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: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
8 dispatch table if the from-char is a dispatch macro character.
9 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
10 by the wordsize no longer ignore the last word. (reported by Lutz
12 * fixed bug in type checking of dynamic-extent variables. (reported
14 * optimization: sequence traversal functions use their freedom to
15 coerce function designators to functions.
16 * optimization: code with many calls to CLOS methods specialized on
17 CLOS classes has had redundant type checks removed.
18 * optimization: added declarations to speed up operations that access
19 the internal character database (for example STRING-UPCASE)
20 * fixed some bugs related to Unicode integration:
21 ** portions of multibyte characters at the end of buffers for
22 character-based file input are correctly transferred to the
23 start of the buffer at the next read.
24 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
25 passing it through to OPEN.
26 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
27 argument given to internal calls to OPEN.
28 ** the built-in comment readers (introduced by character sequences
29 ";" and "#|") are more forgiving to encoding errors; they will
30 STYLE-WARN and then attempt to resync the stream at a character
31 boundary. (thanks to Teemu Kalvas)
32 * fixed some bugs in the x86-64 port:
33 ** Negative short int return values from c-calls are sign-extended
35 ** The stack is aligned to 16-bytes for c-calls, as required by
36 the ABI. (thanks to Cheuksan Wang)
37 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
38 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
39 for enums. (thanks to Vincent Arkesteijn)
40 * fixed some bugs revealed by Paul Dietz' test suite:
41 ** Space, Tab, Linefeed, Return and Page have the invalid
42 secondary constituent character trait.
43 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
45 ** WITH-INPUT-FROM-STRING should only update the index place on
48 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
49 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
50 platform. The port supports 61-bit fixnums, large memory spaces
51 and reloading shared object files.
52 * enhancement: saving cores with foreign code loaded is now
53 supported on ppc/Darwin in addition to the previously supported
55 * enhancement: the statistical profiler now walks deeper into the
56 call stack for more meaningful call-graphs and accrued time
57 reports (x86/x86-64 only). It also now reports time spent in
59 * enhancement: it is now possible to trace most individual methods
60 of a generic function in addition to tracing the generic function
62 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
64 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
65 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
66 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
67 produces an error. (thanks to Vincent Arkesteijn)
68 * bug fix: NAMESTRING on pathnames with :WILD components in their
69 directories works correctly. (thanks to Artem V. Andreev)
70 * fixed bug 125: compiler preserves identity of closures. (reported
72 * bug fixed: functions with &REST arguments sometimes failed with
73 "Undefined function" when compiled with (DEBUG 3). (reported by
75 * bug fix: overflow during compiling of setting element of a bit
76 vector with constant index and value. (reported by Timmy Douglas)
77 * build fix: fixed the dependence on *LOAD-PATHNAME* and
78 *COMPILE-FILE-PATHNAME* being absolute pathnames.
79 * on x86 compiler partially supports stack allocation of dynamic-extent
81 * GO and RETURN-FROM do not check the extent of their exit points
82 when compiled with SAFETY 0.
83 * fixed some bugs related to Unicode integration:
84 ** encoding and decoding errors are now much more robustly
85 handled; it should now be possible to recover even from invalid
86 input or output to the terminal. (thanks to Teemu Kalvas)
87 ** provided a first cut at implementing STRING-TO-OCTETS and
88 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
89 ** altered the SB-MD5 contributed module slightly, changing the
90 interface just enough for it to be supportable for builds where
91 lisp characters are not eight bits.
92 * fixed some bugs revealed by Paul Dietz' test suite:
93 ** the FORMATTER-generated functions for ~V[ conditionals require
94 the correct number of arguments.
95 ** READ-FROM-STRING returns the mandated second value when applied
97 ** the #\Rubout and #\Backspace characters are treated as invalid
98 constituent characters by the tokenizer.
100 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
101 * new feature: reloading changed shared object files with
102 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
103 * new feature: references to foreign variables and functions
104 can now be compiled and loaded before the corresponding shared
105 object file is loaded, as long as the foreign definitions are
106 available at runtime.
107 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
108 supported. (thanks to Dan Debertin)
109 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
110 just keywords. (thanks to Vincent Arkesteijn)
111 * fixed bug #331: structure-class instances corresponding to
112 DEFSTRUCT forms are now created eagerly.
113 * fixed bug #345: backtraces from calls to undefined functions work
114 on x86 as well. Related bug #61 is now also partially fixed on x86
115 (backtraces from throws to unknown catch tags.)
116 * bug fix: lambda-list parsing is now stricter vrt. order and number
117 of lambda-list keywords.
118 * bug fix: as specified by AMOP, an error is signalled if a
119 class-option appears multiple times in a DEFCLASS form. (reported
121 * bug fix: RANDOM can be compiled when the compiler derives the type
122 of its numeric argument as a disjoint set of small integers.
123 (reported by Paul Dietz)
124 * bug fix: starting a core saved with shared objects loaded when
125 those objects are not available no longer causes threaded SBCL to
126 hang. (reported by Sean Ross)
127 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
129 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
130 argument, which must always be NIL. (reported by Kalle Niemitalo)
131 * bug fix: printing 1.0d+23 no longer results in an error.
132 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
133 * bug fix: structure slot setters preserve evaluation order. (thanks
135 * fixed some bugs related to Unicode integration:
136 ** RUN-PROGRAM can allow its child to take input from a Lisp
137 stream. (reported by Stefan Scholl)
138 ** ASDF-INSTALL successfully downloads .tar.gz files in all
139 locales. (reported by Ken Causey)
140 * fixed some bugs revealed by Paul Dietz' test suite:
141 ** INCF, DECF and REMF evaluate their place form as specified in
143 ** FORMATTER expands ~{ ~} iteration directives with V or #
144 parameters correctly.
145 ** FORMATTER deals with the ~@[ ~] conditional directive where the
146 consequent uses no arguments correctly.
147 ** the system has a partial understanding of the (COMPLEX RATIO)
149 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
152 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
153 * new feature: a build-time option (controlled by the :SB-UNICODE
154 keyword feature, enabled by default) for building the system with
155 support for the entire 21-bit character space defined by the
157 * new feature: the system now has rudimentary external-format
158 support; the primary user-visible change at this time is that
159 characters with the high bit set (such as lower-case-e-acute) will
160 print correctly to a terminal in a UTF-8 environment.
161 * minor incompatible change: BASE-CHAR no longer names a class;
162 however, CHARACTER continues to do so, as required by ANSI.
163 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
164 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
165 has been moved to the SB-EXT package (temporarily re-exported from
167 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
168 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
170 * fixed bug #351: better error-handlind and reporting for malformed
172 * fixed bug #350: bignum-printing is now more memory-efficient,
173 allowing printing of very large bignums, eg. (expt 2 10000000).
174 (reported by Bruno Haible)
175 * fixed bug #302: better primitive-type selection for intersection
177 * fixed bug #308: non-graphic characters now all have names, as
178 required. (reported by Bruno Haible)
179 * bug fix: redefining a class with different superclasses now correctly
180 removes it from the direct-subclasses of its previous superclasses.
181 (reported by David Morse)
182 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
183 new value now works. (reported by Bruno Haible)
184 * bug fix: correct canonicalization of multiple non-standard slot
185 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
186 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
187 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
188 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
189 now exists, an signals an error.
190 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
191 during macroexpansion for non-iteration variables as well. (reported
192 by Bruno Haible for CMUCL)
193 * bug fix: Cyclic structures and unprintable objects in compiler
194 messages no longer cause errors. (reported by Bruno Haible)
195 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
196 and READ-FROM-STRING all now return a primary value of NIL if
197 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
198 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
199 (reported by Bruno Haible for CMUCL)
200 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
201 do not designate a condition. (reported by Bruno Haible for
203 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
204 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
205 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
206 permitting resolution in favour of any of the conflicting symbols.
207 (reported by Bruno Haible for CMUCL)
208 * bug fix: EQUAL compiler optimizations is less aggressive on
209 strings which can potentially compare true despite having distinct
210 specialized array element types.
211 * bug fix: unit enumerations can be defined without dividing by
212 zero. (thanks to Vincent Arkesteijn)
213 * FORMAT compile-time argument count checking has been enhanced.
214 (report from Bruno Haible for CMUCL)
215 * a partial workaround for the bug 262: the compiler does not try to
216 inline-expand a local function doing RETURN-FROM from a deleted
217 BLOCK. (thanks to Peter Denno for the bug report and to David
218 Wragg for the simple test case)
219 * fixed some bugs revealed by Paul Dietz' test suite:
220 ** RENAME-PACKAGE allows all package designators as new package
222 ** constraint propagation and lambda variable substitution are
223 more cautious in dealing with partially deleted code.
224 ** compiler sometimes lost reoptimization passes.
225 ** CERROR, given a condition as condition designator, passes its
226 remaining arguments to the continue format control without
228 ** Case-altering FORMAT directives work correctly on non-ASCII
230 ** The REAL type specifier handles bounds outside the floating
231 point ranges without signalling FLOATING-POINT-OVERFLOW.
232 ** Functions with IR1-transformations can create intercomponent
233 references to global functions.
234 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
236 ** FORMAT ~R treats a nil value for its first parameter correctly.
238 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
239 * enhancement: saving cores with foreign code loaded is now
240 supported on x86/NetBSD and sparc/Linux in addition to the previously
242 * bug fix: on some platforms repeated installations caused multiple
243 copies of HTML documentation to be installed -- should not happen
244 any more. (reported by Stefan Scholl)
245 * bug fix: parsing self-recursive alien record types multiple times
246 no longer causes infinite recursion. (reported by Thomas F. Burdick,
247 original patch by Helmut Eller for CMUCL)
248 * bug fix: stack-exhaustion detection works now on NetBSD as well.
249 (thanks to Richard Kreuter)
250 * bug fix: defining classes whose accessors are methods on existing
251 generic functions in other (locked) packages no longer signals
252 bogus package lock violations. (reported by François-René Rideau)
253 * bug fix: special variables as DEFMETHOD parameters no longer have
254 associated bogus type declarations. (reported by David Wragg and
256 * bug fix: read-write consistency on streams of element-type
257 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
258 * bug fix: redefiniton of the only method of a generic function with
259 no DEFGENERIC no longer emits a full WARNING. In addition,
260 redefinition of generic functions with no DEFGENERIC to an
261 incompatible lambda list now signals an error. (thanks to Zach
263 * bug fix: DEFGENERIC now works even when there's a function of the
264 same name in an enclosing lexical environment. (thanks to Zach
266 * fixed compiler failure, caused by instrumenting code during
267 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
268 * optimization: added loop analysis and improved register allocation
269 * fixed some bugs revealed by Paul Dietz' test suite:
270 ** POSITION on displaced vectors with non-zero displacement
271 returns the right answer.
272 ** (SIMPLE-STRING) is a valid type specifier for sequence
274 ** *PRINT-LEVEL* handling for slotless structures is pedantically
276 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
277 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
279 ** The justification version of the FORMAT ~< directive treats
280 non-zero minpad parameter correctly.
282 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
283 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
284 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
285 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
286 the supported interface.
287 * new feature: Single-stepping of code compiled with DEBUG 2 or
288 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
289 * new feature: saving cores with foreign code loaded is now
290 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
291 Timothy Moore's work for CMUCL)
292 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
293 parameters to * instead of NIL if no initform is supplied.
294 (reported by Johan Bockgaard)
295 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
296 correctly when FUNCALL appears as the car of the form. Note:
297 despite this FUNCALL forms are not currently subject to
298 compiler-macro expansion. (port of Raymond Toy's fix for the
299 same from CMUCL, reported by Johan Bockgaard)
300 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
301 (thanks for Teemu Kalvas)
302 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
303 case instead of terminating the iteration (thanks for Julian
304 Squires, Sean Champ and Raymond Toy)
305 * bug fix: incorrect expansion of defgeneric that caused a style
306 warning. (thanks for Zach Beane)
307 * bug fix: slot accessor effective method computation works properly
308 for classes with multiple non-standard applicable methods on
309 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
310 * on x86 compiler supports stack allocation of results of LIST and
311 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
312 CMUCL implementation by Gerd Moellmann)
313 * fixed some bugs revealed by Paul Dietz' test suite:
314 ** FORMAT strings with both the ~<~:;~> form of the justification
315 directive and pretty-printing directives cause an error.
317 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
318 * incompatible change: the internal functions
319 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
320 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
321 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
322 instead of the old functions.
323 * new feature: on platforms where "dladdr" is available foreign
324 function names now appear in backtraces. (based on Helmut Eller's
326 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
327 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
329 * documentation: ASDF manual is now distributed with SBCL.
330 * bug fix: SBCL can now load its contributed modules with REQUIRE
331 even if the system-provided entries have been removed from
332 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
334 * bug fix: backtraces involving undefined functions or assembly
335 routines are more informative. (thanks to Brian Downing)
336 * bug fix: mutually referent alien structures now work correctly.
337 (reported by Rick Taube)
338 * bug fix: structures defined by WITH-ALIEN can be referred to
339 within other definitions in the same WITH-ALIEN.
340 * bug fix: division operators (MOD, TRUNCATE and the like) with
341 constant zero divisors and integer dividends no longer generate
343 * bug fix: provide default methods for INPUT-STREAM-P and
344 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
345 * bug fix: improve the detection and resolution of MOP metacycles
346 (where effective method computation depends on the generic function
347 having its effective method computed). (reported by Bruno Haible)
348 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
349 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
350 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
351 * bug fix: pathnames with relative directory components can be
352 represented relative to default pathnames.
353 * optimization: in taking the GCD of bignums, reduce the two bignums
354 to approximately the same size (using Euclid's algorithm) before
355 applying the more sophisticated binary GCD. (thanks to Juho
357 * optimization: COUNT on bitvectors now operates word-at-a-time.
358 * optimization: ASH with a positive, but not necessarily constant,
359 (leftwards) shift, when in a modular context, is compiled to a
361 * fixed some bugs revealed by Paul Dietz' test suite:
362 ** FORMAT variable parameters ("~V<char>") are defaulted properly
363 if the corresponding argument is NIL.
364 ** FORMAT directives accepting a minpad parameter treat negative
366 ** literal commas as character parameters to FORMAT directives are
368 ** literal spaces directly after ~<Newline> directives within a
369 format-logical-block (~:< ~@:>) do not induce :FILL-style
370 conditional newlines.
371 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
372 as specified: it no longer includes conditional newlines).
373 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
375 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
376 :PER-LINE-PREFIX argument does not evaluate to a string.
378 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
379 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
380 the manual for details; for now, package locks can be disabled by
381 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
382 if no major problems are found then it is likely that they will be
383 compiled in unconditionally.
384 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
385 unsupported operators on all platforms. To load a shared library
386 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
387 object file, link it into a shared library outside of SBCL and
388 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
390 * bug workaround: ROOM T and the GC were not working together
391 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
392 reporting facility aren't preserved. That reporting has been
393 disabled, so now until and unless someone figures out how to make it
394 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
395 * minor incompatible change: as threatened around sbcl-0.8.0, the
396 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
397 The symbols are also exported from SB-PCL for backwards
398 compatibility, but more so than before SB-PCL should be treated as
399 an implementation-internal package.
400 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
402 * the SB-SPROF contrib now works on (most) non-x86 architectures.
403 It is known as of this release not to work on the Alpha, however.
404 * fixed bug #167: errors signalled due to illegal syntax in method
405 bodies are now more legible.
406 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
407 designators and can hence be used with TYPEP.
408 * fixed bug #333: CHECK-TYPE now ensures that the type error
409 signalled, if any, has the right object to be accessed by
410 TYPE-ERROR-DATUM. (reported by Tony Martinez)
411 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
412 5.1.2.3 for multiple-value place subforms. (reported by Kalle
414 * fixed bug #334: programmatic addition of slots using specialized
415 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
416 and :ALLOCATION :CLASS slots.
417 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
418 float exponents. (rereported by Peter Seibel)
419 * fixed a bug: #\Space (and other whitespace characters) are no
420 longer considered to be macro characters in standard syntax by
422 * fixed bug: initialization of condition class metaobjects no longer
423 causes an instance of the condition to be created. (reported by
425 * fixed bug: it is now possible to have slots such that
426 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
427 description is neither :INSTANCE nor :CLASS.
428 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
429 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
430 SLOT-VALUE-USING-CLASS), and no longer causes errors with
431 non-standard SLOT-DEFINITION-ALLOCATION values.
432 * fixed bugs: various race conditions handling exiting threads.
433 CL-PPCRE's thread torture test now passes.
434 * fixed arguably-a-bug: GC time proportional to number of threads in
435 system even when most of them are idle
436 * optimization: improved performance of BIT and SBIT on bit-vectors.
437 * .fasl file incompatibility: The fasl file version number has
438 been incremented because of changes associated with package locks.
440 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
441 * minor incompatible change: the system no longer provides
442 optimization hints (conditions which are TYPEP
443 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
444 mode; these hints are still emitted when higher SPEED optimization
446 * new contrib module: a sampling profiler (profiling by statistical
447 sampling, rather than by instrumenting functions) is available as
448 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
449 * the behaviour of the standard function ED is now customizeable by
450 third parties through a hook variable: see ED's documentation
451 string for information on the protocol.
452 * the compiler no longer emits efficiency notes for (FUNCALL X)
453 when the type of X is uncertain under default optimization
455 * fixed bug 276: mutating a binding of a specialized parameter to a
456 method to something that is not TYPEP the specializer is now
458 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
459 exhibit the required behaviour on the x86 platform. (thanks to
460 Peter van Eynde, Eric Marsden and Bruno Haible)
461 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
462 even for difficult arguments. (reported by Peter Graves)
463 * fixed bug 141a: the backquote printer now descends quoted
465 * fixed another bug in backquote printing: no more destructive
466 modification of the form's list structure. (reported by Brian
468 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
469 move between its address being taken and the call to
470 interrupt_thread, fixing a crashing race condition.
471 * the SB-POSIX contrib implementation has been adjusted so that it
472 no longer exhibits ridiculously poor performance when constructing
473 instances corresponding to C structs.
475 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
476 * minor incompatible change: the sb-grovel contrib now treats C
477 structures as alien (in the sense of SB-ALIEN) objects rather than
478 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
479 has implications for memory management of client code
480 (sb-grovel-returned objects must now be manually managed) and for
481 type safety (alien objects now have full types).
482 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
483 used to control emission of compiler diagnostics, rather than the
484 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
485 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
486 quality should be considered deprecated.
487 * install.sh now installs the user manual as well
488 * (not quite a new documentable feature, but worth considering in
489 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
490 beginnings of a semantically meaningful condition hierarchy is
491 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
493 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
494 designator as the defaults argument.
495 * fixed bug: Displaced arrays whose displaced-to array has become
496 too small now cause ARRAY-DIMENSION to signal an error, providing
497 for safer bounds-checking. (reported by Bruno Haible)
498 * fixed bug: DEFCLASS slot definitions with identical :READER and
499 :WRITER names now signal a reasonable error. (reported by Thomas
501 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
503 * fixed bug: Invalid slot specification errors now print correctly.
504 (thanks to Zach Beane)
505 * fixed bug 320: Shared to local slot value transfers in class
506 redefinitions now happen corrently from superclasses as
507 well. (reported by Bruno Haible)
508 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
510 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
511 improper lists correctly. (reported by Bruno Haible)
512 * fixed bug 313: source-transform for <fun-name> was erroneously
513 applied to a call of a value of a variable with name <fun-name>.
514 (reported by Antonio Menezes Leitao)
515 * fixed bug 307: The obsolete instance protocol ensures that
516 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
517 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
518 remove dead unknown-values globs from the middle of the stack.
519 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
520 easier to resume long computations after using *BREAK-ON-SIGNALS*
521 to diagnose and fix failures. (thanks to Nikodemus Siivola)
522 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
523 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
524 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
525 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
526 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
528 * fixed a (fairly theoretical) bug in string printing: if
529 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
530 does not have array-element-type equal to the most general string
532 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
533 function name. (thanks to Nikodemus Siivola)
534 * fixed bug: compile-time format string checker failed on
535 non-closed ~{. (reported by Thomas F Burdick)
536 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
537 don't warn on legal loop code involving more than one aggregate
538 boolean. (thanks to Nikodemus Siivola)
539 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
540 in keyword position to MAKE-PACKAGE was not being reported as
542 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
543 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
545 * fixed bugs 280 and 312: the checking for multiple definitions in a
546 file is less likely to become confused by uses of inline
548 * fixed bug: the #S reader macro performs the keyword coercion
549 specified for slot names. (reported by Kalle Niemitalo)
550 * fixed bug: lambda lists may contain symbols whose names start with
551 & but are not lambda-list-keywords; their occurrence triggers a
553 * fixed bug 321: define-method-combination argument lambda lists do
554 not cause invalid code to be generated when &optional or &aux
555 variables have default values. (reported by Bruno Haible)
556 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
557 have CLOS classes; this appears also to have fixed reported
558 instability in stack exhaustion detection.
559 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
560 MUFFLE-WARNING when no associated restart is present is now
562 * optimization: rearranged the expansion of various defining macros
563 so that each expands into only one top-level form in a
564 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
566 * optimization: used a previously little-used slot in symbols to
567 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
569 * fixed some bugs revealed by Paul Dietz' test suite:
570 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
572 ** Fixed an optimization bug related to inheritance of initforms
573 from local to shared slots.
574 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
575 ** CAST optimizer forgot to flush argument derived type.
576 ** print/read consistency on floats is now orders of magnitude
577 more likely. (thanks also to Bruno Haible for a similar report
579 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
580 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
581 for objects of type REAL. Make it so.
582 ** SXHASH is specified (infelicitously) to respect similarity,
583 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
584 it so. (thanks to Markus Ziegler)
585 ** on the Alpha, the compiler succeeds in compiling functions
586 returning a known number of arguments greater than 63.
587 ** fixed handling of invalid NIL arguments in keyword position
589 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
591 ** provided a readably-printable representation for RANDOM-STATE
593 ** ensured that pathnames, where they have a namestring, always
594 print using #P"..." syntax.
596 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
597 * Support for the forthcoming 2.0 version of the NetBSD kernel
598 running on x86 hardware has been added. (thanks to Perry
599 E. Metzger most immediately, and others for their past work)
600 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
601 current release version; SBCL's OpenBSD support had been broken
602 since about the time of OpenBSD's switch to ELF binary format.)
603 * [placeholder for DX summary]
604 ** user code with &REST lists declared dynamic-extent, under high
605 speed or space and low safety and debug optimization policy.
606 * The manual has been converted to Texinfo format and the debugger
607 chapter from the cmucl manual has been added.
608 * A facility has been added to extract documentation strings from
609 sbcl and store them as Texinfo-formatted snippets for inclusion in
610 the manual (via Texinfo's @include directive)
611 * bug fix: compiler emitted division in optimized DEREF. (thanks for
612 the test case to Dave Roberts)
613 * bug fix: multidimensional simple arrays loaded from FASLs had fill
614 pointers. (reported by Sean Ross)
615 * bug fix: PROFILE output is printed nicely even for large numerical
616 values. (thanks to Zach Beane)
617 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
618 greater than 32 handle EOF correctly.
619 * bug fix: on X86 an immediate argument of the IMUL instruction is
620 correctly printed in disassembly. (thanks to Lutz Euler)
621 * bug fix: class slots in redefined classes preserve their old
622 values. (thanks to Bruno Haible and Nikodemus Siivola)
623 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
624 succeeds. (reported by Marco Baringer)
625 * fixed some bugs revealed by Paul Dietz' test suite:
626 ** READ-SEQUENCE now works on ECHO-STREAMs.
627 ** RATIONALIZE works more according to its specification. (thanks
630 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
631 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
632 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
633 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
634 should matter to you only if you rebind the printer control
635 variables and then find you want different bindings in the
636 debugger than in the ordinary execution of your program.)
637 * The runtime build system has been tweaked to support building
638 (on SPARC/SunOS) using a C compiler which invokes Sun's own
639 assembler and linker. (thanks to Nikodemus Siivola)
640 * Unbound, undefined, undeclared variables now trigger full
641 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
642 more often programmer error than deliberate exploitation of undefined
644 * optimization: the hash algorithm for strings has changed to one
645 that is less vulnerable to spurious collisions. (thanks to Juho
647 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
648 needless bounds checking. (thanks to Juho Snellman)
649 * optimization: implemented multiplication as a modular
650 (UNSIGNED-BYTE 32) operation on the PPC backend.
651 * fixed some bugs revealed by Paul Dietz' test suite:
652 ** ADJUST-ARRAY now copies the datum in a zero rank array if
654 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
655 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
656 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
658 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
660 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
662 ** ECHO-STREAMs no longer attempt to echo the end of file value to
663 their output stream on EOF from read.
664 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
665 have been read to end-of-file.
666 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
668 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
669 description of determination of which consecutive characters
671 ** Printing the "Space" character with escaping on now yields "#\\ ",
672 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
673 ** Reading floating-point numbers with *READ-BASE* set to a number
674 less than 10 works correctly.
675 ** Reading floating-point numbers with *READ-BASE* set to a number
676 more than 10 works correctly.
677 ** Printing with *PRINT-READABLY* targets the standard readtable, not
678 the readtable currently in effect.
680 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
681 * minor incompatible change: parsing of namestrings on a physical
682 (Unix) host has changed; numbers after the final #\. in a
683 namestring are no longer interpreted as a version field. This is
684 intented to be largely invisible to the user, except that the
685 meaning of the namestring "*.*.*" has changed: it now refers to a
686 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
687 should usually be replaced by
688 (make-pathname :name :wild :type :wild :version :wild)
689 with the added benefit that this is more likely to be portable.
690 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
691 option to OPEN now signals an error if the file being opened
692 exists; this may have an impact on existing code.
693 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
694 signals received, on the PowerPC platforms (both Linux and
695 Darwin). (thanks to Pierre Mai for pointing out the location of
697 * several fixes on OS X: The system now builds and runs cleanly on
698 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
699 (thanks to Brian Mastenbrook)
700 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
701 second-resolution: integer multiples of 1/3600 between -24 and 24.
702 (thanks to Vincent Arkesteijn)
703 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
704 type when called with 1 argument; PEEK-CHAR checked type of
705 PEEK-TYPE only after having read first character from a
706 stream. (reported by Peter Graves)
707 * bug fix: the garbage collector now has much better locality
708 behaviour, and in particular no longer treats all memory as being
709 exhausted when many small objects point to each other in a deeply
711 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
712 recognized as being TYPEP their class.
713 * bug fix: the PUSHNEW documentation string has been corrected.
714 (thanks to Vincent Arkesteijn)
715 * bug fix: defaulting of the value for the last of an atypically
716 large number of multiple values being bound was not being
717 performed correctly on the Alpha or PPC platforms
718 * optimization: implemented multiplication as a modular
719 (UNSIGNED-BYTE 32) operation on the x86 backend.
720 * optimization: SEARCH on simple-base-strings can now be open-coded.
721 (see also contrib/compiler-extras.lisp for inspiration for
722 teaching the compiler about the Boyer-Moore algorithm).
723 * value, returned by MAX (and MIN) called with several EQUALP, but
724 not EQL, arguments now does not depend on compiler settings.
725 * fixed some bugs revealed by Paul Dietz' test suite:
726 ** in stack analysis liveness information is propagated from
727 non-local entry points.
728 ** pathwise CAST removing failed when the CAST node did not start
730 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
731 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
733 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
734 host is already defined.
735 ** RENAME-FILE works on streams instead of signalling an internal
737 ** PEEK-CHAR uses the current readtable when determining whether
738 or not a character is whitespace.
739 ** MERGE-PATHNAMES handles the case when the pathname does not
740 specify a name while the default-pathname specifies a version.
741 ** Pathnames now stand a better chance of respecting print/read
743 ** Attempting to use standardized file system operators with a
744 pathname with invalid :DIRECTORY components signals a
746 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
748 ** DIRECTORY on logical pathnames is more correct.
749 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
750 signal a TYPE-ERROR if their argument is not a stream.
751 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
752 designator argument does not designate a stream.
753 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
754 examining the synonym.
755 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
757 ** OPEN and WITH-OPEN-STREAM allow opening streams with
758 element-type larger than ([UN]SIGNED-BYTE 32).
760 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
761 * When built with the :SB-FUTEX feature, threaded builds now take
762 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
763 for faster/more reliable mutex and condition variable support.
764 * Incompatible change (but one you probably shouldn't have been using
765 anyway): the interface and code for arbitrating between multiple
766 threads in the same user session has been redesigned.
767 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
768 error when called without an explicit environment argument.
769 (thanks to Nikodemus Siivola)
770 * bug fix: buffered :DIRECTION :IO streams are less likely to become
771 confused about their position. (thanks to Adam Warner and Gerd
773 * bug fix: Pretty printing backquoted forms with unquotations in the
774 argument list position of various code constructs such as LAMBDA
775 now works correctly. (reported by Paul Dietz)
776 * bug fix: Pretty printing unquotations no longer loses all
777 stream position information.
778 * optimization: performance of string output streams is now less
779 poor for multiple small sequence writes.
780 * optimization: performance of CSUBTYPEP in the presence of complex
781 expressions involving CONS and NOT many times has been improved.
782 (reported by Paul Dietz)
783 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
785 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
786 streams for the REPL, for the debugger, and so on. (thanks to
788 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
789 * fixed some bugs revealed by Paul Dietz' test suite:
790 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
791 is ignored if it is too small, rather than propagating through
792 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
793 ** extremely complex negations of CONS types were not being
794 sufficiently canonicalized, leading to inconsistencies in
796 ** VALUES tranformer lost derived type.
798 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
799 * fixed a bootstrapping bug: the build process no longer assumes
800 that the various BOOLE-related constants have the same value in
801 host and target lisps. (noted by Paul Dietz' test suite on an
802 SBCL binary built from CLISP)
803 * The system can now be dynamically linked on the MIPS platform,
804 which enables dynamic loading of foreign code from Lisp. (thanks
805 to Ralf Baechle for discussions on the MIPS ABI)
806 * The system now records debugging information for its own source
807 files in a filesystem-position-independent manner, relative to
808 the "SYS" logical host.
809 * fixed a compiler bug: MV-LET convertion did not check references
810 to the "max args" entry point. (reported by Brian Downing)
811 * tweaked disassembly notes to be less confident about proclaiming
812 some instruction as an LRA. (thanks to Brian Downing)
813 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
814 now each have their own history, command character, and other
815 characteristics. (thanks to David Lichteblau)
816 * fixed some bugs revealed by Paul Dietz' test suite:
817 ** compiler failure in compiling LOGAND expressions including a
819 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
820 shift greater than 32.
821 ** FLUSH-DEST did not mark blocks for type check regeneration.
822 ** HANDLER-CASE failed to accept declarations in handler clauses
823 in some circumstances.
825 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
826 * New code in contrib/sb-introspect (still probably not entirely
827 stable yet) provides some support for smart Lisp development
828 environments like SLIME.
829 * The conditions signalled for errors occurring when loading .fasl
830 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
831 in a way which should help ASDF recover gracefully.
832 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
833 stuff has been cleaned up. If you code contrib/ stuff, this might
834 affect you, and you can look at contrib/README, contrib/STANDARDS,
835 and/or the 0.8.4.27 diff to check.
836 * In full calls the compiler now does not generate checks for declared
837 argument types for all arguments.
838 * various threading fixes
839 ** and some experimental patches which didn't make it into
840 the main tree for this release, but which are shipped in
841 contrib/experimental-thread.patch as a possible fix for some
842 failures (deadlock, spinning...) in GC-intensive multithreaded
844 * fixed PPC build problem (source code incompatibility of different
845 library versions): added offsetof() hackery which attempts to divine
846 where glibc maintainers put uc_mcontext today
847 * fixed bug 282: compiler does not trust type assertions while passing
848 arguments to a full call.
849 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
850 &OPTIONAL VALUES type parameter.
851 * fix bug 214: algorithm for noting rejected templates is now more
852 similar to that of template seletion. (also reported by rydis on
854 * fixed bug 141b: printing backquoted information readably and prettily
855 inserts a space where necessary.
856 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
857 (CHARACTER 10) are now reported as errors, rather than propagated
858 as unknown types. (reported by piso on #lisp)
859 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
860 with values NIL and :ERROR. (thanks to Milan Zamazal)
861 * fixed bug 191c: CLOS now does proper keyword argument checking as
862 described in CLHS 7.6.5 and 7.6.5.1.
863 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
864 longer raise an error; further, using a list as a for-as-arithmetic
865 counter now raises a meaningful error.
866 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
867 now understood by sequence creation functions such as MAKE-SEQUENCE
869 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
870 a STRING-INPUT-STREAM.
871 * compiler enhancement: SIGNUM is now better able to derive the type
873 * type declarations inside WITH-SLOTS are checked. (reported by
875 * fixed some bugs revealed by Paul Dietz' test suite:
876 ** incorrect optimization of TRUNCATE for a positive first
877 argument and negative second.
878 ** compiler failure in let-convertion during flushing dead code.
879 ** compiler failure while deriving type of TRUNCATE on an
880 interval, containing 0.
881 ** ASH of a negative bignum by a negative bignum count now returns
883 ** intersection of CONS types now canonicalizes properly, fixing
884 inconsistencies in SUBTYPEP.
886 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
887 * incompatible change: The --disable-debugger command line
888 option now clobbers the debugger at a more fundamental
889 level, by redefining #'INVOKE-DEBUGGER instead of by
890 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
891 is specified by ANSI to ignore *DEBUGGER-HOOK* and
892 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
893 enter the debugger REPL and then suffer recursive errors
894 because *DEBUG-IO* is also messed up in --disable-debugger mode;
895 while under the new system, BREAK in --disable-debugger mode
896 terminates the system just as an unhandled error would.
897 * fixed compiler performance when processing loops with a step >1;
898 * bug fix: DOCUMENTATION now retrieves generic function
899 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
900 support has been systematized, and now supports the methods
901 specified by ANSI, along with a default method and a method for
902 slot documentation. (reported by Nathan Froyd)
903 * bug fix: effective methods associated with a generic function are
904 no longer cached over a change of that generic function's method
905 combination. (reported by Andreas Fuchs)
906 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
907 names a non-existent directory. (thanks to Andreas Fuchs)
908 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
909 the PPC platform, lead to stack corruption; this has been fixed.
910 (reported by Rainer Joswig)
911 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
912 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
913 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
914 arithmetic in general, is now much more reliable on the PPC
916 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
917 platform now returns the right answer.
918 * optimization: restored some effective method precomputation in
919 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
920 precomputation is now tunable.
921 * optimization: compiler-internal data structure use has been
922 reviewed, and changes have been made that should improve the
923 performance of the compiler by about 20%.
924 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
925 simple-base-strings and simple-bit-vectors is improved.
926 * optimization: the optimization of 32-bit logical and arithmetic
927 functions introduced in version 0.8.3 on the x86 has been
928 implemented on the mips, ppc and sparc platforms; an
929 implementation of the same facility, but for 64-bit arithmetic,
930 has been added for the alpha.
931 * microoptimization: the compiler is better able to make use of the
932 x86 LEA instruction for multiplication by constants.
933 * bug fix: in some situations compiler did not report usage of
934 generic arithmetic in (SPEED 3) policy.
935 * bug 145b fix: compiler used wrong type specifier while converting
936 MEMBER-types to numeric.
937 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
939 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
940 index argument now works properly on the Alpha platform.
941 * bug fix: floating point exception treatment on the Alpha platform
943 * bug fix: FILE-POSITION works much better on string input and
944 output streams. (thanks to Nikodemus Siivola)
945 * bug fix: many threading/garbage collection symptoms sorted.
946 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
947 might be pseudo-atomic.
948 * internal change: Stopping for GC is now done with signals not
949 ptrace. GC is now done in whichever thread wanted it, instead of
951 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
953 * bug fix: result form in DO is not contained in the implicit
955 * incompatible change: ICR structure is changed; the value part of
956 CONTINUATION is now called LVAR; corresponding functions are
957 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
958 * added type deriver for ISQRT (thanks to Robert E. Brown).
959 * bug fix: better support for loading from the command line when an
960 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
962 * fixed some bugs revealed by Paul Dietz' test suite:
963 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
964 ** ROUND and FROUND now give the right answer when given very
965 small float arguments.
966 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
968 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
969 ** (GCD 0 <negative-integer>) returned <negative-integer>.
970 ** LCM should return a non-negative integer.
971 ** PARSE-INTEGER returned the index of a terminator instead of the
972 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
973 ** PARSE-INTEGER returned an incorrect index being applied to a
975 ** LCM with two arguments of 0 returns 0 rather than signalling
977 ** unsigned addition of a 32-bit constant with the high bit set no
978 longer causes an internal compiler error on the x86.
979 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
980 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
981 with negative last argument.
982 ** byte specifiers with zero size and position no longer cause
983 an error during type derivation.
984 ** bignum multiplication on the Alpha platform now returns the
986 * porting: The system now builds on SuSE AMD64, although it still
987 generates a 32-bit binary.
988 * .fasl file incompatibility: The fasl file version number has
989 been incremented (because of the changes to internal compiler
990 data structures referred to above).
992 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
993 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
994 more accurately, on the Darwin kernel running on PowerPC hardware.
995 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
996 * Compiler code deletion notes now signal a condition of type
997 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
998 an associated MUFFLE-WARNING restart.
999 * The compiler now performs limited argument count validation of
1000 constant format strings in FORMAT, and where appropriate in ERROR,
1001 CERROR and WARN. (thanks to Gerd Moellmann)
1002 * New ASDF-INSTALL contrib can be used for automatic download and
1003 installation of third-party Lisp code from CCLAN or other sites
1005 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1006 which forces another thread to execute a function supplied by the
1008 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1009 now accept and act upon their :ELEMENT-TYPE keyword argument.
1010 (reported by Martin Atzmueller, Edi Weitz)
1011 * bug fix: FILE-POSITION now accepts position designators up to
1012 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1013 is the greater. (thanks to Patrik Nordebo)
1014 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1015 Kalle Olavi Niemitalo)
1016 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1017 printable. (reported by Eric Marsden)
1018 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1019 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1020 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1021 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1022 (thanks to Stig Sandoe)
1023 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1024 UNREAD-CHAR now correctly includes the unread character in the
1025 target sequence. (thanks to Gerd Moellmann)
1026 * bug fix in threaded builds: the system can now be suspended and
1027 resumed by shell job control with minimal disruption.
1028 * bug fixes in times and timezones >2038AD
1029 * better handling of "where is GNU make?" problem in build scripts
1030 (thanks to Nikodemus Siivola)
1031 * new optimization: inside a named function any reference to a
1032 function with the same name is considered to be a self-reference;
1033 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1034 optimization quality.
1035 * new optimization on x86: logical functions and + now have
1036 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1037 used when the result is truncated to 32 bits.
1038 * VALUES declaration is partially enabled.
1039 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1040 * fixed some bugs revealed by Paul Dietz' test suite:
1041 ** The system now obeys the constraint imposed by
1042 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1043 types form a lattice under type intersection.
1044 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1045 ** ASSOC now ignores NIL elements in an alist.
1046 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1047 and (1+ MOST-POSITIVE-FIXNUM) answers.
1048 ** The addition of a method with invalid qualifiers to a generic
1049 function does not cause an error to be signalled immediately;
1050 a warning is signalled, and the error is generated only on
1051 calling the generic function.
1052 * changed .fasl file version number, in part to add type codes for
1053 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1054 obscure ANSI requirements
1056 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1057 * fixed bug 148: failure to inline-expand a local function left
1058 garbage, confusing the compiler.
1059 * fixed bugs 3cd: structure slot readers perform type check if the
1060 slot can have an invalid value (i.e. it is either not initialized
1061 or can be written with a less specific slot writer).
1062 * bug fix: the compiler now traps array references to elements off
1063 the end of an array; previously, the bounds checking in some
1064 circumstances could go off-by-one.
1065 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1067 * type declarations for array element types now obey the description
1068 on the CLHS page "Declaration TYPE", as per discussions on
1069 sbcl-help around 2003-05-08. This means that a declaration
1070 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1071 declaration, all references to BAR will be asserted or assumed
1072 (with THE, so dependent on compiler policy) to involve objects of
1073 type FOO. Note that no such declaration is implied in
1074 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1075 * declared types of functions from the "Conditions"
1076 chapter. (reported by Paul Dietz)
1077 * bug fix: CERROR accepts a function as its first argument.
1078 * bug fix: NTH an NTHCDR accept a bignum as index
1079 arguments. (reported by Adam Warner)
1080 * optimization: character compare routines now optimize comparing
1081 against a constant character. (reported by Gilbert Baumann)
1082 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1083 argument now works properly on the MIPS platform.
1084 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1085 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1086 declared to have a functional type, does not cause an error
1088 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1089 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1090 * fixed bug: if last continuation of a deleted block has a
1091 destination, this destination should be deleted too. (reported by
1093 * fixed a bug in the bootstrap process: the host compiler's values
1094 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1095 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1096 test case from Patrik Nordebo)
1097 * improved the ability of the disassembler on the PPC platform to
1098 provide helpful disassembly notes.
1099 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1100 the class in more cases than previously.
1101 * bug fix: FILE-POSITION now understands :START and :END for
1102 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1103 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1104 8) data. (thanks to David Lichteblau)
1105 * bug fix: it is possible to add a method to a generic function
1106 without lambda list.
1107 * bug fix: reader failed to signal END-OF-FILE inside an
1108 object representation. (reported by Nikodemus Siivola)
1109 * fixed some bugs revealed by Paul Dietz' test suite:
1110 ** LAST and [N]BUTLAST should accept a bignum.
1111 ** condition slot accessors are methods.
1112 ** (VECTOR NIL) is a subtype of STRING.
1114 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1115 * minor incompatible change: some nonsensical specialized lambda
1116 lists (used in DEFMETHOD) which were previously ignored now signal
1118 * minor incompatible change: the system is now aware of the types of
1119 variables in the COMMON-LISP package, and will signal errors for
1120 most violations of these type constraints (where previously they
1121 were silently accepted).
1122 * minor incompatible change: COMPILE-FILE now uses the freedom
1123 afforded (ANSI 3.2.2.3) to use derived function types for
1124 functions defined in the same file. This also permits the system
1125 to warn on static type mismatches and function
1126 redefinition. (Currently it does not work with high DEBUG level.)
1127 * minor incompatible change: VALUES declaration is disabled.
1128 * When issuing notes, the compiler now signals a condition of type
1129 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1130 restart for use in user handlers. It is expected that the
1131 COMPILER-NOTE condition will eventually become a condition
1132 supertype to a hierarchy of note types, which will then be
1133 handleable in a similar fashion. However, at the moment, no such
1134 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1135 but it's an implementation detail, not a classification for the
1137 * Changes in type checking closed the following bugs:
1138 ** type checking of unused values (192b, 194d, 203);
1139 ** template selection based on unsafe type assertions (192c, 236);
1140 ** type checking in branches (194bc).
1141 * A short form of VALUES type specifier has ANSI meaning (it has
1142 increased the number of situations when SBCL cannot perform type
1144 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1145 keywords or constants is permissible.
1146 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1147 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1148 classes. (thanks to Antonio Martinez)
1149 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1150 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1151 argument to a pretty-print stream. Instead, it leaves any such
1152 operations to DESCRIBE-OBJECT methods.
1153 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1155 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1156 lambda lists. (thanks to David Lichteblau)
1157 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1158 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1159 (thanks to Gerd Moellmann)
1160 * fixed bug 246: increased compilation speed of long
1161 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1163 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1165 * DEFCONSTANT now throws a condition of type
1166 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1167 constant to a non-EQL value; CONTINUE and ABORT restarts
1168 respectively change and preserve the value.
1169 * fixed bug 63: The code walker, part of the implementation of CLOS,
1170 is now better at handling symbol macros.
1171 * bug fix: There is no longer an internal implementation type named
1172 CL:LENGTH. (reported by Raymond Toy)
1173 * bug fix: In macro-like defining macros/special operators the
1174 implicit block does not enclose the lambda list.
1175 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1176 enclosed in parentheses) are not suitable as type specifiers, and
1177 their use properly signals an error now.
1178 * bug fix: An argument count mismatch for a type specifier in code
1179 being compiled no longer causes an unhandled error at compile
1180 time, but signals a compile-time warning.
1181 * fixed simple vector readable printing
1182 * bug fix: DESCRIBE takes more care over whether the class
1183 precedence list slot of a class is bound before accessing it.
1184 (reported by Markus Krummenacker)
1185 * bug fix: FORMATTER can successfully compile pretty-printer format
1186 strings which use variants of the ~* directive inside.
1187 * bug fix: SEARCH now applies its TEST predicate to the elements of
1188 the arguments in the correct order. (thanks to Wolfhard Buss)
1189 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1190 in type inference. (thanks to Robert E. Brown)
1191 * bug fix: Reading in symbols with an explicit package name of ""
1192 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1193 * fixed some bugs revealed by Paul Dietz' test suite:
1194 ** NIL is now allowed as a structure slot name.
1195 ** Arbitrary numbers, not just REALs, are allowed in certain
1196 circumstances in LOOP for-as-arithmetic clauses.
1197 ** Multiple class redefinitions before slot access no longer
1198 causes a type error.
1199 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1200 association between the name and a class.
1201 ** Generic functions with non-standard method-combination and over
1202 five methods all of which return constants no longer return NIL
1203 after the first few invocations. (thanks to Gerd Moellmann)
1204 ** CALL-NEXT-METHOD with no arguments now passes the original
1205 values of the arguments, even in the presence of assignment.
1206 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1208 ** DISASSEMBLE works with closures and funcallable instances.
1209 ** ADD-METHOD now returns the generic function, not the new method.
1210 ** FIND-METHOD signals an error if the lengths of the specializers
1211 is incompatible with the generic function, even if the ERRORP
1213 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1214 which its argument is a member.
1215 ** DEFCLASS only redefines the class named by its class-name
1216 argument if that name is the proper name of the class;
1217 otherwise, it creates a new class.
1218 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1219 of the UNBOUND-SLOT condition to the name of the slot.
1220 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1221 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1222 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1223 SLOT-MAKUNBOUND in the specified fashion.
1225 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1226 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1227 cross-compilation host. As a consequence, we can now bootstrap our
1228 way up to SBCL starting with a bare gcc toolchain and human-readable
1229 source code (first the source to CLISP, then the source to SBCL).
1230 * A contributed module containing a partial implementation of the
1231 simple-streams interface has been included. (thanks to Rudi
1233 * A contributed module implementing the RFC1321 Message Digest
1234 Algorithm, known as MD5, has been included.
1235 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1236 no longer has any effect, as the code controlled by this feature
1237 has been deleted. (As far as we know, no-one has ever built using
1238 this feature, and its semantics were confused in any case).
1239 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1240 work on conditions (as required by the ANSI specification),
1241 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1242 expected behaviour on conditions. Users should note, however,
1243 that such behaviour is not required by the ANSI specification,
1244 and so use of this behaviour may render their code unportable.
1245 * fixed some bugs revealed by Paul Dietz' test suite:
1246 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1248 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1249 ** SXHASH on condition objects no longer returns NIL.
1250 ** :ALLOCATION :CLASS slots are better treated; their values are
1251 updated on class redefinition, and initforms inherited from
1252 superclasses are applied.
1253 ** REMOVE-METHOD returns its generic function argument even when
1254 no method was removed.
1255 ** SHARED-INITIALIZE now initializes the values of the requested
1256 slots, including those with :ALLOCATION :CLASS.
1257 ** ALLOCATE-INSTANCE now works on structure classes defined via
1258 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1260 ** SLOT-EXISTS-P now works on conditions, as well as structures
1262 ** MAKE-LOAD-FORM now has the required methods on
1263 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1264 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1265 keyword, and now implements the SLOT-NAMES argument.
1266 ** methods with &OPTIONAL arguments no longer allow too many
1267 arguments to be passed in the call without error.
1268 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1269 option is consistent with the required arguments of the generic
1270 function lambda list.
1271 * bug fix: REQUIRE accepts a string designator. (Thanks to
1273 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1274 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1275 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1276 * bug fix: The compiler now checks for duplicated variables in macro
1278 * bug fix: SETQ on globals returns the correct value.
1279 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1280 not a COMPILER-ERROR (followed by some other strange error on
1281 choosing the CONTINUE restart).
1282 * bug fix: make.sh and friends are now more consistent in the way that
1283 they look for GNU "make".
1285 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1286 * experimental native threads support (on x86 Linux >=2.4 only).
1287 This is not compiled in by default: you need to add :SB-THREAD to
1288 the target features. See the "Beyond ANSI" chapter of the manual
1290 * fix for longstanding nonANSIism: The old distinction between
1291 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1292 The return value from CL:FIND-CLASS is now a CLOS class, and
1293 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1294 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1295 * An interface to the MetaObject Protocol, as described in Kiczales,
1296 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1297 MIT Press, 1991, is available from the SB-MOP package.
1298 * incompatible change: the SB-PCL package should now be considered
1299 a private implementation detail, and no longer a semi-private MOP
1301 * minor incompatible change: due to rearrangement for threads, the
1302 control stack and binding stack are now allocated at arbitrary
1303 addresses instead of being hardcoded per-port. Users affected by
1304 this probably have to be doing advanced things with shared
1305 libraries, and will know who they are.
1306 * minor incompatible change: Previously, all --eval forms used were
1307 processed with READ before any of them were processed with EVAL.
1308 Now each --eval form is processed with both READ and EVAL before
1309 the next --eval form is processed. (Thus package operations like
1310 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1311 work as the user might reasonably expect.)
1312 * minor incompatible change: *STANDARD-INPUT* is now only an
1313 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1315 * minor incompatible change: Y-OR-N-P is now character-oriented,
1316 not line oriented. Also, YES-OR-NO-P now works without errors.
1317 (thanks to Antonio Martinez)
1318 * sb-aclrepl module improvements: an integrated inspector, added
1319 repl features, and a bug fix to :trace command.
1320 * Known functions, which cannot be open coded by the backend, are
1321 considered to be able to check types of their arguments. (fixing
1322 a bug report by Nathan J. Froyd)
1323 * fixed a bug in computing method discriminating functions: It is
1324 now possible to define methods specialized on classes which have
1325 forward-referenced superclasses. (thanks to Gerd Moellmann)
1326 * fixed evaluation order in optional entries (reported by Gilbert
1328 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1329 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1330 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1331 required-by-AMOP NAME argument, as well as CLASS and
1332 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1333 * fixed bug 20: DEFMETHOD can define methods using names that are
1334 not the proper names of classes to designate class specializers.
1335 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1336 Unix file descriptors, instead of blowing up. (thanks to Antonio
1338 * Garbage collection refactoring: user-visible change is that a
1339 call to the GC function during WITHOUT-GCING will not do garbage
1340 collection until the end of the WITHOUT-GCING. If you were doing
1341 this you were probably losing anyway.
1342 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1343 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1344 TYPEP the latter but not the former.
1345 * The compiler issues a full WARNING for calls to undefined functions
1346 with names from the CL package.
1347 * MAP-INTO for a vector destination is open coded. (reported by
1348 Brian Downing on c.l.l)
1349 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1350 documentation string.
1351 * fixed some bugs revealed by Paul Dietz' test suite:
1352 ** COPY-ALIST now signals an error if its argument is a dotted
1354 ** Condition slots are now accessed more correctly in the presence
1355 of multiple initargs for a given slot.
1356 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1357 exclude restarts of the same name associated with a different
1359 ** DEFCLASS of forward-referenced classes with another
1360 forward-referenced class in the superclasses list no longer
1362 ** Condition slots are now initialized once each, not multiple
1363 times. (thanks to Gerd Moellmann)
1364 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1366 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1367 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1368 arguments contain duplicated elements.
1369 ** RESTART-CASE understands local macros.
1370 ** RESTART-CASE associates exactly its own restarts with a condition.
1371 ** ENDP in safe mode checks its argument to be of type LIST.
1372 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1373 in question is unbound.
1374 ** Optimized MAKE-INSTANCE functions no longer cause internal
1375 assertion failures in the presence of duplicate initargs.
1376 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1377 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1378 propagated even in the ctor optimized implementation of
1380 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1382 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1383 * a better implementation of SXHASH on (simple) bit vectors,
1384 measured both in execution speed and in distribution of results
1385 over the positive fixnums, has been installed. Likewise, a better
1386 implementation of EQUAL for simple bit vectors is now available.
1387 * fixed CEILING optimization for a divisor of form 2^k.
1388 * fixed bug 240 (emitting extra style warnings "using the lexical
1389 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1390 by Antonio Martinez)
1391 * fixed SXHASH, giving different results for NIL depending on type
1392 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1393 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1394 variable. (found by Rolf Wester)
1395 * SBCL does not ignore type declarations for special
1396 variables. (reported by rif on c.l.l 2003-03-05)
1397 * some bug fixes in contrib/sb-aclrepl/
1398 * fixed some bugs revealed by Paul Dietz' test suite:
1399 ** a bug in the CONS type specifier, whereby the CAR and CDR
1400 types got intertwined, has been fixed;
1401 ** the type system is now able to reason about the interaction
1402 between INTEGER and RATIO types more completely;
1403 ** APPEND, [N]REVERSE and NRECONC check that those their
1404 arguments, which must be proper lists, are really so;
1405 ** An array specialized to be unable to hold elements has been
1406 implemented, as required -- yes, really -- by ANSI;
1407 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1408 on malformed property lists;
1410 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1411 * incompatible packaging change: in line with Unix convention,
1412 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1413 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1414 * REQUIRE and PROVIDE are now optionally capable of doing something
1415 useful. See the documentation string for REQUIRE.
1416 * infrastructure for a managed SBCL contrib system: contributed
1417 modules in this release include:
1418 ** the ASDF system definition facility;
1419 ** an interface to the BSD Sockets API;
1420 ** an ACL-like convenience interface to the repl;
1421 (thanks to Kevin Rosenberg)
1422 ** an implementation of ROTATE-BYTE, with efficient implementation
1424 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1425 gives a closer approximation to the right answer than previously.
1426 (thanks to Raymond Toy)
1427 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1428 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1429 optional environment arguments, as required by ANSI.
1430 * fixed bugs in other functions taking environment objects, allowing
1431 calls with an explicit NIL environment argument to be compiled
1433 * fixed bug 228: primary return values from
1434 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1435 COMPILE or FUNCTION.
1436 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1437 :TYPE will no longer signal errors on innocuous objects.
1438 * fixed bug 231b: SETQ is better at respecting type declarations in
1439 the lexical environment.
1440 * fixed a bug in DEFCLASS: classes named by symbols with no or
1441 unprintable packages can now be defined.
1442 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1443 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1444 * fixed compiler failure related to checking types of functions.
1445 (reported by Robert E. Brown)
1446 * the compiler is now much more consistent in its error-checking
1447 treatment of bounding index arguments to sequence functions: in
1448 (SAFETY 3) code, errors will be signalled in almost all cases if
1449 invalid sequence bounding indices are passed to functions defined
1450 by ANSI to operate on sequences.
1451 * fixed a bug in the build procedure: documentation of SBCL-specific
1452 packages is now preserved and available in the final Lisp image.
1453 * lifted FDEFINITION lookup out of loops in the implementation of
1454 many list operations. (thanks to Robert E. Brown)
1455 * fixed a bug in the reader: the #n# reader macro now works for
1456 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1457 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1458 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1459 E. Brown and Tony Martinez respectively)
1460 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1461 SLOT-BOUNDP: in particular, it is now able to optimize them much
1462 better, and is now not vulnerable to having packages renamed.
1463 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1464 always called when a slot is not present in an instance. (thanks
1466 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1467 longer signal an error on unreading a character following EOF on
1468 the previous constituent stream. (thanks to Tony Martinez)
1469 * fixed some bugs revealed by Paul Dietz' test suite:
1470 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1471 not just nonnegative fixnums;
1472 ** the logical bit-array operators such as BIT-AND now accept an
1473 explicit NIL for their "opt-arg" argument (to indicate a
1474 freshly-consed result bit-array);
1475 ** ELT now signals an error on an invalid sequence index in safe
1477 ** the type system is now cleverer about negations of numeric
1478 types, and consequently understands the BIGNUM and RATIO types
1480 ** the type system is now cleverer about the interaction between
1481 INTEGER and RATIO types: while bugs still remain, many more
1482 cases are accurately computed;
1483 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1484 if it is in the last clause;
1485 ** CONSTANTLY now correctly returns a side-effect-free function in
1487 ** DECLARE is no longer treated as a special-operator; in
1488 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1489 * incremented fasl file version number due to the change in the
1490 DEFSTRUCT-SLOT-DESCRIPTION structure.
1492 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1493 * minor incompatible change: code processed by the "interpreter" or
1494 EVAL now has a compilation optimization policy of (DEBUG 2)
1495 (changed from (DEBUG 1)) to improve debuggability of interactive
1496 development, and to allow the use of the debug RETURN command in
1498 * an experimental implementation of the RETURN command for the
1499 debugger has been included. (thanks to Frederik Kuivinen)
1500 * fixed bug 62: constraints were not propagated into a loop.
1501 * fixed bug in embedded calls of SORT (reported and investigated by
1503 * fixed some bugs revealed by Paul Dietz' test suite:
1504 ** printing and reading of arrays with some dimensions having
1505 length 0 (thanks to Gerd Moellmann);
1506 ** BOA constructor with &AUX argument without a default value does
1507 not cause a type error;
1508 ** CONSTANTP now returns true for all self-evaluating objects.
1510 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1511 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1512 accessors that are related by inheritance, as specified in the
1513 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1514 Valtteri Vuorikoski)
1515 * The compiler is now able to inline functions that were defined in
1516 a complex lexical environment (e.g. inside a MACROLET).
1517 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1519 * Support for the upcoming FreeBSD-5.0 release has been included.
1520 (thanks to Dag-Erling Smorgrav)
1521 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1522 effect when it is not in a toplevel context.
1523 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1524 stand a better chance of being correct. (thanks to Gerd
1526 * fixed bug in COERCE, which now signals an error on coercing a
1527 rational to a bounded real type which excludes the expected
1529 * The compiler is now able to derive types more accurately from the
1530 COERCE and COMPILE functions.
1531 * fixed bug 223: functional binding is considered to be constant
1532 only for symbols in the CL package.
1533 * fixed bug 231: SETQ did not check the type of a variable being set
1534 (reported by Robert E. Brown)
1535 * A new optimization for MAKE-INSTANCE has been included, fixing
1536 various bugs (including relating to :ALLOCATION :CLASS slots and
1537 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1539 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1540 ** As required by ANSI, LOOP now disallows anonymous collection
1541 clauses such as COLLECT I in conjunction with aggregate boolean
1542 clauses such as THEREIS (= I 1);
1543 ** LOOP now signals an error when any variable is reused in the
1544 same loop (including the potentially useful construct analogous
1545 to WITH A = 1 WITH A = (1+ A);
1546 ** IT is only a special loop symbol within the first clause of a
1547 conditional loop clause;
1548 ** LOOP with a typed iteration variable over a hashtable now
1549 signals a type error iff it should.
1550 * fixed some other bugs revealed by Paul Dietz' test suite:
1551 ** FILE-STREAM now names the class previously known as FD-STREAM;
1552 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1553 argument) no longer signals an error;
1554 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1555 of the slot symbol, rather than using the current package
1556 ((:CONC-NAME "") continues to intern the slot's name in the
1558 * incremented fasl file version number, because of the incompatible
1559 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1560 of the new implementation of DEFINE-COMPILER-MACRO.
1562 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1563 * Support for building SBCL for MIPS platforms running in
1564 little-endian mode has now been checked in, and basic
1565 functionality on said platforms verified.
1566 * minor incompatible change: PCL now records the pathname of a file
1567 in which methods and the like are defined, rather than its
1569 * minor incompatible change: TRUENAME now considers the truename of
1570 a file naming a directory to be the pathname with :DIRECTORY
1571 component indicating that directory.
1572 * minor incompatible change: a NAMED clause in the extended form of
1573 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1574 reason for the previous behaviour is unclear.
1575 * more systematization and improvement of CLOS and MOP conformance
1576 in PCL (thanks to Gerd Moellman and Pierre Mai):
1577 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1579 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1580 primary methods with no specializers;
1581 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1583 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1584 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1585 FORWARD-REFERENCED-CLASSes; error reporting on
1586 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1588 ** SXHASH on CLOS instances now uses a slot internal to the
1589 instance to return different numbers on distinct instances,
1590 while preserving the same return value through invocations of
1592 ** DEFMETHOD signals errors when methods with longer incongruent
1593 lambda lists are added to generic functions;
1594 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1595 CLASS, as specified in AMOP;
1596 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1597 on the order returned by the primary method for classes of
1598 class STANDARD-CLASS;
1599 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1600 * fixed some bugs shown by Paul Dietz' test suite:
1601 ** DOLIST puts its body in TAGBODY;
1602 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1604 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1605 value producing form;
1606 ** if more variables are given to PROGV than values, extra
1607 variables are bound and made to have no value;
1608 ** NSUBSTITUTE on list arguments gets the right answer with
1610 ** ELT signals an error of type TYPE-ERROR when the index argument
1611 is not a valid sequence index;
1612 ** LOOP signals (at macroexpansion time) an error of type
1613 PROGRAM-ERROR when duplicate variable names are found;
1614 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1615 ** FUNCALL of special-operators now cause an error of type
1617 ** PSETQ now works as required in the presence of side-effecting
1618 symbol-macro places;
1619 ** NCONC accepts any object as its last argument;
1620 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1622 ** loop-for-as-package does not require a package to be explicitely
1624 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1626 * fixed bug 166: compiler preserves "there is a way to go"
1627 invariant when deleting code.
1628 * fixed bug 172: macro lambda lists with required arguments after
1629 &REST arguments now cause an error to be signalled. (thanks to
1631 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1632 bug. (thanks to Matthew Danish)
1633 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1635 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1636 arguments are lexically rebound. (thanks to Gerd Moellmann and
1638 * fixed bug 194: error messages are now more informative when there
1639 is no primary method applicable in a call to a generic
1640 function. (thanks to Gerd Moellmann)
1641 * fixed bug in command line argument checking (thanks to Julian
1643 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1644 (thanks to Matthew Danish)
1645 * incremented fasl file version number, because of the
1646 SXHASH-related changes in the layout of CLOS data structures
1648 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1649 * minor incompatible change: The runtime (the Unix executable named
1650 "sbcl") is now much pickier about the .core files it will load.
1651 Essentially it now requires .core files to descend from the same
1652 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1653 as the runtime does. (The intent is to prevent the crashes which
1654 can occur, and which can even be reported as mysterious failures,
1655 when people patch the sources or change the build parameters
1656 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1657 sbcl and .core files.)
1658 * fixed bug: VALUES-LIST is no longer optimized away.
1659 * fixed bug 142: The FFI conversion of C string values to Lisp
1660 string values no longer conses excessively. (thanks to Nathan
1661 Froyd porting Raymond Toy's fix to CMU CL)
1662 * began to systematize and improve MOP conformance in PCL (thanks to
1663 Nathan Froyd, Gerd Moellman and Pierre Mai):
1664 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1666 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1667 ** FINALIZE-INHERITANCE is now called on class finalization;
1668 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1669 argument precedence order.
1670 * fixed bug 202: The compiler no longer fails on functions whose
1671 derived types contradict their declared type.
1672 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1673 so it can be non-toplevel.
1674 * The fasl file version number has changed (because of the new
1675 implementation of DEFMACRO).
1676 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1677 safe code, that any length requirement by their type-specifier
1678 argument is valid. The exceptions to this are described in bug
1680 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1681 functions now check the types of their inputs as required by ANSI.
1682 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1683 introduced symbol is DECLAREd to be SPECIAL.
1684 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1685 * fixed bug: :COUNT argument to sequence functions may be negative
1686 * fixed bug: body of DO-SYMBOLS may contain declarations
1687 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1688 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1689 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1690 symbol macro only once
1691 * fixed printing of call frame when argument list is unavailable
1692 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1693 * compiler no longer signals WARNING on unknown keyword
1696 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1697 * A beta-quality port to the mips architecture running Linux,
1698 based on the old CMUCL backend, has been made. It has been tested
1699 on a big-endian kernel, and works sufficiently well to be able to
1700 rebuild itself; it has not been tested in little-endian mode.
1701 * fixed an inconsistency between gencgc.c and purify.c which made
1702 dumping/loading .core files unreliable
1703 * fixed bug 120a: The compiler now deals correctly with IFs where
1704 the consequent is the same as the alternative, instead of
1705 misderiving the return type. (thanks to Alexey Dejneka)
1706 * fixed bug 113: Logical pathnames are now dumpable (the logical
1707 host is resolved at load-time, throwing an error if it is not
1709 * fixed bug 174: FORMAT's error message is slightly clearer when a
1710 non-printing character is used in a format directive.
1711 * fixed several bugs in compiler checking of type declarations, i.e.
1712 violations of the Python "declarations are assertions" principle
1713 (thanks to Alexey Dejneka)
1714 * fixed several bugs in PCL's error checking (thanks to Gerd
1716 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1718 * fixed bug in compilation of functions as first class values
1719 (thanks to Antonio Martinez-Shotton)
1720 * The compiler's handling TYPE-ERRORs which it can prove will
1721 inevitably happen at runtime has been cleaned up and corrected
1722 in several ways. (thanks to Alexey Dejneka)
1723 * improved argument type checking for various basic arithmetic
1724 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1725 lavished on them in the past that they can be compiled in many
1726 ways in different special cases
1727 * fixed bug 181: compiler checks validity of user supplied type
1729 * cleaned up code flushing in optimization: Function calls which
1730 should signal errors for safety purposes (e.g. which ANSI says
1731 should signal errors when their arguments are of incorrect type)
1732 are no longer optimized away.
1733 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1734 * incremented fasl file version number, because changes in the
1735 implementation of sequence functions like COERCE caused
1736 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1737 in old inline expansions) to become undefined. (Actually these
1738 changes were later undone, so we might very well be binary
1739 compatible with 0.7.7 after all, but leaving the version number
1740 incremented seemed like the simplest and most conservative
1743 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1744 * An alpha-quality port to the parisc architecture running Linux,
1745 based on the old CMUCL backend, has been made. This, even more so
1746 than the other backends, should be considered still a work in
1747 progress; known problems include that the Linux kernel in 64-bit
1748 mode does not propagate the correct sigcontext structure to
1749 userspace, and consequently SBCL on a parisc64 kernel will not
1751 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1752 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1753 suppressed.) Also now that the compiler is looking at declarations
1754 in the environment, it checks optimization declarations as well,
1755 and suppresses inlining when (> DEBUG SPEED).
1756 * More fixes have been made to treatment of floating point exception
1757 treatment and other Unix signals. In particular, floating point
1758 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1759 * The detection and handling of control stack exhaustion (infinite
1760 or very deeply nested recursion) has changed. Stack exhaustion
1761 detection is now done by write-protecting pages at the OS level
1762 and applies at all optimization settings; when found, a
1763 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1764 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1765 be caught using IGNORE-ERRORS.
1766 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1767 correctly when transforming recursive LABELS functions to LETs.
1768 (thanks to Alexey Dejneka)
1769 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1770 that are names of constants or global variables.
1771 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1772 alien routines with docstrings.
1773 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1774 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1776 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1777 (thanks to Eric Marsden)
1778 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1779 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1780 * Bug fix: The SPARC backend can now compile functions involving
1781 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1782 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1783 to a functions expecting a PATHNAME-DESIGNATOR.
1784 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1785 lambda lists. (thanks to Alexey Dejneka)
1786 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1787 functions so that EXPORTing the name of the class doesn't cause
1788 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1789 to Antonio Martinez for reporting this)
1790 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1791 constant-folded without failing an assertion. (thanks to Einar
1792 Floystad Dorum for reporting this)
1793 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1794 types are now dealt with more correctly by the compiler.
1795 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1796 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1797 behaviour analogous to RENAME-FILE. This puts its behaviour more
1798 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1800 * The fasl file version number has changed again. (because of the
1801 bug fix involving the names of PCL MAKE-INSTANCE functions)
1803 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1804 * bug fix: Floating point exceptions are treated much more
1805 consistently on the x86/Linux and PPC/Linux platforms.
1806 * Array initialization with :INITIAL-ELEMENT is now much faster for
1807 cases when the compiler cannot open code the array creation, but
1808 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1809 array accesses have also seen a speed increase.
1810 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1811 specified. (This was at the root of some bad interactions between
1812 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1814 * bug fix: Internal error arguments for undefined functions are now
1815 computed correctly on the PPC/Linux platform.
1816 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1817 Raymond Toy's patch for CMU CL)
1818 * Support for the Solaris 9 operating environment has been included
1819 (thanks to Daniel Merritt)
1820 * A very ugly but hopefully complete draft of the missing FFI chapter
1821 of the manual has been created by reformatting the corresponding
1822 CMU CL manual chapter into (currently very ugly and incoherent)
1823 DocBook and bringing it up to date for SBCL behavior. Thus, the
1824 manual is now essentially complete, at least by my extreme
1825 once-and-only-once standards, whereby it's acceptable to refer to
1826 the doc strings of SB-EXT functions as the primary documentation.
1827 * The fasl file version number has changed again, due to cleanup of
1828 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1829 is no longer a static symbol.)
1831 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1832 * SBCL now builds with OpenMCL (version 0.12) as the
1833 cross-compilation host; also, more progress has been made toward
1834 bootstrapping under CLISP.
1835 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1837 * bug 158 fixed: The compiler can now deal with integer loop
1838 increments different from 1; fixing this turned out also to fix
1840 * bug 169 fixed: no more bogus warnings about using lexical bindings
1841 despite the presence of perfectly good SPECIAL declarations (thanks
1842 to David Lichteblau)
1843 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1844 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1845 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1847 * bug 180 fixed: Method combination specifications no longer ignore
1848 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1849 * bug fix: Structure type predicate functions now check their argument
1850 count as they should.
1851 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1852 correctly. (thanks to Pierre Mai)
1853 * minor incompatible change: The --noprogrammer option is deprecated
1854 in favor of the new --disable-debugger option, which is very similar.
1855 (The major difference is that it takes effect at a slightly different
1856 time at startup, causing handling of errors in --sysinit and
1857 --userinit files will be affected differently.) The
1858 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1859 been added to allow this functionality to be controlled from ordinary
1860 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1861 maintainers, who might want to run non-interactive scripts to
1862 build SBCL cores which will later be used interactively.)
1863 * minor incompatible change: The LOAD function no longer, when given
1864 a wild pathname to load, loads all files matching that pathname.
1865 Instead, an error of type FILE-ERROR is signalled.
1867 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1868 * bug 147 fixed: The compiler preserves its block link/count
1869 invariants more correctly now so that it doesn't crash. (thanks
1871 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1873 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1874 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1875 no-value-for-this-character as NIL (as specified by ANSI).
1876 * HOST-NAMESTRING on physical pathnames now returns a string that is
1877 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1878 (thanks to Christophe Rhodes)
1879 * The Alpha port handles icache flushing more correctly. (thanks to
1881 * More progress has been made toward bootstrapping under CLISP. (thanks
1882 to Christophe Rhodes)
1883 * The fasl file format has changed again, because dynamic loading
1884 on OpenBSD (which has non-ELF object files) motivated some cleanups
1885 in the way that foreign symbols are transformed and passed around.
1886 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1887 is no longer treated as whitespace by the reader, but instead as
1888 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1889 |A
\7fB|, instead of A as it used to.
1891 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1892 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1893 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1894 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1895 this as of 0.7.1.45, but was left out of the previous news section
1896 (thanks to Dan Barlow)
1897 * SBCL now runs on the Solaris operating system on SPARC architectures
1898 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1899 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1900 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1901 * SPARC backend cleanups, allowing builds of cores optimized for V8
1902 and V9 SPARCS, and also emission of code targeted to a particular
1903 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1905 * SBCL is closer to bootstrapping under CLISP, thanks to various
1906 fixes by Christophe Rhodes.
1907 * The fasl file format has changed again, to allow the compiler's
1908 INFO database to support symbol macros.
1909 * The user manual (in doc/) is formatted into HTML more nicely.
1910 (thanks to coreythomas)
1911 * The system is smarter about SUBTYPEP relationships, especially
1912 those involving NOT types (including types such as ATOM which are
1913 represented internally using NOT types). Thus SUBTYPEP is less
1914 likely to return (VALUES NIL NIL) in general, and in particular
1915 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1917 * The fasl file format has changed again, because the internal
1918 representation of types now includes a new slot to support the new
1919 SUBTYPEP-of-NOT-types logic.
1920 * (not a change in the main branch of SBCL, but a related prototype
1921 which can hopefully be merged into the main branch of SBCL in the
1922 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1923 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1924 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1925 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1926 dispatch macro character combinations. (thanks to Alexey Dejneka)
1927 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1928 that superficially look like logical namestrings correctly.
1929 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1931 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1932 default supplied-p) for &optional and &key arguments. (thanks to
1935 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1936 * incompatible change: The compiler is now less aggressive about
1937 tail call optimization, doing it only when (> SPACE DEBUG) or
1938 (> SPEED DEBUG). (This is an incompatible change because there are
1939 programs which relied on the old CMU-CL-style behavior to optimize
1940 away their unbounded recursion which will now die of stack overflow.)
1941 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1942 for non-GENCGC systems has been increased to 20M (since that
1943 seems much closer to the likely performance optimum for modern
1944 systems than the old 4M value was)
1945 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1946 and *DEBUG-PRINT-LEVEL*
1947 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1948 of CMU CL's support for SPARC, and various endianness and other
1949 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1950 * new syntactic sugar for the Unix command line: --load foo.bar is now
1951 an alternate notation for --eval '(load "foo.bar")'.
1953 ** The system now detects stack overflow and handles it gracefully,
1954 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1955 optimization settings. (This is a good thing in general, and
1956 its introduction in this version should be particularly timely
1957 for anyone whose code fails because of suppression of tail
1959 ** The system now hunts for the C variable "environ" in a more
1960 devious way, to avoid segfaults when the C library version differs
1961 between compile time and run time. (thanks to Christophe Rhodes)
1962 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1963 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1964 ** The compiler no longer issues bogus style warnings for undefined
1965 classes in the same source file as the DEFCLASSes which defined
1966 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1968 ** fixes in CONDITION class precedence list for undefined function
1969 errors (thanks to Alexei Dejneka)
1970 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1971 correctly. (thanks to Dan Barlow)
1972 ** portability fixes aiming at bootstrapping under CLISP (thanks
1973 to Dave McDonald and Christophe Rhodes)
1974 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1975 ** fixes in type translation and and type inference (thanks to
1977 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1978 ** various fixes in the new ports (thanks to Dan Barlow)
1979 * several changes related to debugging:
1980 ** suppression of tail recursion, as noted above
1981 ** stack overflow detection, as noted above
1982 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1983 is now the default. (For some time encapsulation has been more
1984 reliable than the breakpoint-based :ENCAPSULATE NIL
1985 implementation, at least on X86 systems; and I just noticed that
1986 encapsulation also seems closer to the spirit of the ANSI
1989 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
1991 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
1992 up properly again. (There was a packaging bug in 0.7.0 which
1993 left their definitions in SB-SYS::LOAD-FOREIGN and
1994 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
1995 vital for most things which interface to C-level interfaces,
1996 like extensions working with sockets or databases or
1997 Perl-compatible regexes or whatever, and the need to fix
1998 this bug is the main reason that 0.7.1 was released so
2000 ** DEFGENERIC is now choosier about the methods it redefines, so that
2001 reLOADing a previously-LOADed file containing DEFGENERICs does
2002 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2003 cycle now works as it should. (thanks to Alexey Dejneka)
2004 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2005 fixed by Christophe Rhodes. (He actually submitted this patch
2006 months ago, and I delayed until after 0.7.0.)
2007 ** Bug 111 (internal compiler confusion about runtime checks on
2008 FUNCTION types) was fixed by Alexey Dejneka.
2009 * Some internal cleanups (getting rid of variables which aren't
2010 needed now that the byte interpreter is gone) caused the fasl
2011 file format number to change again.
2013 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2014 * major incompatible change: The default fasl file extension, i.e. the
2015 default extension for files produced by COMPILE-FILE, has changed
2016 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2018 ** There are many changes in the implementation of the compiler.
2019 SBCL is now essentially a compiler-only implementation of ANSI
2020 Common Lisp. EVAL still "interprets" a few special cases, but
2021 almost all the interesting cases are handled by creating
2022 a LAMBDA expression, calling COMPILE on it, then calling
2023 FUNCALL on the result.
2024 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2025 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2026 Since the code is newer, there might still be some new bugs
2027 (though not as many as before Martin Atzmueller's fixes:-). But
2028 the new code is substantially simpler and clearer, and hopefully
2029 any remaining bugs will be simpler, less fundamental, and more
2030 fixable then the bugs in the old code.
2031 ** The revised compiler is still a little unsteady on its feet.
2033 *** The debugging information it produces (particularly the names
2034 of FUNCTION objects) is sometimes much less useful than what
2035 the old compiler produced.
2036 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2037 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2038 within a MACROLET) has been temporarily weakened.
2039 ** There are new compiler optimizations for various functions:
2040 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2041 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2042 *** the math functions TRUNCATE, FLOOR, and CEILING
2043 *** the function-of-all-trades COERCE
2044 Mostly these should be transparent, but there's one
2045 potentially-annoying problem (bug 117): when the compiler
2046 inline-expands a function and does type analysis on the result,
2047 it can create control paths which have type mismatches, and
2048 when it can't prove that those control paths aren't taken,
2049 it will issue WARNINGs about the type mismatches. This is
2050 a particular problem in practice for the new sequence functions.
2051 It's not clear how this should be fixed, and for now, a
2052 workaround is given in the entry for 117 in the BUGS file.
2053 ** (Because of the interaction between the two previous items --
2054 occasional inlining problems and new inline expansions -- some
2055 of the new sequence function optimizations won't really kick in
2056 completely until debugging information, and then inlining, are
2057 straightened out in some future version.)
2058 * minor incompatible changes:
2059 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2060 DIRECTORY no longer implicitly promotes NIL slots of its
2061 pathname argument to :WILD. In particular, when you ask for the
2062 contents of a directory (which you used to be able to do without
2063 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2064 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2065 ** changes in behavior that ANSI explicitly defines to be
2066 implementation dependent:
2067 *** The new compiler-only implementation still conforms with ANSI,
2068 but acts a little different than before. Besides the obvious
2069 changes in performance tradeoffs (that the cost per form passed
2070 to EVAL has gone up, and the cost per form executed by EVAL
2071 has gone down), the behavior of the system changes a little
2072 because there are no longer any interpreted function objects.
2073 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2074 e.g. doing COMPILE on the output of interactive DEFUN is
2076 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2078 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2079 has changed from (:CL) to NIL.
2080 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2081 e.g. #\Newline and #\Tab, don't have names specified in the
2082 ANSI Common Lisp standard, is now based on their ASCII symbolic
2083 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2084 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2085 are no longer used for output.
2086 ** changes in internal implementation constants:
2087 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2088 4 million. (If your application spends a lot of time GCing and
2089 you have a lot of RAM, you might want to experiment with
2090 increasing it even more.)
2091 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2092 However, almost all old code should still continue to work without
2093 immediate update, as SB-C-CALL is now a (deprecated) nickname
2095 ** Old operator names in the style DEF-FOO are now deprecated in
2096 favor of new corresponding names DEFINE-FOO, for consistency with
2097 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2098 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2099 internal symbols, but a few supported extensions like
2100 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2101 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2102 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2103 etc. as you get deeper into recursive calls to the debugger
2104 command loop, instead of the old "5]", "5]]", "5]]]"
2105 sequence. (I was motivated to do this when squabbles between
2106 ILISP and SBCL left me very deeply nested in the debugger. In the
2107 short term, this change will probably provoke more ILISP/SBCL
2108 squabbles, but hopefully it will be an improvement in the long run.)
2109 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2110 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2111 function, since declaiming return types involving aliens is
2112 (1) annoyingly messy to do by hand and (2) vital to efficient
2113 compilation of code which calls such functions.
2114 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2115 longer reexported by the SB-EXT package. They're solely useful
2116 for alien code, so it seems more logical that you should get
2117 them from the SB-ALIEN package, not in SB-EXT.
2118 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2119 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2120 features. Instead, the code that they used to control is always
2121 built into the system.
2122 * many other bug fixes
2123 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2124 advantage of the new EVAL-WHEN stuff and to clean them up in
2125 general, and they are now more ANSI-compliant in a number of
2126 ways. Martin Atzmueller is responsible for a lot of this.
2127 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2129 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2130 *** correct ERROR type for various file operations
2131 *** some fixes for Lisp streams
2132 *** DEFMETHOD syntax checking
2133 *** changing old weird representation of debug information as
2134 strings (which, among their other deficiencies, don't transform
2135 correctly when you rename packages, and don't change their
2136 print representation when you change things like *PACKAGE*
2137 and *PRINT-LENGTH*) to symbols and lists of symbols
2138 He also made several improvements and fixed several bugs in DESCRIBE.
2139 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2141 *** misbehavior of WRITE-STRING/WRITE-LINE
2142 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2143 and several other LOOP problems as well
2144 *** DIRECTORY when similar filenames are present
2145 *** DEFGENERIC with :METHOD options
2146 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2147 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2148 *** argument ordering in FIND with :TEST option
2149 *** mishandled package designator argument in APROPOS-LIST
2150 *** various problems in the backquote readmacro
2151 *** a bug in APROPOS
2152 *** probably some others that I'm not describing very well here,
2153 since the CVS log documents them by reference to sbcl-devel
2154 messages, and the SourceForge archives aren't working well.:-(
2155 ** Dan Barlow improved the Alpha port (and is making progress on the
2156 PPC port, for those of you who think different).
2157 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2158 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2159 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2160 continues to work on the SPARC port (for those of you in a position
2161 to look down upon our little PC-compatible boxes from a great height).
2162 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2163 and uses that copy, rather than the current dynamic value, when
2164 it's trying to decide whether to truncate output. Thus e.g.
2165 (let ((*print-lines* 50))
2166 (pprint-logical-block (stream nil)
2168 (let ((*print-lines* 8))
2169 (print (aref possiblybigthings i) stream)))))
2170 should now truncate the logical block only at 50 lines, instead of
2171 often truncating it at 8 lines, as it did before.
2172 * The doc/cmucl/ directory, containing old CMU CL documentation
2173 from the time of the fork, is no longer part of the base system.
2174 SourceForge has shut down its anonymous FTP service, and with it
2175 my original plan for distributing the old CMU CL documentation
2176 there. For now, if you need these files you can download an old
2177 SBCL source release and extract them from it.
2178 * The fasl file version number changed again, for dozens of reasons,
2179 some of which are apparent above.
2181 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2182 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2183 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2184 patch, so that DISASSEMBLE output is much nicer.
2185 * The code in the SB-PROFILE package now seems reasonably stable.
2186 I still haven't decided what the final interface should look like
2187 (I'd like PROFILE to interact cleanly with TRACE, since both
2188 facilities use function encapsulation) but if you have a need
2189 for profiling now, you can probably use it successfully with
2190 the current CMU-CL-style interface.
2191 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2192 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2193 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2194 link now returns the dangling link itself, and for similar
2195 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2196 link itself. (In these cases the old code signalled an error and
2197 looped endlessly, respectively.) Thus, DIRECTORY now works even
2198 in the presence of dangling and cyclic symbolic links.
2199 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2200 is now a supported extension again, since the consensus on
2201 sbcl-devel was that it can be useful for ordinary development
2202 work, not just for debugging SBCL itself.
2203 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2204 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2205 that currently-defined functions might be redefined later with
2206 different return types.
2207 * Hash tables can be printed readably, as inspired by CMU CL code
2208 of Eric Marsden and SBCL code of Martin Atzmueller.
2209 * better error handling in CLOS method combination, thanks to
2210 Martin Atzmueller porting Pierre Mai's CMU CL patches
2211 * more overflow fixes for >16Mbyte I/O buffers
2212 * A bug in READ has been fixed, so that now a single Ctrl-D
2213 character suffices to cause end-of-file on character streams.
2214 In particular, now you only need one Ctrl-D at the command
2215 line (not two) to exit SBCL.
2216 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2218 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2219 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2220 does the right thing.
2221 * The implementation of some type tests, especially for CONDITION
2222 types, is now tidier and maybe faster, due to CMU CL code
2223 originally by Douglas Crosher, ported by Martin Atzmueller.
2224 * Some math functions have been fixed, and there are new
2225 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2226 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2227 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2228 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2229 likely to be built into sbcl-0.7.0.)
2230 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2231 CL, ported to SBCL by Martin Atzmueller)
2232 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2233 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2234 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2235 * There's a new slam.sh hack to shorten the edit/compile/debug
2236 cycle for low-level changes to SBCL itself, and a new
2237 :SB-AFTER-XC-CORE target feature to control the generation of
2238 the after-xc.core file needed by slam.sh.
2239 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2240 COMPILE-FILE is no longer supported, so that now every function
2241 gets an entry point, so that block compilation looks a little
2242 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2243 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2244 returns the number of bytes consed since the system started,
2245 rather than the number consed since the first time the function
2246 was called. (The new definition parallels ANSI functions like
2247 CL:GET-INTERNAL-RUN-TIME.)
2248 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2249 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2250 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2251 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2252 * Fasl file version numbers are now independent of the target CPU,
2253 since historically most system changes which required version
2254 number changes have affected all CPUs equally. Similarly,
2255 the byte fasl file version is now equal to the ordinary
2258 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2259 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2260 is no longer recognized. I apologize for this, because it was
2261 listed in SB-EXT as a supported extension, but I found that
2262 its existing behavior was poorly specified, as well as incorrectly
2263 specified, and it looked like too much of a mess to straighten it
2264 out. I have enough on my hands trying to get ANSI stuff to work..
2265 * many patches ported from CMU CL by Martin Atzmueller, with
2266 half a dozen bug fixes in pretty-printing and the debugger, and
2267 half a dozen others elsewhere
2268 * fixed bug 13: Floating point infinities are now supported again.
2269 They might still be a little bit flaky, but thanks to bug reports
2270 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2271 as flaky as they were.
2272 * The --noprogrammer command line option is now supported. (Its
2273 behavior is slightly different in detail from what the old man
2274 page claimed it would do, but it's still appropriate under the
2275 same circumstances that the man page talks about.)
2276 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2277 are now supported, and enabled by default. Thus, the compiler can
2278 handle many floating point and complex operations much less
2279 inefficiently. (Thus e.g. you can implement a complex FFT
2281 * The compiler now detects type mismatches between DECLAIM FTYPE
2282 and DEFUN better, and implements CHECK-TYPE more correctly, and
2283 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2284 thanks to patches from Martin Atzmueller.
2285 * various fixes to make the cross-compiler more portable to
2286 ANSI-conforming-but-different cross-compilation hosts (notably
2287 Lispworks for Windows, following bug reports from Arthur Lemmens)
2288 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2289 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2290 fixed thanks to Pierre Mai's CMU CL patches.
2291 * fixes to keep the system from overflowing internal counters when
2292 it tries to use i/o buffers larger than 16M bytes
2293 * fixed bug 45a: Various internal functions required to support
2294 complex special functions have been merged from CMU CL sources.
2295 (When I was first setting up SBCL, I misunderstood a compile-time
2296 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2297 * improved support for type intersection and union, fixing bug 12
2298 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2299 more obscure bugs as well
2300 * some steps toward byte-compiling non-performance-critical
2301 parts of the system, courtesy of patches from Martin Atzmueller
2302 * Christophe Rhodes has made some debian packages of sbcl at
2303 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2304 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2305 stable, but are nonetheless usable. When he's ready, I'd be happy
2306 to add them to the SourceForge "File Releases" section. (And if
2307 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2308 * new fasl file format version number (because of changes in
2309 internal representation of (OR ..) types to accommodate the new
2310 support for (AND ..) types, among other things)
2312 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2313 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2315 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2317 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2318 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2319 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2320 * DESCRIBE now works on structure objects again.
2321 * Most function call argument type mismatches are now handled as
2322 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2323 know whether the function will be redefined before the call is
2324 executed. (The compiler could flag local calls with full WARNINGs,
2325 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2326 it doesn't keep track of enough information to know whether calls
2327 are local in this sense.)
2328 * Compiler output is now more verbose, with messages truncated
2329 later than before. (There should be some supported way for users
2330 to override the default verbosity, but I haven't decided how to
2331 provide it yet, so this behavior is still controlled by the internal
2332 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2333 src/compiler/ir1util.lisp.)
2334 * Fasl file format version numbers have increased again, because
2335 support for the Gray streams extension changes the layout of the
2336 system's STREAM objects.
2337 * The Gray subclassable streams extension now works, thanks to a
2338 patch from Martin Atzmueller.
2339 * The full LOAD-FOREIGN extension (not just the primitive
2340 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2341 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2342 but like most other programs, it defaults to copying the Unix
2343 environment from the original process instead of starting the
2344 new process in an empty environment.
2345 * Extensions which manipulate the Unix environment now support
2346 an :ENVIRONMENT keyword option which doesn't smash case or
2347 do other bad things. The CMU-CL-style :ENV option is retained
2348 for porting convenience.
2349 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2350 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2352 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2354 * A patch from Martin Atzmueller seems to have solved the SIGINT
2355 problem, and as far as we know, signal-handling now works cleanly.
2356 (If you find any new bugs, please report them!)
2357 * The system no longer defaults Lisp source file names to types
2358 ".l", ".cl", or ".lsp", but only to ".lisp".
2359 * The compiler no longer uses special default file extensions for
2360 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2361 seems to expect a single default extension for all compiled code,
2362 and there's no compelling reason to try to stretch the standard
2363 to allow two different extensions.) Instead, byte-compiled files
2364 default to the same extension as native-compiled files.
2365 * Fasl file format version numbers have increased again, because
2366 a rearrangement of internal implementation packages made some
2367 dumped symbols in old fasl files unreadable in new cores.
2368 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2369 many fewer weird special cases.
2370 * Bug #17 (differing COMPILE-FILE behavior between logical and
2371 physical pathnames) has been fixed, and some related misbehavior too,
2372 thanks to a patch from Martin Atzmueller.
2373 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2374 by Tim Moore, ported to SBCL by Martin Atzmueller.
2375 * Martin Atzmueller fixed several filesystem-related problems,
2376 including bug #36, in part by porting CMU CL patches, which were
2377 written in part by Paul Werkowski.
2378 * More compiler warnings in src/runtime/ are gone, thanks to
2379 more patches from Martin Atzmueller.
2380 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2383 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2385 * DESCRIBE now works on CONDITION objects.
2386 * The debugger now handles errors which arise when trying to print
2387 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2389 * The build system now uses an additional file, customize-target-features.lisp,
2390 to allow local modifications to the target *FEATURES* list. (The point of
2391 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2392 debugging features enabled, without having to worry about propagating it
2393 into everyone's system when I do a "cvs update".) When no
2394 customize-target-features.lisp file exists, the target *FEATURES* list
2395 should be constructed the same way as before.
2396 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2397 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2398 warn about behavior which is undefined under the ANSI spec. Note: This
2399 is specified by ANSI, but it's not very popular with programmers.
2400 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2401 macro in the SBCL sources for an example of a workaround which you
2402 might use to make portable ANSI-standard code which does what you want.
2403 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2404 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2405 when it's used as a non-toplevel form. (This is required in order
2406 to implement the DEFCONSTANT-EQX macro.)
2407 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2408 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2409 funny things, due to bugs in the implementation of EVAL-WHEN
2410 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2411 non-toplevel DEFCONSTANTs are uncommon.)
2412 * The core file version number and fasl file version number have been
2413 incremented, because the old noncompliant DEFCONSTANT behavior involved
2414 calling functions which no longer exist, and because I also took the
2415 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2416 * fixed bug 1 (error handling before read-eval-print loop starts), and
2417 redid debugger restarts and related debugger commands somewhat while
2419 ** The QUIT debugger command is gone, since it did something
2420 rather different than the SB-EXT:QUIT command, and since it never
2421 worked properly outside the main toplevel read/eval/print loop.
2422 Invoking the new TOPLEVEL restart provides the same functionality.
2423 ** The GO debugger command is also gone, since you can just invoke
2424 the CONTINUE restart directly instead.
2425 ** The TOP debugger command is also gone, since it's redundant with the
2426 FRAME 0 command, and since it interfered with abbreviations for the
2428 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2430 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2431 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2432 some porting work by Martin Atzmueller)
2433 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2434 with a boatload of patches which he ported to SBCL. Now that those
2436 ** The system tries to make sure that its low-priority messages
2437 are prefixed by semicolons, to help people who like to use
2438 syntax highlighting in their ILISP buffer. (This patch
2439 was originally due to Raymond Toy.)
2440 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2441 patches originally written by Raymond Toy.
2442 ** The compiler understands coercion between single-value and
2443 multiple-VALUES type expressions better, getting rid of some very
2444 weird behavior, thanks to patches originally by Robert MacLachlan
2445 and Douglas Crosher.
2446 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2447 lambda lists, thanks to a patch originally by Pierre Mai.
2448 ** The system no longer bogusly warns about "abbreviated type
2450 ** The compiler gets less confused by inlining and RETURN-FROM,
2451 thanks to some patches originally by Tim Moore.
2452 ** The system no longer hangs when dumping circular lists to fasl
2453 files, thanks to a patch originally from Douglas Crosher.
2454 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2455 undefined function error.
2456 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2457 in structure slot initforms) for now, documented workaround instead:-|
2458 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2459 * fixed bug 5: added stubs for various Gray stream functions called
2460 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2461 installed, at least appropriate type errors are generated
2462 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2463 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2464 consistently in DEFMETHOD forms.
2465 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2466 it doesn't seem to affect SBCL after all
2467 * The C runtime system now builds with better optimization and many
2468 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2470 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2472 * The system is now under CVS at SourceForge (instead of the
2473 CVS repository on my home machine).
2474 * The new signal handling code has been tweaked to treat register
2475 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2476 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2477 downstream, e.g. in debug-int.lisp, has implicit dependencies
2478 on the unsignedness of integer representation of machine words,
2479 and that caused the system to bomb out with infinite regress
2480 when trying to recover from type errors involving signed values,
2481 e.g. (BUTLAST '(1 2 3) -1).)
2482 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2483 (This was one of the bugs Peter Van Eynde reported back in July.)
2484 * The system now uses code inspired by Colin Walters' O(N)
2485 implementation of MAP (from the cmucl-imp@cons.org mailing
2486 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2487 inline the MAP operation, and there is more than one
2488 sequence argument to the MAP call (so that it can't just
2489 do ETYPECASE once and for all based on the type of the
2490 single sequence argument). (The old non-inline implementation
2491 of the general M-argument sequence-of-length-N case required
2492 O(M*N*N) time when any of the sequence arguments were LISTs.)
2493 * The QUIT :UNIX-CODE keyword argument has been renamed to
2494 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2495 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2497 * Raymond Wiker's patches to port dynamic loading from Linux to
2498 FreeBSD have been added.
2499 * The BUGS file is now more nearly up to date, thanks in large part
2500 to Martin Atzmueller's review of it.
2501 * The debugger now flushes standard output streams before it begins
2502 its output ("debugger invoked" and so forth).
2503 * The core version number and fasl file version number have both
2504 been incremented, because of incompatible changes in the layout
2506 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2507 used to not be called for a saved Lisp image.)
2508 * Martin Atzmueller's version of a patch to fix a compiler crash,
2509 as posted on sbcl-devel 13 September 2000, has been installed.
2510 * Instead of installing Martin Atzmueller's patch for the
2511 compiler transform for SUBSEQ, I deleted the compiler transform,
2512 and transforms for some similar consing operations.
2513 * A bug in signal handling which kept TRACE from working on OpenBSD
2515 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2516 away by constant folding
2517 * The system now defines its address space constants in one place
2518 (in the Lisp sources), and propagates them automatically elsewhere
2519 (through GENESIS and the sbcl.h file). Therefore, patching the
2520 address map is less unnecessarily tedious and error-prone. The
2521 Lisp names of address space constants have also been systematized.
2522 * CVS tags like dollar-Header-dollar have been removed from
2523 the sources, because they have never saved me trouble and
2524 they've been source of trouble working with patches and other
2525 diff-related operations.
2526 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2527 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2529 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2531 * The system has been ported to OpenBSD.
2532 * The system now compiles with a simple "sh make.sh" on the systems
2533 that it's supported on. I.e., now you no longer need to tweak
2534 text in the target-features.lisp-expr and symlinks in src/runtime/
2535 by hand, the make.sh takes care of it for you.
2536 * The system is no longer so grossly inefficient when compiling code
2537 involving vectors implemented as general (not simple) vectors (VECTOR T),
2538 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2539 which dares to use the various sequence functions on non-simple
2540 vectors, takes less of a performance hit.
2541 * There is now a primitive type predicate VECTOR-T-P
2542 to test for the (VECTOR T) type, so that e.g.
2543 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2544 can now be compiled with some semblance of efficiency. (The old code
2545 turned the type declaration into a full call to %TYPEP at runtime!)
2546 * AREF on (VECTOR T) is still not fast, since it's still compiled
2547 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2548 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2549 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2550 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2551 without making full calls to SUBTYPEP at runtime.
2552 (Some analogous efficiency issues for non-simple vectors specialized to
2553 element types other than T, or for non-simple multidimensional arrays,
2554 have not been addressed. They could almost certainly be handled the
2555 same way if anyone is motivated to do so.)
2556 * The changes in array handling break binary compatibility, so
2557 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2558 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2559 instead of (VALUES T T).
2560 * By following the instructions that Dan Barlow posted to sbcl-devel
2561 on 2 July 2000, I was able to enable primitive dynamic object
2562 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2563 functionality is not implemented (since it calls ld to resolve
2564 library references automatically, requiring RUN-PROGRAM for its
2565 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2566 try to resolve library references) is now supported.
2567 * The system now flushes the standard output streams when it terminates,
2568 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2569 them at several other probably-convenient times, e.g. in each pass of
2570 the toplevel read-eval-print loop, and after evaluating a form given
2571 as an "--eval" command-line option. (These changes were motivated by a
2572 discussion of stream flushing issues on cmucl-imp in August 2000.)
2573 * The source transform for TYPEP of array types no longer assumes
2574 that an array whose element type is a not-yet-defined type
2575 is implemented as an array of T, but instead punts, so that the
2576 type will be interpreted at runtime.
2577 * There is now some support for cross-compiling in make.sh: each of
2578 the phases of make.sh has its own script. (This should be transparent
2579 to people doing ordinary, non-cross-compile builds.)
2580 * Since my laptop doesn't have hundreds of megabytes of memory like
2581 my desktop machine, I became more motivated to do some items on
2582 my to-do list in order to reduce the size of the system a little:
2583 ** Arrange for various needed-only-at-cold-init things to be
2584 uninterned after cold init. To support this, those things have
2585 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2586 symbols with such names are now uninterned after cold init).
2587 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2588 fasl files for cold load.
2589 ** Remove the old compiler structure pooling code (which used to
2590 be conditional on the target feature :SB-ALLOC) completely.
2591 ** Redo the representation of some data in cold init to be more compact.
2592 (I also looked into supporting byte compiled code at bootstrap time,
2593 which would probably reduce the size of the system a lot, but that
2594 looked too complicated, so I punted for now.)
2595 * The maximum signal nesting depth in the src/runtime/ support code has
2596 been reduced from 4096 to 256. (I don't know any reason for the very
2597 large old value. If the new smaller value turns out to break something,
2598 I'll probably just bump it back up.)
2599 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2601 * Many, many bugs reported by Peter Van Eynde have been added to
2602 the BUGS list; some have even been fixed.
2603 * While enabling dynamic object file loading, I tried to make the
2604 code easier to understand, renaming various functions and variables
2605 with less ambiguous names, and changing some function calling
2606 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2607 * While trying to figure out how to do the OpenBSD port, I tried to
2608 clean up some of the code in src/runtime/. In particular, I dropped
2609 support for non-POSIX signal handling, added various comments,
2610 tweaked the code to reduce the number of compilation warnings, and
2611 renamed some files to increase consistency.
2612 * To support the new automatic configuration functionality in make.sh,
2613 the source file target-features.lisp-expr has been replaced with the
2614 source file base-target-features.lisp-expr and the machine-generated
2615 file local-target-features.lisp-expr.
2616 * fixed a stupid quoting error in make.sh so that using CMU CL
2617 "lisp -batch" as cross-compilation host works again
2619 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2621 * DESCRIBE no longer tries to call itself recursively to describe
2622 bound/fbound values, so that it no longer fails on symbols which are
2623 bound to themselves (like keywords, T, and NIL).
2624 * DESCRIBE now works on generic functions.
2625 * The printer now prints less-screwed-up representations of closures
2626 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2627 * A private symbol is used instead of the :EMPTY keyword previously
2628 used to mark empty slots in hash tables. Thus
2629 (DEFVAR *HT* (MAKE-HASH-TABLE))
2630 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2631 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2632 now does what ANSI says that it should. (You can still get
2633 similar noncompliant behavior if bang on the hash table
2634 implementation with all the symbols you get back from
2635 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2636 This breaks binary compatibility, since tests for equality to
2637 :EMPTY are wired into things like the macroexpansion of
2638 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2640 * There's now a minimal placeholder implementation for CL:STEP,
2641 as required by ANSI.
2642 * An obscure bug in the interaction of the normal compiler, the byte
2643 compiler, inlining, and structure predicates has been patched
2644 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2645 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2647 * Missing ordinary arguments in a macro call are now detected even
2648 when the macro lambda list contains &KEY or &REST.
2649 * The debugger no longer complains about encountering the top of the
2650 stack when you type "FRAME 0" to explicitly instruct it to go to
2651 the top of the stack. And it now prints the frame you request even
2652 if it's the current frame (instead of saying "You are here.").
2653 * As specified by ANSI, the system now always prints keywords
2654 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2656 * The default initial SIZE of HASH-TABLEs is now smaller.
2657 * Type information from CLOS class dispatch is now propagated
2658 into DEFMETHOD bodies, so that e.g.
2659 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2661 is now basically equivalent to
2662 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2663 (DECLARE (TYPE SINGLE-FLOAT X))
2665 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2666 operation, without having to do run-time type dispatch.
2667 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2668 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2669 * Since I don't seem to be making big file reorganizations very often
2670 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2671 is overflowing my ability to conveniently back them up), I've finally
2672 checked the system into CVS. (The CVS repository is on my home system,
2673 not at SourceForge -- putting it on SourceForge might come later.)
2674 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2675 high-level GC-NOTIFY-FOO functions send their output. (There's
2676 still very little control of where low-level verbose GC functions
2677 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2678 less than it used to -- the GC-NOTIFY-FOO functions are now under
2679 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2680 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2681 in only one place in the source code, and propagates it automatically
2682 everywhere that it's needed. Thus e.g. when I bump the version from
2683 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2684 * The C source files now include boilerplate legalese and documentation
2685 at the head of each file (just as the Lisp source files already did).
2686 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2687 to his page will be replaced with a link to his new CLiki service.
2689 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2691 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2692 * The build process now looks for GNU make under the default name "gmake",
2693 instead of "make" as it used to. If GNU make is not available as "gmake"
2694 on your system, you can change this default behavior by setting the
2695 GNUMAKE environment variable.
2696 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2697 can build without error under CMU CL.
2699 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2701 * There is now a partial SBCL user manual (with some new text and some
2702 text cribbed from the CMU CL manual).
2703 * The beginnings of a profiler have been added (starting with the
2704 CMU CL profiler and simplifying and cleaning up). Eventually the
2705 main interface should be through the TRACE macro, but for now,
2706 it's still accessed through vaguely CMU-CL-style functions and macros
2707 exported from the package SB-PROFILE.
2708 * Some problems left over from porting CMU CL to the new
2709 cross-compilation bootstrap process have been cleaned up:
2710 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2711 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2712 part when getting rid of the compiler *BACKEND* stuff.)
2713 ** The value of *NULL-TYPE* was screwed up, because it was
2714 being initialized before the type system knew the final
2715 definition of the 'NULL type. This screwed up several key
2716 optimizations in the compiler, causing inefficiency in all sorts
2717 of places. (I found it because I wanted to understand why
2718 GET-INTERNAL-RUN-TIME was consing.)
2719 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2720 PROCLAIM FTYPE information. Unfortunately this broke binary
2721 compatibility again, since now the forms output by DEFGENERIC
2722 to refer to functions which didn't exist in 0.6.3.
2723 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2724 can use the new (as of 0.6.3) transform for SOME into MAP into
2726 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2727 values to (INTEGER 0 1000000), so that the time code will no longer
2728 occasionally get blown up by Linux returning 1000000 microseconds
2729 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2730 its output conform to the ANSI spec. (Alas, this makes its output
2731 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2732 * A full call to MAP NIL with a single sequence argument no longer conses.
2733 * fixes to problems pointed out by Martin Atzmueller:
2734 * The manual page no longer talks about multiprocessing as though
2735 it were currently supported.
2736 * The ILISP support patches have been removed from the distribution,
2737 because as of version 5.10.1, ILISP now supports SBCL without us
2738 having to maintain patches.
2739 * added a modified version of Raymond Toy's recent CMU CL patch for
2740 EQUALP comparison of HASH-TABLE
2742 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2744 * The system still can't cross-compile itself with
2745 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2746 entails), but at least it can compile more of itself that way
2747 than it used to be able to, and various buglets which were uncovered
2748 by trying to cross-compile itself that way have now been fixed.
2749 * This release breaks binary compatibility again. This time
2750 at least I've incremented the FASL file format version to 2, so that the
2751 problem can be detected reliably instead of just causing weird errors.
2752 * various new style warnings:
2753 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2754 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2755 ** using the lexical binding of a variable named in the *FOO* style
2756 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2757 as specified by ANSI.
2758 * *RANDOM-STATE* is no longer automatically initialized from
2759 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2760 default behavior of the system is to repeat its behavior every time
2761 it's run. If you'd like to change this behavior, you can always
2762 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2763 old convention there was no comparably easy way to get the system to
2764 repeat its behavior every time it was run.
2765 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2766 FTYPE declarations has been removed, in favor of their ANSI
2767 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2768 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2769 the types of their sequence arguments can be determined at compile time.
2770 This is done through a new open code expansion for MAP which eliminates
2771 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2772 argument types can be determined at compile time.
2773 * The optimizer now transforms COERCE into an identity operation when it
2774 can prove that the coerced object is already of the correct type. (This
2775 can be a win for machine generated code, including the output of other
2776 optimization transforms, such as the MAP transform above.)
2777 * Credit information has been moved from source file headers into CREDITS.
2778 * Source file headers have been made more standard.
2779 * The CASE macro now compiles without complaining even when it has
2782 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2784 * (Note that the way that the PCL macroexpansions were rewritten
2785 to accommodate the change in DEFGENERIC below breaks binary
2786 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2787 not run under sbcl-0.6.2. Once we get out of alpha releases,
2788 i.e. hit release 1.0.0, we'll probably try to maintain binary
2789 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2790 and sbcl-1.4.4. Until then, however, it might be fairly common
2791 for maintenance releases to break binary compatibility.)
2792 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2794 * The reporting of unhandled signals has been changed to print some
2795 explanatory text as well as the report form. (Previously only
2796 the report form was printed.)
2797 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2798 it defines, so that the compiler no longer issues undefined function
2799 warnings for compiled-but-not-yet-loaded generic functions.
2800 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2801 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2802 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2803 section of the ANSI spec are supported.
2804 * The "" nickname for the "KEYWORD" package has been removed.
2805 The reader still handles symbol tokens which begin with a package marker
2806 as keywords, but it doesn't expose its mechanism for doing so in the
2807 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2808 * The system now issues STYLE-WARNINGs for contradictory TYPE
2809 proclamations. (Warnings for contradictory FTYPE proclamations would
2810 be nice too, but those can't be done usefully unless the type system
2811 is made smarter about FUNCTION types.)
2812 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2813 systematized, so that "*target-*.lisp is supposed to exist only on the
2814 target and imply that there's a related file which exists on the
2815 host, and *host-*.lisp is supposed to exist only on the host and imply
2816 that there's a related file which exists on the target. This involves a
2817 lot of renaming. Hopefully the acute confusion caused by the renaming
2818 will be justified by the reduction in chronic confusion..
2819 ** runtime-type.lisp -> early-target-type.lisp
2820 ** target-type.lisp -> late-target-type.lisp
2821 ** early-host-format.lisp -> early-format.lisp
2822 ** late-host-format.lisp -> late-format.lisp
2823 ** host-error.lisp -> misc-error.lisp
2824 ** early-error.lisp -> early-target-error.lisp
2825 ** late-error.lisp -> late-target-error.lisp
2826 ** host-defboot.lisp -> early-defboot.lisp
2827 ** code/misc.lisp -> code/target-misc.lisp
2828 ** code/host-misc.lisp -> code/misc.lisp
2829 ** code/numbers.lisp -> code/target-numbers.lisp
2830 ** code/early-numbers.lisp -> numbers.lisp
2831 ** early-host-type.lisp -> early-type.lisp
2832 ** late-host-type.lisp -> late-type.lisp
2833 ** host-typep.lisp -> typep.lisp
2834 ** load.lisp -> target-load.lisp
2835 ** host-load.lisp -> load.lisp
2836 ** host-disassem.lisp -> disassem.lisp
2837 ** host-insts.lisp -> insts.lisp
2838 ** byte-comp.lisp -> target-byte-comp.lisp
2839 ** host-byte-comp.lisp -> byte-comp.lisp
2840 ** host-signal.lisp -> signal.lisp
2841 ** host-defstruct.lisp -> defstruct.lisp
2842 ** late-target-type.lisp -> deftypes-for-target.lisp
2843 Furthermore, several other previously target-only files foo.lisp (e.g.
2844 hash-table.lisp and random.lisp) have been split into a target-and-host
2845 foo.lisp file and a target-only target-foo.lisp file, with their key type
2846 definitions in the target-and-host part, so that the cross-compiler will
2847 know more about target types.
2848 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2849 gone away. In their place are various *BACKEND-FOO* variables
2850 corresponding to the slots of the old structure.
2851 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2852 into the target SBCL has been fixed.
2853 * The chill.lisp system for loading cold code into a running SBCL
2855 * Support for the CMU CL "scavenger hook" extension has been removed.
2856 (It was undocumented and unused in the CMU CL sources that SBCL was
2857 derived from, and stale in sbcl-0.6.1.)
2858 * Various errors in the cross-compiler type system were detected
2859 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2860 (enabling various consistency checks). Many of them were fixed,
2861 but some hard problems remain, so the compiler is back to
2862 running without *TYPE-SYSTEM-INITIALIZED* for now.
2863 * As part of the cross-compiler type system cleanup, I implemented
2864 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2865 * I have started adding UNCROSS calls throughout the type system
2866 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2867 on cross-compiler input files will be able to go away, and various
2869 * CONSTANTP now returns true for quoted forms (as explicitly required
2872 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2874 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2875 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2876 (MAKE-LIST -1). In the longer term, it ought to become true
2877 that declarations are assertions even at SAFETY 1. For now, it's not
2878 quite true even at SAFETY 3, but it's at least more nearly true..
2879 (Note that this change seems to increases the size of the system by
2880 O(5%) and to decrease the speed of the compiler by 20% or more.)
2881 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2882 for the problem of printing dozens of lines of distracting information
2883 about low-level system machinery as part of the top stack frame
2884 on entry to the debugger when an undefined function was called.
2885 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2886 is not reset to COMMON-LISP-USER.
2887 * Compilation of stuff related to dyncount.lisp has been made conditional
2888 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2889 smaller. The various dyncount-related symbols have been moved into
2890 a new "SB-DYNCOUNT" package.
2891 * tty-inspect.lisp has been renamed to inspect.lisp.
2892 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2893 feature has gone away. (When we eventually port to other flavors of
2894 libc and/or Unix, we'll try to make the differences between flavors
2895 invisible at the user level.)
2896 * Various other *FEATURES* tags, and/or their associated conditionals,
2897 have been removed if obsolescent, or given better documentation, or
2898 sometimes given more-mnemonic names.
2900 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2902 * tidied up "make.sh" script
2903 * tidied up system directory structure
2904 * better "clean.sh" behavior
2905 * added doc/FOR-CMUCL-DEVELOPERS
2906 * many many small tweaks to output format, e.g. removing possibly-confusing
2907 trailing #\. character in DESCRIBE-INSTANCE
2908 * (EQUALP #\A 'A) no longer signals an error.
2909 * new hashing code, including EQUALP hashing
2910 * tidied up Lisp initialization and toplevel
2911 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2912 * command line argument processing
2913 * added POSIX-GETENV function to deal with Unix-ish environment variables
2914 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2915 terminating SBCL on EOF
2916 * non-verbose GC by default
2917 * There is no more "sbcl" shell script; the sbcl file is now the C
2918 runtime executable (just like CMU CL).
2919 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2921 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2922 symbol and package information as Lisp native symbol and package objects
2923 instead of strings naming symbols and strings naming packages. This way,
2924 whenever packages are renamed (as in warm init), debug information is
2925 transformed along with everything else.
2926 * tweaked the optimization policy declarations which control the building
2927 of SBCL itself. Now, among other things, the system no longer saves
2928 source location debugging information. (This helps two problems at once
2929 by reducing SBCL size and by keeping SBCL from trying to look for its
2930 sources -- which may not exist -- when reporting errors.)
2931 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2932 debugging and testing purposes
2933 * cleaned up printing, making the printer call PRINT-OBJECT for
2934 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2935 methods, giving nearly-ANSI behavior
2936 * converted almost all special variables to use *FOO* naming convention
2937 * deleted PARSE-TIME functionality, since it can be done portably
2938 * moved some files out of cold init into warm init
2939 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2941 * regularized formatting of source files
2942 * added an install.sh script
2943 * fixed ridiculous memory usage of cross-compiler by making
2944 compiler/alloc.lisp not try to do pooling unless it can hook
2945 itself into the GC of the cross-compilation host. Now the system
2946 builds nicely on my old laptop.
2947 * added :SB-ALLOC in target-features.lisp-expr
2948 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2949 was not implemented)
2950 * re-did condition handling and note reporting in the compiler. Notes
2951 are no longer handled by signalling conditions. Style warnings
2952 and warnings are handled more correctly and reported in such a way
2953 that it's easy to find one or the other in your output (so that you
2954 can e.g. figure out which of many problems caused COMPILE-FILE to
2956 * changed the severity of several compiler warnings from full WARNING
2957 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2958 compiler note reporting so that it doesn't use the condition system
2959 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2960 COMPILE-FILE command)
2961 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2962 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2963 no longer supported, since their implementation was deeply intertwingled
2964 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2965 * removed bogus "support" for compiler macros named (SETF FOO), and
2966 removed the compiler macro for SETF INFO (but only after making a fool
2967 of myself on the cmucl-imp mailing list by posting a bogus patch for
2968 DEFINE-COMPILER-MACRO..)
2969 * Compiled files containing forms which have side effects on the Lisp
2970 reader (such as DEFPACKAGE forms) are now handled more correctly.
2971 (Compiler queuing of top level lambdas has been suppressed by setting
2972 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2973 * deleted various currently-unused source files, e.g. gengc.lisp. They
2974 may be added back at some point e.g. when porting to other architectures,
2975 but until they are it's distracting to distribute them and to try to
2977 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2978 there were so many of them they're just distractions, and UNCROSS is
2979 known to be able to handle the current sources
2980 * moved PROFILE functionality into TRACE, so that it will be clear
2981 how the wrapping and unwrapping of functions when you profile them
2982 interacts with the wrapping and unwrapping of functions when you
2983 trace them. (Actually, the functionality isn't there yet, but at least
2984 the interface specification is there. Hopefully, the functionality will
2985 arrive with some future maintenance release.)
2986 * removed host-oops.lisp
2987 * changed signature of QUIT function to allow UNIX-CODE argument
2988 * fixed READ-SEQUENCE bug
2989 * tweaked verbose GC output so that it looks more like the progress
2990 output that ANSI specifies for functions like LOAD
2991 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
2992 * added <http://sbcl.sourceforge.com> to the banner information printed by