *FEATURES* list. (The point of
2397 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2398 debugging features enabled, without having to worry about propagating it
2399 into everyone's system when I do a "cvs update".) When no
2400 customize-target-features.lisp file exists, the target *FEATURES* list
2401 should be constructed the same way as before.
2402 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2403 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2404 warn about behavior which is undefined under the ANSI spec. Note: This
2405 is specified by ANSI, but it's not very popular with programmers.
2406 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2407 macro in the SBCL sources for an example of a workaround which you
2408 might use to make portable ANSI-standard code which does what you want.
2409 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2410 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2411 when it's used as a non-toplevel form. (This is required in order
2412 to implement the DEFCONSTANT-EQX macro.)
2413 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2414 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2415 funny things, due to bugs in the implementation of EVAL-WHEN
2416 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2417 non-toplevel DEFCONSTANTs are uncommon.)
2418 * The core file version number and fasl file version number have been
2419 incremented, because the old noncompliant DEFCONSTANT behavior involved
2420 calling functions which no longer exist, and because I also took the
2421 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2422 * fixed bug 1 (error handling before read-eval-print loop starts), and
2423 redid debugger restarts and related debugger commands somewhat while
2425 ** The QUIT debugger command is gone, since it did something
2426 rather different than the SB-EXT:QUIT command, and since it never
2427 worked properly outside the main toplevel read/eval/print loop.
2428 Invoking the new TOPLEVEL restart provides the same functionality.
2429 ** The GO debugger command is also gone, since you can just invoke
2430 the CONTINUE restart directly instead.
2431 ** The TOP debugger command is also gone, since it's redundant with the
2432 FRAME 0 command, and since it interfered with abbreviations for the
2434 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2436 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2437 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2438 some porting work by Martin Atzmueller)
2439 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2440 with a boatload of patches which he ported to SBCL. Now that those
2442 ** The system tries to make sure that its low-priority messages
2443 are prefixed by semicolons, to help people who like to use
2444 syntax highlighting in their ILISP buffer. (This patch
2445 was originally due to Raymond Toy.)
2446 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2447 patches originally written by Raymond Toy.
2448 ** The compiler understands coercion between single-value and
2449 multiple-VALUES type expressions better, getting rid of some very
2450 weird behavior, thanks to patches originally by Robert MacLachlan
2451 and Douglas Crosher.
2452 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2453 lambda lists, thanks to a patch originally by Pierre Mai.
2454 ** The system no longer bogusly warns about "abbreviated type
2456 ** The compiler gets less confused by inlining and RETURN-FROM,
2457 thanks to some patches originally by Tim Moore.
2458 ** The system no longer hangs when dumping circular lists to fasl
2459 files, thanks to a patch originally from Douglas Crosher.
2460 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2461 undefined function error.
2462 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2463 in structure slot initforms) for now, documented workaround instead:-|
2464 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2465 * fixed bug 5: added stubs for various Gray stream functions called
2466 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2467 installed, at least appropriate type errors are generated
2468 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2469 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2470 consistently in DEFMETHOD forms.
2471 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2472 it doesn't seem to affect SBCL after all
2473 * The C runtime system now builds with better optimization and many
2474 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2476 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2478 * The system is now under CVS at SourceForge (instead of the
2479 CVS repository on my home machine).
2480 * The new signal handling code has been tweaked to treat register
2481 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2482 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2483 downstream, e.g. in debug-int.lisp, has implicit dependencies
2484 on the unsignedness of integer representation of machine words,
2485 and that caused the system to bomb out with infinite regress
2486 when trying to recover from type errors involving signed values,
2487 e.g. (BUTLAST '(1 2 3) -1).)
2488 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2489 (This was one of the bugs Peter Van Eynde reported back in July.)
2490 * The system now uses code inspired by Colin Walters' O(N)
2491 implementation of MAP (from the cmucl-imp@cons.org mailing
2492 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2493 inline the MAP operation, and there is more than one
2494 sequence argument to the MAP call (so that it can't just
2495 do ETYPECASE once and for all based on the type of the
2496 single sequence argument). (The old non-inline implementation
2497 of the general M-argument sequence-of-length-N case required
2498 O(M*N*N) time when any of the sequence arguments were LISTs.)
2499 * The QUIT :UNIX-CODE keyword argument has been renamed to
2500 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2501 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2503 * Raymond Wiker's patches to port dynamic loading from Linux to
2504 FreeBSD have been added.
2505 * The BUGS file is now more nearly up to date, thanks in large part
2506 to Martin Atzmueller's review of it.
2507 * The debugger now flushes standard output streams before it begins
2508 its output ("debugger invoked" and so forth).
2509 * The core version number and fasl file version number have both
2510 been incremented, because of incompatible changes in the layout
2512 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2513 used to not be called for a saved Lisp image.)
2514 * Martin Atzmueller's version of a patch to fix a compiler crash,
2515 as posted on sbcl-devel 13 September 2000, has been installed.
2516 * Instead of installing Martin Atzmueller's patch for the
2517 compiler transform for SUBSEQ, I deleted the compiler transform,
2518 and transforms for some similar consing operations.
2519 * A bug in signal handling which kept TRACE from working on OpenBSD
2521 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2522 away by constant folding
2523 * The system now defines its address space constants in one place
2524 (in the Lisp sources), and propagates them automatically elsewhere
2525 (through GENESIS and the sbcl.h file). Therefore, patching the
2526 address map is less unnecessarily tedious and error-prone. The
2527 Lisp names of address space constants have also been systematized.
2528 * CVS tags like dollar-Header-dollar have been removed from
2529 the sources, because they have never saved me trouble and
2530 they've been source of trouble working with patches and other
2531 diff-related operations.
2532 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2533 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2535 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2537 * The system has been ported to OpenBSD.
2538 * The system now compiles with a simple "sh make.sh" on the systems
2539 that it's supported on. I.e., now you no longer need to tweak
2540 text in the target-features.lisp-expr and symlinks in src/runtime/
2541 by hand, the make.sh takes care of it for you.
2542 * The system is no longer so grossly inefficient when compiling code
2543 involving vectors implemented as general (not simple) vectors (VECTOR T),
2544 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2545 which dares to use the various sequence functions on non-simple
2546 vectors, takes less of a performance hit.
2547 * There is now a primitive type predicate VECTOR-T-P
2548 to test for the (VECTOR T) type, so that e.g.
2549 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2550 can now be compiled with some semblance of efficiency. (The old code
2551 turned the type declaration into a full call to %TYPEP at runtime!)
2552 * AREF on (VECTOR T) is still not fast, since it's still compiled
2553 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2554 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2555 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2556 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2557 without making full calls to SUBTYPEP at runtime.
2558 (Some analogous efficiency issues for non-simple vectors specialized to
2559 element types other than T, or for non-simple multidimensional arrays,
2560 have not been addressed. They could almost certainly be handled the
2561 same way if anyone is motivated to do so.)
2562 * The changes in array handling break binary compatibility, so
2563 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2564 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2565 instead of (VALUES T T).
2566 * By following the instructions that Dan Barlow posted to sbcl-devel
2567 on 2 July 2000, I was able to enable primitive dynamic object
2568 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2569 functionality is not implemented (since it calls ld to resolve
2570 library references automatically, requiring RUN-PROGRAM for its
2571 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2572 try to resolve library references) is now supported.
2573 * The system now flushes the standard output streams when it terminates,
2574 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2575 them at several other probably-convenient times, e.g. in each pass of
2576 the toplevel read-eval-print loop, and after evaluating a form given
2577 as an "--eval" command-line option. (These changes were motivated by a
2578 discussion of stream flushing issues on cmucl-imp in August 2000.)
2579 * The source transform for TYPEP of array types no longer assumes
2580 that an array whose element type is a not-yet-defined type
2581 is implemented as an array of T, but instead punts, so that the
2582 type will be interpreted at runtime.
2583 * There is now some support for cross-compiling in make.sh: each of
2584 the phases of make.sh has its own script. (This should be transparent
2585 to people doing ordinary, non-cross-compile builds.)
2586 * Since my laptop doesn't have hundreds of megabytes of memory like
2587 my desktop machine, I became more motivated to do some items on
2588 my to-do list in order to reduce the size of the system a little:
2589 ** Arrange for various needed-only-at-cold-init things to be
2590 uninterned after cold init. To support this, those things have
2591 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2592 symbols with such names are now uninterned after cold init).
2593 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2594 fasl files for cold load.
2595 ** Remove the old compiler structure pooling code (which used to
2596 be conditional on the target feature :SB-ALLOC) completely.
2597 ** Redo the representation of some data in cold init to be more compact.
2598 (I also looked into supporting byte compiled code at bootstrap time,
2599 which would probably reduce the size of the system a lot, but that
2600 looked too complicated, so I punted for now.)
2601 * The maximum signal nesting depth in the src/runtime/ support code has
2602 been reduced from 4096 to 256. (I don't know any reason for the very
2603 large old value. If the new smaller value turns out to break something,
2604 I'll probably just bump it back up.)
2605 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2607 * Many, many bugs reported by Peter Van Eynde have been added to
2608 the BUGS list; some have even been fixed.
2609 * While enabling dynamic object file loading, I tried to make the
2610 code easier to understand, renaming various functions and variables
2611 with less ambiguous names, and changing some function calling
2612 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2613 * While trying to figure out how to do the OpenBSD port, I tried to
2614 clean up some of the code in src/runtime/. In particular, I dropped
2615 support for non-POSIX signal handling, added various comments,
2616 tweaked the code to reduce the number of compilation warnings, and
2617 renamed some files to increase consistency.
2618 * To support the new automatic configuration functionality in make.sh,
2619 the source file target-features.lisp-expr has been replaced with the
2620 source file base-target-features.lisp-expr and the machine-generated
2621 file local-target-features.lisp-expr.
2622 * fixed a stupid quoting error in make.sh so that using CMU CL
2623 "lisp -batch" as cross-compilation host works again
2625 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2627 * DESCRIBE no longer tries to call itself recursively to describe
2628 bound/fbound values, so that it no longer fails on symbols which are
2629 bound to themselves (like keywords, T, and NIL).
2630 * DESCRIBE now works on generic functions.
2631 * The printer now prints less-screwed-up representations of closures
2632 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2633 * A private symbol is used instead of the :EMPTY keyword previously
2634 used to mark empty slots in hash tables. Thus
2635 (DEFVAR *HT* (MAKE-HASH-TABLE))
2636 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2637 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2638 now does what ANSI says that it should. (You can still get
2639 similar noncompliant behavior if bang on the hash table
2640 implementation with all the symbols you get back from
2641 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2642 This breaks binary compatibility, since tests for equality to
2643 :EMPTY are wired into things like the macroexpansion of
2644 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2646 * There's now a minimal placeholder implementation for CL:STEP,
2647 as required by ANSI.
2648 * An obscure bug in the interaction of the normal compiler, the byte
2649 compiler, inlining, and structure predicates has been patched
2650 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2651 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2653 * Missing ordinary arguments in a macro call are now detected even
2654 when the macro lambda list contains &KEY or &REST.
2655 * The debugger no longer complains about encountering the top of the
2656 stack when you type "FRAME 0" to explicitly instruct it to go to
2657 the top of the stack. And it now prints the frame you request even
2658 if it's the current frame (instead of saying "You are here.").
2659 * As specified by ANSI, the system now always prints keywords
2660 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2662 * The default initial SIZE of HASH-TABLEs is now smaller.
2663 * Type information from CLOS class dispatch is now propagated
2664 into DEFMETHOD bodies, so that e.g.
2665 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2667 is now basically equivalent to
2668 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2669 (DECLARE (TYPE SINGLE-FLOAT X))
2671 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2672 operation, without having to do run-time type dispatch.
2673 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2674 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2675 * Since I don't seem to be making big file reorganizations very often
2676 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2677 is overflowing my ability to conveniently back them up), I've finally
2678 checked the system into CVS. (The CVS repository is on my home system,
2679 not at SourceForge -- putting it on SourceForge might come later.)
2680 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2681 high-level GC-NOTIFY-FOO functions send their output. (There's
2682 still very little control of where low-level verbose GC functions
2683 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2684 less than it used to -- the GC-NOTIFY-FOO functions are now under
2685 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2686 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2687 in only one place in the source code, and propagates it automatically
2688 everywhere that it's needed. Thus e.g. when I bump the version from
2689 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2690 * The C source files now include boilerplate legalese and documentation
2691 at the head of each file (just as the Lisp source files already did).
2692 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2693 to his page will be replaced with a link to his new CLiki service.
2695 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2697 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2698 * The build process now looks for GNU make under the default name "gmake",
2699 instead of "make" as it used to. If GNU make is not available as "gmake"
2700 on your system, you can change this default behavior by setting the
2701 GNUMAKE environment variable.
2702 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2703 can build without error under CMU CL.
2705 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2707 * There is now a partial SBCL user manual (with some new text and some
2708 text cribbed from the CMU CL manual).
2709 * The beginnings of a profiler have been added (starting with the
2710 CMU CL profiler and simplifying and cleaning up). Eventually the
2711 main interface should be through the TRACE macro, but for now,
2712 it's still accessed through vaguely CMU-CL-style functions and macros
2713 exported from the package SB-PROFILE.
2714 * Some problems left over from porting CMU CL to the new
2715 cross-compilation bootstrap process have been cleaned up:
2716 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2717 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2718 part when getting rid of the compiler *BACKEND* stuff.)
2719 ** The value of *NULL-TYPE* was screwed up, because it was
2720 being initialized before the type system knew the final
2721 definition of the 'NULL type. This screwed up several key
2722 optimizations in the compiler, causing inefficiency in all sorts
2723 of places. (I found it because I wanted to understand why
2724 GET-INTERNAL-RUN-TIME was consing.)
2725 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2726 PROCLAIM FTYPE information. Unfortunately this broke binary
2727 compatibility again, since now the forms output by DEFGENERIC
2728 to refer to functions which didn't exist in 0.6.3.
2729 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2730 can use the new (as of 0.6.3) transform for SOME into MAP into
2732 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2733 values to (INTEGER 0 1000000), so that the time code will no longer
2734 occasionally get blown up by Linux returning 1000000 microseconds
2735 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2736 its output conform to the ANSI spec. (Alas, this makes its output
2737 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2738 * A full call to MAP NIL with a single sequence argument no longer conses.
2739 * fixes to problems pointed out by Martin Atzmueller:
2740 * The manual page no longer talks about multiprocessing as though
2741 it were currently supported.
2742 * The ILISP support patches have been removed from the distribution,
2743 because as of version 5.10.1, ILISP now supports SBCL without us
2744 having to maintain patches.
2745 * added a modified version of Raymond Toy's recent CMU CL patch for
2746 EQUALP comparison of HASH-TABLE
2748 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2750 * The system still can't cross-compile itself with
2751 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2752 entails), but at least it can compile more of itself that way
2753 than it used to be able to, and various buglets which were uncovered
2754 by trying to cross-compile itself that way have now been fixed.
2755 * This release breaks binary compatibility again. This time
2756 at least I've incremented the FASL file format version to 2, so that the
2757 problem can be detected reliably instead of just causing weird errors.
2758 * various new style warnings:
2759 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2760 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2761 ** using the lexical binding of a variable named in the *FOO* style
2762 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2763 as specified by ANSI.
2764 * *RANDOM-STATE* is no longer automatically initialized from
2765 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2766 default behavior of the system is to repeat its behavior every time
2767 it's run. If you'd like to change this behavior, you can always
2768 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2769 old convention there was no comparably easy way to get the system to
2770 repeat its behavior every time it was run.
2771 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2772 FTYPE declarations has been removed, in favor of their ANSI
2773 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2774 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2775 the types of their sequence arguments can be determined at compile time.
2776 This is done through a new open code expansion for MAP which eliminates
2777 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2778 argument types can be determined at compile time.
2779 * The optimizer now transforms COERCE into an identity operation when it
2780 can prove that the coerced object is already of the correct type. (This
2781 can be a win for machine generated code, including the output of other
2782 optimization transforms, such as the MAP transform above.)
2783 * Credit information has been moved from source file headers into CREDITS.
2784 * Source file headers have been made more standard.
2785 * The CASE macro now compiles without complaining even when it has
2788 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2790 * (Note that the way that the PCL macroexpansions were rewritten
2791 to accommodate the change in DEFGENERIC below breaks binary
2792 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2793 not run under sbcl-0.6.2. Once we get out of alpha releases,
2794 i.e. hit release 1.0.0, we'll probably try to maintain binary
2795 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2796 and sbcl-1.4.4. Until then, however, it might be fairly common
2797 for maintenance releases to break binary compatibility.)
2798 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2800 * The reporting of unhandled signals has been changed to print some
2801 explanatory text as well as the report form. (Previously only
2802 the report form was printed.)
2803 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2804 it defines, so that the compiler no longer issues undefined function
2805 warnings for compiled-but-not-yet-loaded generic functions.
2806 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2807 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2808 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2809 section of the ANSI spec are supported.
2810 * The "" nickname for the "KEYWORD" package has been removed.
2811 The reader still handles symbol tokens which begin with a package marker
2812 as keywords, but it doesn't expose its mechanism for doing so in the
2813 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2814 * The system now issues STYLE-WARNINGs for contradictory TYPE
2815 proclamations. (Warnings for contradictory FTYPE proclamations would
2816 be nice too, but those can't be done usefully unless the type system
2817 is made smarter about FUNCTION types.)
2818 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2819 systematized, so that "*target-*.lisp is supposed to exist only on the
2820 target and imply that there's a related file which exists on the
2821 host, and *host-*.lisp is supposed to exist only on the host and imply
2822 that there's a related file which exists on the target. This involves a
2823 lot of renaming. Hopefully the acute confusion caused by the renaming
2824 will be justified by the reduction in chronic confusion..
2825 ** runtime-type.lisp -> early-target-type.lisp
2826 ** target-type.lisp -> late-target-type.lisp
2827 ** early-host-format.lisp -> early-format.lisp
2828 ** late-host-format.lisp -> late-format.lisp
2829 ** host-error.lisp -> misc-error.lisp
2830 ** early-error.lisp -> early-target-error.lisp
2831 ** late-error.lisp -> late-target-error.lisp
2832 ** host-defboot.lisp -> early-defboot.lisp
2833 ** code/misc.lisp -> code/target-misc.lisp
2834 ** code/host-misc.lisp -> code/misc.lisp
2835 ** code/numbers.lisp -> code/target-numbers.lisp
2836 ** code/early-numbers.lisp -> numbers.lisp
2837 ** early-host-type.lisp -> early-type.lisp
2838 ** late-host-type.lisp -> late-type.lisp
2839 ** host-typep.lisp -> typep.lisp
2840 ** load.lisp -> target-load.lisp
2841 ** host-load.lisp -> load.lisp
2842 ** host-disassem.lisp -> disassem.lisp
2843 ** host-insts.lisp -> insts.lisp
2844 ** byte-comp.lisp -> target-byte-comp.lisp
2845 ** host-byte-comp.lisp -> byte-comp.lisp
2846 ** host-signal.lisp -> signal.lisp
2847 ** host-defstruct.lisp -> defstruct.lisp
2848 ** late-target-type.lisp -> deftypes-for-target.lisp
2849 Furthermore, several other previously target-only files foo.lisp (e.g.
2850 hash-table.lisp and random.lisp) have been split into a target-and-host
2851 foo.lisp file and a target-only target-foo.lisp file, with their key type
2852 definitions in the target-and-host part, so that the cross-compiler will
2853 know more about target types.
2854 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2855 gone away. In their place are various *BACKEND-FOO* variables
2856 corresponding to the slots of the old structure.
2857 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2858 into the target SBCL has been fixed.
2859 * The chill.lisp system for loading cold code into a running SBCL
2861 * Support for the CMU CL "scavenger hook" extension has been removed.
2862 (It was undocumented and unused in the CMU CL sources that SBCL was
2863 derived from, and stale in sbcl-0.6.1.)
2864 * Various errors in the cross-compiler type system were detected
2865 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2866 (enabling various consistency checks). Many of them were fixed,
2867 but some hard problems remain, so the compiler is back to
2868 running without *TYPE-SYSTEM-INITIALIZED* for now.
2869 * As part of the cross-compiler type system cleanup, I implemented
2870 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2871 * I have started adding UNCROSS calls throughout the type system
2872 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2873 on cross-compiler input files will be able to go away, and various
2875 * CONSTANTP now returns true for quoted forms (as explicitly required
2878 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2880 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2881 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2882 (MAKE-LIST -1). In the longer term, it ought to become true
2883 that declarations are assertions even at SAFETY 1. For now, it's not
2884 quite true even at SAFETY 3, but it's at least more nearly true..
2885 (Note that this change seems to increases the size of the system by
2886 O(5%) and to decrease the speed of the compiler by 20% or more.)
2887 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2888 for the problem of printing dozens of lines of distracting information
2889 about low-level system machinery as part of the top stack frame
2890 on entry to the debugger when an undefined function was called.
2891 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2892 is not reset to COMMON-LISP-USER.
2893 * Compilation of stuff related to dyncount.lisp has been made conditional
2894 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2895 smaller. The various dyncount-related symbols have been moved into
2896 a new "SB-DYNCOUNT" package.
2897 * tty-inspect.lisp has been renamed to inspect.lisp.
2898 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2899 feature has gone away. (When we eventually port to other flavors of
2900 libc and/or Unix, we'll try to make the differences between flavors
2901 invisible at the user level.)
2902 * Various other *FEATURES* tags, and/or their associated conditionals,
2903 have been removed if obsolescent, or given better documentation, or
2904 sometimes given more-mnemonic names.
2906 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2908 * tidied up "make.sh" script
2909 * tidied up system directory structure
2910 * better "clean.sh" behavior
2911 * added doc/FOR-CMUCL-DEVELOPERS
2912 * many many small tweaks to output format, e.g. removing possibly-confusing
2913 trailing #\. character in DESCRIBE-INSTANCE
2914 * (EQUALP #\A 'A) no longer signals an error.
2915 * new hashing code, including EQUALP hashing
2916 * tidied up Lisp initialization and toplevel
2917 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2918 * command line argument processing
2919 * added POSIX-GETENV function to deal with Unix-ish environment variables
2920 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2921 terminating SBCL on EOF
2922 * non-verbose GC by default
2923 * There is no more "sbcl" shell script; the sbcl file is now the C
2924 runtime executable (just like CMU CL).
2925 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2927 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2928 symbol and package information as Lisp native symbol and package objects
2929 instead of strings naming symbols and strings naming packages. This way,
2930 whenever packages are renamed (as in warm init), debug information is
2931 transformed along with everything else.
2932 * tweaked the optimization policy declarations which control the building
2933 of SBCL itself. Now, among other things, the system no longer saves
2934 source location debugging information. (This helps two problems at once
2935 by reducing SBCL size and by keeping SBCL from trying to look for its
2936 sources -- which may not exist -- when reporting errors.)
2937 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2938 debugging and testing purposes
2939 * cleaned up printing, making the printer call PRINT-OBJECT for
2940 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2941 methods, giving nearly-ANSI behavior
2942 * converted almost all special variables to use *FOO* naming convention
2943 * deleted PARSE-TIME functionality, since it can be done portably
2944 * moved some files out of cold init into warm init
2945 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2947 * regularized formatting of source files
2948 * added an install.sh script
2949 * fixed ridiculous memory usage of cross-compiler by making
2950 compiler/alloc.lisp not try to do pooling unless it can hook
2951 itself into the GC of the cross-compilation host. Now the system
2952 builds nicely on my old laptop.
2953 * added :SB-ALLOC in target-features.lisp-expr
2954 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2955 was not implemented)
2956 * re-did condition handling and note reporting in the compiler. Notes
2957 are no longer handled by signalling conditions. Style warnings
2958 and warnings are handled more correctly and reported in such a way
2959 that it's easy to find one or the other in your output (so that you
2960 can e.g. figure out which of many problems caused COMPILE-FILE to
2962 * changed the severity of several compiler warnings from full WARNING
2963 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2964 compiler note reporting so that it doesn't use the condition system
2965 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2966 COMPILE-FILE command)
2967 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2968 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2969 no longer supported, since their implementation was deeply intertwingled
2970 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2971 * removed bogus "support" for compiler macros named (SETF FOO), and
2972 removed the compiler macro for SETF INFO (but only after making a fool
2973 of myself on the cmucl-imp mailing list by posting a bogus patch for
2974 DEFINE-COMPILER-MACRO..)
2975 * Compiled files containing forms which have side effects on the Lisp
2976 reader (such as DEFPACKAGE forms) are now handled more correctly.
2977 (Compiler queuing of top level lambdas has been suppressed by setting
2978 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2979 * deleted various currently-unused source files, e.g. gengc.lisp. They
2980 may be added back at some point e.g. when porting to other architectures,
2981 but until they are it's distracting to distribute them and to try to
2983 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2984 there were so many of them they're just distractions, and UNCROSS is
2985 known to be able to handle the current sources
2986 * moved PROFILE functionality into TRACE, so that it will be clear
2987 how the wrapping and unwrapping of functions when you profile them
2988 interacts with the wrapping and unwrapping of functions when you
2989 trace them. (Actually, the functionality isn't there yet, but at least
2990 the interface specification is there. Hopefully, the functionality will
2991 arrive with some future maintenance release.)
2992 * removed host-oops.lisp
2993 * changed signature of QUIT function to allow UNIX-CODE argument
2994 * fixed READ-SEQUENCE bug
2995 * tweaked verbose GC output so that it looks more like the progress
2996 output that ANSI specifies for functions like LOAD
2997 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2998 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob