1 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2 * internal entry point details and argument counts no longer appear
3 in backtraces unless explicitly requested by setting
4 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
5 * built-in and standard functions no longer have names like "top
6 level local call to FOO".
7 * fixed bug 32: functions defined in non-null lexical environments
8 now have more legible printed representation
9 * fixed bug 33: functions defined in non-null lexical environemnts
10 are now more amenable to inspection by INSPECT.
11 * workaround for bug 354: XEPs no longer appear in backtraces unless
13 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
14 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
15 (reported by Lutz Euler)
16 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
17 compiled code. (reported by Paul Dietz)
18 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
19 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
20 STRING-STREAM and FILE-STREAM, respectively.
22 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
23 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
24 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
26 * fixed loading of multiply forward-referenced layouts.
27 (thanks to Cheuksan Wang)
28 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
29 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
30 (see COMPILE-FILE documentation for details of :PRINT option)
31 * fixed bugs 19 and 317: fixed-format floating point printing is
32 more accurate. This also fixes a bug reported by Adam Warner
33 related to the ~@F format directive.
34 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
36 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
37 dispatch table if the from-char is a dispatch macro character.
38 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
39 by the wordsize no longer ignore the last word. (reported by Lutz
41 * fixed bug in type checking of dynamic-extent variables. (reported
43 * optimization: sequence traversal functions use their freedom to
44 coerce function designators to functions.
45 * optimization: code with many calls to CLOS methods specialized on
46 CLOS classes has had redundant type checks removed.
47 * optimization: added declarations to speed up operations that access
48 the internal character database (for example STRING-UPCASE)
49 * optimization: comparison operations between floats and sufficiently small
50 fixnums no longer create extra rationals
51 * fixed some bugs related to Unicode integration:
52 ** portions of multibyte characters at the end of buffers for
53 character-based file input are correctly transferred to the
54 start of the buffer at the next read.
55 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
56 passing it through to OPEN.
57 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
58 argument given to internal calls to OPEN.
59 ** the built-in comment readers (introduced by character sequences
60 ";" and "#|") are more forgiving to encoding errors; they will
61 STYLE-WARN and then attempt to resync the stream at a character
62 boundary. (thanks to Teemu Kalvas)
63 * fixed some bugs in the x86-64 port:
64 ** Negative short int return values from c-calls are sign-extended
66 ** The stack is aligned to 16-bytes for c-calls, as required by
67 the ABI. (thanks to Cheuksan Wang)
68 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
69 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
70 for enums. (thanks to Vincent Arkesteijn)
71 ** Multiple small optimizations and bugfixes for floating point
73 * fixed some bugs revealed by Paul Dietz' test suite:
74 ** Space, Tab, Linefeed, Return and Page have the invalid
75 secondary constituent character trait.
76 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
78 ** WITH-INPUT-FROM-STRING should only update the index place on
80 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
82 ** Bit-array operations (BIT-AND and similar) worked incorrectly
83 with one-dimensional arrays with fill pointers.
84 ** TYPE-OF failed on a complex with an integer realpart and a
86 ** compiler failure during type inference for the code of form
87 (IF (EQL X (THE ...)) ...) (MISC.535).
89 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
90 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
91 platform. The port supports 61-bit fixnums, large memory spaces
92 and reloading shared object files.
93 * enhancement: saving cores with foreign code loaded is now
94 supported on ppc/Darwin in addition to the previously supported
96 * enhancement: the statistical profiler now walks deeper into the
97 call stack for more meaningful call-graphs and accrued time
98 reports (x86/x86-64 only). It also now reports time spent in
100 * enhancement: it is now possible to trace most individual methods
101 of a generic function in addition to tracing the generic function
103 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
105 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
106 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
107 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
108 produces an error. (thanks to Vincent Arkesteijn)
109 * bug fix: NAMESTRING on pathnames with :WILD components in their
110 directories works correctly. (thanks to Artem V. Andreev)
111 * fixed bug 125: compiler preserves identity of closures. (reported
113 * bug fixed: functions with &REST arguments sometimes failed with
114 "Undefined function" when compiled with (DEBUG 3). (reported by
116 * bug fix: overflow during compiling of setting element of a bit
117 vector with constant index and value. (reported by Timmy Douglas)
118 * build fix: fixed the dependence on *LOAD-PATHNAME* and
119 *COMPILE-FILE-PATHNAME* being absolute pathnames.
120 * on x86 compiler partially supports stack allocation of dynamic-extent
122 * GO and RETURN-FROM do not check the extent of their exit points
123 when compiled with SAFETY 0.
124 * fixed some bugs related to Unicode integration:
125 ** encoding and decoding errors are now much more robustly
126 handled; it should now be possible to recover even from invalid
127 input or output to the terminal. (thanks to Teemu Kalvas)
128 ** provided a first cut at implementing STRING-TO-OCTETS and
129 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
130 ** altered the SB-MD5 contributed module slightly, changing the
131 interface just enough for it to be supportable for builds where
132 lisp characters are not eight bits.
133 * fixed some bugs revealed by Paul Dietz' test suite:
134 ** the FORMATTER-generated functions for ~V[ conditionals require
135 the correct number of arguments.
136 ** READ-FROM-STRING returns the mandated second value when applied
137 to displaced strings.
138 ** the #\Rubout and #\Backspace characters are treated as invalid
139 constituent characters by the tokenizer.
141 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
142 * new feature: reloading changed shared object files with
143 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
144 * new feature: references to foreign variables and functions
145 can now be compiled and loaded before the corresponding shared
146 object file is loaded, as long as the foreign definitions are
147 available at runtime.
148 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
149 supported. (thanks to Dan Debertin)
150 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
151 just keywords. (thanks to Vincent Arkesteijn)
152 * fixed bug #331: structure-class instances corresponding to
153 DEFSTRUCT forms are now created eagerly.
154 * fixed bug #345: backtraces from calls to undefined functions work
155 on x86 as well. Related bug #61 is now also partially fixed on x86
156 (backtraces from throws to unknown catch tags.)
157 * bug fix: lambda-list parsing is now stricter vrt. order and number
158 of lambda-list keywords.
159 * bug fix: as specified by AMOP, an error is signalled if a
160 class-option appears multiple times in a DEFCLASS form. (reported
162 * bug fix: RANDOM can be compiled when the compiler derives the type
163 of its numeric argument as a disjoint set of small integers.
164 (reported by Paul Dietz)
165 * bug fix: starting a core saved with shared objects loaded when
166 those objects are not available no longer causes threaded SBCL to
167 hang. (reported by Sean Ross)
168 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
170 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
171 argument, which must always be NIL. (reported by Kalle Niemitalo)
172 * bug fix: printing 1.0d+23 no longer results in an error.
173 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
174 * bug fix: structure slot setters preserve evaluation order. (thanks
176 * fixed some bugs related to Unicode integration:
177 ** RUN-PROGRAM can allow its child to take input from a Lisp
178 stream. (reported by Stefan Scholl)
179 ** ASDF-INSTALL successfully downloads .tar.gz files in all
180 locales. (reported by Ken Causey)
181 * fixed some bugs revealed by Paul Dietz' test suite:
182 ** INCF, DECF and REMF evaluate their place form as specified in
184 ** FORMATTER expands ~{ ~} iteration directives with V or #
185 parameters correctly.
186 ** FORMATTER deals with the ~@[ ~] conditional directive where the
187 consequent uses no arguments correctly.
188 ** the system has a partial understanding of the (COMPLEX RATIO)
190 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
193 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
194 * new feature: a build-time option (controlled by the :SB-UNICODE
195 keyword feature, enabled by default) for building the system with
196 support for the entire 21-bit character space defined by the
198 * new feature: the system now has rudimentary external-format
199 support; the primary user-visible change at this time is that
200 characters with the high bit set (such as lower-case-e-acute) will
201 print correctly to a terminal in a UTF-8 environment.
202 * minor incompatible change: BASE-CHAR no longer names a class;
203 however, CHARACTER continues to do so, as required by ANSI.
204 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
205 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
206 has been moved to the SB-EXT package (temporarily re-exported from
208 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
209 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
211 * fixed bug #351: better error-handlind and reporting for malformed
213 * fixed bug #350: bignum-printing is now more memory-efficient,
214 allowing printing of very large bignums, eg. (expt 2 10000000).
215 (reported by Bruno Haible)
216 * fixed bug #302: better primitive-type selection for intersection
218 * fixed bug #308: non-graphic characters now all have names, as
219 required. (reported by Bruno Haible)
220 * bug fix: redefining a class with different superclasses now correctly
221 removes it from the direct-subclasses of its previous superclasses.
222 (reported by David Morse)
223 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
224 new value now works. (reported by Bruno Haible)
225 * bug fix: correct canonicalization of multiple non-standard slot
226 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
227 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
228 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
229 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
230 now exists, an signals an error.
231 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
232 during macroexpansion for non-iteration variables as well. (reported
233 by Bruno Haible for CMUCL)
234 * bug fix: Cyclic structures and unprintable objects in compiler
235 messages no longer cause errors. (reported by Bruno Haible)
236 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
237 and READ-FROM-STRING all now return a primary value of NIL if
238 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
239 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
240 (reported by Bruno Haible for CMUCL)
241 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
242 do not designate a condition. (reported by Bruno Haible for
244 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
245 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
246 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
247 permitting resolution in favour of any of the conflicting symbols.
248 (reported by Bruno Haible for CMUCL)
249 * bug fix: EQUAL compiler optimizations is less aggressive on
250 strings which can potentially compare true despite having distinct
251 specialized array element types.
252 * bug fix: unit enumerations can be defined without dividing by
253 zero. (thanks to Vincent Arkesteijn)
254 * FORMAT compile-time argument count checking has been enhanced.
255 (report from Bruno Haible for CMUCL)
256 * a partial workaround for the bug 262: the compiler does not try to
257 inline-expand a local function doing RETURN-FROM from a deleted
258 BLOCK. (thanks to Peter Denno for the bug report and to David
259 Wragg for the simple test case)
260 * fixed some bugs revealed by Paul Dietz' test suite:
261 ** RENAME-PACKAGE allows all package designators as new package
263 ** constraint propagation and lambda variable substitution are
264 more cautious in dealing with partially deleted code.
265 ** compiler sometimes lost reoptimization passes.
266 ** CERROR, given a condition as condition designator, passes its
267 remaining arguments to the continue format control without
269 ** Case-altering FORMAT directives work correctly on non-ASCII
271 ** The REAL type specifier handles bounds outside the floating
272 point ranges without signalling FLOATING-POINT-OVERFLOW.
273 ** Functions with IR1-transformations can create intercomponent
274 references to global functions.
275 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
277 ** FORMAT ~R treats a nil value for its first parameter correctly.
279 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
280 * enhancement: saving cores with foreign code loaded is now
281 supported on x86/NetBSD and sparc/Linux in addition to the previously
283 * bug fix: on some platforms repeated installations caused multiple
284 copies of HTML documentation to be installed -- should not happen
285 any more. (reported by Stefan Scholl)
286 * bug fix: parsing self-recursive alien record types multiple times
287 no longer causes infinite recursion. (reported by Thomas F. Burdick,
288 original patch by Helmut Eller for CMUCL)
289 * bug fix: stack-exhaustion detection works now on NetBSD as well.
290 (thanks to Richard Kreuter)
291 * bug fix: defining classes whose accessors are methods on existing
292 generic functions in other (locked) packages no longer signals
293 bogus package lock violations. (reported by François-René Rideau)
294 * bug fix: special variables as DEFMETHOD parameters no longer have
295 associated bogus type declarations. (reported by David Wragg and
297 * bug fix: read-write consistency on streams of element-type
298 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
299 * bug fix: redefiniton of the only method of a generic function with
300 no DEFGENERIC no longer emits a full WARNING. In addition,
301 redefinition of generic functions with no DEFGENERIC to an
302 incompatible lambda list now signals an error. (thanks to Zach
304 * bug fix: DEFGENERIC now works even when there's a function of the
305 same name in an enclosing lexical environment. (thanks to Zach
307 * fixed compiler failure, caused by instrumenting code during
308 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
309 * optimization: added loop analysis and improved register allocation
310 * fixed some bugs revealed by Paul Dietz' test suite:
311 ** POSITION on displaced vectors with non-zero displacement
312 returns the right answer.
313 ** (SIMPLE-STRING) is a valid type specifier for sequence
315 ** *PRINT-LEVEL* handling for slotless structures is pedantically
317 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
318 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
320 ** The justification version of the FORMAT ~< directive treats
321 non-zero minpad parameter correctly.
323 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
324 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
325 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
326 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
327 the supported interface.
328 * new feature: Single-stepping of code compiled with DEBUG 2 or
329 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
330 * new feature: saving cores with foreign code loaded is now
331 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
332 Timothy Moore's work for CMUCL)
333 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
334 parameters to * instead of NIL if no initform is supplied.
335 (reported by Johan Bockgaard)
336 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
337 correctly when FUNCALL appears as the car of the form. Note:
338 despite this FUNCALL forms are not currently subject to
339 compiler-macro expansion. (port of Raymond Toy's fix for the
340 same from CMUCL, reported by Johan Bockgaard)
341 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
342 (thanks for Teemu Kalvas)
343 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
344 case instead of terminating the iteration (thanks for Julian
345 Squires, Sean Champ and Raymond Toy)
346 * bug fix: incorrect expansion of defgeneric that caused a style
347 warning. (thanks for Zach Beane)
348 * bug fix: slot accessor effective method computation works properly
349 for classes with multiple non-standard applicable methods on
350 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
351 * on x86 compiler supports stack allocation of results of LIST and
352 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
353 CMUCL implementation by Gerd Moellmann)
354 * fixed some bugs revealed by Paul Dietz' test suite:
355 ** FORMAT strings with both the ~<~:;~> form of the justification
356 directive and pretty-printing directives cause an error.
358 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
359 * incompatible change: the internal functions
360 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
361 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
362 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
363 instead of the old functions.
364 * new feature: on platforms where "dladdr" is available foreign
365 function names now appear in backtraces. (based on Helmut Eller's
367 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
368 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
370 * documentation: ASDF manual is now distributed with SBCL.
371 * bug fix: SBCL can now load its contributed modules with REQUIRE
372 even if the system-provided entries have been removed from
373 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
375 * bug fix: backtraces involving undefined functions or assembly
376 routines are more informative. (thanks to Brian Downing)
377 * bug fix: mutually referent alien structures now work correctly.
378 (reported by Rick Taube)
379 * bug fix: structures defined by WITH-ALIEN can be referred to
380 within other definitions in the same WITH-ALIEN.
381 * bug fix: division operators (MOD, TRUNCATE and the like) with
382 constant zero divisors and integer dividends no longer generate
384 * bug fix: provide default methods for INPUT-STREAM-P and
385 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
386 * bug fix: improve the detection and resolution of MOP metacycles
387 (where effective method computation depends on the generic function
388 having its effective method computed). (reported by Bruno Haible)
389 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
390 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
391 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
392 * bug fix: pathnames with relative directory components can be
393 represented relative to default pathnames.
394 * optimization: in taking the GCD of bignums, reduce the two bignums
395 to approximately the same size (using Euclid's algorithm) before
396 applying the more sophisticated binary GCD. (thanks to Juho
398 * optimization: COUNT on bitvectors now operates word-at-a-time.
399 * optimization: ASH with a positive, but not necessarily constant,
400 (leftwards) shift, when in a modular context, is compiled to a
402 * fixed some bugs revealed by Paul Dietz' test suite:
403 ** FORMAT variable parameters ("~V<char>") are defaulted properly
404 if the corresponding argument is NIL.
405 ** FORMAT directives accepting a minpad parameter treat negative
407 ** literal commas as character parameters to FORMAT directives are
409 ** literal spaces directly after ~<Newline> directives within a
410 format-logical-block (~:< ~@:>) do not induce :FILL-style
411 conditional newlines.
412 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
413 as specified: it no longer includes conditional newlines).
414 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
416 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
417 :PER-LINE-PREFIX argument does not evaluate to a string.
419 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
420 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
421 the manual for details; for now, package locks can be disabled by
422 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
423 if no major problems are found then it is likely that they will be
424 compiled in unconditionally.
425 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
426 unsupported operators on all platforms. To load a shared library
427 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
428 object file, link it into a shared library outside of SBCL and
429 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
431 * bug workaround: ROOM T and the GC were not working together
432 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
433 reporting facility aren't preserved. That reporting has been
434 disabled, so now until and unless someone figures out how to make it
435 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
436 * minor incompatible change: as threatened around sbcl-0.8.0, the
437 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
438 The symbols are also exported from SB-PCL for backwards
439 compatibility, but more so than before SB-PCL should be treated as
440 an implementation-internal package.
441 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
443 * the SB-SPROF contrib now works on (most) non-x86 architectures.
444 It is known as of this release not to work on the Alpha, however.
445 * fixed bug #167: errors signalled due to illegal syntax in method
446 bodies are now more legible.
447 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
448 designators and can hence be used with TYPEP.
449 * fixed bug #333: CHECK-TYPE now ensures that the type error
450 signalled, if any, has the right object to be accessed by
451 TYPE-ERROR-DATUM. (reported by Tony Martinez)
452 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
453 5.1.2.3 for multiple-value place subforms. (reported by Kalle
455 * fixed bug #334: programmatic addition of slots using specialized
456 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
457 and :ALLOCATION :CLASS slots.
458 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
459 float exponents. (rereported by Peter Seibel)
460 * fixed a bug: #\Space (and other whitespace characters) are no
461 longer considered to be macro characters in standard syntax by
463 * fixed bug: initialization of condition class metaobjects no longer
464 causes an instance of the condition to be created. (reported by
466 * fixed bug: it is now possible to have slots such that
467 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
468 description is neither :INSTANCE nor :CLASS.
469 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
470 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
471 SLOT-VALUE-USING-CLASS), and no longer causes errors with
472 non-standard SLOT-DEFINITION-ALLOCATION values.
473 * fixed bugs: various race conditions handling exiting threads.
474 CL-PPCRE's thread torture test now passes.
475 * fixed arguably-a-bug: GC time proportional to number of threads in
476 system even when most of them are idle
477 * optimization: improved performance of BIT and SBIT on bit-vectors.
478 * .fasl file incompatibility: The fasl file version number has
479 been incremented because of changes associated with package locks.
481 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
482 * minor incompatible change: the system no longer provides
483 optimization hints (conditions which are TYPEP
484 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
485 mode; these hints are still emitted when higher SPEED optimization
487 * new contrib module: a sampling profiler (profiling by statistical
488 sampling, rather than by instrumenting functions) is available as
489 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
490 * the behaviour of the standard function ED is now customizeable by
491 third parties through a hook variable: see ED's documentation
492 string for information on the protocol.
493 * the compiler no longer emits efficiency notes for (FUNCALL X)
494 when the type of X is uncertain under default optimization
496 * fixed bug 276: mutating a binding of a specialized parameter to a
497 method to something that is not TYPEP the specializer is now
499 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
500 exhibit the required behaviour on the x86 platform. (thanks to
501 Peter van Eynde, Eric Marsden and Bruno Haible)
502 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
503 even for difficult arguments. (reported by Peter Graves)
504 * fixed bug 141a: the backquote printer now descends quoted
506 * fixed another bug in backquote printing: no more destructive
507 modification of the form's list structure. (reported by Brian
509 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
510 move between its address being taken and the call to
511 interrupt_thread, fixing a crashing race condition.
512 * the SB-POSIX contrib implementation has been adjusted so that it
513 no longer exhibits ridiculously poor performance when constructing
514 instances corresponding to C structs.
516 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
517 * minor incompatible change: the sb-grovel contrib now treats C
518 structures as alien (in the sense of SB-ALIEN) objects rather than
519 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
520 has implications for memory management of client code
521 (sb-grovel-returned objects must now be manually managed) and for
522 type safety (alien objects now have full types).
523 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
524 used to control emission of compiler diagnostics, rather than the
525 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
526 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
527 quality should be considered deprecated.
528 * install.sh now installs the user manual as well
529 * (not quite a new documentable feature, but worth considering in
530 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
531 beginnings of a semantically meaningful condition hierarchy is
532 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
534 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
535 designator as the defaults argument.
536 * fixed bug: Displaced arrays whose displaced-to array has become
537 too small now cause ARRAY-DIMENSION to signal an error, providing
538 for safer bounds-checking. (reported by Bruno Haible)
539 * fixed bug: DEFCLASS slot definitions with identical :READER and
540 :WRITER names now signal a reasonable error. (reported by Thomas
542 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
544 * fixed bug: Invalid slot specification errors now print correctly.
545 (thanks to Zach Beane)
546 * fixed bug 320: Shared to local slot value transfers in class
547 redefinitions now happen corrently from superclasses as
548 well. (reported by Bruno Haible)
549 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
551 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
552 improper lists correctly. (reported by Bruno Haible)
553 * fixed bug 313: source-transform for <fun-name> was erroneously
554 applied to a call of a value of a variable with name <fun-name>.
555 (reported by Antonio Menezes Leitao)
556 * fixed bug 307: The obsolete instance protocol ensures that
557 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
558 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
559 remove dead unknown-values globs from the middle of the stack.
560 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
561 easier to resume long computations after using *BREAK-ON-SIGNALS*
562 to diagnose and fix failures. (thanks to Nikodemus Siivola)
563 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
564 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
565 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
566 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
567 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
569 * fixed a (fairly theoretical) bug in string printing: if
570 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
571 does not have array-element-type equal to the most general string
573 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
574 function name. (thanks to Nikodemus Siivola)
575 * fixed bug: compile-time format string checker failed on
576 non-closed ~{. (reported by Thomas F Burdick)
577 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
578 don't warn on legal loop code involving more than one aggregate
579 boolean. (thanks to Nikodemus Siivola)
580 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
581 in keyword position to MAKE-PACKAGE was not being reported as
583 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
584 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
586 * fixed bugs 280 and 312: the checking for multiple definitions in a
587 file is less likely to become confused by uses of inline
589 * fixed bug: the #S reader macro performs the keyword coercion
590 specified for slot names. (reported by Kalle Niemitalo)
591 * fixed bug: lambda lists may contain symbols whose names start with
592 & but are not lambda-list-keywords; their occurrence triggers a
594 * fixed bug 321: define-method-combination argument lambda lists do
595 not cause invalid code to be generated when &optional or &aux
596 variables have default values. (reported by Bruno Haible)
597 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
598 have CLOS classes; this appears also to have fixed reported
599 instability in stack exhaustion detection.
600 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
601 MUFFLE-WARNING when no associated restart is present is now
603 * optimization: rearranged the expansion of various defining macros
604 so that each expands into only one top-level form in a
605 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
607 * optimization: used a previously little-used slot in symbols to
608 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
610 * fixed some bugs revealed by Paul Dietz' test suite:
611 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
613 ** Fixed an optimization bug related to inheritance of initforms
614 from local to shared slots.
615 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
616 ** CAST optimizer forgot to flush argument derived type.
617 ** print/read consistency on floats is now orders of magnitude
618 more likely. (thanks also to Bruno Haible for a similar report
620 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
621 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
622 for objects of type REAL. Make it so.
623 ** SXHASH is specified (infelicitously) to respect similarity,
624 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
625 it so. (thanks to Markus Ziegler)
626 ** on the Alpha, the compiler succeeds in compiling functions
627 returning a known number of arguments greater than 63.
628 ** fixed handling of invalid NIL arguments in keyword position
630 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
632 ** provided a readably-printable representation for RANDOM-STATE
634 ** ensured that pathnames, where they have a namestring, always
635 print using #P"..." syntax.
637 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
638 * Support for the forthcoming 2.0 version of the NetBSD kernel
639 running on x86 hardware has been added. (thanks to Perry
640 E. Metzger most immediately, and others for their past work)
641 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
642 current release version; SBCL's OpenBSD support had been broken
643 since about the time of OpenBSD's switch to ELF binary format.)
644 * [placeholder for DX summary]
645 ** user code with &REST lists declared dynamic-extent, under high
646 speed or space and low safety and debug optimization policy.
647 * The manual has been converted to Texinfo format and the debugger
648 chapter from the cmucl manual has been added.
649 * A facility has been added to extract documentation strings from
650 sbcl and store them as Texinfo-formatted snippets for inclusion in
651 the manual (via Texinfo's @include directive)
652 * bug fix: compiler emitted division in optimized DEREF. (thanks for
653 the test case to Dave Roberts)
654 * bug fix: multidimensional simple arrays loaded from FASLs had fill
655 pointers. (reported by Sean Ross)
656 * bug fix: PROFILE output is printed nicely even for large numerical
657 values. (thanks to Zach Beane)
658 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
659 greater than 32 handle EOF correctly.
660 * bug fix: on X86 an immediate argument of the IMUL instruction is
661 correctly printed in disassembly. (thanks to Lutz Euler)
662 * bug fix: class slots in redefined classes preserve their old
663 values. (thanks to Bruno Haible and Nikodemus Siivola)
664 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
665 succeeds. (reported by Marco Baringer)
666 * fixed some bugs revealed by Paul Dietz' test suite:
667 ** READ-SEQUENCE now works on ECHO-STREAMs.
668 ** RATIONALIZE works more according to its specification. (thanks
671 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
672 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
673 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
674 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
675 should matter to you only if you rebind the printer control
676 variables and then find you want different bindings in the
677 debugger than in the ordinary execution of your program.)
678 * The runtime build system has been tweaked to support building
679 (on SPARC/SunOS) using a C compiler which invokes Sun's own
680 assembler and linker. (thanks to Nikodemus Siivola)
681 * Unbound, undefined, undeclared variables now trigger full
682 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
683 more often programmer error than deliberate exploitation of undefined
685 * optimization: the hash algorithm for strings has changed to one
686 that is less vulnerable to spurious collisions. (thanks to Juho
688 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
689 needless bounds checking. (thanks to Juho Snellman)
690 * optimization: implemented multiplication as a modular
691 (UNSIGNED-BYTE 32) operation on the PPC backend.
692 * fixed some bugs revealed by Paul Dietz' test suite:
693 ** ADJUST-ARRAY now copies the datum in a zero rank array if
695 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
696 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
697 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
699 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
701 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
703 ** ECHO-STREAMs no longer attempt to echo the end of file value to
704 their output stream on EOF from read.
705 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
706 have been read to end-of-file.
707 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
709 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
710 description of determination of which consecutive characters
712 ** Printing the "Space" character with escaping on now yields "#\\ ",
713 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
714 ** Reading floating-point numbers with *READ-BASE* set to a number
715 less than 10 works correctly.
716 ** Reading floating-point numbers with *READ-BASE* set to a number
717 more than 10 works correctly.
718 ** Printing with *PRINT-READABLY* targets the standard readtable, not
719 the readtable currently in effect.
721 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
722 * minor incompatible change: parsing of namestrings on a physical
723 (Unix) host has changed; numbers after the final #\. in a
724 namestring are no longer interpreted as a version field. This is
725 intented to be largely invisible to the user, except that the
726 meaning of the namestring "*.*.*" has changed: it now refers to a
727 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
728 should usually be replaced by
729 (make-pathname :name :wild :type :wild :version :wild)
730 with the added benefit that this is more likely to be portable.
731 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
732 option to OPEN now signals an error if the file being opened
733 exists; this may have an impact on existing code.
734 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
735 signals received, on the PowerPC platforms (both Linux and
736 Darwin). (thanks to Pierre Mai for pointing out the location of
738 * several fixes on OS X: The system now builds and runs cleanly on
739 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
740 (thanks to Brian Mastenbrook)
741 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
742 second-resolution: integer multiples of 1/3600 between -24 and 24.
743 (thanks to Vincent Arkesteijn)
744 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
745 type when called with 1 argument; PEEK-CHAR checked type of
746 PEEK-TYPE only after having read first character from a
747 stream. (reported by Peter Graves)
748 * bug fix: the garbage collector now has much better locality
749 behaviour, and in particular no longer treats all memory as being
750 exhausted when many small objects point to each other in a deeply
752 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
753 recognized as being TYPEP their class.
754 * bug fix: the PUSHNEW documentation string has been corrected.
755 (thanks to Vincent Arkesteijn)
756 * bug fix: defaulting of the value for the last of an atypically
757 large number of multiple values being bound was not being
758 performed correctly on the Alpha or PPC platforms
759 * optimization: implemented multiplication as a modular
760 (UNSIGNED-BYTE 32) operation on the x86 backend.
761 * optimization: SEARCH on simple-base-strings can now be open-coded.
762 (see also contrib/compiler-extras.lisp for inspiration for
763 teaching the compiler about the Boyer-Moore algorithm).
764 * value, returned by MAX (and MIN) called with several EQUALP, but
765 not EQL, arguments now does not depend on compiler settings.
766 * fixed some bugs revealed by Paul Dietz' test suite:
767 ** in stack analysis liveness information is propagated from
768 non-local entry points.
769 ** pathwise CAST removing failed when the CAST node did not start
771 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
772 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
774 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
775 host is already defined.
776 ** RENAME-FILE works on streams instead of signalling an internal
778 ** PEEK-CHAR uses the current readtable when determining whether
779 or not a character is whitespace.
780 ** MERGE-PATHNAMES handles the case when the pathname does not
781 specify a name while the default-pathname specifies a version.
782 ** Pathnames now stand a better chance of respecting print/read
784 ** Attempting to use standardized file system operators with a
785 pathname with invalid :DIRECTORY components signals a
787 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
789 ** DIRECTORY on logical pathnames is more correct.
790 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
791 signal a TYPE-ERROR if their argument is not a stream.
792 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
793 designator argument does not designate a stream.
794 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
795 examining the synonym.
796 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
798 ** OPEN and WITH-OPEN-STREAM allow opening streams with
799 element-type larger than ([UN]SIGNED-BYTE 32).
801 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
802 * When built with the :SB-FUTEX feature, threaded builds now take
803 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
804 for faster/more reliable mutex and condition variable support.
805 * Incompatible change (but one you probably shouldn't have been using
806 anyway): the interface and code for arbitrating between multiple
807 threads in the same user session has been redesigned.
808 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
809 error when called without an explicit environment argument.
810 (thanks to Nikodemus Siivola)
811 * bug fix: buffered :DIRECTION :IO streams are less likely to become
812 confused about their position. (thanks to Adam Warner and Gerd
814 * bug fix: Pretty printing backquoted forms with unquotations in the
815 argument list position of various code constructs such as LAMBDA
816 now works correctly. (reported by Paul Dietz)
817 * bug fix: Pretty printing unquotations no longer loses all
818 stream position information.
819 * optimization: performance of string output streams is now less
820 poor for multiple small sequence writes.
821 * optimization: performance of CSUBTYPEP in the presence of complex
822 expressions involving CONS and NOT many times has been improved.
823 (reported by Paul Dietz)
824 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
826 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
827 streams for the REPL, for the debugger, and so on. (thanks to
829 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
830 * fixed some bugs revealed by Paul Dietz' test suite:
831 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
832 is ignored if it is too small, rather than propagating through
833 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
834 ** extremely complex negations of CONS types were not being
835 sufficiently canonicalized, leading to inconsistencies in
837 ** VALUES tranformer lost derived type.
839 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
840 * fixed a bootstrapping bug: the build process no longer assumes
841 that the various BOOLE-related constants have the same value in
842 host and target lisps. (noted by Paul Dietz' test suite on an
843 SBCL binary built from CLISP)
844 * The system can now be dynamically linked on the MIPS platform,
845 which enables dynamic loading of foreign code from Lisp. (thanks
846 to Ralf Baechle for discussions on the MIPS ABI)
847 * The system now records debugging information for its own source
848 files in a filesystem-position-independent manner, relative to
849 the "SYS" logical host.
850 * fixed a compiler bug: MV-LET convertion did not check references
851 to the "max args" entry point. (reported by Brian Downing)
852 * tweaked disassembly notes to be less confident about proclaiming
853 some instruction as an LRA. (thanks to Brian Downing)
854 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
855 now each have their own history, command character, and other
856 characteristics. (thanks to David Lichteblau)
857 * fixed some bugs revealed by Paul Dietz' test suite:
858 ** compiler failure in compiling LOGAND expressions including a
860 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
861 shift greater than 32.
862 ** FLUSH-DEST did not mark blocks for type check regeneration.
863 ** HANDLER-CASE failed to accept declarations in handler clauses
864 in some circumstances.
866 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
867 * New code in contrib/sb-introspect (still probably not entirely
868 stable yet) provides some support for smart Lisp development
869 environments like SLIME.
870 * The conditions signalled for errors occurring when loading .fasl
871 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
872 in a way which should help ASDF recover gracefully.
873 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
874 stuff has been cleaned up. If you code contrib/ stuff, this might
875 affect you, and you can look at contrib/README, contrib/STANDARDS,
876 and/or the 0.8.4.27 diff to check.
877 * In full calls the compiler now does not generate checks for declared
878 argument types for all arguments.
879 * various threading fixes
880 ** and some experimental patches which didn't make it into
881 the main tree for this release, but which are shipped in
882 contrib/experimental-thread.patch as a possible fix for some
883 failures (deadlock, spinning...) in GC-intensive multithreaded
885 * fixed PPC build problem (source code incompatibility of different
886 library versions): added offsetof() hackery which attempts to divine
887 where glibc maintainers put uc_mcontext today
888 * fixed bug 282: compiler does not trust type assertions while passing
889 arguments to a full call.
890 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
891 &OPTIONAL VALUES type parameter.
892 * fix bug 214: algorithm for noting rejected templates is now more
893 similar to that of template seletion. (also reported by rydis on
895 * fixed bug 141b: printing backquoted information readably and prettily
896 inserts a space where necessary.
897 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
898 (CHARACTER 10) are now reported as errors, rather than propagated
899 as unknown types. (reported by piso on #lisp)
900 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
901 with values NIL and :ERROR. (thanks to Milan Zamazal)
902 * fixed bug 191c: CLOS now does proper keyword argument checking as
903 described in CLHS 7.6.5 and 7.6.5.1.
904 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
905 longer raise an error; further, using a list as a for-as-arithmetic
906 counter now raises a meaningful error.
907 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
908 now understood by sequence creation functions such as MAKE-SEQUENCE
910 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
911 a STRING-INPUT-STREAM.
912 * compiler enhancement: SIGNUM is now better able to derive the type
914 * type declarations inside WITH-SLOTS are checked. (reported by
916 * fixed some bugs revealed by Paul Dietz' test suite:
917 ** incorrect optimization of TRUNCATE for a positive first
918 argument and negative second.
919 ** compiler failure in let-convertion during flushing dead code.
920 ** compiler failure while deriving type of TRUNCATE on an
921 interval, containing 0.
922 ** ASH of a negative bignum by a negative bignum count now returns
924 ** intersection of CONS types now canonicalizes properly, fixing
925 inconsistencies in SUBTYPEP.
927 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
928 * incompatible change: The --disable-debugger command line
929 option now clobbers the debugger at a more fundamental
930 level, by redefining #'INVOKE-DEBUGGER instead of by
931 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
932 is specified by ANSI to ignore *DEBUGGER-HOOK* and
933 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
934 enter the debugger REPL and then suffer recursive errors
935 because *DEBUG-IO* is also messed up in --disable-debugger mode;
936 while under the new system, BREAK in --disable-debugger mode
937 terminates the system just as an unhandled error would.
938 * fixed compiler performance when processing loops with a step >1;
939 * bug fix: DOCUMENTATION now retrieves generic function
940 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
941 support has been systematized, and now supports the methods
942 specified by ANSI, along with a default method and a method for
943 slot documentation. (reported by Nathan Froyd)
944 * bug fix: effective methods associated with a generic function are
945 no longer cached over a change of that generic function's method
946 combination. (reported by Andreas Fuchs)
947 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
948 names a non-existent directory. (thanks to Andreas Fuchs)
949 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
950 the PPC platform, lead to stack corruption; this has been fixed.
951 (reported by Rainer Joswig)
952 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
953 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
954 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
955 arithmetic in general, is now much more reliable on the PPC
957 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
958 platform now returns the right answer.
959 * optimization: restored some effective method precomputation in
960 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
961 precomputation is now tunable.
962 * optimization: compiler-internal data structure use has been
963 reviewed, and changes have been made that should improve the
964 performance of the compiler by about 20%.
965 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
966 simple-base-strings and simple-bit-vectors is improved.
967 * optimization: the optimization of 32-bit logical and arithmetic
968 functions introduced in version 0.8.3 on the x86 has been
969 implemented on the mips, ppc and sparc platforms; an
970 implementation of the same facility, but for 64-bit arithmetic,
971 has been added for the alpha.
972 * microoptimization: the compiler is better able to make use of the
973 x86 LEA instruction for multiplication by constants.
974 * bug fix: in some situations compiler did not report usage of
975 generic arithmetic in (SPEED 3) policy.
976 * bug 145b fix: compiler used wrong type specifier while converting
977 MEMBER-types to numeric.
978 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
980 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
981 index argument now works properly on the Alpha platform.
982 * bug fix: floating point exception treatment on the Alpha platform
984 * bug fix: FILE-POSITION works much better on string input and
985 output streams. (thanks to Nikodemus Siivola)
986 * bug fix: many threading/garbage collection symptoms sorted.
987 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
988 might be pseudo-atomic.
989 * internal change: Stopping for GC is now done with signals not
990 ptrace. GC is now done in whichever thread wanted it, instead of
992 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
994 * bug fix: result form in DO is not contained in the implicit
996 * incompatible change: ICR structure is changed; the value part of
997 CONTINUATION is now called LVAR; corresponding functions are
998 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
999 * added type deriver for ISQRT (thanks to Robert E. Brown).
1000 * bug fix: better support for loading from the command line when an
1001 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1003 * fixed some bugs revealed by Paul Dietz' test suite:
1004 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1005 ** ROUND and FROUND now give the right answer when given very
1006 small float arguments.
1007 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1009 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1010 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1011 ** LCM should return a non-negative integer.
1012 ** PARSE-INTEGER returned the index of a terminator instead of the
1013 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1014 ** PARSE-INTEGER returned an incorrect index being applied to a
1016 ** LCM with two arguments of 0 returns 0 rather than signalling
1018 ** unsigned addition of a 32-bit constant with the high bit set no
1019 longer causes an internal compiler error on the x86.
1020 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1021 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1022 with negative last argument.
1023 ** byte specifiers with zero size and position no longer cause
1024 an error during type derivation.
1025 ** bignum multiplication on the Alpha platform now returns the
1027 * porting: The system now builds on SuSE AMD64, although it still
1028 generates a 32-bit binary.
1029 * .fasl file incompatibility: The fasl file version number has
1030 been incremented (because of the changes to internal compiler
1031 data structures referred to above).
1033 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1034 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1035 more accurately, on the Darwin kernel running on PowerPC hardware.
1036 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1037 * Compiler code deletion notes now signal a condition of type
1038 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1039 an associated MUFFLE-WARNING restart.
1040 * The compiler now performs limited argument count validation of
1041 constant format strings in FORMAT, and where appropriate in ERROR,
1042 CERROR and WARN. (thanks to Gerd Moellmann)
1043 * New ASDF-INSTALL contrib can be used for automatic download and
1044 installation of third-party Lisp code from CCLAN or other sites
1046 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1047 which forces another thread to execute a function supplied by the
1049 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1050 now accept and act upon their :ELEMENT-TYPE keyword argument.
1051 (reported by Martin Atzmueller, Edi Weitz)
1052 * bug fix: FILE-POSITION now accepts position designators up to
1053 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1054 is the greater. (thanks to Patrik Nordebo)
1055 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1056 Kalle Olavi Niemitalo)
1057 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1058 printable. (reported by Eric Marsden)
1059 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1060 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1061 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1062 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1063 (thanks to Stig Sandoe)
1064 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1065 UNREAD-CHAR now correctly includes the unread character in the
1066 target sequence. (thanks to Gerd Moellmann)
1067 * bug fix in threaded builds: the system can now be suspended and
1068 resumed by shell job control with minimal disruption.
1069 * bug fixes in times and timezones >2038AD
1070 * better handling of "where is GNU make?" problem in build scripts
1071 (thanks to Nikodemus Siivola)
1072 * new optimization: inside a named function any reference to a
1073 function with the same name is considered to be a self-reference;
1074 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1075 optimization quality.
1076 * new optimization on x86: logical functions and + now have
1077 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1078 used when the result is truncated to 32 bits.
1079 * VALUES declaration is partially enabled.
1080 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1081 * fixed some bugs revealed by Paul Dietz' test suite:
1082 ** The system now obeys the constraint imposed by
1083 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1084 types form a lattice under type intersection.
1085 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1086 ** ASSOC now ignores NIL elements in an alist.
1087 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1088 and (1+ MOST-POSITIVE-FIXNUM) answers.
1089 ** The addition of a method with invalid qualifiers to a generic
1090 function does not cause an error to be signalled immediately;
1091 a warning is signalled, and the error is generated only on
1092 calling the generic function.
1093 * changed .fasl file version number, in part to add type codes for
1094 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1095 obscure ANSI requirements
1097 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1098 * fixed bug 148: failure to inline-expand a local function left
1099 garbage, confusing the compiler.
1100 * fixed bugs 3cd: structure slot readers perform type check if the
1101 slot can have an invalid value (i.e. it is either not initialized
1102 or can be written with a less specific slot writer).
1103 * bug fix: the compiler now traps array references to elements off
1104 the end of an array; previously, the bounds checking in some
1105 circumstances could go off-by-one.
1106 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1108 * type declarations for array element types now obey the description
1109 on the CLHS page "Declaration TYPE", as per discussions on
1110 sbcl-help around 2003-05-08. This means that a declaration
1111 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1112 declaration, all references to BAR will be asserted or assumed
1113 (with THE, so dependent on compiler policy) to involve objects of
1114 type FOO. Note that no such declaration is implied in
1115 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1116 * declared types of functions from the "Conditions"
1117 chapter. (reported by Paul Dietz)
1118 * bug fix: CERROR accepts a function as its first argument.
1119 * bug fix: NTH an NTHCDR accept a bignum as index
1120 arguments. (reported by Adam Warner)
1121 * optimization: character compare routines now optimize comparing
1122 against a constant character. (reported by Gilbert Baumann)
1123 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1124 argument now works properly on the MIPS platform.
1125 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1126 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1127 declared to have a functional type, does not cause an error
1129 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1130 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1131 * fixed bug: if last continuation of a deleted block has a
1132 destination, this destination should be deleted too. (reported by
1134 * fixed a bug in the bootstrap process: the host compiler's values
1135 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1136 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1137 test case from Patrik Nordebo)
1138 * improved the ability of the disassembler on the PPC platform to
1139 provide helpful disassembly notes.
1140 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1141 the class in more cases than previously.
1142 * bug fix: FILE-POSITION now understands :START and :END for
1143 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1144 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1145 8) data. (thanks to David Lichteblau)
1146 * bug fix: it is possible to add a method to a generic function
1147 without lambda list.
1148 * bug fix: reader failed to signal END-OF-FILE inside an
1149 object representation. (reported by Nikodemus Siivola)
1150 * fixed some bugs revealed by Paul Dietz' test suite:
1151 ** LAST and [N]BUTLAST should accept a bignum.
1152 ** condition slot accessors are methods.
1153 ** (VECTOR NIL) is a subtype of STRING.
1155 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1156 * minor incompatible change: some nonsensical specialized lambda
1157 lists (used in DEFMETHOD) which were previously ignored now signal
1159 * minor incompatible change: the system is now aware of the types of
1160 variables in the COMMON-LISP package, and will signal errors for
1161 most violations of these type constraints (where previously they
1162 were silently accepted).
1163 * minor incompatible change: COMPILE-FILE now uses the freedom
1164 afforded (ANSI 3.2.2.3) to use derived function types for
1165 functions defined in the same file. This also permits the system
1166 to warn on static type mismatches and function
1167 redefinition. (Currently it does not work with high DEBUG level.)
1168 * minor incompatible change: VALUES declaration is disabled.
1169 * When issuing notes, the compiler now signals a condition of type
1170 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1171 restart for use in user handlers. It is expected that the
1172 COMPILER-NOTE condition will eventually become a condition
1173 supertype to a hierarchy of note types, which will then be
1174 handleable in a similar fashion. However, at the moment, no such
1175 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1176 but it's an implementation detail, not a classification for the
1178 * Changes in type checking closed the following bugs:
1179 ** type checking of unused values (192b, 194d, 203);
1180 ** template selection based on unsafe type assertions (192c, 236);
1181 ** type checking in branches (194bc).
1182 * A short form of VALUES type specifier has ANSI meaning (it has
1183 increased the number of situations when SBCL cannot perform type
1185 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1186 keywords or constants is permissible.
1187 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1188 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1189 classes. (thanks to Antonio Martinez)
1190 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1191 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1192 argument to a pretty-print stream. Instead, it leaves any such
1193 operations to DESCRIBE-OBJECT methods.
1194 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1196 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1197 lambda lists. (thanks to David Lichteblau)
1198 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1199 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1200 (thanks to Gerd Moellmann)
1201 * fixed bug 246: increased compilation speed of long
1202 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1204 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1206 * DEFCONSTANT now throws a condition of type
1207 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1208 constant to a non-EQL value; CONTINUE and ABORT restarts
1209 respectively change and preserve the value.
1210 * fixed bug 63: The code walker, part of the implementation of CLOS,
1211 is now better at handling symbol macros.
1212 * bug fix: There is no longer an internal implementation type named
1213 CL:LENGTH. (reported by Raymond Toy)
1214 * bug fix: In macro-like defining macros/special operators the
1215 implicit block does not enclose the lambda list.
1216 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1217 enclosed in parentheses) are not suitable as type specifiers, and
1218 their use properly signals an error now.
1219 * bug fix: An argument count mismatch for a type specifier in code
1220 being compiled no longer causes an unhandled error at compile
1221 time, but signals a compile-time warning.
1222 * fixed simple vector readable printing
1223 * bug fix: DESCRIBE takes more care over whether the class
1224 precedence list slot of a class is bound before accessing it.
1225 (reported by Markus Krummenacker)
1226 * bug fix: FORMATTER can successfully compile pretty-printer format
1227 strings which use variants of the ~* directive inside.
1228 * bug fix: SEARCH now applies its TEST predicate to the elements of
1229 the arguments in the correct order. (thanks to Wolfhard Buss)
1230 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1231 in type inference. (thanks to Robert E. Brown)
1232 * bug fix: Reading in symbols with an explicit package name of ""
1233 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1234 * fixed some bugs revealed by Paul Dietz' test suite:
1235 ** NIL is now allowed as a structure slot name.
1236 ** Arbitrary numbers, not just REALs, are allowed in certain
1237 circumstances in LOOP for-as-arithmetic clauses.
1238 ** Multiple class redefinitions before slot access no longer
1239 causes a type error.
1240 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1241 association between the name and a class.
1242 ** Generic functions with non-standard method-combination and over
1243 five methods all of which return constants no longer return NIL
1244 after the first few invocations. (thanks to Gerd Moellmann)
1245 ** CALL-NEXT-METHOD with no arguments now passes the original
1246 values of the arguments, even in the presence of assignment.
1247 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1249 ** DISASSEMBLE works with closures and funcallable instances.
1250 ** ADD-METHOD now returns the generic function, not the new method.
1251 ** FIND-METHOD signals an error if the lengths of the specializers
1252 is incompatible with the generic function, even if the ERRORP
1254 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1255 which its argument is a member.
1256 ** DEFCLASS only redefines the class named by its class-name
1257 argument if that name is the proper name of the class;
1258 otherwise, it creates a new class.
1259 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1260 of the UNBOUND-SLOT condition to the name of the slot.
1261 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1262 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1263 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1264 SLOT-MAKUNBOUND in the specified fashion.
1266 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1267 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1268 cross-compilation host. As a consequence, we can now bootstrap our
1269 way up to SBCL starting with a bare gcc toolchain and human-readable
1270 source code (first the source to CLISP, then the source to SBCL).
1271 * A contributed module containing a partial implementation of the
1272 simple-streams interface has been included. (thanks to Rudi
1274 * A contributed module implementing the RFC1321 Message Digest
1275 Algorithm, known as MD5, has been included.
1276 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1277 no longer has any effect, as the code controlled by this feature
1278 has been deleted. (As far as we know, no-one has ever built using
1279 this feature, and its semantics were confused in any case).
1280 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1281 work on conditions (as required by the ANSI specification),
1282 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1283 expected behaviour on conditions. Users should note, however,
1284 that such behaviour is not required by the ANSI specification,
1285 and so use of this behaviour may render their code unportable.
1286 * fixed some bugs revealed by Paul Dietz' test suite:
1287 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1289 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1290 ** SXHASH on condition objects no longer returns NIL.
1291 ** :ALLOCATION :CLASS slots are better treated; their values are
1292 updated on class redefinition, and initforms inherited from
1293 superclasses are applied.
1294 ** REMOVE-METHOD returns its generic function argument even when
1295 no method was removed.
1296 ** SHARED-INITIALIZE now initializes the values of the requested
1297 slots, including those with :ALLOCATION :CLASS.
1298 ** ALLOCATE-INSTANCE now works on structure classes defined via
1299 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1301 ** SLOT-EXISTS-P now works on conditions, as well as structures
1303 ** MAKE-LOAD-FORM now has the required methods on
1304 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1305 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1306 keyword, and now implements the SLOT-NAMES argument.
1307 ** methods with &OPTIONAL arguments no longer allow too many
1308 arguments to be passed in the call without error.
1309 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1310 option is consistent with the required arguments of the generic
1311 function lambda list.
1312 * bug fix: REQUIRE accepts a string designator. (Thanks to
1314 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1315 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1316 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1317 * bug fix: The compiler now checks for duplicated variables in macro
1319 * bug fix: SETQ on globals returns the correct value.
1320 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1321 not a COMPILER-ERROR (followed by some other strange error on
1322 choosing the CONTINUE restart).
1323 * bug fix: make.sh and friends are now more consistent in the way that
1324 they look for GNU "make".
1326 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1327 * experimental native threads support (on x86 Linux >=2.4 only).
1328 This is not compiled in by default: you need to add :SB-THREAD to
1329 the target features. See the "Beyond ANSI" chapter of the manual
1331 * fix for longstanding nonANSIism: The old distinction between
1332 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1333 The return value from CL:FIND-CLASS is now a CLOS class, and
1334 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1335 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1336 * An interface to the MetaObject Protocol, as described in Kiczales,
1337 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1338 MIT Press, 1991, is available from the SB-MOP package.
1339 * incompatible change: the SB-PCL package should now be considered
1340 a private implementation detail, and no longer a semi-private MOP
1342 * minor incompatible change: due to rearrangement for threads, the
1343 control stack and binding stack are now allocated at arbitrary
1344 addresses instead of being hardcoded per-port. Users affected by
1345 this probably have to be doing advanced things with shared
1346 libraries, and will know who they are.
1347 * minor incompatible change: Previously, all --eval forms used were
1348 processed with READ before any of them were processed with EVAL.
1349 Now each --eval form is processed with both READ and EVAL before
1350 the next --eval form is processed. (Thus package operations like
1351 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1352 work as the user might reasonably expect.)
1353 * minor incompatible change: *STANDARD-INPUT* is now only an
1354 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1356 * minor incompatible change: Y-OR-N-P is now character-oriented,
1357 not line oriented. Also, YES-OR-NO-P now works without errors.
1358 (thanks to Antonio Martinez)
1359 * sb-aclrepl module improvements: an integrated inspector, added
1360 repl features, and a bug fix to :trace command.
1361 * Known functions, which cannot be open coded by the backend, are
1362 considered to be able to check types of their arguments. (fixing
1363 a bug report by Nathan J. Froyd)
1364 * fixed a bug in computing method discriminating functions: It is
1365 now possible to define methods specialized on classes which have
1366 forward-referenced superclasses. (thanks to Gerd Moellmann)
1367 * fixed evaluation order in optional entries (reported by Gilbert
1369 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1370 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1371 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1372 required-by-AMOP NAME argument, as well as CLASS and
1373 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1374 * fixed bug 20: DEFMETHOD can define methods using names that are
1375 not the proper names of classes to designate class specializers.
1376 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1377 Unix file descriptors, instead of blowing up. (thanks to Antonio
1379 * Garbage collection refactoring: user-visible change is that a
1380 call to the GC function during WITHOUT-GCING will not do garbage
1381 collection until the end of the WITHOUT-GCING. If you were doing
1382 this you were probably losing anyway.
1383 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1384 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1385 TYPEP the latter but not the former.
1386 * The compiler issues a full WARNING for calls to undefined functions
1387 with names from the CL package.
1388 * MAP-INTO for a vector destination is open coded. (reported by
1389 Brian Downing on c.l.l)
1390 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1391 documentation string.
1392 * fixed some bugs revealed by Paul Dietz' test suite:
1393 ** COPY-ALIST now signals an error if its argument is a dotted
1395 ** Condition slots are now accessed more correctly in the presence
1396 of multiple initargs for a given slot.
1397 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1398 exclude restarts of the same name associated with a different
1400 ** DEFCLASS of forward-referenced classes with another
1401 forward-referenced class in the superclasses list no longer
1403 ** Condition slots are now initialized once each, not multiple
1404 times. (thanks to Gerd Moellmann)
1405 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1407 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1408 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1409 arguments contain duplicated elements.
1410 ** RESTART-CASE understands local macros.
1411 ** RESTART-CASE associates exactly its own restarts with a condition.
1412 ** ENDP in safe mode checks its argument to be of type LIST.
1413 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1414 in question is unbound.
1415 ** Optimized MAKE-INSTANCE functions no longer cause internal
1416 assertion failures in the presence of duplicate initargs.
1417 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1418 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1419 propagated even in the ctor optimized implementation of
1421 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1423 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1424 * a better implementation of SXHASH on (simple) bit vectors,
1425 measured both in execution speed and in distribution of results
1426 over the positive fixnums, has been installed. Likewise, a better
1427 implementation of EQUAL for simple bit vectors is now available.
1428 * fixed CEILING optimization for a divisor of form 2^k.
1429 * fixed bug 240 (emitting extra style warnings "using the lexical
1430 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1431 by Antonio Martinez)
1432 * fixed SXHASH, giving different results for NIL depending on type
1433 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1434 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1435 variable. (found by Rolf Wester)
1436 * SBCL does not ignore type declarations for special
1437 variables. (reported by rif on c.l.l 2003-03-05)
1438 * some bug fixes in contrib/sb-aclrepl/
1439 * fixed some bugs revealed by Paul Dietz' test suite:
1440 ** a bug in the CONS type specifier, whereby the CAR and CDR
1441 types got intertwined, has been fixed;
1442 ** the type system is now able to reason about the interaction
1443 between INTEGER and RATIO types more completely;
1444 ** APPEND, [N]REVERSE and NRECONC check that those their
1445 arguments, which must be proper lists, are really so;
1446 ** An array specialized to be unable to hold elements has been
1447 implemented, as required -- yes, really -- by ANSI;
1448 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1449 on malformed property lists;
1451 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1452 * incompatible packaging change: in line with Unix convention,
1453 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1454 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1455 * REQUIRE and PROVIDE are now optionally capable of doing something
1456 useful. See the documentation string for REQUIRE.
1457 * infrastructure for a managed SBCL contrib system: contributed
1458 modules in this release include:
1459 ** the ASDF system definition facility;
1460 ** an interface to the BSD Sockets API;
1461 ** an ACL-like convenience interface to the repl;
1462 (thanks to Kevin Rosenberg)
1463 ** an implementation of ROTATE-BYTE, with efficient implementation
1465 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1466 gives a closer approximation to the right answer than previously.
1467 (thanks to Raymond Toy)
1468 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1469 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1470 optional environment arguments, as required by ANSI.
1471 * fixed bugs in other functions taking environment objects, allowing
1472 calls with an explicit NIL environment argument to be compiled
1474 * fixed bug 228: primary return values from
1475 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1476 COMPILE or FUNCTION.
1477 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1478 :TYPE will no longer signal errors on innocuous objects.
1479 * fixed bug 231b: SETQ is better at respecting type declarations in
1480 the lexical environment.
1481 * fixed a bug in DEFCLASS: classes named by symbols with no or
1482 unprintable packages can now be defined.
1483 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1484 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1485 * fixed compiler failure related to checking types of functions.
1486 (reported by Robert E. Brown)
1487 * the compiler is now much more consistent in its error-checking
1488 treatment of bounding index arguments to sequence functions: in
1489 (SAFETY 3) code, errors will be signalled in almost all cases if
1490 invalid sequence bounding indices are passed to functions defined
1491 by ANSI to operate on sequences.
1492 * fixed a bug in the build procedure: documentation of SBCL-specific
1493 packages is now preserved and available in the final Lisp image.
1494 * lifted FDEFINITION lookup out of loops in the implementation of
1495 many list operations. (thanks to Robert E. Brown)
1496 * fixed a bug in the reader: the #n# reader macro now works for
1497 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1498 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1499 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1500 E. Brown and Tony Martinez respectively)
1501 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1502 SLOT-BOUNDP: in particular, it is now able to optimize them much
1503 better, and is now not vulnerable to having packages renamed.
1504 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1505 always called when a slot is not present in an instance. (thanks
1507 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1508 longer signal an error on unreading a character following EOF on
1509 the previous constituent stream. (thanks to Tony Martinez)
1510 * fixed some bugs revealed by Paul Dietz' test suite:
1511 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1512 not just nonnegative fixnums;
1513 ** the logical bit-array operators such as BIT-AND now accept an
1514 explicit NIL for their "opt-arg" argument (to indicate a
1515 freshly-consed result bit-array);
1516 ** ELT now signals an error on an invalid sequence index in safe
1518 ** the type system is now cleverer about negations of numeric
1519 types, and consequently understands the BIGNUM and RATIO types
1521 ** the type system is now cleverer about the interaction between
1522 INTEGER and RATIO types: while bugs still remain, many more
1523 cases are accurately computed;
1524 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1525 if it is in the last clause;
1526 ** CONSTANTLY now correctly returns a side-effect-free function in
1528 ** DECLARE is no longer treated as a special-operator; in
1529 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1530 * incremented fasl file version number due to the change in the
1531 DEFSTRUCT-SLOT-DESCRIPTION structure.
1533 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1534 * minor incompatible change: code processed by the "interpreter" or
1535 EVAL now has a compilation optimization policy of (DEBUG 2)
1536 (changed from (DEBUG 1)) to improve debuggability of interactive
1537 development, and to allow the use of the debug RETURN command in
1539 * an experimental implementation of the RETURN command for the
1540 debugger has been included. (thanks to Frederik Kuivinen)
1541 * fixed bug 62: constraints were not propagated into a loop.
1542 * fixed bug in embedded calls of SORT (reported and investigated by
1544 * fixed some bugs revealed by Paul Dietz' test suite:
1545 ** printing and reading of arrays with some dimensions having
1546 length 0 (thanks to Gerd Moellmann);
1547 ** BOA constructor with &AUX argument without a default value does
1548 not cause a type error;
1549 ** CONSTANTP now returns true for all self-evaluating objects.
1551 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1552 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1553 accessors that are related by inheritance, as specified in the
1554 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1555 Valtteri Vuorikoski)
1556 * The compiler is now able to inline functions that were defined in
1557 a complex lexical environment (e.g. inside a MACROLET).
1558 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1560 * Support for the upcoming FreeBSD-5.0 release has been included.
1561 (thanks to Dag-Erling Smorgrav)
1562 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1563 effect when it is not in a toplevel context.
1564 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1565 stand a better chance of being correct. (thanks to Gerd
1567 * fixed bug in COERCE, which now signals an error on coercing a
1568 rational to a bounded real type which excludes the expected
1570 * The compiler is now able to derive types more accurately from the
1571 COERCE and COMPILE functions.
1572 * fixed bug 223: functional binding is considered to be constant
1573 only for symbols in the CL package.
1574 * fixed bug 231: SETQ did not check the type of a variable being set
1575 (reported by Robert E. Brown)
1576 * A new optimization for MAKE-INSTANCE has been included, fixing
1577 various bugs (including relating to :ALLOCATION :CLASS slots and
1578 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1580 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1581 ** As required by ANSI, LOOP now disallows anonymous collection
1582 clauses such as COLLECT I in conjunction with aggregate boolean
1583 clauses such as THEREIS (= I 1);
1584 ** LOOP now signals an error when any variable is reused in the
1585 same loop (including the potentially useful construct analogous
1586 to WITH A = 1 WITH A = (1+ A);
1587 ** IT is only a special loop symbol within the first clause of a
1588 conditional loop clause;
1589 ** LOOP with a typed iteration variable over a hashtable now
1590 signals a type error iff it should.
1591 * fixed some other bugs revealed by Paul Dietz' test suite:
1592 ** FILE-STREAM now names the class previously known as FD-STREAM;
1593 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1594 argument) no longer signals an error;
1595 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1596 of the slot symbol, rather than using the current package
1597 ((:CONC-NAME "") continues to intern the slot's name in the
1599 * incremented fasl file version number, because of the incompatible
1600 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1601 of the new implementation of DEFINE-COMPILER-MACRO.
1603 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1604 * Support for building SBCL for MIPS platforms running in
1605 little-endian mode has now been checked in, and basic
1606 functionality on said platforms verified.
1607 * minor incompatible change: PCL now records the pathname of a file
1608 in which methods and the like are defined, rather than its
1610 * minor incompatible change: TRUENAME now considers the truename of
1611 a file naming a directory to be the pathname with :DIRECTORY
1612 component indicating that directory.
1613 * minor incompatible change: a NAMED clause in the extended form of
1614 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1615 reason for the previous behaviour is unclear.
1616 * more systematization and improvement of CLOS and MOP conformance
1617 in PCL (thanks to Gerd Moellman and Pierre Mai):
1618 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1620 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1621 primary methods with no specializers;
1622 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1624 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1625 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1626 FORWARD-REFERENCED-CLASSes; error reporting on
1627 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1629 ** SXHASH on CLOS instances now uses a slot internal to the
1630 instance to return different numbers on distinct instances,
1631 while preserving the same return value through invocations of
1633 ** DEFMETHOD signals errors when methods with longer incongruent
1634 lambda lists are added to generic functions;
1635 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1636 CLASS, as specified in AMOP;
1637 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1638 on the order returned by the primary method for classes of
1639 class STANDARD-CLASS;
1640 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1641 * fixed some bugs shown by Paul Dietz' test suite:
1642 ** DOLIST puts its body in TAGBODY;
1643 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1645 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1646 value producing form;
1647 ** if more variables are given to PROGV than values, extra
1648 variables are bound and made to have no value;
1649 ** NSUBSTITUTE on list arguments gets the right answer with
1651 ** ELT signals an error of type TYPE-ERROR when the index argument
1652 is not a valid sequence index;
1653 ** LOOP signals (at macroexpansion time) an error of type
1654 PROGRAM-ERROR when duplicate variable names are found;
1655 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1656 ** FUNCALL of special-operators now cause an error of type
1658 ** PSETQ now works as required in the presence of side-effecting
1659 symbol-macro places;
1660 ** NCONC accepts any object as its last argument;
1661 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1663 ** loop-for-as-package does not require a package to be explicitely
1665 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1667 * fixed bug 166: compiler preserves "there is a way to go"
1668 invariant when deleting code.
1669 * fixed bug 172: macro lambda lists with required arguments after
1670 &REST arguments now cause an error to be signalled. (thanks to
1672 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1673 bug. (thanks to Matthew Danish)
1674 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1676 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1677 arguments are lexically rebound. (thanks to Gerd Moellmann and
1679 * fixed bug 194: error messages are now more informative when there
1680 is no primary method applicable in a call to a generic
1681 function. (thanks to Gerd Moellmann)
1682 * fixed bug in command line argument checking (thanks to Julian
1684 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1685 (thanks to Matthew Danish)
1686 * incremented fasl file version number, because of the
1687 SXHASH-related changes in the layout of CLOS data structures
1689 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1690 * minor incompatible change: The runtime (the Unix executable named
1691 "sbcl") is now much pickier about the .core files it will load.
1692 Essentially it now requires .core files to descend from the same
1693 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1694 as the runtime does. (The intent is to prevent the crashes which
1695 can occur, and which can even be reported as mysterious failures,
1696 when people patch the sources or change the build parameters
1697 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1698 sbcl and .core files.)
1699 * fixed bug: VALUES-LIST is no longer optimized away.
1700 * fixed bug 142: The FFI conversion of C string values to Lisp
1701 string values no longer conses excessively. (thanks to Nathan
1702 Froyd porting Raymond Toy's fix to CMU CL)
1703 * began to systematize and improve MOP conformance in PCL (thanks to
1704 Nathan Froyd, Gerd Moellman and Pierre Mai):
1705 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1707 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1708 ** FINALIZE-INHERITANCE is now called on class finalization;
1709 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1710 argument precedence order.
1711 * fixed bug 202: The compiler no longer fails on functions whose
1712 derived types contradict their declared type.
1713 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1714 so it can be non-toplevel.
1715 * The fasl file version number has changed (because of the new
1716 implementation of DEFMACRO).
1717 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1718 safe code, that any length requirement by their type-specifier
1719 argument is valid. The exceptions to this are described in bug
1721 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1722 functions now check the types of their inputs as required by ANSI.
1723 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1724 introduced symbol is DECLAREd to be SPECIAL.
1725 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1726 * fixed bug: :COUNT argument to sequence functions may be negative
1727 * fixed bug: body of DO-SYMBOLS may contain declarations
1728 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1729 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1730 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1731 symbol macro only once
1732 * fixed printing of call frame when argument list is unavailable
1733 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1734 * compiler no longer signals WARNING on unknown keyword
1737 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1738 * A beta-quality port to the mips architecture running Linux,
1739 based on the old CMUCL backend, has been made. It has been tested
1740 on a big-endian kernel, and works sufficiently well to be able to
1741 rebuild itself; it has not been tested in little-endian mode.
1742 * fixed an inconsistency between gencgc.c and purify.c which made
1743 dumping/loading .core files unreliable
1744 * fixed bug 120a: The compiler now deals correctly with IFs where
1745 the consequent is the same as the alternative, instead of
1746 misderiving the return type. (thanks to Alexey Dejneka)
1747 * fixed bug 113: Logical pathnames are now dumpable (the logical
1748 host is resolved at load-time, throwing an error if it is not
1750 * fixed bug 174: FORMAT's error message is slightly clearer when a
1751 non-printing character is used in a format directive.
1752 * fixed several bugs in compiler checking of type declarations, i.e.
1753 violations of the Python "declarations are assertions" principle
1754 (thanks to Alexey Dejneka)
1755 * fixed several bugs in PCL's error checking (thanks to Gerd
1757 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1759 * fixed bug in compilation of functions as first class values
1760 (thanks to Antonio Martinez-Shotton)
1761 * The compiler's handling TYPE-ERRORs which it can prove will
1762 inevitably happen at runtime has been cleaned up and corrected
1763 in several ways. (thanks to Alexey Dejneka)
1764 * improved argument type checking for various basic arithmetic
1765 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1766 lavished on them in the past that they can be compiled in many
1767 ways in different special cases
1768 * fixed bug 181: compiler checks validity of user supplied type
1770 * cleaned up code flushing in optimization: Function calls which
1771 should signal errors for safety purposes (e.g. which ANSI says
1772 should signal errors when their arguments are of incorrect type)
1773 are no longer optimized away.
1774 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1775 * incremented fasl file version number, because changes in the
1776 implementation of sequence functions like COERCE caused
1777 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1778 in old inline expansions) to become undefined. (Actually these
1779 changes were later undone, so we might very well be binary
1780 compatible with 0.7.7 after all, but leaving the version number
1781 incremented seemed like the simplest and most conservative
1784 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1785 * An alpha-quality port to the parisc architecture running Linux,
1786 based on the old CMUCL backend, has been made. This, even more so
1787 than the other backends, should be considered still a work in
1788 progress; known problems include that the Linux kernel in 64-bit
1789 mode does not propagate the correct sigcontext structure to
1790 userspace, and consequently SBCL on a parisc64 kernel will not
1792 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1793 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1794 suppressed.) Also now that the compiler is looking at declarations
1795 in the environment, it checks optimization declarations as well,
1796 and suppresses inlining when (> DEBUG SPEED).
1797 * More fixes have been made to treatment of floating point exception
1798 treatment and other Unix signals. In particular, floating point
1799 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1800 * The detection and handling of control stack exhaustion (infinite
1801 or very deeply nested recursion) has changed. Stack exhaustion
1802 detection is now done by write-protecting pages at the OS level
1803 and applies at all optimization settings; when found, a
1804 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1805 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1806 be caught using IGNORE-ERRORS.
1807 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1808 correctly when transforming recursive LABELS functions to LETs.
1809 (thanks to Alexey Dejneka)
1810 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1811 that are names of constants or global variables.
1812 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1813 alien routines with docstrings.
1814 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1815 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1817 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1818 (thanks to Eric Marsden)
1819 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1820 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1821 * Bug fix: The SPARC backend can now compile functions involving
1822 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1823 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1824 to a functions expecting a PATHNAME-DESIGNATOR.
1825 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1826 lambda lists. (thanks to Alexey Dejneka)
1827 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1828 functions so that EXPORTing the name of the class doesn't cause
1829 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1830 to Antonio Martinez for reporting this)
1831 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1832 constant-folded without failing an assertion. (thanks to Einar
1833 Floystad Dorum for reporting this)
1834 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1835 types are now dealt with more correctly by the compiler.
1836 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1837 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1838 behaviour analogous to RENAME-FILE. This puts its behaviour more
1839 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1841 * The fasl file version number has changed again. (because of the
1842 bug fix involving the names of PCL MAKE-INSTANCE functions)
1844 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1845 * bug fix: Floating point exceptions are treated much more
1846 consistently on the x86/Linux and PPC/Linux platforms.
1847 * Array initialization with :INITIAL-ELEMENT is now much faster for
1848 cases when the compiler cannot open code the array creation, but
1849 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1850 array accesses have also seen a speed increase.
1851 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1852 specified. (This was at the root of some bad interactions between
1853 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1855 * bug fix: Internal error arguments for undefined functions are now
1856 computed correctly on the PPC/Linux platform.
1857 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1858 Raymond Toy's patch for CMU CL)
1859 * Support for the Solaris 9 operating environment has been included
1860 (thanks to Daniel Merritt)
1861 * A very ugly but hopefully complete draft of the missing FFI chapter
1862 of the manual has been created by reformatting the corresponding
1863 CMU CL manual chapter into (currently very ugly and incoherent)
1864 DocBook and bringing it up to date for SBCL behavior. Thus, the
1865 manual is now essentially complete, at least by my extreme
1866 once-and-only-once standards, whereby it's acceptable to refer to
1867 the doc strings of SB-EXT functions as the primary documentation.
1868 * The fasl file version number has changed again, due to cleanup of
1869 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1870 is no longer a static symbol.)
1872 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1873 * SBCL now builds with OpenMCL (version 0.12) as the
1874 cross-compilation host; also, more progress has been made toward
1875 bootstrapping under CLISP.
1876 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1878 * bug 158 fixed: The compiler can now deal with integer loop
1879 increments different from 1; fixing this turned out also to fix
1881 * bug 169 fixed: no more bogus warnings about using lexical bindings
1882 despite the presence of perfectly good SPECIAL declarations (thanks
1883 to David Lichteblau)
1884 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1885 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1886 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1888 * bug 180 fixed: Method combination specifications no longer ignore
1889 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1890 * bug fix: Structure type predicate functions now check their argument
1891 count as they should.
1892 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1893 correctly. (thanks to Pierre Mai)
1894 * minor incompatible change: The --noprogrammer option is deprecated
1895 in favor of the new --disable-debugger option, which is very similar.
1896 (The major difference is that it takes effect at a slightly different
1897 time at startup, causing handling of errors in --sysinit and
1898 --userinit files will be affected differently.) The
1899 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1900 been added to allow this functionality to be controlled from ordinary
1901 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1902 maintainers, who might want to run non-interactive scripts to
1903 build SBCL cores which will later be used interactively.)
1904 * minor incompatible change: The LOAD function no longer, when given
1905 a wild pathname to load, loads all files matching that pathname.
1906 Instead, an error of type FILE-ERROR is signalled.
1908 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1909 * bug 147 fixed: The compiler preserves its block link/count
1910 invariants more correctly now so that it doesn't crash. (thanks
1912 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1914 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1915 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1916 no-value-for-this-character as NIL (as specified by ANSI).
1917 * HOST-NAMESTRING on physical pathnames now returns a string that is
1918 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1919 (thanks to Christophe Rhodes)
1920 * The Alpha port handles icache flushing more correctly. (thanks to
1922 * More progress has been made toward bootstrapping under CLISP. (thanks
1923 to Christophe Rhodes)
1924 * The fasl file format has changed again, because dynamic loading
1925 on OpenBSD (which has non-ELF object files) motivated some cleanups
1926 in the way that foreign symbols are transformed and passed around.
1927 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1928 is no longer treated as whitespace by the reader, but instead as
1929 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1930 |A
\7fB|, instead of A as it used to.
1932 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1933 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1934 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1935 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1936 this as of 0.7.1.45, but was left out of the previous news section
1937 (thanks to Dan Barlow)
1938 * SBCL now runs on the Solaris operating system on SPARC architectures
1939 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1940 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1941 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1942 * SPARC backend cleanups, allowing builds of cores optimized for V8
1943 and V9 SPARCS, and also emission of code targeted to a particular
1944 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1946 * SBCL is closer to bootstrapping under CLISP, thanks to various
1947 fixes by Christophe Rhodes.
1948 * The fasl file format has changed again, to allow the compiler's
1949 INFO database to support symbol macros.
1950 * The user manual (in doc/) is formatted into HTML more nicely.
1951 (thanks to coreythomas)
1952 * The system is smarter about SUBTYPEP relationships, especially
1953 those involving NOT types (including types such as ATOM which are
1954 represented internally using NOT types). Thus SUBTYPEP is less
1955 likely to return (VALUES NIL NIL) in general, and in particular
1956 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1958 * The fasl file format has changed again, because the internal
1959 representation of types now includes a new slot to support the new
1960 SUBTYPEP-of-NOT-types logic.
1961 * (not a change in the main branch of SBCL, but a related prototype
1962 which can hopefully be merged into the main branch of SBCL in the
1963 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1964 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1965 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1966 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1967 dispatch macro character combinations. (thanks to Alexey Dejneka)
1968 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1969 that superficially look like logical namestrings correctly.
1970 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1972 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1973 default supplied-p) for &optional and &key arguments. (thanks to
1976 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1977 * incompatible change: The compiler is now less aggressive about
1978 tail call optimization, doing it only when (> SPACE DEBUG) or
1979 (> SPEED DEBUG). (This is an incompatible change because there are
1980 programs which relied on the old CMU-CL-style behavior to optimize
1981 away their unbounded recursion which will now die of stack overflow.)
1982 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1983 for non-GENCGC systems has been increased to 20M (since that
1984 seems much closer to the likely performance optimum for modern
1985 systems than the old 4M value was)
1986 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1987 and *DEBUG-PRINT-LEVEL*
1988 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1989 of CMU CL's support for SPARC, and various endianness and other
1990 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1991 * new syntactic sugar for the Unix command line: --load foo.bar is now
1992 an alternate notation for --eval '(load "foo.bar")'.
1994 ** The system now detects stack overflow and handles it gracefully,
1995 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1996 optimization settings. (This is a good thing in general, and
1997 its introduction in this version should be particularly timely
1998 for anyone whose code fails because of suppression of tail
2000 ** The system now hunts for the C variable "environ" in a more
2001 devious way, to avoid segfaults when the C library version differs
2002 between compile time and run time. (thanks to Christophe Rhodes)
2003 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2004 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2005 ** The compiler no longer issues bogus style warnings for undefined
2006 classes in the same source file as the DEFCLASSes which defined
2007 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2009 ** fixes in CONDITION class precedence list for undefined function
2010 errors (thanks to Alexei Dejneka)
2011 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2012 correctly. (thanks to Dan Barlow)
2013 ** portability fixes aiming at bootstrapping under CLISP (thanks
2014 to Dave McDonald and Christophe Rhodes)
2015 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2016 ** fixes in type translation and and type inference (thanks to
2018 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2019 ** various fixes in the new ports (thanks to Dan Barlow)
2020 * several changes related to debugging:
2021 ** suppression of tail recursion, as noted above
2022 ** stack overflow detection, as noted above
2023 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2024 is now the default. (For some time encapsulation has been more
2025 reliable than the breakpoint-based :ENCAPSULATE NIL
2026 implementation, at least on X86 systems; and I just noticed that
2027 encapsulation also seems closer to the spirit of the ANSI
2030 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2032 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2033 up properly again. (There was a packaging bug in 0.7.0 which
2034 left their definitions in SB-SYS::LOAD-FOREIGN and
2035 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2036 vital for most things which interface to C-level interfaces,
2037 like extensions working with sockets or databases or
2038 Perl-compatible regexes or whatever, and the need to fix
2039 this bug is the main reason that 0.7.1 was released so
2041 ** DEFGENERIC is now choosier about the methods it redefines, so that
2042 reLOADing a previously-LOADed file containing DEFGENERICs does
2043 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2044 cycle now works as it should. (thanks to Alexey Dejneka)
2045 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2046 fixed by Christophe Rhodes. (He actually submitted this patch
2047 months ago, and I delayed until after 0.7.0.)
2048 ** Bug 111 (internal compiler confusion about runtime checks on
2049 FUNCTION types) was fixed by Alexey Dejneka.
2050 * Some internal cleanups (getting rid of variables which aren't
2051 needed now that the byte interpreter is gone) caused the fasl
2052 file format number to change again.
2054 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2055 * major incompatible change: The default fasl file extension, i.e. the
2056 default extension for files produced by COMPILE-FILE, has changed
2057 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2059 ** There are many changes in the implementation of the compiler.
2060 SBCL is now essentially a compiler-only implementation of ANSI
2061 Common Lisp. EVAL still "interprets" a few special cases, but
2062 almost all the interesting cases are handled by creating
2063 a LAMBDA expression, calling COMPILE on it, then calling
2064 FUNCALL on the result.
2065 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2066 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2067 Since the code is newer, there might still be some new bugs
2068 (though not as many as before Martin Atzmueller's fixes:-). But
2069 the new code is substantially simpler and clearer, and hopefully
2070 any remaining bugs will be simpler, less fundamental, and more
2071 fixable then the bugs in the old code.
2072 ** The revised compiler is still a little unsteady on its feet.
2074 *** The debugging information it produces (particularly the names
2075 of FUNCTION objects) is sometimes much less useful than what
2076 the old compiler produced.
2077 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2078 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2079 within a MACROLET) has been temporarily weakened.
2080 ** There are new compiler optimizations for various functions:
2081 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2082 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2083 *** the math functions TRUNCATE, FLOOR, and CEILING
2084 *** the function-of-all-trades COERCE
2085 Mostly these should be transparent, but there's one
2086 potentially-annoying problem (bug 117): when the compiler
2087 inline-expands a function and does type analysis on the result,
2088 it can create control paths which have type mismatches, and
2089 when it can't prove that those control paths aren't taken,
2090 it will issue WARNINGs about the type mismatches. This is
2091 a particular problem in practice for the new sequence functions.
2092 It's not clear how this should be fixed, and for now, a
2093 workaround is given in the entry for 117 in the BUGS file.
2094 ** (Because of the interaction between the two previous items --
2095 occasional inlining problems and new inline expansions -- some
2096 of the new sequence function optimizations won't really kick in
2097 completely until debugging information, and then inlining, are
2098 straightened out in some future version.)
2099 * minor incompatible changes:
2100 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2101 DIRECTORY no longer implicitly promotes NIL slots of its
2102 pathname argument to :WILD. In particular, when you ask for the
2103 contents of a directory (which you used to be able to do without
2104 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2105 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2106 ** changes in behavior that ANSI explicitly defines to be
2107 implementation dependent:
2108 *** The new compiler-only implementation still conforms with ANSI,
2109 but acts a little different than before. Besides the obvious
2110 changes in performance tradeoffs (that the cost per form passed
2111 to EVAL has gone up, and the cost per form executed by EVAL
2112 has gone down), the behavior of the system changes a little
2113 because there are no longer any interpreted function objects.
2114 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2115 e.g. doing COMPILE on the output of interactive DEFUN is
2117 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2119 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2120 has changed from (:CL) to NIL.
2121 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2122 e.g. #\Newline and #\Tab, don't have names specified in the
2123 ANSI Common Lisp standard, is now based on their ASCII symbolic
2124 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2125 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2126 are no longer used for output.
2127 ** changes in internal implementation constants:
2128 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2129 4 million. (If your application spends a lot of time GCing and
2130 you have a lot of RAM, you might want to experiment with
2131 increasing it even more.)
2132 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2133 However, almost all old code should still continue to work without
2134 immediate update, as SB-C-CALL is now a (deprecated) nickname
2136 ** Old operator names in the style DEF-FOO are now deprecated in
2137 favor of new corresponding names DEFINE-FOO, for consistency with
2138 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2139 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2140 internal symbols, but a few supported extensions like
2141 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2142 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2143 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2144 etc. as you get deeper into recursive calls to the debugger
2145 command loop, instead of the old "5]", "5]]", "5]]]"
2146 sequence. (I was motivated to do this when squabbles between
2147 ILISP and SBCL left me very deeply nested in the debugger. In the
2148 short term, this change will probably provoke more ILISP/SBCL
2149 squabbles, but hopefully it will be an improvement in the long run.)
2150 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2151 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2152 function, since declaiming return types involving aliens is
2153 (1) annoyingly messy to do by hand and (2) vital to efficient
2154 compilation of code which calls such functions.
2155 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2156 longer reexported by the SB-EXT package. They're solely useful
2157 for alien code, so it seems more logical that you should get
2158 them from the SB-ALIEN package, not in SB-EXT.
2159 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2160 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2161 features. Instead, the code that they used to control is always
2162 built into the system.
2163 * many other bug fixes
2164 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2165 advantage of the new EVAL-WHEN stuff and to clean them up in
2166 general, and they are now more ANSI-compliant in a number of
2167 ways. Martin Atzmueller is responsible for a lot of this.
2168 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2170 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2171 *** correct ERROR type for various file operations
2172 *** some fixes for Lisp streams
2173 *** DEFMETHOD syntax checking
2174 *** changing old weird representation of debug information as
2175 strings (which, among their other deficiencies, don't transform
2176 correctly when you rename packages, and don't change their
2177 print representation when you change things like *PACKAGE*
2178 and *PRINT-LENGTH*) to symbols and lists of symbols
2179 He also made several improvements and fixed several bugs in DESCRIBE.
2180 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2182 *** misbehavior of WRITE-STRING/WRITE-LINE
2183 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2184 and several other LOOP problems as well
2185 *** DIRECTORY when similar filenames are present
2186 *** DEFGENERIC with :METHOD options
2187 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2188 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2189 *** argument ordering in FIND with :TEST option
2190 *** mishandled package designator argument in APROPOS-LIST
2191 *** various problems in the backquote readmacro
2192 *** a bug in APROPOS
2193 *** probably some others that I'm not describing very well here,
2194 since the CVS log documents them by reference to sbcl-devel
2195 messages, and the SourceForge archives aren't working well.:-(
2196 ** Dan Barlow improved the Alpha port (and is making progress on the
2197 PPC port, for those of you who think different).
2198 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2199 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2200 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2201 continues to work on the SPARC port (for those of you in a position
2202 to look down upon our little PC-compatible boxes from a great height).
2203 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2204 and uses that copy, rather than the current dynamic value, when
2205 it's trying to decide whether to truncate output. Thus e.g.
2206 (let ((*print-lines* 50))
2207 (pprint-logical-block (stream nil)
2209 (let ((*print-lines* 8))
2210 (print (aref possiblybigthings i) stream)))))
2211 should now truncate the logical block only at 50 lines, instead of
2212 often truncating it at 8 lines, as it did before.
2213 * The doc/cmucl/ directory, containing old CMU CL documentation
2214 from the time of the fork, is no longer part of the base system.
2215 SourceForge has shut down its anonymous FTP service, and with it
2216 my original plan for distributing the old CMU CL documentation
2217 there. For now, if you need these files you can download an old
2218 SBCL source release and extract them from it.
2219 * The fasl file version number changed again, for dozens of reasons,
2220 some of which are apparent above.
2222 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2223 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2224 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2225 patch, so that DISASSEMBLE output is much nicer.
2226 * The code in the SB-PROFILE package now seems reasonably stable.
2227 I still haven't decided what the final interface should look like
2228 (I'd like PROFILE to interact cleanly with TRACE, since both
2229 facilities use function encapsulation) but if you have a need
2230 for profiling now, you can probably use it successfully with
2231 the current CMU-CL-style interface.
2232 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2233 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2234 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2235 link now returns the dangling link itself, and for similar
2236 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2237 link itself. (In these cases the old code signalled an error and
2238 looped endlessly, respectively.) Thus, DIRECTORY now works even
2239 in the presence of dangling and cyclic symbolic links.
2240 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2241 is now a supported extension again, since the consensus on
2242 sbcl-devel was that it can be useful for ordinary development
2243 work, not just for debugging SBCL itself.
2244 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2245 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2246 that currently-defined functions might be redefined later with
2247 different return types.
2248 * Hash tables can be printed readably, as inspired by CMU CL code
2249 of Eric Marsden and SBCL code of Martin Atzmueller.
2250 * better error handling in CLOS method combination, thanks to
2251 Martin Atzmueller porting Pierre Mai's CMU CL patches
2252 * more overflow fixes for >16Mbyte I/O buffers
2253 * A bug in READ has been fixed, so that now a single Ctrl-D
2254 character suffices to cause end-of-file on character streams.
2255 In particular, now you only need one Ctrl-D at the command
2256 line (not two) to exit SBCL.
2257 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2259 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2260 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2261 does the right thing.
2262 * The implementation of some type tests, especially for CONDITION
2263 types, is now tidier and maybe faster, due to CMU CL code
2264 originally by Douglas Crosher, ported by Martin Atzmueller.
2265 * Some math functions have been fixed, and there are new
2266 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2267 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2268 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2269 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2270 likely to be built into sbcl-0.7.0.)
2271 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2272 CL, ported to SBCL by Martin Atzmueller)
2273 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2274 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2275 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2276 * There's a new slam.sh hack to shorten the edit/compile/debug
2277 cycle for low-level changes to SBCL itself, and a new
2278 :SB-AFTER-XC-CORE target feature to control the generation of
2279 the after-xc.core file needed by slam.sh.
2280 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2281 COMPILE-FILE is no longer supported, so that now every function
2282 gets an entry point, so that block compilation looks a little
2283 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2284 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2285 returns the number of bytes consed since the system started,
2286 rather than the number consed since the first time the function
2287 was called. (The new definition parallels ANSI functions like
2288 CL:GET-INTERNAL-RUN-TIME.)
2289 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2290 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2291 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2292 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2293 * Fasl file version numbers are now independent of the target CPU,
2294 since historically most system changes which required version
2295 number changes have affected all CPUs equally. Similarly,
2296 the byte fasl file version is now equal to the ordinary
2299 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2300 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2301 is no longer recognized. I apologize for this, because it was
2302 listed in SB-EXT as a supported extension, but I found that
2303 its existing behavior was poorly specified, as well as incorrectly
2304 specified, and it looked like too much of a mess to straighten it
2305 out. I have enough on my hands trying to get ANSI stuff to work..
2306 * many patches ported from CMU CL by Martin Atzmueller, with
2307 half a dozen bug fixes in pretty-printing and the debugger, and
2308 half a dozen others elsewhere
2309 * fixed bug 13: Floating point infinities are now supported again.
2310 They might still be a little bit flaky, but thanks to bug reports
2311 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2312 as flaky as they were.
2313 * The --noprogrammer command line option is now supported. (Its
2314 behavior is slightly different in detail from what the old man
2315 page claimed it would do, but it's still appropriate under the
2316 same circumstances that the man page talks about.)
2317 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2318 are now supported, and enabled by default. Thus, the compiler can
2319 handle many floating point and complex operations much less
2320 inefficiently. (Thus e.g. you can implement a complex FFT
2322 * The compiler now detects type mismatches between DECLAIM FTYPE
2323 and DEFUN better, and implements CHECK-TYPE more correctly, and
2324 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2325 thanks to patches from Martin Atzmueller.
2326 * various fixes to make the cross-compiler more portable to
2327 ANSI-conforming-but-different cross-compilation hosts (notably
2328 Lispworks for Windows, following bug reports from Arthur Lemmens)
2329 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2330 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2331 fixed thanks to Pierre Mai's CMU CL patches.
2332 * fixes to keep the system from overflowing internal counters when
2333 it tries to use i/o buffers larger than 16M bytes
2334 * fixed bug 45a: Various internal functions required to support
2335 complex special functions have been merged from CMU CL sources.
2336 (When I was first setting up SBCL, I misunderstood a compile-time
2337 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2338 * improved support for type intersection and union, fixing bug 12
2339 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2340 more obscure bugs as well
2341 * some steps toward byte-compiling non-performance-critical
2342 parts of the system, courtesy of patches from Martin Atzmueller
2343 * Christophe Rhodes has made some debian packages of sbcl at
2344 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2345 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2346 stable, but are nonetheless usable. When he's ready, I'd be happy
2347 to add them to the SourceForge "File Releases" section. (And if
2348 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2349 * new fasl file format version number (because of changes in
2350 internal representation of (OR ..) types to accommodate the new
2351 support for (AND ..) types, among other things)
2353 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2354 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2356 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2358 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2359 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2360 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2361 * DESCRIBE now works on structure objects again.
2362 * Most function call argument type mismatches are now handled as
2363 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2364 know whether the function will be redefined before the call is
2365 executed. (The compiler could flag local calls with full WARNINGs,
2366 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2367 it doesn't keep track of enough information to know whether calls
2368 are local in this sense.)
2369 * Compiler output is now more verbose, with messages truncated
2370 later than before. (There should be some supported way for users
2371 to override the default verbosity, but I haven't decided how to
2372 provide it yet, so this behavior is still controlled by the internal
2373 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2374 src/compiler/ir1util.lisp.)
2375 * Fasl file format version numbers have increased again, because
2376 support for the Gray streams extension changes the layout of the
2377 system's STREAM objects.
2378 * The Gray subclassable streams extension now works, thanks to a
2379 patch from Martin Atzmueller.
2380 * The full LOAD-FOREIGN extension (not just the primitive
2381 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2382 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2383 but like most other programs, it defaults to copying the Unix
2384 environment from the original process instead of starting the
2385 new process in an empty environment.
2386 * Extensions which manipulate the Unix environment now support
2387 an :ENVIRONMENT keyword option which doesn't smash case or
2388 do other bad things. The CMU-CL-style :ENV option is retained
2389 for porting convenience.
2390 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2391 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2393 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2395 * A patch from Martin Atzmueller seems to have solved the SIGINT
2396 problem, and as far as we know, signal-handling now works cleanly.
2397 (If you find any new bugs, please report them!)
2398 * The system no longer defaults Lisp source file names to types
2399 ".l", ".cl", or ".lsp", but only to ".lisp".
2400 * The compiler no longer uses special default file extensions for
2401 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2402 seems to expect a single default extension for all compiled code,
2403 and there's no compelling reason to try to stretch the standard
2404 to allow two different extensions.) Instead, byte-compiled files
2405 default to the same extension as native-compiled files.
2406 * Fasl file format version numbers have increased again, because
2407 a rearrangement of internal implementation packages made some
2408 dumped symbols in old fasl files unreadable in new cores.
2409 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2410 many fewer weird special cases.
2411 * Bug #17 (differing COMPILE-FILE behavior between logical and
2412 physical pathnames) has been fixed, and some related misbehavior too,
2413 thanks to a patch from Martin Atzmueller.
2414 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2415 by Tim Moore, ported to SBCL by Martin Atzmueller.
2416 * Martin Atzmueller fixed several filesystem-related problems,
2417 including bug #36, in part by porting CMU CL patches, which were
2418 written in part by Paul Werkowski.
2419 * More compiler warnings in src/runtime/ are gone, thanks to
2420 more patches from Martin Atzmueller.
2421 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2424 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2426 * DESCRIBE now works on CONDITION objects.
2427 * The debugger now handles errors which arise when trying to print
2428 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2430 * The build system now uses an additional file, customize-target-features.lisp,
2431 to allow local modifications to the target *FEATURES* list. (The point of
2432 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2433 debugging features enabled, without having to worry about propagating it
2434 into everyone's system when I do a "cvs update".) When no
2435 customize-target-features.lisp file exists, the target *FEATURES* list
2436 should be constructed the same way as before.
2437 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2438 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2439 warn about behavior which is undefined under the ANSI spec. Note: This
2440 is specified by ANSI, but it's not very popular with programmers.
2441 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2442 macro in the SBCL sources for an example of a workaround which you
2443 might use to make portable ANSI-standard code which does what you want.
2444 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2445 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2446 when it's used as a non-toplevel form. (This is required in order
2447 to implement the DEFCONSTANT-EQX macro.)
2448 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2449 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2450 funny things, due to bugs in the implementation of EVAL-WHEN
2451 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2452 non-toplevel DEFCONSTANTs are uncommon.)
2453 * The core file version number and fasl file version number have been
2454 incremented, because the old noncompliant DEFCONSTANT behavior involved
2455 calling functions which no longer exist, and because I also took the
2456 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2457 * fixed bug 1 (error handling before read-eval-print loop starts), and
2458 redid debugger restarts and related debugger commands somewhat while
2460 ** The QUIT debugger command is gone, since it did something
2461 rather different than the SB-EXT:QUIT command, and since it never
2462 worked properly outside the main toplevel read/eval/print loop.
2463 Invoking the new TOPLEVEL restart provides the same functionality.
2464 ** The GO debugger command is also gone, since you can just invoke
2465 the CONTINUE restart directly instead.
2466 ** The TOP debugger command is also gone, since it's redundant with the
2467 FRAME 0 command, and since it interfered with abbreviations for the
2469 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2471 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2472 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2473 some porting work by Martin Atzmueller)
2474 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2475 with a boatload of patches which he ported to SBCL. Now that those
2477 ** The system tries to make sure that its low-priority messages
2478 are prefixed by semicolons, to help people who like to use
2479 syntax highlighting in their ILISP buffer. (This patch
2480 was originally due to Raymond Toy.)
2481 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2482 patches originally written by Raymond Toy.
2483 ** The compiler understands coercion between single-value and
2484 multiple-VALUES type expressions better, getting rid of some very
2485 weird behavior, thanks to patches originally by Robert MacLachlan
2486 and Douglas Crosher.
2487 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2488 lambda lists, thanks to a patch originally by Pierre Mai.
2489 ** The system no longer bogusly warns about "abbreviated type
2491 ** The compiler gets less confused by inlining and RETURN-FROM,
2492 thanks to some patches originally by Tim Moore.
2493 ** The system no longer hangs when dumping circular lists to fasl
2494 files, thanks to a patch originally from Douglas Crosher.
2495 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2496 undefined function error.
2497 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2498 in structure slot initforms) for now, documented workaround instead:-|
2499 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2500 * fixed bug 5: added stubs for various Gray stream functions called
2501 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2502 installed, at least appropriate type errors are generated
2503 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2504 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2505 consistently in DEFMETHOD forms.
2506 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2507 it doesn't seem to affect SBCL after all
2508 * The C runtime system now builds with better optimization and many
2509 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2511 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2513 * The system is now under CVS at SourceForge (instead of the
2514 CVS repository on my home machine).
2515 * The new signal handling code has been tweaked to treat register
2516 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2517 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2518 downstream, e.g. in debug-int.lisp, has implicit dependencies
2519 on the unsignedness of integer representation of machine words,
2520 and that caused the system to bomb out with infinite regress
2521 when trying to recover from type errors involving signed values,
2522 e.g. (BUTLAST '(1 2 3) -1).)
2523 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2524 (This was one of the bugs Peter Van Eynde reported back in July.)
2525 * The system now uses code inspired by Colin Walters' O(N)
2526 implementation of MAP (from the cmucl-imp@cons.org mailing
2527 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2528 inline the MAP operation, and there is more than one
2529 sequence argument to the MAP call (so that it can't just
2530 do ETYPECASE once and for all based on the type of the
2531 single sequence argument). (The old non-inline implementation
2532 of the general M-argument sequence-of-length-N case required
2533 O(M*N*N) time when any of the sequence arguments were LISTs.)
2534 * The QUIT :UNIX-CODE keyword argument has been renamed to
2535 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2536 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2538 * Raymond Wiker's patches to port dynamic loading from Linux to
2539 FreeBSD have been added.
2540 * The BUGS file is now more nearly up to date, thanks in large part
2541 to Martin Atzmueller's review of it.
2542 * The debugger now flushes standard output streams before it begins
2543 its output ("debugger invoked" and so forth).
2544 * The core version number and fasl file version number have both
2545 been incremented, because of incompatible changes in the layout
2547 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2548 used to not be called for a saved Lisp image.)
2549 * Martin Atzmueller's version of a patch to fix a compiler crash,
2550 as posted on sbcl-devel 13 September 2000, has been installed.
2551 * Instead of installing Martin Atzmueller's patch for the
2552 compiler transform for SUBSEQ, I deleted the compiler transform,
2553 and transforms for some similar consing operations.
2554 * A bug in signal handling which kept TRACE from working on OpenBSD
2556 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2557 away by constant folding
2558 * The system now defines its address space constants in one place
2559 (in the Lisp sources), and propagates them automatically elsewhere
2560 (through GENESIS and the sbcl.h file). Therefore, patching the
2561 address map is less unnecessarily tedious and error-prone. The
2562 Lisp names of address space constants have also been systematized.
2563 * CVS tags like dollar-Header-dollar have been removed from
2564 the sources, because they have never saved me trouble and
2565 they've been source of trouble working with patches and other
2566 diff-related operations.
2567 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2568 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2570 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2572 * The system has been ported to OpenBSD.
2573 * The system now compiles with a simple "sh make.sh" on the systems
2574 that it's supported on. I.e., now you no longer need to tweak
2575 text in the target-features.lisp-expr and symlinks in src/runtime/
2576 by hand, the make.sh takes care of it for you.
2577 * The system is no longer so grossly inefficient when compiling code
2578 involving vectors implemented as general (not simple) vectors (VECTOR T),
2579 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2580 which dares to use the various sequence functions on non-simple
2581 vectors, takes less of a performance hit.
2582 * There is now a primitive type predicate VECTOR-T-P
2583 to test for the (VECTOR T) type, so that e.g.
2584 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2585 can now be compiled with some semblance of efficiency. (The old code
2586 turned the type declaration into a full call to %TYPEP at runtime!)
2587 * AREF on (VECTOR T) is still not fast, since it's still compiled
2588 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2589 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2590 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2591 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2592 without making full calls to SUBTYPEP at runtime.
2593 (Some analogous efficiency issues for non-simple vectors specialized to
2594 element types other than T, or for non-simple multidimensional arrays,
2595 have not been addressed. They could almost certainly be handled the
2596 same way if anyone is motivated to do so.)
2597 * The changes in array handling break binary compatibility, so
2598 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2599 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2600 instead of (VALUES T T).
2601 * By following the instructions that Dan Barlow posted to sbcl-devel
2602 on 2 July 2000, I was able to enable primitive dynamic object
2603 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2604 functionality is not implemented (since it calls ld to resolve
2605 library references automatically, requiring RUN-PROGRAM for its
2606 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2607 try to resolve library references) is now supported.
2608 * The system now flushes the standard output streams when it terminates,
2609 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2610 them at several other probably-convenient times, e.g. in each pass of
2611 the toplevel read-eval-print loop, and after evaluating a form given
2612 as an "--eval" command-line option. (These changes were motivated by a
2613 discussion of stream flushing issues on cmucl-imp in August 2000.)
2614 * The source transform for TYPEP of array types no longer assumes
2615 that an array whose element type is a not-yet-defined type
2616 is implemented as an array of T, but instead punts, so that the
2617 type will be interpreted at runtime.
2618 * There is now some support for cross-compiling in make.sh: each of
2619 the phases of make.sh has its own script. (This should be transparent
2620 to people doing ordinary, non-cross-compile builds.)
2621 * Since my laptop doesn't have hundreds of megabytes of memory like
2622 my desktop machine, I became more motivated to do some items on
2623 my to-do list in order to reduce the size of the system a little:
2624 ** Arrange for various needed-only-at-cold-init things to be
2625 uninterned after cold init. To support this, those things have
2626 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2627 symbols with such names are now uninterned after cold init).
2628 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2629 fasl files for cold load.
2630 ** Remove the old compiler structure pooling code (which used to
2631 be conditional on the target feature :SB-ALLOC) completely.
2632 ** Redo the representation of some data in cold init to be more compact.
2633 (I also looked into supporting byte compiled code at bootstrap time,
2634 which would probably reduce the size of the system a lot, but that
2635 looked too complicated, so I punted for now.)
2636 * The maximum signal nesting depth in the src/runtime/ support code has
2637 been reduced from 4096 to 256. (I don't know any reason for the very
2638 large old value. If the new smaller value turns out to break something,
2639 I'll probably just bump it back up.)
2640 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2642 * Many, many bugs reported by Peter Van Eynde have been added to
2643 the BUGS list; some have even been fixed.
2644 * While enabling dynamic object file loading, I tried to make the
2645 code easier to understand, renaming various functions and variables
2646 with less ambiguous names, and changing some function calling
2647 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2648 * While trying to figure out how to do the OpenBSD port, I tried to
2649 clean up some of the code in src/runtime/. In particular, I dropped
2650 support for non-POSIX signal handling, added various comments,
2651 tweaked the code to reduce the number of compilation warnings, and
2652 renamed some files to increase consistency.
2653 * To support the new automatic configuration functionality in make.sh,
2654 the source file target-features.lisp-expr has been replaced with the
2655 source file base-target-features.lisp-expr and the machine-generated
2656 file local-target-features.lisp-expr.
2657 * fixed a stupid quoting error in make.sh so that using CMU CL
2658 "lisp -batch" as cross-compilation host works again
2660 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2662 * DESCRIBE no longer tries to call itself recursively to describe
2663 bound/fbound values, so that it no longer fails on symbols which are
2664 bound to themselves (like keywords, T, and NIL).
2665 * DESCRIBE now works on generic functions.
2666 * The printer now prints less-screwed-up representations of closures
2667 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2668 * A private symbol is used instead of the :EMPTY keyword previously
2669 used to mark empty slots in hash tables. Thus
2670 (DEFVAR *HT* (MAKE-HASH-TABLE))
2671 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2672 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2673 now does what ANSI says that it should. (You can still get
2674 similar noncompliant behavior if bang on the hash table
2675 implementation with all the symbols you get back from
2676 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2677 This breaks binary compatibility, since tests for equality to
2678 :EMPTY are wired into things like the macroexpansion of
2679 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2681 * There's now a minimal placeholder implementation for CL:STEP,
2682 as required by ANSI.
2683 * An obscure bug in the interaction of the normal compiler, the byte
2684 compiler, inlining, and structure predicates has been patched
2685 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2686 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2688 * Missing ordinary arguments in a macro call are now detected even
2689 when the macro lambda list contains &KEY or &REST.
2690 * The debugger no longer complains about encountering the top of the
2691 stack when you type "FRAME 0" to explicitly instruct it to go to
2692 the top of the stack. And it now prints the frame you request even
2693 if it's the current frame (instead of saying "You are here.").
2694 * As specified by ANSI, the system now always prints keywords
2695 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2697 * The default initial SIZE of HASH-TABLEs is now smaller.
2698 * Type information from CLOS class dispatch is now propagated
2699 into DEFMETHOD bodies, so that e.g.
2700 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2702 is now basically equivalent to
2703 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2704 (DECLARE (TYPE SINGLE-FLOAT X))
2706 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2707 operation, without having to do run-time type dispatch.
2708 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2709 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2710 * Since I don't seem to be making big file reorganizations very often
2711 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2712 is overflowing my ability to conveniently back them up), I've finally
2713 checked the system into CVS. (The CVS repository is on my home system,
2714 not at SourceForge -- putting it on SourceForge might come later.)
2715 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2716 high-level GC-NOTIFY-FOO functions send their output. (There's
2717 still very little control of where low-level verbose GC functions
2718 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2719 less than it used to -- the GC-NOTIFY-FOO functions are now under
2720 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2721 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2722 in only one place in the source code, and propagates it automatically
2723 everywhere that it's needed. Thus e.g. when I bump the version from
2724 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2725 * The C source files now include boilerplate legalese and documentation
2726 at the head of each file (just as the Lisp source files already did).
2727 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2728 to his page will be replaced with a link to his new CLiki service.
2730 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2732 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2733 * The build process now looks for GNU make under the default name "gmake",
2734 instead of "make" as it used to. If GNU make is not available as "gmake"
2735 on your system, you can change this default behavior by setting the
2736 GNUMAKE environment variable.
2737 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2738 can build without error under CMU CL.
2740 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2742 * There is now a partial SBCL user manual (with some new text and some
2743 text cribbed from the CMU CL manual).
2744 * The beginnings of a profiler have been added (starting with the
2745 CMU CL profiler and simplifying and cleaning up). Eventually the
2746 main interface should be through the TRACE macro, but for now,
2747 it's still accessed through vaguely CMU-CL-style functions and macros
2748 exported from the package SB-PROFILE.
2749 * Some problems left over from porting CMU CL to the new
2750 cross-compilation bootstrap process have been cleaned up:
2751 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2752 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2753 part when getting rid of the compiler *BACKEND* stuff.)
2754 ** The value of *NULL-TYPE* was screwed up, because it was
2755 being initialized before the type system knew the final
2756 definition of the 'NULL type. This screwed up several key
2757 optimizations in the compiler, causing inefficiency in all sorts
2758 of places. (I found it because I wanted to understand why
2759 GET-INTERNAL-RUN-TIME was consing.)
2760 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2761 PROCLAIM FTYPE information. Unfortunately this broke binary
2762 compatibility again, since now the forms output by DEFGENERIC
2763 to refer to functions which didn't exist in 0.6.3.
2764 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2765 can use the new (as of 0.6.3) transform for SOME into MAP into
2767 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2768 values to (INTEGER 0 1000000), so that the time code will no longer
2769 occasionally get blown up by Linux returning 1000000 microseconds
2770 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2771 its output conform to the ANSI spec. (Alas, this makes its output
2772 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2773 * A full call to MAP NIL with a single sequence argument no longer conses.
2774 * fixes to problems pointed out by Martin Atzmueller:
2775 * The manual page no longer talks about multiprocessing as though
2776 it were currently supported.
2777 * The ILISP support patches have been removed from the distribution,
2778 because as of version 5.10.1, ILISP now supports SBCL without us
2779 having to maintain patches.
2780 * added a modified version of Raymond Toy's recent CMU CL patch for
2781 EQUALP comparison of HASH-TABLE
2783 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2785 * The system still can't cross-compile itself with
2786 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2787 entails), but at least it can compile more of itself that way
2788 than it used to be able to, and various buglets which were uncovered
2789 by trying to cross-compile itself that way have now been fixed.
2790 * This release breaks binary compatibility again. This time
2791 at least I've incremented the FASL file format version to 2, so that the
2792 problem can be detected reliably instead of just causing weird errors.
2793 * various new style warnings:
2794 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2795 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2796 ** using the lexical binding of a variable named in the *FOO* style
2797 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2798 as specified by ANSI.
2799 * *RANDOM-STATE* is no longer automatically initialized from
2800 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2801 default behavior of the system is to repeat its behavior every time
2802 it's run. If you'd like to change this behavior, you can always
2803 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2804 old convention there was no comparably easy way to get the system to
2805 repeat its behavior every time it was run.
2806 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2807 FTYPE declarations has been removed, in favor of their ANSI
2808 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2809 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2810 the types of their sequence arguments can be determined at compile time.
2811 This is done through a new open code expansion for MAP which eliminates
2812 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2813 argument types can be determined at compile time.
2814 * The optimizer now transforms COERCE into an identity operation when it
2815 can prove that the coerced object is already of the correct type. (This
2816 can be a win for machine generated code, including the output of other
2817 optimization transforms, such as the MAP transform above.)
2818 * Credit information has been moved from source file headers into CREDITS.
2819 * Source file headers have been made more standard.
2820 * The CASE macro now compiles without complaining even when it has
2823 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2825 * (Note that the way that the PCL macroexpansions were rewritten
2826 to accommodate the change in DEFGENERIC below breaks binary
2827 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2828 not run under sbcl-0.6.2. Once we get out of alpha releases,
2829 i.e. hit release 1.0.0, we'll probably try to maintain binary
2830 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2831 and sbcl-1.4.4. Until then, however, it might be fairly common
2832 for maintenance releases to break binary compatibility.)
2833 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2835 * The reporting of unhandled signals has been changed to print some
2836 explanatory text as well as the report form. (Previously only
2837 the report form was printed.)
2838 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2839 it defines, so that the compiler no longer issues undefined function
2840 warnings for compiled-but-not-yet-loaded generic functions.
2841 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2842 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2843 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2844 section of the ANSI spec are supported.
2845 * The "" nickname for the "KEYWORD" package has been removed.
2846 The reader still handles symbol tokens which begin with a package marker
2847 as keywords, but it doesn't expose its mechanism for doing so in the
2848 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2849 * The system now issues STYLE-WARNINGs for contradictory TYPE
2850 proclamations. (Warnings for contradictory FTYPE proclamations would
2851 be nice too, but those can't be done usefully unless the type system
2852 is made smarter about FUNCTION types.)
2853 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2854 systematized, so that "*target-*.lisp is supposed to exist only on the
2855 target and imply that there's a related file which exists on the
2856 host, and *host-*.lisp is supposed to exist only on the host and imply
2857 that there's a related file which exists on the target. This involves a
2858 lot of renaming. Hopefully the acute confusion caused by the renaming
2859 will be justified by the reduction in chronic confusion..
2860 ** runtime-type.lisp -> early-target-type.lisp
2861 ** target-type.lisp -> late-target-type.lisp
2862 ** early-host-format.lisp -> early-format.lisp
2863 ** late-host-format.lisp -> late-format.lisp
2864 ** host-error.lisp -> misc-error.lisp
2865 ** early-error.lisp -> early-target-error.lisp
2866 ** late-error.lisp -> late-target-error.lisp
2867 ** host-defboot.lisp -> early-defboot.lisp
2868 ** code/misc.lisp -> code/target-misc.lisp
2869 ** code/host-misc.lisp -> code/misc.lisp
2870 ** code/numbers.lisp -> code/target-numbers.lisp
2871 ** code/early-numbers.lisp -> numbers.lisp
2872 ** early-host-type.lisp -> early-type.lisp
2873 ** late-host-type.lisp -> late-type.lisp
2874 ** host-typep.lisp -> typep.lisp
2875 ** load.lisp -> target-load.lisp
2876 ** host-load.lisp -> load.lisp
2877 ** host-disassem.lisp -> disassem.lisp
2878 ** host-insts.lisp -> insts.lisp
2879 ** byte-comp.lisp -> target-byte-comp.lisp
2880 ** host-byte-comp.lisp -> byte-comp.lisp
2881 ** host-signal.lisp -> signal.lisp
2882 ** host-defstruct.lisp -> defstruct.lisp
2883 ** late-target-type.lisp -> deftypes-for-target.lisp
2884 Furthermore, several other previously target-only files foo.lisp (e.g.
2885 hash-table.lisp and random.lisp) have been split into a target-and-host
2886 foo.lisp file and a target-only target-foo.lisp file, with their key type
2887 definitions in the target-and-host part, so that the cross-compiler will
2888 know more about target types.
2889 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2890 gone away. In their place are various *BACKEND-FOO* variables
2891 corresponding to the slots of the old structure.
2892 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2893 into the target SBCL has been fixed.
2894 * The chill.lisp system for loading cold code into a running SBCL
2896 * Support for the CMU CL "scavenger hook" extension has been removed.
2897 (It was undocumented and unused in the CMU CL sources that SBCL was
2898 derived from, and stale in sbcl-0.6.1.)
2899 * Various errors in the cross-compiler type system were detected
2900 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2901 (enabling various consistency checks). Many of them were fixed,
2902 but some hard problems remain, so the compiler is back to
2903 running without *TYPE-SYSTEM-INITIALIZED* for now.
2904 * As part of the cross-compiler type system cleanup, I implemented
2905 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2906 * I have started adding UNCROSS calls throughout the type system
2907 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2908 on cross-compiler input files will be able to go away, and various
2910 * CONSTANTP now returns true for quoted forms (as explicitly required
2913 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2915 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2916 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2917 (MAKE-LIST -1). In the longer term, it ought to become true
2918 that declarations are assertions even at SAFETY 1. For now, it's not
2919 quite true even at SAFETY 3, but it's at least more nearly true..
2920 (Note that this change seems to increases the size of the system by
2921 O(5%) and to decrease the speed of the compiler by 20% or more.)
2922 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2923 for the problem of printing dozens of lines of distracting information
2924 about low-level system machinery as part of the top stack frame
2925 on entry to the debugger when an undefined function was called.
2926 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2927 is not reset to COMMON-LISP-USER.
2928 * Compilation of stuff related to dyncount.lisp has been made conditional
2929 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2930 smaller. The various dyncount-related symbols have been moved into
2931 a new "SB-DYNCOUNT" package.
2932 * tty-inspect.lisp has been renamed to inspect.lisp.
2933 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2934 feature has gone away. (When we eventually port to other flavors of
2935 libc and/or Unix, we'll try to make the differences between flavors
2936 invisible at the user level.)
2937 * Various other *FEATURES* tags, and/or their associated conditionals,
2938 have been removed if obsolescent, or given better documentation, or
2939 sometimes given more-mnemonic names.
2941 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2943 * tidied up "make.sh" script
2944 * tidied up system directory structure
2945 * better "clean.sh" behavior
2946 * added doc/FOR-CMUCL-DEVELOPERS
2947 * many many small tweaks to output format, e.g. removing possibly-confusing
2948 trailing #\. character in DESCRIBE-INSTANCE
2949 * (EQUALP #\A 'A) no longer signals an error.
2950 * new hashing code, including EQUALP hashing
2951 * tidied up Lisp initialization and toplevel
2952 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2953 * command line argument processing
2954 * added POSIX-GETENV function to deal with Unix-ish environment variables
2955 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2956 terminating SBCL on EOF
2957 * non-verbose GC by default
2958 * There is no more "sbcl" shell script; the sbcl file is now the C
2959 runtime executable (just like CMU CL).
2960 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2962 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2963 symbol and package information as Lisp native symbol and package objects
2964 instead of strings naming symbols and strings naming packages. This way,
2965 whenever packages are renamed (as in warm init), debug information is
2966 transformed along with everything else.
2967 * tweaked the optimization policy declarations which control the building
2968 of SBCL itself. Now, among other things, the system no longer saves
2969 source location debugging information. (This helps two problems at once
2970 by reducing SBCL size and by keeping SBCL from trying to look for its
2971 sources -- which may not exist -- when reporting errors.)
2972 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2973 debugging and testing purposes
2974 * cleaned up printing, making the printer call PRINT-OBJECT for
2975 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2976 methods, giving nearly-ANSI behavior
2977 * converted almost all special variables to use *FOO* naming convention
2978 * deleted PARSE-TIME functionality, since it can be done portably
2979 * moved some files out of cold init into warm init
2980 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2982 * regularized formatting of source files
2983 * added an install.sh script
2984 * fixed ridiculous memory usage of cross-compiler by making
2985 compiler/alloc.lisp not try to do pooling unless it can hook
2986 itself into the GC of the cross-compilation host. Now the system
2987 builds nicely on my old laptop.
2988 * added :SB-ALLOC in target-features.lisp-expr
2989 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2990 was not implemented)
2991 * re-did condition handling and note reporting in the compiler. Notes
2992 are no longer handled by signalling conditions. Style warnings
2993 and warnings are handled more correctly and reported in such a way
2994 that it's easy to find one or the other in your output (so that you
2995 can e.g. figure out which of many problems caused COMPILE-FILE to
2997 * changed the severity of several compiler warnings from full WARNING
2998 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2999 compiler note reporting so that it doesn't use the condition system
3000 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3001 COMPILE-FILE command)
3002 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3003 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3004 no longer supported, since their implementation was deeply intertwingled
3005 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3006 * removed bogus "support" for compiler macros named (SETF FOO), and
3007 removed the compiler macro for SETF INFO (but only after making a fool
3008 of myself on the cmucl-imp mailing list by posting a bogus patch for
3009 DEFINE-COMPILER-MACRO..)
3010 * Compiled files containing forms which have side effects on the Lisp
3011 reader (such as DEFPACKAGE forms) are now handled more correctly.
3012 (Compiler queuing of top level lambdas has been suppressed by setting
3013 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3014 * deleted various currently-unused source files, e.g. gengc.lisp. They
3015 may be added back at some point e.g. when porting to other architectures,
3016 but until they are it's distracting to distribute them and to try to
3018 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3019 there were so many of them they're just distractions, and UNCROSS is
3020 known to be able to handle the current sources
3021 * moved PROFILE functionality into TRACE, so that it will be clear
3022 how the wrapping and unwrapping of functions when you profile them
3023 interacts with the wrapping and unwrapping of functions when you
3024 trace them. (Actually, the functionality isn't there yet, but at least
3025 the interface specification is there. Hopefully, the functionality will
3026 arrive with some future maintenance release.)
3027 * removed host-oops.lisp
3028 * changed signature of QUIT function to allow UNIX-CODE argument
3029 * fixed READ-SEQUENCE bug
3030 * tweaked verbose GC output so that it looks more like the progress
3031 output that ANSI specifies for functions like LOAD
3032 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3033 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob