1 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
3 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
5 * fixed loading of multiply forward-referenced layouts.
6 (thanks to Cheuksan Wang)
7 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
8 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
9 (see COMPILE-FILE documentation for details of :PRINT option)
10 * fixed bugs 19 and 317: fixed-format floating point printing is
11 more accurate. This also fixes a bug reported by Adam Warner
12 related to the ~@F format directive.
13 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
15 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
16 dispatch table if the from-char is a dispatch macro character.
17 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
18 by the wordsize no longer ignore the last word. (reported by Lutz
20 * fixed bug in type checking of dynamic-extent variables. (reported
22 * optimization: sequence traversal functions use their freedom to
23 coerce function designators to functions.
24 * optimization: code with many calls to CLOS methods specialized on
25 CLOS classes has had redundant type checks removed.
26 * optimization: added declarations to speed up operations that access
27 the internal character database (for example STRING-UPCASE)
28 * optimization: comparison operations between floats and sufficiently small
29 fixnums no longer create extra rationals
30 * fixed some bugs related to Unicode integration:
31 ** portions of multibyte characters at the end of buffers for
32 character-based file input are correctly transferred to the
33 start of the buffer at the next read.
34 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
35 passing it through to OPEN.
36 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
37 argument given to internal calls to OPEN.
38 ** the built-in comment readers (introduced by character sequences
39 ";" and "#|") are more forgiving to encoding errors; they will
40 STYLE-WARN and then attempt to resync the stream at a character
41 boundary. (thanks to Teemu Kalvas)
42 * fixed some bugs in the x86-64 port:
43 ** Negative short int return values from c-calls are sign-extended
45 ** The stack is aligned to 16-bytes for c-calls, as required by
46 the ABI. (thanks to Cheuksan Wang)
47 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
48 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
49 for enums. (thanks to Vincent Arkesteijn)
50 ** Multiple small optimizations and bugfixes for floating point
52 * fixed some bugs revealed by Paul Dietz' test suite:
53 ** Space, Tab, Linefeed, Return and Page have the invalid
54 secondary constituent character trait.
55 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
57 ** WITH-INPUT-FROM-STRING should only update the index place on
59 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
61 ** Bit-array operations (BIT-AND and similar) worked incorrectly
62 with one-dimensional arrays with fill pointers.
64 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
65 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
66 platform. The port supports 61-bit fixnums, large memory spaces
67 and reloading shared object files.
68 * enhancement: saving cores with foreign code loaded is now
69 supported on ppc/Darwin in addition to the previously supported
71 * enhancement: the statistical profiler now walks deeper into the
72 call stack for more meaningful call-graphs and accrued time
73 reports (x86/x86-64 only). It also now reports time spent in
75 * enhancement: it is now possible to trace most individual methods
76 of a generic function in addition to tracing the generic function
78 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
80 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
81 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
82 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
83 produces an error. (thanks to Vincent Arkesteijn)
84 * bug fix: NAMESTRING on pathnames with :WILD components in their
85 directories works correctly. (thanks to Artem V. Andreev)
86 * fixed bug 125: compiler preserves identity of closures. (reported
88 * bug fixed: functions with &REST arguments sometimes failed with
89 "Undefined function" when compiled with (DEBUG 3). (reported by
91 * bug fix: overflow during compiling of setting element of a bit
92 vector with constant index and value. (reported by Timmy Douglas)
93 * build fix: fixed the dependence on *LOAD-PATHNAME* and
94 *COMPILE-FILE-PATHNAME* being absolute pathnames.
95 * on x86 compiler partially supports stack allocation of dynamic-extent
97 * GO and RETURN-FROM do not check the extent of their exit points
98 when compiled with SAFETY 0.
99 * fixed some bugs related to Unicode integration:
100 ** encoding and decoding errors are now much more robustly
101 handled; it should now be possible to recover even from invalid
102 input or output to the terminal. (thanks to Teemu Kalvas)
103 ** provided a first cut at implementing STRING-TO-OCTETS and
104 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
105 ** altered the SB-MD5 contributed module slightly, changing the
106 interface just enough for it to be supportable for builds where
107 lisp characters are not eight bits.
108 * fixed some bugs revealed by Paul Dietz' test suite:
109 ** the FORMATTER-generated functions for ~V[ conditionals require
110 the correct number of arguments.
111 ** READ-FROM-STRING returns the mandated second value when applied
112 to displaced strings.
113 ** the #\Rubout and #\Backspace characters are treated as invalid
114 constituent characters by the tokenizer.
116 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
117 * new feature: reloading changed shared object files with
118 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
119 * new feature: references to foreign variables and functions
120 can now be compiled and loaded before the corresponding shared
121 object file is loaded, as long as the foreign definitions are
122 available at runtime.
123 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
124 supported. (thanks to Dan Debertin)
125 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
126 just keywords. (thanks to Vincent Arkesteijn)
127 * fixed bug #331: structure-class instances corresponding to
128 DEFSTRUCT forms are now created eagerly.
129 * fixed bug #345: backtraces from calls to undefined functions work
130 on x86 as well. Related bug #61 is now also partially fixed on x86
131 (backtraces from throws to unknown catch tags.)
132 * bug fix: lambda-list parsing is now stricter vrt. order and number
133 of lambda-list keywords.
134 * bug fix: as specified by AMOP, an error is signalled if a
135 class-option appears multiple times in a DEFCLASS form. (reported
137 * bug fix: RANDOM can be compiled when the compiler derives the type
138 of its numeric argument as a disjoint set of small integers.
139 (reported by Paul Dietz)
140 * bug fix: starting a core saved with shared objects loaded when
141 those objects are not available no longer causes threaded SBCL to
142 hang. (reported by Sean Ross)
143 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
145 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
146 argument, which must always be NIL. (reported by Kalle Niemitalo)
147 * bug fix: printing 1.0d+23 no longer results in an error.
148 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
149 * bug fix: structure slot setters preserve evaluation order. (thanks
151 * fixed some bugs related to Unicode integration:
152 ** RUN-PROGRAM can allow its child to take input from a Lisp
153 stream. (reported by Stefan Scholl)
154 ** ASDF-INSTALL successfully downloads .tar.gz files in all
155 locales. (reported by Ken Causey)
156 * fixed some bugs revealed by Paul Dietz' test suite:
157 ** INCF, DECF and REMF evaluate their place form as specified in
159 ** FORMATTER expands ~{ ~} iteration directives with V or #
160 parameters correctly.
161 ** FORMATTER deals with the ~@[ ~] conditional directive where the
162 consequent uses no arguments correctly.
163 ** the system has a partial understanding of the (COMPLEX RATIO)
165 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
168 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
169 * new feature: a build-time option (controlled by the :SB-UNICODE
170 keyword feature, enabled by default) for building the system with
171 support for the entire 21-bit character space defined by the
173 * new feature: the system now has rudimentary external-format
174 support; the primary user-visible change at this time is that
175 characters with the high bit set (such as lower-case-e-acute) will
176 print correctly to a terminal in a UTF-8 environment.
177 * minor incompatible change: BASE-CHAR no longer names a class;
178 however, CHARACTER continues to do so, as required by ANSI.
179 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
180 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
181 has been moved to the SB-EXT package (temporarily re-exported from
183 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
184 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
186 * fixed bug #351: better error-handlind and reporting for malformed
188 * fixed bug #350: bignum-printing is now more memory-efficient,
189 allowing printing of very large bignums, eg. (expt 2 10000000).
190 (reported by Bruno Haible)
191 * fixed bug #302: better primitive-type selection for intersection
193 * fixed bug #308: non-graphic characters now all have names, as
194 required. (reported by Bruno Haible)
195 * bug fix: redefining a class with different superclasses now correctly
196 removes it from the direct-subclasses of its previous superclasses.
197 (reported by David Morse)
198 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
199 new value now works. (reported by Bruno Haible)
200 * bug fix: correct canonicalization of multiple non-standard slot
201 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
202 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
203 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
204 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
205 now exists, an signals an error.
206 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
207 during macroexpansion for non-iteration variables as well. (reported
208 by Bruno Haible for CMUCL)
209 * bug fix: Cyclic structures and unprintable objects in compiler
210 messages no longer cause errors. (reported by Bruno Haible)
211 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
212 and READ-FROM-STRING all now return a primary value of NIL if
213 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
214 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
215 (reported by Bruno Haible for CMUCL)
216 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
217 do not designate a condition. (reported by Bruno Haible for
219 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
220 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
221 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
222 permitting resolution in favour of any of the conflicting symbols.
223 (reported by Bruno Haible for CMUCL)
224 * bug fix: EQUAL compiler optimizations is less aggressive on
225 strings which can potentially compare true despite having distinct
226 specialized array element types.
227 * bug fix: unit enumerations can be defined without dividing by
228 zero. (thanks to Vincent Arkesteijn)
229 * FORMAT compile-time argument count checking has been enhanced.
230 (report from Bruno Haible for CMUCL)
231 * a partial workaround for the bug 262: the compiler does not try to
232 inline-expand a local function doing RETURN-FROM from a deleted
233 BLOCK. (thanks to Peter Denno for the bug report and to David
234 Wragg for the simple test case)
235 * fixed some bugs revealed by Paul Dietz' test suite:
236 ** RENAME-PACKAGE allows all package designators as new package
238 ** constraint propagation and lambda variable substitution are
239 more cautious in dealing with partially deleted code.
240 ** compiler sometimes lost reoptimization passes.
241 ** CERROR, given a condition as condition designator, passes its
242 remaining arguments to the continue format control without
244 ** Case-altering FORMAT directives work correctly on non-ASCII
246 ** The REAL type specifier handles bounds outside the floating
247 point ranges without signalling FLOATING-POINT-OVERFLOW.
248 ** Functions with IR1-transformations can create intercomponent
249 references to global functions.
250 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
252 ** FORMAT ~R treats a nil value for its first parameter correctly.
254 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
255 * enhancement: saving cores with foreign code loaded is now
256 supported on x86/NetBSD and sparc/Linux in addition to the previously
258 * bug fix: on some platforms repeated installations caused multiple
259 copies of HTML documentation to be installed -- should not happen
260 any more. (reported by Stefan Scholl)
261 * bug fix: parsing self-recursive alien record types multiple times
262 no longer causes infinite recursion. (reported by Thomas F. Burdick,
263 original patch by Helmut Eller for CMUCL)
264 * bug fix: stack-exhaustion detection works now on NetBSD as well.
265 (thanks to Richard Kreuter)
266 * bug fix: defining classes whose accessors are methods on existing
267 generic functions in other (locked) packages no longer signals
268 bogus package lock violations. (reported by François-René Rideau)
269 * bug fix: special variables as DEFMETHOD parameters no longer have
270 associated bogus type declarations. (reported by David Wragg and
272 * bug fix: read-write consistency on streams of element-type
273 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
274 * bug fix: redefiniton of the only method of a generic function with
275 no DEFGENERIC no longer emits a full WARNING. In addition,
276 redefinition of generic functions with no DEFGENERIC to an
277 incompatible lambda list now signals an error. (thanks to Zach
279 * bug fix: DEFGENERIC now works even when there's a function of the
280 same name in an enclosing lexical environment. (thanks to Zach
282 * fixed compiler failure, caused by instrumenting code during
283 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
284 * optimization: added loop analysis and improved register allocation
285 * fixed some bugs revealed by Paul Dietz' test suite:
286 ** POSITION on displaced vectors with non-zero displacement
287 returns the right answer.
288 ** (SIMPLE-STRING) is a valid type specifier for sequence
290 ** *PRINT-LEVEL* handling for slotless structures is pedantically
292 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
293 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
295 ** The justification version of the FORMAT ~< directive treats
296 non-zero minpad parameter correctly.
298 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
299 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
300 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
301 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
302 the supported interface.
303 * new feature: Single-stepping of code compiled with DEBUG 2 or
304 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
305 * new feature: saving cores with foreign code loaded is now
306 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
307 Timothy Moore's work for CMUCL)
308 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
309 parameters to * instead of NIL if no initform is supplied.
310 (reported by Johan Bockgaard)
311 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
312 correctly when FUNCALL appears as the car of the form. Note:
313 despite this FUNCALL forms are not currently subject to
314 compiler-macro expansion. (port of Raymond Toy's fix for the
315 same from CMUCL, reported by Johan Bockgaard)
316 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
317 (thanks for Teemu Kalvas)
318 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
319 case instead of terminating the iteration (thanks for Julian
320 Squires, Sean Champ and Raymond Toy)
321 * bug fix: incorrect expansion of defgeneric that caused a style
322 warning. (thanks for Zach Beane)
323 * bug fix: slot accessor effective method computation works properly
324 for classes with multiple non-standard applicable methods on
325 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
326 * on x86 compiler supports stack allocation of results of LIST and
327 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
328 CMUCL implementation by Gerd Moellmann)
329 * fixed some bugs revealed by Paul Dietz' test suite:
330 ** FORMAT strings with both the ~<~:;~> form of the justification
331 directive and pretty-printing directives cause an error.
333 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
334 * incompatible change: the internal functions
335 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
336 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
337 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
338 instead of the old functions.
339 * new feature: on platforms where "dladdr" is available foreign
340 function names now appear in backtraces. (based on Helmut Eller's
342 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
343 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
345 * documentation: ASDF manual is now distributed with SBCL.
346 * bug fix: SBCL can now load its contributed modules with REQUIRE
347 even if the system-provided entries have been removed from
348 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
350 * bug fix: backtraces involving undefined functions or assembly
351 routines are more informative. (thanks to Brian Downing)
352 * bug fix: mutually referent alien structures now work correctly.
353 (reported by Rick Taube)
354 * bug fix: structures defined by WITH-ALIEN can be referred to
355 within other definitions in the same WITH-ALIEN.
356 * bug fix: division operators (MOD, TRUNCATE and the like) with
357 constant zero divisors and integer dividends no longer generate
359 * bug fix: provide default methods for INPUT-STREAM-P and
360 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
361 * bug fix: improve the detection and resolution of MOP metacycles
362 (where effective method computation depends on the generic function
363 having its effective method computed). (reported by Bruno Haible)
364 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
365 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
366 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
367 * bug fix: pathnames with relative directory components can be
368 represented relative to default pathnames.
369 * optimization: in taking the GCD of bignums, reduce the two bignums
370 to approximately the same size (using Euclid's algorithm) before
371 applying the more sophisticated binary GCD. (thanks to Juho
373 * optimization: COUNT on bitvectors now operates word-at-a-time.
374 * optimization: ASH with a positive, but not necessarily constant,
375 (leftwards) shift, when in a modular context, is compiled to a
377 * fixed some bugs revealed by Paul Dietz' test suite:
378 ** FORMAT variable parameters ("~V<char>") are defaulted properly
379 if the corresponding argument is NIL.
380 ** FORMAT directives accepting a minpad parameter treat negative
382 ** literal commas as character parameters to FORMAT directives are
384 ** literal spaces directly after ~<Newline> directives within a
385 format-logical-block (~:< ~@:>) do not induce :FILL-style
386 conditional newlines.
387 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
388 as specified: it no longer includes conditional newlines).
389 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
391 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
392 :PER-LINE-PREFIX argument does not evaluate to a string.
394 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
395 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
396 the manual for details; for now, package locks can be disabled by
397 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
398 if no major problems are found then it is likely that they will be
399 compiled in unconditionally.
400 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
401 unsupported operators on all platforms. To load a shared library
402 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
403 object file, link it into a shared library outside of SBCL and
404 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
406 * bug workaround: ROOM T and the GC were not working together
407 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
408 reporting facility aren't preserved. That reporting has been
409 disabled, so now until and unless someone figures out how to make it
410 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
411 * minor incompatible change: as threatened around sbcl-0.8.0, the
412 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
413 The symbols are also exported from SB-PCL for backwards
414 compatibility, but more so than before SB-PCL should be treated as
415 an implementation-internal package.
416 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
418 * the SB-SPROF contrib now works on (most) non-x86 architectures.
419 It is known as of this release not to work on the Alpha, however.
420 * fixed bug #167: errors signalled due to illegal syntax in method
421 bodies are now more legible.
422 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
423 designators and can hence be used with TYPEP.
424 * fixed bug #333: CHECK-TYPE now ensures that the type error
425 signalled, if any, has the right object to be accessed by
426 TYPE-ERROR-DATUM. (reported by Tony Martinez)
427 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
428 5.1.2.3 for multiple-value place subforms. (reported by Kalle
430 * fixed bug #334: programmatic addition of slots using specialized
431 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
432 and :ALLOCATION :CLASS slots.
433 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
434 float exponents. (rereported by Peter Seibel)
435 * fixed a bug: #\Space (and other whitespace characters) are no
436 longer considered to be macro characters in standard syntax by
438 * fixed bug: initialization of condition class metaobjects no longer
439 causes an instance of the condition to be created. (reported by
441 * fixed bug: it is now possible to have slots such that
442 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
443 description is neither :INSTANCE nor :CLASS.
444 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
445 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
446 SLOT-VALUE-USING-CLASS), and no longer causes errors with
447 non-standard SLOT-DEFINITION-ALLOCATION values.
448 * fixed bugs: various race conditions handling exiting threads.
449 CL-PPCRE's thread torture test now passes.
450 * fixed arguably-a-bug: GC time proportional to number of threads in
451 system even when most of them are idle
452 * optimization: improved performance of BIT and SBIT on bit-vectors.
453 * .fasl file incompatibility: The fasl file version number has
454 been incremented because of changes associated with package locks.
456 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
457 * minor incompatible change: the system no longer provides
458 optimization hints (conditions which are TYPEP
459 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
460 mode; these hints are still emitted when higher SPEED optimization
462 * new contrib module: a sampling profiler (profiling by statistical
463 sampling, rather than by instrumenting functions) is available as
464 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
465 * the behaviour of the standard function ED is now customizeable by
466 third parties through a hook variable: see ED's documentation
467 string for information on the protocol.
468 * the compiler no longer emits efficiency notes for (FUNCALL X)
469 when the type of X is uncertain under default optimization
471 * fixed bug 276: mutating a binding of a specialized parameter to a
472 method to something that is not TYPEP the specializer is now
474 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
475 exhibit the required behaviour on the x86 platform. (thanks to
476 Peter van Eynde, Eric Marsden and Bruno Haible)
477 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
478 even for difficult arguments. (reported by Peter Graves)
479 * fixed bug 141a: the backquote printer now descends quoted
481 * fixed another bug in backquote printing: no more destructive
482 modification of the form's list structure. (reported by Brian
484 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
485 move between its address being taken and the call to
486 interrupt_thread, fixing a crashing race condition.
487 * the SB-POSIX contrib implementation has been adjusted so that it
488 no longer exhibits ridiculously poor performance when constructing
489 instances corresponding to C structs.
491 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
492 * minor incompatible change: the sb-grovel contrib now treats C
493 structures as alien (in the sense of SB-ALIEN) objects rather than
494 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
495 has implications for memory management of client code
496 (sb-grovel-returned objects must now be manually managed) and for
497 type safety (alien objects now have full types).
498 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
499 used to control emission of compiler diagnostics, rather than the
500 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
501 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
502 quality should be considered deprecated.
503 * install.sh now installs the user manual as well
504 * (not quite a new documentable feature, but worth considering in
505 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
506 beginnings of a semantically meaningful condition hierarchy is
507 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
509 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
510 designator as the defaults argument.
511 * fixed bug: Displaced arrays whose displaced-to array has become
512 too small now cause ARRAY-DIMENSION to signal an error, providing
513 for safer bounds-checking. (reported by Bruno Haible)
514 * fixed bug: DEFCLASS slot definitions with identical :READER and
515 :WRITER names now signal a reasonable error. (reported by Thomas
517 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
519 * fixed bug: Invalid slot specification errors now print correctly.
520 (thanks to Zach Beane)
521 * fixed bug 320: Shared to local slot value transfers in class
522 redefinitions now happen corrently from superclasses as
523 well. (reported by Bruno Haible)
524 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
526 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
527 improper lists correctly. (reported by Bruno Haible)
528 * fixed bug 313: source-transform for <fun-name> was erroneously
529 applied to a call of a value of a variable with name <fun-name>.
530 (reported by Antonio Menezes Leitao)
531 * fixed bug 307: The obsolete instance protocol ensures that
532 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
533 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
534 remove dead unknown-values globs from the middle of the stack.
535 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
536 easier to resume long computations after using *BREAK-ON-SIGNALS*
537 to diagnose and fix failures. (thanks to Nikodemus Siivola)
538 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
539 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
540 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
541 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
542 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
544 * fixed a (fairly theoretical) bug in string printing: if
545 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
546 does not have array-element-type equal to the most general string
548 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
549 function name. (thanks to Nikodemus Siivola)
550 * fixed bug: compile-time format string checker failed on
551 non-closed ~{. (reported by Thomas F Burdick)
552 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
553 don't warn on legal loop code involving more than one aggregate
554 boolean. (thanks to Nikodemus Siivola)
555 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
556 in keyword position to MAKE-PACKAGE was not being reported as
558 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
559 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
561 * fixed bugs 280 and 312: the checking for multiple definitions in a
562 file is less likely to become confused by uses of inline
564 * fixed bug: the #S reader macro performs the keyword coercion
565 specified for slot names. (reported by Kalle Niemitalo)
566 * fixed bug: lambda lists may contain symbols whose names start with
567 & but are not lambda-list-keywords; their occurrence triggers a
569 * fixed bug 321: define-method-combination argument lambda lists do
570 not cause invalid code to be generated when &optional or &aux
571 variables have default values. (reported by Bruno Haible)
572 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
573 have CLOS classes; this appears also to have fixed reported
574 instability in stack exhaustion detection.
575 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
576 MUFFLE-WARNING when no associated restart is present is now
578 * optimization: rearranged the expansion of various defining macros
579 so that each expands into only one top-level form in a
580 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
582 * optimization: used a previously little-used slot in symbols to
583 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
585 * fixed some bugs revealed by Paul Dietz' test suite:
586 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
588 ** Fixed an optimization bug related to inheritance of initforms
589 from local to shared slots.
590 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
591 ** CAST optimizer forgot to flush argument derived type.
592 ** print/read consistency on floats is now orders of magnitude
593 more likely. (thanks also to Bruno Haible for a similar report
595 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
596 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
597 for objects of type REAL. Make it so.
598 ** SXHASH is specified (infelicitously) to respect similarity,
599 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
600 it so. (thanks to Markus Ziegler)
601 ** on the Alpha, the compiler succeeds in compiling functions
602 returning a known number of arguments greater than 63.
603 ** fixed handling of invalid NIL arguments in keyword position
605 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
607 ** provided a readably-printable representation for RANDOM-STATE
609 ** ensured that pathnames, where they have a namestring, always
610 print using #P"..." syntax.
612 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
613 * Support for the forthcoming 2.0 version of the NetBSD kernel
614 running on x86 hardware has been added. (thanks to Perry
615 E. Metzger most immediately, and others for their past work)
616 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
617 current release version; SBCL's OpenBSD support had been broken
618 since about the time of OpenBSD's switch to ELF binary format.)
619 * [placeholder for DX summary]
620 ** user code with &REST lists declared dynamic-extent, under high
621 speed or space and low safety and debug optimization policy.
622 * The manual has been converted to Texinfo format and the debugger
623 chapter from the cmucl manual has been added.
624 * A facility has been added to extract documentation strings from
625 sbcl and store them as Texinfo-formatted snippets for inclusion in
626 the manual (via Texinfo's @include directive)
627 * bug fix: compiler emitted division in optimized DEREF. (thanks for
628 the test case to Dave Roberts)
629 * bug fix: multidimensional simple arrays loaded from FASLs had fill
630 pointers. (reported by Sean Ross)
631 * bug fix: PROFILE output is printed nicely even for large numerical
632 values. (thanks to Zach Beane)
633 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
634 greater than 32 handle EOF correctly.
635 * bug fix: on X86 an immediate argument of the IMUL instruction is
636 correctly printed in disassembly. (thanks to Lutz Euler)
637 * bug fix: class slots in redefined classes preserve their old
638 values. (thanks to Bruno Haible and Nikodemus Siivola)
639 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
640 succeeds. (reported by Marco Baringer)
641 * fixed some bugs revealed by Paul Dietz' test suite:
642 ** READ-SEQUENCE now works on ECHO-STREAMs.
643 ** RATIONALIZE works more according to its specification. (thanks
646 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
647 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
648 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
649 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
650 should matter to you only if you rebind the printer control
651 variables and then find you want different bindings in the
652 debugger than in the ordinary execution of your program.)
653 * The runtime build system has been tweaked to support building
654 (on SPARC/SunOS) using a C compiler which invokes Sun's own
655 assembler and linker. (thanks to Nikodemus Siivola)
656 * Unbound, undefined, undeclared variables now trigger full
657 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
658 more often programmer error than deliberate exploitation of undefined
660 * optimization: the hash algorithm for strings has changed to one
661 that is less vulnerable to spurious collisions. (thanks to Juho
663 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
664 needless bounds checking. (thanks to Juho Snellman)
665 * optimization: implemented multiplication as a modular
666 (UNSIGNED-BYTE 32) operation on the PPC backend.
667 * fixed some bugs revealed by Paul Dietz' test suite:
668 ** ADJUST-ARRAY now copies the datum in a zero rank array if
670 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
671 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
672 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
674 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
676 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
678 ** ECHO-STREAMs no longer attempt to echo the end of file value to
679 their output stream on EOF from read.
680 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
681 have been read to end-of-file.
682 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
684 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
685 description of determination of which consecutive characters
687 ** Printing the "Space" character with escaping on now yields "#\\ ",
688 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
689 ** Reading floating-point numbers with *READ-BASE* set to a number
690 less than 10 works correctly.
691 ** Reading floating-point numbers with *READ-BASE* set to a number
692 more than 10 works correctly.
693 ** Printing with *PRINT-READABLY* targets the standard readtable, not
694 the readtable currently in effect.
696 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
697 * minor incompatible change: parsing of namestrings on a physical
698 (Unix) host has changed; numbers after the final #\. in a
699 namestring are no longer interpreted as a version field. This is
700 intented to be largely invisible to the user, except that the
701 meaning of the namestring "*.*.*" has changed: it now refers to a
702 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
703 should usually be replaced by
704 (make-pathname :name :wild :type :wild :version :wild)
705 with the added benefit that this is more likely to be portable.
706 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
707 option to OPEN now signals an error if the file being opened
708 exists; this may have an impact on existing code.
709 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
710 signals received, on the PowerPC platforms (both Linux and
711 Darwin). (thanks to Pierre Mai for pointing out the location of
713 * several fixes on OS X: The system now builds and runs cleanly on
714 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
715 (thanks to Brian Mastenbrook)
716 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
717 second-resolution: integer multiples of 1/3600 between -24 and 24.
718 (thanks to Vincent Arkesteijn)
719 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
720 type when called with 1 argument; PEEK-CHAR checked type of
721 PEEK-TYPE only after having read first character from a
722 stream. (reported by Peter Graves)
723 * bug fix: the garbage collector now has much better locality
724 behaviour, and in particular no longer treats all memory as being
725 exhausted when many small objects point to each other in a deeply
727 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
728 recognized as being TYPEP their class.
729 * bug fix: the PUSHNEW documentation string has been corrected.
730 (thanks to Vincent Arkesteijn)
731 * bug fix: defaulting of the value for the last of an atypically
732 large number of multiple values being bound was not being
733 performed correctly on the Alpha or PPC platforms
734 * optimization: implemented multiplication as a modular
735 (UNSIGNED-BYTE 32) operation on the x86 backend.
736 * optimization: SEARCH on simple-base-strings can now be open-coded.
737 (see also contrib/compiler-extras.lisp for inspiration for
738 teaching the compiler about the Boyer-Moore algorithm).
739 * value, returned by MAX (and MIN) called with several EQUALP, but
740 not EQL, arguments now does not depend on compiler settings.
741 * fixed some bugs revealed by Paul Dietz' test suite:
742 ** in stack analysis liveness information is propagated from
743 non-local entry points.
744 ** pathwise CAST removing failed when the CAST node did not start
746 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
747 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
749 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
750 host is already defined.
751 ** RENAME-FILE works on streams instead of signalling an internal
753 ** PEEK-CHAR uses the current readtable when determining whether
754 or not a character is whitespace.
755 ** MERGE-PATHNAMES handles the case when the pathname does not
756 specify a name while the default-pathname specifies a version.
757 ** Pathnames now stand a better chance of respecting print/read
759 ** Attempting to use standardized file system operators with a
760 pathname with invalid :DIRECTORY components signals a
762 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
764 ** DIRECTORY on logical pathnames is more correct.
765 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
766 signal a TYPE-ERROR if their argument is not a stream.
767 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
768 designator argument does not designate a stream.
769 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
770 examining the synonym.
771 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
773 ** OPEN and WITH-OPEN-STREAM allow opening streams with
774 element-type larger than ([UN]SIGNED-BYTE 32).
776 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
777 * When built with the :SB-FUTEX feature, threaded builds now take
778 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
779 for faster/more reliable mutex and condition variable support.
780 * Incompatible change (but one you probably shouldn't have been using
781 anyway): the interface and code for arbitrating between multiple
782 threads in the same user session has been redesigned.
783 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
784 error when called without an explicit environment argument.
785 (thanks to Nikodemus Siivola)
786 * bug fix: buffered :DIRECTION :IO streams are less likely to become
787 confused about their position. (thanks to Adam Warner and Gerd
789 * bug fix: Pretty printing backquoted forms with unquotations in the
790 argument list position of various code constructs such as LAMBDA
791 now works correctly. (reported by Paul Dietz)
792 * bug fix: Pretty printing unquotations no longer loses all
793 stream position information.
794 * optimization: performance of string output streams is now less
795 poor for multiple small sequence writes.
796 * optimization: performance of CSUBTYPEP in the presence of complex
797 expressions involving CONS and NOT many times has been improved.
798 (reported by Paul Dietz)
799 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
801 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
802 streams for the REPL, for the debugger, and so on. (thanks to
804 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
805 * fixed some bugs revealed by Paul Dietz' test suite:
806 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
807 is ignored if it is too small, rather than propagating through
808 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
809 ** extremely complex negations of CONS types were not being
810 sufficiently canonicalized, leading to inconsistencies in
812 ** VALUES tranformer lost derived type.
814 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
815 * fixed a bootstrapping bug: the build process no longer assumes
816 that the various BOOLE-related constants have the same value in
817 host and target lisps. (noted by Paul Dietz' test suite on an
818 SBCL binary built from CLISP)
819 * The system can now be dynamically linked on the MIPS platform,
820 which enables dynamic loading of foreign code from Lisp. (thanks
821 to Ralf Baechle for discussions on the MIPS ABI)
822 * The system now records debugging information for its own source
823 files in a filesystem-position-independent manner, relative to
824 the "SYS" logical host.
825 * fixed a compiler bug: MV-LET convertion did not check references
826 to the "max args" entry point. (reported by Brian Downing)
827 * tweaked disassembly notes to be less confident about proclaiming
828 some instruction as an LRA. (thanks to Brian Downing)
829 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
830 now each have their own history, command character, and other
831 characteristics. (thanks to David Lichteblau)
832 * fixed some bugs revealed by Paul Dietz' test suite:
833 ** compiler failure in compiling LOGAND expressions including a
835 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
836 shift greater than 32.
837 ** FLUSH-DEST did not mark blocks for type check regeneration.
838 ** HANDLER-CASE failed to accept declarations in handler clauses
839 in some circumstances.
841 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
842 * New code in contrib/sb-introspect (still probably not entirely
843 stable yet) provides some support for smart Lisp development
844 environments like SLIME.
845 * The conditions signalled for errors occurring when loading .fasl
846 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
847 in a way which should help ASDF recover gracefully.
848 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
849 stuff has been cleaned up. If you code contrib/ stuff, this might
850 affect you, and you can look at contrib/README, contrib/STANDARDS,
851 and/or the 0.8.4.27 diff to check.
852 * In full calls the compiler now does not generate checks for declared
853 argument types for all arguments.
854 * various threading fixes
855 ** and some experimental patches which didn't make it into
856 the main tree for this release, but which are shipped in
857 contrib/experimental-thread.patch as a possible fix for some
858 failures (deadlock, spinning...) in GC-intensive multithreaded
860 * fixed PPC build problem (source code incompatibility of different
861 library versions): added offsetof() hackery which attempts to divine
862 where glibc maintainers put uc_mcontext today
863 * fixed bug 282: compiler does not trust type assertions while passing
864 arguments to a full call.
865 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
866 &OPTIONAL VALUES type parameter.
867 * fix bug 214: algorithm for noting rejected templates is now more
868 similar to that of template seletion. (also reported by rydis on
870 * fixed bug 141b: printing backquoted information readably and prettily
871 inserts a space where necessary.
872 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
873 (CHARACTER 10) are now reported as errors, rather than propagated
874 as unknown types. (reported by piso on #lisp)
875 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
876 with values NIL and :ERROR. (thanks to Milan Zamazal)
877 * fixed bug 191c: CLOS now does proper keyword argument checking as
878 described in CLHS 7.6.5 and 7.6.5.1.
879 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
880 longer raise an error; further, using a list as a for-as-arithmetic
881 counter now raises a meaningful error.
882 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
883 now understood by sequence creation functions such as MAKE-SEQUENCE
885 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
886 a STRING-INPUT-STREAM.
887 * compiler enhancement: SIGNUM is now better able to derive the type
889 * type declarations inside WITH-SLOTS are checked. (reported by
891 * fixed some bugs revealed by Paul Dietz' test suite:
892 ** incorrect optimization of TRUNCATE for a positive first
893 argument and negative second.
894 ** compiler failure in let-convertion during flushing dead code.
895 ** compiler failure while deriving type of TRUNCATE on an
896 interval, containing 0.
897 ** ASH of a negative bignum by a negative bignum count now returns
899 ** intersection of CONS types now canonicalizes properly, fixing
900 inconsistencies in SUBTYPEP.
902 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
903 * incompatible change: The --disable-debugger command line
904 option now clobbers the debugger at a more fundamental
905 level, by redefining #'INVOKE-DEBUGGER instead of by
906 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
907 is specified by ANSI to ignore *DEBUGGER-HOOK* and
908 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
909 enter the debugger REPL and then suffer recursive errors
910 because *DEBUG-IO* is also messed up in --disable-debugger mode;
911 while under the new system, BREAK in --disable-debugger mode
912 terminates the system just as an unhandled error would.
913 * fixed compiler performance when processing loops with a step >1;
914 * bug fix: DOCUMENTATION now retrieves generic function
915 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
916 support has been systematized, and now supports the methods
917 specified by ANSI, along with a default method and a method for
918 slot documentation. (reported by Nathan Froyd)
919 * bug fix: effective methods associated with a generic function are
920 no longer cached over a change of that generic function's method
921 combination. (reported by Andreas Fuchs)
922 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
923 names a non-existent directory. (thanks to Andreas Fuchs)
924 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
925 the PPC platform, lead to stack corruption; this has been fixed.
926 (reported by Rainer Joswig)
927 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
928 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
929 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
930 arithmetic in general, is now much more reliable on the PPC
932 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
933 platform now returns the right answer.
934 * optimization: restored some effective method precomputation in
935 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
936 precomputation is now tunable.
937 * optimization: compiler-internal data structure use has been
938 reviewed, and changes have been made that should improve the
939 performance of the compiler by about 20%.
940 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
941 simple-base-strings and simple-bit-vectors is improved.
942 * optimization: the optimization of 32-bit logical and arithmetic
943 functions introduced in version 0.8.3 on the x86 has been
944 implemented on the mips, ppc and sparc platforms; an
945 implementation of the same facility, but for 64-bit arithmetic,
946 has been added for the alpha.
947 * microoptimization: the compiler is better able to make use of the
948 x86 LEA instruction for multiplication by constants.
949 * bug fix: in some situations compiler did not report usage of
950 generic arithmetic in (SPEED 3) policy.
951 * bug 145b fix: compiler used wrong type specifier while converting
952 MEMBER-types to numeric.
953 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
955 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
956 index argument now works properly on the Alpha platform.
957 * bug fix: floating point exception treatment on the Alpha platform
959 * bug fix: FILE-POSITION works much better on string input and
960 output streams. (thanks to Nikodemus Siivola)
961 * bug fix: many threading/garbage collection symptoms sorted.
962 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
963 might be pseudo-atomic.
964 * internal change: Stopping for GC is now done with signals not
965 ptrace. GC is now done in whichever thread wanted it, instead of
967 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
969 * bug fix: result form in DO is not contained in the implicit
971 * incompatible change: ICR structure is changed; the value part of
972 CONTINUATION is now called LVAR; corresponding functions are
973 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
974 * added type deriver for ISQRT (thanks to Robert E. Brown).
975 * bug fix: better support for loading from the command line when an
976 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
978 * fixed some bugs revealed by Paul Dietz' test suite:
979 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
980 ** ROUND and FROUND now give the right answer when given very
981 small float arguments.
982 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
984 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
985 ** (GCD 0 <negative-integer>) returned <negative-integer>.
986 ** LCM should return a non-negative integer.
987 ** PARSE-INTEGER returned the index of a terminator instead of the
988 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
989 ** PARSE-INTEGER returned an incorrect index being applied to a
991 ** LCM with two arguments of 0 returns 0 rather than signalling
993 ** unsigned addition of a 32-bit constant with the high bit set no
994 longer causes an internal compiler error on the x86.
995 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
996 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
997 with negative last argument.
998 ** byte specifiers with zero size and position no longer cause
999 an error during type derivation.
1000 ** bignum multiplication on the Alpha platform now returns the
1002 * porting: The system now builds on SuSE AMD64, although it still
1003 generates a 32-bit binary.
1004 * .fasl file incompatibility: The fasl file version number has
1005 been incremented (because of the changes to internal compiler
1006 data structures referred to above).
1008 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1009 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1010 more accurately, on the Darwin kernel running on PowerPC hardware.
1011 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1012 * Compiler code deletion notes now signal a condition of type
1013 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1014 an associated MUFFLE-WARNING restart.
1015 * The compiler now performs limited argument count validation of
1016 constant format strings in FORMAT, and where appropriate in ERROR,
1017 CERROR and WARN. (thanks to Gerd Moellmann)
1018 * New ASDF-INSTALL contrib can be used for automatic download and
1019 installation of third-party Lisp code from CCLAN or other sites
1021 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1022 which forces another thread to execute a function supplied by the
1024 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1025 now accept and act upon their :ELEMENT-TYPE keyword argument.
1026 (reported by Martin Atzmueller, Edi Weitz)
1027 * bug fix: FILE-POSITION now accepts position designators up to
1028 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1029 is the greater. (thanks to Patrik Nordebo)
1030 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1031 Kalle Olavi Niemitalo)
1032 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1033 printable. (reported by Eric Marsden)
1034 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1035 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1036 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1037 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1038 (thanks to Stig Sandoe)
1039 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1040 UNREAD-CHAR now correctly includes the unread character in the
1041 target sequence. (thanks to Gerd Moellmann)
1042 * bug fix in threaded builds: the system can now be suspended and
1043 resumed by shell job control with minimal disruption.
1044 * bug fixes in times and timezones >2038AD
1045 * better handling of "where is GNU make?" problem in build scripts
1046 (thanks to Nikodemus Siivola)
1047 * new optimization: inside a named function any reference to a
1048 function with the same name is considered to be a self-reference;
1049 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1050 optimization quality.
1051 * new optimization on x86: logical functions and + now have
1052 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1053 used when the result is truncated to 32 bits.
1054 * VALUES declaration is partially enabled.
1055 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1056 * fixed some bugs revealed by Paul Dietz' test suite:
1057 ** The system now obeys the constraint imposed by
1058 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1059 types form a lattice under type intersection.
1060 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1061 ** ASSOC now ignores NIL elements in an alist.
1062 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1063 and (1+ MOST-POSITIVE-FIXNUM) answers.
1064 ** The addition of a method with invalid qualifiers to a generic
1065 function does not cause an error to be signalled immediately;
1066 a warning is signalled, and the error is generated only on
1067 calling the generic function.
1068 * changed .fasl file version number, in part to add type codes for
1069 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1070 obscure ANSI requirements
1072 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1073 * fixed bug 148: failure to inline-expand a local function left
1074 garbage, confusing the compiler.
1075 * fixed bugs 3cd: structure slot readers perform type check if the
1076 slot can have an invalid value (i.e. it is either not initialized
1077 or can be written with a less specific slot writer).
1078 * bug fix: the compiler now traps array references to elements off
1079 the end of an array; previously, the bounds checking in some
1080 circumstances could go off-by-one.
1081 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1083 * type declarations for array element types now obey the description
1084 on the CLHS page "Declaration TYPE", as per discussions on
1085 sbcl-help around 2003-05-08. This means that a declaration
1086 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1087 declaration, all references to BAR will be asserted or assumed
1088 (with THE, so dependent on compiler policy) to involve objects of
1089 type FOO. Note that no such declaration is implied in
1090 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1091 * declared types of functions from the "Conditions"
1092 chapter. (reported by Paul Dietz)
1093 * bug fix: CERROR accepts a function as its first argument.
1094 * bug fix: NTH an NTHCDR accept a bignum as index
1095 arguments. (reported by Adam Warner)
1096 * optimization: character compare routines now optimize comparing
1097 against a constant character. (reported by Gilbert Baumann)
1098 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1099 argument now works properly on the MIPS platform.
1100 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1101 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1102 declared to have a functional type, does not cause an error
1104 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1105 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1106 * fixed bug: if last continuation of a deleted block has a
1107 destination, this destination should be deleted too. (reported by
1109 * fixed a bug in the bootstrap process: the host compiler's values
1110 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1111 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1112 test case from Patrik Nordebo)
1113 * improved the ability of the disassembler on the PPC platform to
1114 provide helpful disassembly notes.
1115 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1116 the class in more cases than previously.
1117 * bug fix: FILE-POSITION now understands :START and :END for
1118 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1119 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1120 8) data. (thanks to David Lichteblau)
1121 * bug fix: it is possible to add a method to a generic function
1122 without lambda list.
1123 * bug fix: reader failed to signal END-OF-FILE inside an
1124 object representation. (reported by Nikodemus Siivola)
1125 * fixed some bugs revealed by Paul Dietz' test suite:
1126 ** LAST and [N]BUTLAST should accept a bignum.
1127 ** condition slot accessors are methods.
1128 ** (VECTOR NIL) is a subtype of STRING.
1130 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1131 * minor incompatible change: some nonsensical specialized lambda
1132 lists (used in DEFMETHOD) which were previously ignored now signal
1134 * minor incompatible change: the system is now aware of the types of
1135 variables in the COMMON-LISP package, and will signal errors for
1136 most violations of these type constraints (where previously they
1137 were silently accepted).
1138 * minor incompatible change: COMPILE-FILE now uses the freedom
1139 afforded (ANSI 3.2.2.3) to use derived function types for
1140 functions defined in the same file. This also permits the system
1141 to warn on static type mismatches and function
1142 redefinition. (Currently it does not work with high DEBUG level.)
1143 * minor incompatible change: VALUES declaration is disabled.
1144 * When issuing notes, the compiler now signals a condition of type
1145 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1146 restart for use in user handlers. It is expected that the
1147 COMPILER-NOTE condition will eventually become a condition
1148 supertype to a hierarchy of note types, which will then be
1149 handleable in a similar fashion. However, at the moment, no such
1150 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1151 but it's an implementation detail, not a classification for the
1153 * Changes in type checking closed the following bugs:
1154 ** type checking of unused values (192b, 194d, 203);
1155 ** template selection based on unsafe type assertions (192c, 236);
1156 ** type checking in branches (194bc).
1157 * A short form of VALUES type specifier has ANSI meaning (it has
1158 increased the number of situations when SBCL cannot perform type
1160 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1161 keywords or constants is permissible.
1162 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1163 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1164 classes. (thanks to Antonio Martinez)
1165 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1166 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1167 argument to a pretty-print stream. Instead, it leaves any such
1168 operations to DESCRIBE-OBJECT methods.
1169 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1171 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1172 lambda lists. (thanks to David Lichteblau)
1173 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1174 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1175 (thanks to Gerd Moellmann)
1176 * fixed bug 246: increased compilation speed of long
1177 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1179 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1181 * DEFCONSTANT now throws a condition of type
1182 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1183 constant to a non-EQL value; CONTINUE and ABORT restarts
1184 respectively change and preserve the value.
1185 * fixed bug 63: The code walker, part of the implementation of CLOS,
1186 is now better at handling symbol macros.
1187 * bug fix: There is no longer an internal implementation type named
1188 CL:LENGTH. (reported by Raymond Toy)
1189 * bug fix: In macro-like defining macros/special operators the
1190 implicit block does not enclose the lambda list.
1191 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1192 enclosed in parentheses) are not suitable as type specifiers, and
1193 their use properly signals an error now.
1194 * bug fix: An argument count mismatch for a type specifier in code
1195 being compiled no longer causes an unhandled error at compile
1196 time, but signals a compile-time warning.
1197 * fixed simple vector readable printing
1198 * bug fix: DESCRIBE takes more care over whether the class
1199 precedence list slot of a class is bound before accessing it.
1200 (reported by Markus Krummenacker)
1201 * bug fix: FORMATTER can successfully compile pretty-printer format
1202 strings which use variants of the ~* directive inside.
1203 * bug fix: SEARCH now applies its TEST predicate to the elements of
1204 the arguments in the correct order. (thanks to Wolfhard Buss)
1205 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1206 in type inference. (thanks to Robert E. Brown)
1207 * bug fix: Reading in symbols with an explicit package name of ""
1208 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1209 * fixed some bugs revealed by Paul Dietz' test suite:
1210 ** NIL is now allowed as a structure slot name.
1211 ** Arbitrary numbers, not just REALs, are allowed in certain
1212 circumstances in LOOP for-as-arithmetic clauses.
1213 ** Multiple class redefinitions before slot access no longer
1214 causes a type error.
1215 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1216 association between the name and a class.
1217 ** Generic functions with non-standard method-combination and over
1218 five methods all of which return constants no longer return NIL
1219 after the first few invocations. (thanks to Gerd Moellmann)
1220 ** CALL-NEXT-METHOD with no arguments now passes the original
1221 values of the arguments, even in the presence of assignment.
1222 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1224 ** DISASSEMBLE works with closures and funcallable instances.
1225 ** ADD-METHOD now returns the generic function, not the new method.
1226 ** FIND-METHOD signals an error if the lengths of the specializers
1227 is incompatible with the generic function, even if the ERRORP
1229 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1230 which its argument is a member.
1231 ** DEFCLASS only redefines the class named by its class-name
1232 argument if that name is the proper name of the class;
1233 otherwise, it creates a new class.
1234 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1235 of the UNBOUND-SLOT condition to the name of the slot.
1236 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1237 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1238 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1239 SLOT-MAKUNBOUND in the specified fashion.
1241 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1242 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1243 cross-compilation host. As a consequence, we can now bootstrap our
1244 way up to SBCL starting with a bare gcc toolchain and human-readable
1245 source code (first the source to CLISP, then the source to SBCL).
1246 * A contributed module containing a partial implementation of the
1247 simple-streams interface has been included. (thanks to Rudi
1249 * A contributed module implementing the RFC1321 Message Digest
1250 Algorithm, known as MD5, has been included.
1251 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1252 no longer has any effect, as the code controlled by this feature
1253 has been deleted. (As far as we know, no-one has ever built using
1254 this feature, and its semantics were confused in any case).
1255 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1256 work on conditions (as required by the ANSI specification),
1257 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1258 expected behaviour on conditions. Users should note, however,
1259 that such behaviour is not required by the ANSI specification,
1260 and so use of this behaviour may render their code unportable.
1261 * fixed some bugs revealed by Paul Dietz' test suite:
1262 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1264 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1265 ** SXHASH on condition objects no longer returns NIL.
1266 ** :ALLOCATION :CLASS slots are better treated; their values are
1267 updated on class redefinition, and initforms inherited from
1268 superclasses are applied.
1269 ** REMOVE-METHOD returns its generic function argument even when
1270 no method was removed.
1271 ** SHARED-INITIALIZE now initializes the values of the requested
1272 slots, including those with :ALLOCATION :CLASS.
1273 ** ALLOCATE-INSTANCE now works on structure classes defined via
1274 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1276 ** SLOT-EXISTS-P now works on conditions, as well as structures
1278 ** MAKE-LOAD-FORM now has the required methods on
1279 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1280 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1281 keyword, and now implements the SLOT-NAMES argument.
1282 ** methods with &OPTIONAL arguments no longer allow too many
1283 arguments to be passed in the call without error.
1284 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1285 option is consistent with the required arguments of the generic
1286 function lambda list.
1287 * bug fix: REQUIRE accepts a string designator. (Thanks to
1289 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1290 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1291 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1292 * bug fix: The compiler now checks for duplicated variables in macro
1294 * bug fix: SETQ on globals returns the correct value.
1295 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1296 not a COMPILER-ERROR (followed by some other strange error on
1297 choosing the CONTINUE restart).
1298 * bug fix: make.sh and friends are now more consistent in the way that
1299 they look for GNU "make".
1301 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1302 * experimental native threads support (on x86 Linux >=2.4 only).
1303 This is not compiled in by default: you need to add :SB-THREAD to
1304 the target features. See the "Beyond ANSI" chapter of the manual
1306 * fix for longstanding nonANSIism: The old distinction between
1307 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1308 The return value from CL:FIND-CLASS is now a CLOS class, and
1309 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1310 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1311 * An interface to the MetaObject Protocol, as described in Kiczales,
1312 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1313 MIT Press, 1991, is available from the SB-MOP package.
1314 * incompatible change: the SB-PCL package should now be considered
1315 a private implementation detail, and no longer a semi-private MOP
1317 * minor incompatible change: due to rearrangement for threads, the
1318 control stack and binding stack are now allocated at arbitrary
1319 addresses instead of being hardcoded per-port. Users affected by
1320 this probably have to be doing advanced things with shared
1321 libraries, and will know who they are.
1322 * minor incompatible change: Previously, all --eval forms used were
1323 processed with READ before any of them were processed with EVAL.
1324 Now each --eval form is processed with both READ and EVAL before
1325 the next --eval form is processed. (Thus package operations like
1326 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1327 work as the user might reasonably expect.)
1328 * minor incompatible change: *STANDARD-INPUT* is now only an
1329 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1331 * minor incompatible change: Y-OR-N-P is now character-oriented,
1332 not line oriented. Also, YES-OR-NO-P now works without errors.
1333 (thanks to Antonio Martinez)
1334 * sb-aclrepl module improvements: an integrated inspector, added
1335 repl features, and a bug fix to :trace command.
1336 * Known functions, which cannot be open coded by the backend, are
1337 considered to be able to check types of their arguments. (fixing
1338 a bug report by Nathan J. Froyd)
1339 * fixed a bug in computing method discriminating functions: It is
1340 now possible to define methods specialized on classes which have
1341 forward-referenced superclasses. (thanks to Gerd Moellmann)
1342 * fixed evaluation order in optional entries (reported by Gilbert
1344 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1345 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1346 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1347 required-by-AMOP NAME argument, as well as CLASS and
1348 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1349 * fixed bug 20: DEFMETHOD can define methods using names that are
1350 not the proper names of classes to designate class specializers.
1351 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1352 Unix file descriptors, instead of blowing up. (thanks to Antonio
1354 * Garbage collection refactoring: user-visible change is that a
1355 call to the GC function during WITHOUT-GCING will not do garbage
1356 collection until the end of the WITHOUT-GCING. If you were doing
1357 this you were probably losing anyway.
1358 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1359 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1360 TYPEP the latter but not the former.
1361 * The compiler issues a full WARNING for calls to undefined functions
1362 with names from the CL package.
1363 * MAP-INTO for a vector destination is open coded. (reported by
1364 Brian Downing on c.l.l)
1365 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1366 documentation string.
1367 * fixed some bugs revealed by Paul Dietz' test suite:
1368 ** COPY-ALIST now signals an error if its argument is a dotted
1370 ** Condition slots are now accessed more correctly in the presence
1371 of multiple initargs for a given slot.
1372 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1373 exclude restarts of the same name associated with a different
1375 ** DEFCLASS of forward-referenced classes with another
1376 forward-referenced class in the superclasses list no longer
1378 ** Condition slots are now initialized once each, not multiple
1379 times. (thanks to Gerd Moellmann)
1380 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1382 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1383 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1384 arguments contain duplicated elements.
1385 ** RESTART-CASE understands local macros.
1386 ** RESTART-CASE associates exactly its own restarts with a condition.
1387 ** ENDP in safe mode checks its argument to be of type LIST.
1388 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1389 in question is unbound.
1390 ** Optimized MAKE-INSTANCE functions no longer cause internal
1391 assertion failures in the presence of duplicate initargs.
1392 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1393 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1394 propagated even in the ctor optimized implementation of
1396 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1398 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1399 * a better implementation of SXHASH on (simple) bit vectors,
1400 measured both in execution speed and in distribution of results
1401 over the positive fixnums, has been installed. Likewise, a better
1402 implementation of EQUAL for simple bit vectors is now available.
1403 * fixed CEILING optimization for a divisor of form 2^k.
1404 * fixed bug 240 (emitting extra style warnings "using the lexical
1405 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1406 by Antonio Martinez)
1407 * fixed SXHASH, giving different results for NIL depending on type
1408 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1409 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1410 variable. (found by Rolf Wester)
1411 * SBCL does not ignore type declarations for special
1412 variables. (reported by rif on c.l.l 2003-03-05)
1413 * some bug fixes in contrib/sb-aclrepl/
1414 * fixed some bugs revealed by Paul Dietz' test suite:
1415 ** a bug in the CONS type specifier, whereby the CAR and CDR
1416 types got intertwined, has been fixed;
1417 ** the type system is now able to reason about the interaction
1418 between INTEGER and RATIO types more completely;
1419 ** APPEND, [N]REVERSE and NRECONC check that those their
1420 arguments, which must be proper lists, are really so;
1421 ** An array specialized to be unable to hold elements has been
1422 implemented, as required -- yes, really -- by ANSI;
1423 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1424 on malformed property lists;
1426 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1427 * incompatible packaging change: in line with Unix convention,
1428 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1429 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1430 * REQUIRE and PROVIDE are now optionally capable of doing something
1431 useful. See the documentation string for REQUIRE.
1432 * infrastructure for a managed SBCL contrib system: contributed
1433 modules in this release include:
1434 ** the ASDF system definition facility;
1435 ** an interface to the BSD Sockets API;
1436 ** an ACL-like convenience interface to the repl;
1437 (thanks to Kevin Rosenberg)
1438 ** an implementation of ROTATE-BYTE, with efficient implementation
1440 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1441 gives a closer approximation to the right answer than previously.
1442 (thanks to Raymond Toy)
1443 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1444 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1445 optional environment arguments, as required by ANSI.
1446 * fixed bugs in other functions taking environment objects, allowing
1447 calls with an explicit NIL environment argument to be compiled
1449 * fixed bug 228: primary return values from
1450 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1451 COMPILE or FUNCTION.
1452 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1453 :TYPE will no longer signal errors on innocuous objects.
1454 * fixed bug 231b: SETQ is better at respecting type declarations in
1455 the lexical environment.
1456 * fixed a bug in DEFCLASS: classes named by symbols with no or
1457 unprintable packages can now be defined.
1458 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1459 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1460 * fixed compiler failure related to checking types of functions.
1461 (reported by Robert E. Brown)
1462 * the compiler is now much more consistent in its error-checking
1463 treatment of bounding index arguments to sequence functions: in
1464 (SAFETY 3) code, errors will be signalled in almost all cases if
1465 invalid sequence bounding indices are passed to functions defined
1466 by ANSI to operate on sequences.
1467 * fixed a bug in the build procedure: documentation of SBCL-specific
1468 packages is now preserved and available in the final Lisp image.
1469 * lifted FDEFINITION lookup out of loops in the implementation of
1470 many list operations. (thanks to Robert E. Brown)
1471 * fixed a bug in the reader: the #n# reader macro now works for
1472 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1473 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1474 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1475 E. Brown and Tony Martinez respectively)
1476 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1477 SLOT-BOUNDP: in particular, it is now able to optimize them much
1478 better, and is now not vulnerable to having packages renamed.
1479 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1480 always called when a slot is not present in an instance. (thanks
1482 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1483 longer signal an error on unreading a character following EOF on
1484 the previous constituent stream. (thanks to Tony Martinez)
1485 * fixed some bugs revealed by Paul Dietz' test suite:
1486 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1487 not just nonnegative fixnums;
1488 ** the logical bit-array operators such as BIT-AND now accept an
1489 explicit NIL for their "opt-arg" argument (to indicate a
1490 freshly-consed result bit-array);
1491 ** ELT now signals an error on an invalid sequence index in safe
1493 ** the type system is now cleverer about negations of numeric
1494 types, and consequently understands the BIGNUM and RATIO types
1496 ** the type system is now cleverer about the interaction between
1497 INTEGER and RATIO types: while bugs still remain, many more
1498 cases are accurately computed;
1499 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1500 if it is in the last clause;
1501 ** CONSTANTLY now correctly returns a side-effect-free function in
1503 ** DECLARE is no longer treated as a special-operator; in
1504 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1505 * incremented fasl file version number due to the change in the
1506 DEFSTRUCT-SLOT-DESCRIPTION structure.
1508 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1509 * minor incompatible change: code processed by the "interpreter" or
1510 EVAL now has a compilation optimization policy of (DEBUG 2)
1511 (changed from (DEBUG 1)) to improve debuggability of interactive
1512 development, and to allow the use of the debug RETURN command in
1514 * an experimental implementation of the RETURN command for the
1515 debugger has been included. (thanks to Frederik Kuivinen)
1516 * fixed bug 62: constraints were not propagated into a loop.
1517 * fixed bug in embedded calls of SORT (reported and investigated by
1519 * fixed some bugs revealed by Paul Dietz' test suite:
1520 ** printing and reading of arrays with some dimensions having
1521 length 0 (thanks to Gerd Moellmann);
1522 ** BOA constructor with &AUX argument without a default value does
1523 not cause a type error;
1524 ** CONSTANTP now returns true for all self-evaluating objects.
1526 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1527 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1528 accessors that are related by inheritance, as specified in the
1529 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1530 Valtteri Vuorikoski)
1531 * The compiler is now able to inline functions that were defined in
1532 a complex lexical environment (e.g. inside a MACROLET).
1533 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1535 * Support for the upcoming FreeBSD-5.0 release has been included.
1536 (thanks to Dag-Erling Smorgrav)
1537 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1538 effect when it is not in a toplevel context.
1539 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1540 stand a better chance of being correct. (thanks to Gerd
1542 * fixed bug in COERCE, which now signals an error on coercing a
1543 rational to a bounded real type which excludes the expected
1545 * The compiler is now able to derive types more accurately from the
1546 COERCE and COMPILE functions.
1547 * fixed bug 223: functional binding is considered to be constant
1548 only for symbols in the CL package.
1549 * fixed bug 231: SETQ did not check the type of a variable being set
1550 (reported by Robert E. Brown)
1551 * A new optimization for MAKE-INSTANCE has been included, fixing
1552 various bugs (including relating to :ALLOCATION :CLASS slots and
1553 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1555 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1556 ** As required by ANSI, LOOP now disallows anonymous collection
1557 clauses such as COLLECT I in conjunction with aggregate boolean
1558 clauses such as THEREIS (= I 1);
1559 ** LOOP now signals an error when any variable is reused in the
1560 same loop (including the potentially useful construct analogous
1561 to WITH A = 1 WITH A = (1+ A);
1562 ** IT is only a special loop symbol within the first clause of a
1563 conditional loop clause;
1564 ** LOOP with a typed iteration variable over a hashtable now
1565 signals a type error iff it should.
1566 * fixed some other bugs revealed by Paul Dietz' test suite:
1567 ** FILE-STREAM now names the class previously known as FD-STREAM;
1568 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1569 argument) no longer signals an error;
1570 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1571 of the slot symbol, rather than using the current package
1572 ((:CONC-NAME "") continues to intern the slot's name in the
1574 * incremented fasl file version number, because of the incompatible
1575 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1576 of the new implementation of DEFINE-COMPILER-MACRO.
1578 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1579 * Support for building SBCL for MIPS platforms running in
1580 little-endian mode has now been checked in, and basic
1581 functionality on said platforms verified.
1582 * minor incompatible change: PCL now records the pathname of a file
1583 in which methods and the like are defined, rather than its
1585 * minor incompatible change: TRUENAME now considers the truename of
1586 a file naming a directory to be the pathname with :DIRECTORY
1587 component indicating that directory.
1588 * minor incompatible change: a NAMED clause in the extended form of
1589 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1590 reason for the previous behaviour is unclear.
1591 * more systematization and improvement of CLOS and MOP conformance
1592 in PCL (thanks to Gerd Moellman and Pierre Mai):
1593 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1595 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1596 primary methods with no specializers;
1597 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1599 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1600 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1601 FORWARD-REFERENCED-CLASSes; error reporting on
1602 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1604 ** SXHASH on CLOS instances now uses a slot internal to the
1605 instance to return different numbers on distinct instances,
1606 while preserving the same return value through invocations of
1608 ** DEFMETHOD signals errors when methods with longer incongruent
1609 lambda lists are added to generic functions;
1610 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1611 CLASS, as specified in AMOP;
1612 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1613 on the order returned by the primary method for classes of
1614 class STANDARD-CLASS;
1615 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1616 * fixed some bugs shown by Paul Dietz' test suite:
1617 ** DOLIST puts its body in TAGBODY;
1618 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1620 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1621 value producing form;
1622 ** if more variables are given to PROGV than values, extra
1623 variables are bound and made to have no value;
1624 ** NSUBSTITUTE on list arguments gets the right answer with
1626 ** ELT signals an error of type TYPE-ERROR when the index argument
1627 is not a valid sequence index;
1628 ** LOOP signals (at macroexpansion time) an error of type
1629 PROGRAM-ERROR when duplicate variable names are found;
1630 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1631 ** FUNCALL of special-operators now cause an error of type
1633 ** PSETQ now works as required in the presence of side-effecting
1634 symbol-macro places;
1635 ** NCONC accepts any object as its last argument;
1636 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1638 ** loop-for-as-package does not require a package to be explicitely
1640 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1642 * fixed bug 166: compiler preserves "there is a way to go"
1643 invariant when deleting code.
1644 * fixed bug 172: macro lambda lists with required arguments after
1645 &REST arguments now cause an error to be signalled. (thanks to
1647 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1648 bug. (thanks to Matthew Danish)
1649 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1651 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1652 arguments are lexically rebound. (thanks to Gerd Moellmann and
1654 * fixed bug 194: error messages are now more informative when there
1655 is no primary method applicable in a call to a generic
1656 function. (thanks to Gerd Moellmann)
1657 * fixed bug in command line argument checking (thanks to Julian
1659 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1660 (thanks to Matthew Danish)
1661 * incremented fasl file version number, because of the
1662 SXHASH-related changes in the layout of CLOS data structures
1664 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1665 * minor incompatible change: The runtime (the Unix executable named
1666 "sbcl") is now much pickier about the .core files it will load.
1667 Essentially it now requires .core files to descend from the same
1668 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1669 as the runtime does. (The intent is to prevent the crashes which
1670 can occur, and which can even be reported as mysterious failures,
1671 when people patch the sources or change the build parameters
1672 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1673 sbcl and .core files.)
1674 * fixed bug: VALUES-LIST is no longer optimized away.
1675 * fixed bug 142: The FFI conversion of C string values to Lisp
1676 string values no longer conses excessively. (thanks to Nathan
1677 Froyd porting Raymond Toy's fix to CMU CL)
1678 * began to systematize and improve MOP conformance in PCL (thanks to
1679 Nathan Froyd, Gerd Moellman and Pierre Mai):
1680 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1682 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1683 ** FINALIZE-INHERITANCE is now called on class finalization;
1684 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1685 argument precedence order.
1686 * fixed bug 202: The compiler no longer fails on functions whose
1687 derived types contradict their declared type.
1688 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1689 so it can be non-toplevel.
1690 * The fasl file version number has changed (because of the new
1691 implementation of DEFMACRO).
1692 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1693 safe code, that any length requirement by their type-specifier
1694 argument is valid. The exceptions to this are described in bug
1696 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1697 functions now check the types of their inputs as required by ANSI.
1698 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1699 introduced symbol is DECLAREd to be SPECIAL.
1700 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1701 * fixed bug: :COUNT argument to sequence functions may be negative
1702 * fixed bug: body of DO-SYMBOLS may contain declarations
1703 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1704 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1705 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1706 symbol macro only once
1707 * fixed printing of call frame when argument list is unavailable
1708 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1709 * compiler no longer signals WARNING on unknown keyword
1712 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1713 * A beta-quality port to the mips architecture running Linux,
1714 based on the old CMUCL backend, has been made. It has been tested
1715 on a big-endian kernel, and works sufficiently well to be able to
1716 rebuild itself; it has not been tested in little-endian mode.
1717 * fixed an inconsistency between gencgc.c and purify.c which made
1718 dumping/loading .core files unreliable
1719 * fixed bug 120a: The compiler now deals correctly with IFs where
1720 the consequent is the same as the alternative, instead of
1721 misderiving the return type. (thanks to Alexey Dejneka)
1722 * fixed bug 113: Logical pathnames are now dumpable (the logical
1723 host is resolved at load-time, throwing an error if it is not
1725 * fixed bug 174: FORMAT's error message is slightly clearer when a
1726 non-printing character is used in a format directive.
1727 * fixed several bugs in compiler checking of type declarations, i.e.
1728 violations of the Python "declarations are assertions" principle
1729 (thanks to Alexey Dejneka)
1730 * fixed several bugs in PCL's error checking (thanks to Gerd
1732 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1734 * fixed bug in compilation of functions as first class values
1735 (thanks to Antonio Martinez-Shotton)
1736 * The compiler's handling TYPE-ERRORs which it can prove will
1737 inevitably happen at runtime has been cleaned up and corrected
1738 in several ways. (thanks to Alexey Dejneka)
1739 * improved argument type checking for various basic arithmetic
1740 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1741 lavished on them in the past that they can be compiled in many
1742 ways in different special cases
1743 * fixed bug 181: compiler checks validity of user supplied type
1745 * cleaned up code flushing in optimization: Function calls which
1746 should signal errors for safety purposes (e.g. which ANSI says
1747 should signal errors when their arguments are of incorrect type)
1748 are no longer optimized away.
1749 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1750 * incremented fasl file version number, because changes in the
1751 implementation of sequence functions like COERCE caused
1752 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1753 in old inline expansions) to become undefined. (Actually these
1754 changes were later undone, so we might very well be binary
1755 compatible with 0.7.7 after all, but leaving the version number
1756 incremented seemed like the simplest and most conservative
1759 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1760 * An alpha-quality port to the parisc architecture running Linux,
1761 based on the old CMUCL backend, has been made. This, even more so
1762 than the other backends, should be considered still a work in
1763 progress; known problems include that the Linux kernel in 64-bit
1764 mode does not propagate the correct sigcontext structure to
1765 userspace, and consequently SBCL on a parisc64 kernel will not
1767 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1768 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1769 suppressed.) Also now that the compiler is looking at declarations
1770 in the environment, it checks optimization declarations as well,
1771 and suppresses inlining when (> DEBUG SPEED).
1772 * More fixes have been made to treatment of floating point exception
1773 treatment and other Unix signals. In particular, floating point
1774 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1775 * The detection and handling of control stack exhaustion (infinite
1776 or very deeply nested recursion) has changed. Stack exhaustion
1777 detection is now done by write-protecting pages at the OS level
1778 and applies at all optimization settings; when found, a
1779 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1780 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1781 be caught using IGNORE-ERRORS.
1782 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1783 correctly when transforming recursive LABELS functions to LETs.
1784 (thanks to Alexey Dejneka)
1785 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1786 that are names of constants or global variables.
1787 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1788 alien routines with docstrings.
1789 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1790 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1792 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1793 (thanks to Eric Marsden)
1794 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1795 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1796 * Bug fix: The SPARC backend can now compile functions involving
1797 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1798 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1799 to a functions expecting a PATHNAME-DESIGNATOR.
1800 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1801 lambda lists. (thanks to Alexey Dejneka)
1802 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1803 functions so that EXPORTing the name of the class doesn't cause
1804 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1805 to Antonio Martinez for reporting this)
1806 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1807 constant-folded without failing an assertion. (thanks to Einar
1808 Floystad Dorum for reporting this)
1809 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1810 types are now dealt with more correctly by the compiler.
1811 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1812 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1813 behaviour analogous to RENAME-FILE. This puts its behaviour more
1814 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1816 * The fasl file version number has changed again. (because of the
1817 bug fix involving the names of PCL MAKE-INSTANCE functions)
1819 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1820 * bug fix: Floating point exceptions are treated much more
1821 consistently on the x86/Linux and PPC/Linux platforms.
1822 * Array initialization with :INITIAL-ELEMENT is now much faster for
1823 cases when the compiler cannot open code the array creation, but
1824 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1825 array accesses have also seen a speed increase.
1826 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1827 specified. (This was at the root of some bad interactions between
1828 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1830 * bug fix: Internal error arguments for undefined functions are now
1831 computed correctly on the PPC/Linux platform.
1832 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1833 Raymond Toy's patch for CMU CL)
1834 * Support for the Solaris 9 operating environment has been included
1835 (thanks to Daniel Merritt)
1836 * A very ugly but hopefully complete draft of the missing FFI chapter
1837 of the manual has been created by reformatting the corresponding
1838 CMU CL manual chapter into (currently very ugly and incoherent)
1839 DocBook and bringing it up to date for SBCL behavior. Thus, the
1840 manual is now essentially complete, at least by my extreme
1841 once-and-only-once standards, whereby it's acceptable to refer to
1842 the doc strings of SB-EXT functions as the primary documentation.
1843 * The fasl file version number has changed again, due to cleanup of
1844 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1845 is no longer a static symbol.)
1847 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1848 * SBCL now builds with OpenMCL (version 0.12) as the
1849 cross-compilation host; also, more progress has been made toward
1850 bootstrapping under CLISP.
1851 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1853 * bug 158 fixed: The compiler can now deal with integer loop
1854 increments different from 1; fixing this turned out also to fix
1856 * bug 169 fixed: no more bogus warnings about using lexical bindings
1857 despite the presence of perfectly good SPECIAL declarations (thanks
1858 to David Lichteblau)
1859 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1860 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1861 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1863 * bug 180 fixed: Method combination specifications no longer ignore
1864 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1865 * bug fix: Structure type predicate functions now check their argument
1866 count as they should.
1867 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1868 correctly. (thanks to Pierre Mai)
1869 * minor incompatible change: The --noprogrammer option is deprecated
1870 in favor of the new --disable-debugger option, which is very similar.
1871 (The major difference is that it takes effect at a slightly different
1872 time at startup, causing handling of errors in --sysinit and
1873 --userinit files will be affected differently.) The
1874 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1875 been added to allow this functionality to be controlled from ordinary
1876 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1877 maintainers, who might want to run non-interactive scripts to
1878 build SBCL cores which will later be used interactively.)
1879 * minor incompatible change: The LOAD function no longer, when given
1880 a wild pathname to load, loads all files matching that pathname.
1881 Instead, an error of type FILE-ERROR is signalled.
1883 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1884 * bug 147 fixed: The compiler preserves its block link/count
1885 invariants more correctly now so that it doesn't crash. (thanks
1887 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1889 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1890 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1891 no-value-for-this-character as NIL (as specified by ANSI).
1892 * HOST-NAMESTRING on physical pathnames now returns a string that is
1893 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1894 (thanks to Christophe Rhodes)
1895 * The Alpha port handles icache flushing more correctly. (thanks to
1897 * More progress has been made toward bootstrapping under CLISP. (thanks
1898 to Christophe Rhodes)
1899 * The fasl file format has changed again, because dynamic loading
1900 on OpenBSD (which has non-ELF object files) motivated some cleanups
1901 in the way that foreign symbols are transformed and passed around.
1902 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1903 is no longer treated as whitespace by the reader, but instead as
1904 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1905 |A
\7fB|, instead of A as it used to.
1907 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1908 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1909 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1910 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1911 this as of 0.7.1.45, but was left out of the previous news section
1912 (thanks to Dan Barlow)
1913 * SBCL now runs on the Solaris operating system on SPARC architectures
1914 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1915 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1916 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1917 * SPARC backend cleanups, allowing builds of cores optimized for V8
1918 and V9 SPARCS, and also emission of code targeted to a particular
1919 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1921 * SBCL is closer to bootstrapping under CLISP, thanks to various
1922 fixes by Christophe Rhodes.
1923 * The fasl file format has changed again, to allow the compiler's
1924 INFO database to support symbol macros.
1925 * The user manual (in doc/) is formatted into HTML more nicely.
1926 (thanks to coreythomas)
1927 * The system is smarter about SUBTYPEP relationships, especially
1928 those involving NOT types (including types such as ATOM which are
1929 represented internally using NOT types). Thus SUBTYPEP is less
1930 likely to return (VALUES NIL NIL) in general, and in particular
1931 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1933 * The fasl file format has changed again, because the internal
1934 representation of types now includes a new slot to support the new
1935 SUBTYPEP-of-NOT-types logic.
1936 * (not a change in the main branch of SBCL, but a related prototype
1937 which can hopefully be merged into the main branch of SBCL in the
1938 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1939 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1940 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1941 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1942 dispatch macro character combinations. (thanks to Alexey Dejneka)
1943 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1944 that superficially look like logical namestrings correctly.
1945 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1947 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1948 default supplied-p) for &optional and &key arguments. (thanks to
1951 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1952 * incompatible change: The compiler is now less aggressive about
1953 tail call optimization, doing it only when (> SPACE DEBUG) or
1954 (> SPEED DEBUG). (This is an incompatible change because there are
1955 programs which relied on the old CMU-CL-style behavior to optimize
1956 away their unbounded recursion which will now die of stack overflow.)
1957 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1958 for non-GENCGC systems has been increased to 20M (since that
1959 seems much closer to the likely performance optimum for modern
1960 systems than the old 4M value was)
1961 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1962 and *DEBUG-PRINT-LEVEL*
1963 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1964 of CMU CL's support for SPARC, and various endianness and other
1965 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1966 * new syntactic sugar for the Unix command line: --load foo.bar is now
1967 an alternate notation for --eval '(load "foo.bar")'.
1969 ** The system now detects stack overflow and handles it gracefully,
1970 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1971 optimization settings. (This is a good thing in general, and
1972 its introduction in this version should be particularly timely
1973 for anyone whose code fails because of suppression of tail
1975 ** The system now hunts for the C variable "environ" in a more
1976 devious way, to avoid segfaults when the C library version differs
1977 between compile time and run time. (thanks to Christophe Rhodes)
1978 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1979 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1980 ** The compiler no longer issues bogus style warnings for undefined
1981 classes in the same source file as the DEFCLASSes which defined
1982 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1984 ** fixes in CONDITION class precedence list for undefined function
1985 errors (thanks to Alexei Dejneka)
1986 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1987 correctly. (thanks to Dan Barlow)
1988 ** portability fixes aiming at bootstrapping under CLISP (thanks
1989 to Dave McDonald and Christophe Rhodes)
1990 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1991 ** fixes in type translation and and type inference (thanks to
1993 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1994 ** various fixes in the new ports (thanks to Dan Barlow)
1995 * several changes related to debugging:
1996 ** suppression of tail recursion, as noted above
1997 ** stack overflow detection, as noted above
1998 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1999 is now the default. (For some time encapsulation has been more
2000 reliable than the breakpoint-based :ENCAPSULATE NIL
2001 implementation, at least on X86 systems; and I just noticed that
2002 encapsulation also seems closer to the spirit of the ANSI
2005 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2007 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2008 up properly again. (There was a packaging bug in 0.7.0 which
2009 left their definitions in SB-SYS::LOAD-FOREIGN and
2010 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2011 vital for most things which interface to C-level interfaces,
2012 like extensions working with sockets or databases or
2013 Perl-compatible regexes or whatever, and the need to fix
2014 this bug is the main reason that 0.7.1 was released so
2016 ** DEFGENERIC is now choosier about the methods it redefines, so that
2017 reLOADing a previously-LOADed file containing DEFGENERICs does
2018 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2019 cycle now works as it should. (thanks to Alexey Dejneka)
2020 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2021 fixed by Christophe Rhodes. (He actually submitted this patch
2022 months ago, and I delayed until after 0.7.0.)
2023 ** Bug 111 (internal compiler confusion about runtime checks on
2024 FUNCTION types) was fixed by Alexey Dejneka.
2025 * Some internal cleanups (getting rid of variables which aren't
2026 needed now that the byte interpreter is gone) caused the fasl
2027 file format number to change again.
2029 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2030 * major incompatible change: The default fasl file extension, i.e. the
2031 default extension for files produced by COMPILE-FILE, has changed
2032 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2034 ** There are many changes in the implementation of the compiler.
2035 SBCL is now essentially a compiler-only implementation of ANSI
2036 Common Lisp. EVAL still "interprets" a few special cases, but
2037 almost all the interesting cases are handled by creating
2038 a LAMBDA expression, calling COMPILE on it, then calling
2039 FUNCALL on the result.
2040 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2041 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2042 Since the code is newer, there might still be some new bugs
2043 (though not as many as before Martin Atzmueller's fixes:-). But
2044 the new code is substantially simpler and clearer, and hopefully
2045 any remaining bugs will be simpler, less fundamental, and more
2046 fixable then the bugs in the old code.
2047 ** The revised compiler is still a little unsteady on its feet.
2049 *** The debugging information it produces (particularly the names
2050 of FUNCTION objects) is sometimes much less useful than what
2051 the old compiler produced.
2052 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2053 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2054 within a MACROLET) has been temporarily weakened.
2055 ** There are new compiler optimizations for various functions:
2056 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2057 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2058 *** the math functions TRUNCATE, FLOOR, and CEILING
2059 *** the function-of-all-trades COERCE
2060 Mostly these should be transparent, but there's one
2061 potentially-annoying problem (bug 117): when the compiler
2062 inline-expands a function and does type analysis on the result,
2063 it can create control paths which have type mismatches, and
2064 when it can't prove that those control paths aren't taken,
2065 it will issue WARNINGs about the type mismatches. This is
2066 a particular problem in practice for the new sequence functions.
2067 It's not clear how this should be fixed, and for now, a
2068 workaround is given in the entry for 117 in the BUGS file.
2069 ** (Because of the interaction between the two previous items --
2070 occasional inlining problems and new inline expansions -- some
2071 of the new sequence function optimizations won't really kick in
2072 completely until debugging information, and then inlining, are
2073 straightened out in some future version.)
2074 * minor incompatible changes:
2075 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2076 DIRECTORY no longer implicitly promotes NIL slots of its
2077 pathname argument to :WILD. In particular, when you ask for the
2078 contents of a directory (which you used to be able to do without
2079 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2080 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2081 ** changes in behavior that ANSI explicitly defines to be
2082 implementation dependent:
2083 *** The new compiler-only implementation still conforms with ANSI,
2084 but acts a little different than before. Besides the obvious
2085 changes in performance tradeoffs (that the cost per form passed
2086 to EVAL has gone up, and the cost per form executed by EVAL
2087 has gone down), the behavior of the system changes a little
2088 because there are no longer any interpreted function objects.
2089 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2090 e.g. doing COMPILE on the output of interactive DEFUN is
2092 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2094 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2095 has changed from (:CL) to NIL.
2096 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2097 e.g. #\Newline and #\Tab, don't have names specified in the
2098 ANSI Common Lisp standard, is now based on their ASCII symbolic
2099 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2100 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2101 are no longer used for output.
2102 ** changes in internal implementation constants:
2103 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2104 4 million. (If your application spends a lot of time GCing and
2105 you have a lot of RAM, you might want to experiment with
2106 increasing it even more.)
2107 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2108 However, almost all old code should still continue to work without
2109 immediate update, as SB-C-CALL is now a (deprecated) nickname
2111 ** Old operator names in the style DEF-FOO are now deprecated in
2112 favor of new corresponding names DEFINE-FOO, for consistency with
2113 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2114 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2115 internal symbols, but a few supported extensions like
2116 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2117 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2118 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2119 etc. as you get deeper into recursive calls to the debugger
2120 command loop, instead of the old "5]", "5]]", "5]]]"
2121 sequence. (I was motivated to do this when squabbles between
2122 ILISP and SBCL left me very deeply nested in the debugger. In the
2123 short term, this change will probably provoke more ILISP/SBCL
2124 squabbles, but hopefully it will be an improvement in the long run.)
2125 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2126 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2127 function, since declaiming return types involving aliens is
2128 (1) annoyingly messy to do by hand and (2) vital to efficient
2129 compilation of code which calls such functions.
2130 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2131 longer reexported by the SB-EXT package. They're solely useful
2132 for alien code, so it seems more logical that you should get
2133 them from the SB-ALIEN package, not in SB-EXT.
2134 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2135 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2136 features. Instead, the code that they used to control is always
2137 built into the system.
2138 * many other bug fixes
2139 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2140 advantage of the new EVAL-WHEN stuff and to clean them up in
2141 general, and they are now more ANSI-compliant in a number of
2142 ways. Martin Atzmueller is responsible for a lot of this.
2143 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2145 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2146 *** correct ERROR type for various file operations
2147 *** some fixes for Lisp streams
2148 *** DEFMETHOD syntax checking
2149 *** changing old weird representation of debug information as
2150 strings (which, among their other deficiencies, don't transform
2151 correctly when you rename packages, and don't change their
2152 print representation when you change things like *PACKAGE*
2153 and *PRINT-LENGTH*) to symbols and lists of symbols
2154 He also made several improvements and fixed several bugs in DESCRIBE.
2155 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2157 *** misbehavior of WRITE-STRING/WRITE-LINE
2158 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2159 and several other LOOP problems as well
2160 *** DIRECTORY when similar filenames are present
2161 *** DEFGENERIC with :METHOD options
2162 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2163 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2164 *** argument ordering in FIND with :TEST option
2165 *** mishandled package designator argument in APROPOS-LIST
2166 *** various problems in the backquote readmacro
2167 *** a bug in APROPOS
2168 *** probably some others that I'm not describing very well here,
2169 since the CVS log documents them by reference to sbcl-devel
2170 messages, and the SourceForge archives aren't working well.:-(
2171 ** Dan Barlow improved the Alpha port (and is making progress on the
2172 PPC port, for those of you who think different).
2173 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2174 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2175 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2176 continues to work on the SPARC port (for those of you in a position
2177 to look down upon our little PC-compatible boxes from a great height).
2178 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2179 and uses that copy, rather than the current dynamic value, when
2180 it's trying to decide whether to truncate output. Thus e.g.
2181 (let ((*print-lines* 50))
2182 (pprint-logical-block (stream nil)
2184 (let ((*print-lines* 8))
2185 (print (aref possiblybigthings i) stream)))))
2186 should now truncate the logical block only at 50 lines, instead of
2187 often truncating it at 8 lines, as it did before.
2188 * The doc/cmucl/ directory, containing old CMU CL documentation
2189 from the time of the fork, is no longer part of the base system.
2190 SourceForge has shut down its anonymous FTP service, and with it
2191 my original plan for distributing the old CMU CL documentation
2192 there. For now, if you need these files you can download an old
2193 SBCL source release and extract them from it.
2194 * The fasl file version number changed again, for dozens of reasons,
2195 some of which are apparent above.
2197 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2198 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2199 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2200 patch, so that DISASSEMBLE output is much nicer.
2201 * The code in the SB-PROFILE package now seems reasonably stable.
2202 I still haven't decided what the final interface should look like
2203 (I'd like PROFILE to interact cleanly with TRACE, since both
2204 facilities use function encapsulation) but if you have a need
2205 for profiling now, you can probably use it successfully with
2206 the current CMU-CL-style interface.
2207 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2208 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2209 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2210 link now returns the dangling link itself, and for similar
2211 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2212 link itself. (In these cases the old code signalled an error and
2213 looped endlessly, respectively.) Thus, DIRECTORY now works even
2214 in the presence of dangling and cyclic symbolic links.
2215 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2216 is now a supported extension again, since the consensus on
2217 sbcl-devel was that it can be useful for ordinary development
2218 work, not just for debugging SBCL itself.
2219 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2220 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2221 that currently-defined functions might be redefined later with
2222 different return types.
2223 * Hash tables can be printed readably, as inspired by CMU CL code
2224 of Eric Marsden and SBCL code of Martin Atzmueller.
2225 * better error handling in CLOS method combination, thanks to
2226 Martin Atzmueller porting Pierre Mai's CMU CL patches
2227 * more overflow fixes for >16Mbyte I/O buffers
2228 * A bug in READ has been fixed, so that now a single Ctrl-D
2229 character suffices to cause end-of-file on character streams.
2230 In particular, now you only need one Ctrl-D at the command
2231 line (not two) to exit SBCL.
2232 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2234 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2235 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2236 does the right thing.
2237 * The implementation of some type tests, especially for CONDITION
2238 types, is now tidier and maybe faster, due to CMU CL code
2239 originally by Douglas Crosher, ported by Martin Atzmueller.
2240 * Some math functions have been fixed, and there are new
2241 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2242 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2243 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2244 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2245 likely to be built into sbcl-0.7.0.)
2246 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2247 CL, ported to SBCL by Martin Atzmueller)
2248 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2249 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2250 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2251 * There's a new slam.sh hack to shorten the edit/compile/debug
2252 cycle for low-level changes to SBCL itself, and a new
2253 :SB-AFTER-XC-CORE target feature to control the generation of
2254 the after-xc.core file needed by slam.sh.
2255 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2256 COMPILE-FILE is no longer supported, so that now every function
2257 gets an entry point, so that block compilation looks a little
2258 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2259 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2260 returns the number of bytes consed since the system started,
2261 rather than the number consed since the first time the function
2262 was called. (The new definition parallels ANSI functions like
2263 CL:GET-INTERNAL-RUN-TIME.)
2264 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2265 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2266 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2267 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2268 * Fasl file version numbers are now independent of the target CPU,
2269 since historically most system changes which required version
2270 number changes have affected all CPUs equally. Similarly,
2271 the byte fasl file version is now equal to the ordinary
2274 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2275 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2276 is no longer recognized. I apologize for this, because it was
2277 listed in SB-EXT as a supported extension, but I found that
2278 its existing behavior was poorly specified, as well as incorrectly
2279 specified, and it looked like too much of a mess to straighten it
2280 out. I have enough on my hands trying to get ANSI stuff to work..
2281 * many patches ported from CMU CL by Martin Atzmueller, with
2282 half a dozen bug fixes in pretty-printing and the debugger, and
2283 half a dozen others elsewhere
2284 * fixed bug 13: Floating point infinities are now supported again.
2285 They might still be a little bit flaky, but thanks to bug reports
2286 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2287 as flaky as they were.
2288 * The --noprogrammer command line option is now supported. (Its
2289 behavior is slightly different in detail from what the old man
2290 page claimed it would do, but it's still appropriate under the
2291 same circumstances that the man page talks about.)
2292 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2293 are now supported, and enabled by default. Thus, the compiler can
2294 handle many floating point and complex operations much less
2295 inefficiently. (Thus e.g. you can implement a complex FFT
2297 * The compiler now detects type mismatches between DECLAIM FTYPE
2298 and DEFUN better, and implements CHECK-TYPE more correctly, and
2299 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2300 thanks to patches from Martin Atzmueller.
2301 * various fixes to make the cross-compiler more portable to
2302 ANSI-conforming-but-different cross-compilation hosts (notably
2303 Lispworks for Windows, following bug reports from Arthur Lemmens)
2304 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2305 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2306 fixed thanks to Pierre Mai's CMU CL patches.
2307 * fixes to keep the system from overflowing internal counters when
2308 it tries to use i/o buffers larger than 16M bytes
2309 * fixed bug 45a: Various internal functions required to support
2310 complex special functions have been merged from CMU CL sources.
2311 (When I was first setting up SBCL, I misunderstood a compile-time
2312 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2313 * improved support for type intersection and union, fixing bug 12
2314 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2315 more obscure bugs as well
2316 * some steps toward byte-compiling non-performance-critical
2317 parts of the system, courtesy of patches from Martin Atzmueller
2318 * Christophe Rhodes has made some debian packages of sbcl at
2319 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2320 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2321 stable, but are nonetheless usable. When he's ready, I'd be happy
2322 to add them to the SourceForge "File Releases" section. (And if
2323 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2324 * new fasl file format version number (because of changes in
2325 internal representation of (OR ..) types to accommodate the new
2326 support for (AND ..) types, among other things)
2328 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2329 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2331 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2333 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2334 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2335 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2336 * DESCRIBE now works on structure objects again.
2337 * Most function call argument type mismatches are now handled as
2338 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2339 know whether the function will be redefined before the call is
2340 executed. (The compiler could flag local calls with full WARNINGs,
2341 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2342 it doesn't keep track of enough information to know whether calls
2343 are local in this sense.)
2344 * Compiler output is now more verbose, with messages truncated
2345 later than before. (There should be some supported way for users
2346 to override the default verbosity, but I haven't decided how to
2347 provide it yet, so this behavior is still controlled by the internal
2348 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2349 src/compiler/ir1util.lisp.)
2350 * Fasl file format version numbers have increased again, because
2351 support for the Gray streams extension changes the layout of the
2352 system's STREAM objects.
2353 * The Gray subclassable streams extension now works, thanks to a
2354 patch from Martin Atzmueller.
2355 * The full LOAD-FOREIGN extension (not just the primitive
2356 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2357 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2358 but like most other programs, it defaults to copying the Unix
2359 environment from the original process instead of starting the
2360 new process in an empty environment.
2361 * Extensions which manipulate the Unix environment now support
2362 an :ENVIRONMENT keyword option which doesn't smash case or
2363 do other bad things. The CMU-CL-style :ENV option is retained
2364 for porting convenience.
2365 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2366 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2368 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2370 * A patch from Martin Atzmueller seems to have solved the SIGINT
2371 problem, and as far as we know, signal-handling now works cleanly.
2372 (If you find any new bugs, please report them!)
2373 * The system no longer defaults Lisp source file names to types
2374 ".l", ".cl", or ".lsp", but only to ".lisp".
2375 * The compiler no longer uses special default file extensions for
2376 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2377 seems to expect a single default extension for all compiled code,
2378 and there's no compelling reason to try to stretch the standard
2379 to allow two different extensions.) Instead, byte-compiled files
2380 default to the same extension as native-compiled files.
2381 * Fasl file format version numbers have increased again, because
2382 a rearrangement of internal implementation packages made some
2383 dumped symbols in old fasl files unreadable in new cores.
2384 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2385 many fewer weird special cases.
2386 * Bug #17 (differing COMPILE-FILE behavior between logical and
2387 physical pathnames) has been fixed, and some related misbehavior too,
2388 thanks to a patch from Martin Atzmueller.
2389 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2390 by Tim Moore, ported to SBCL by Martin Atzmueller.
2391 * Martin Atzmueller fixed several filesystem-related problems,
2392 including bug #36, in part by porting CMU CL patches, which were
2393 written in part by Paul Werkowski.
2394 * More compiler warnings in src/runtime/ are gone, thanks to
2395 more patches from Martin Atzmueller.
2396 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2399 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2401 * DESCRIBE now works on CONDITION objects.
2402 * The debugger now handles errors which arise when trying to print
2403 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2405 * The build system now uses an additional file, customize-target-features.lisp,
2406 to allow local modifications to the target *FEATURES* list. (The point of
2407 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2408 debugging features enabled, without having to worry about propagating it
2409 into everyone's system when I do a "cvs update".) When no
2410 customize-target-features.lisp file exists, the target *FEATURES* list
2411 should be constructed the same way as before.
2412 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2413 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2414 warn about behavior which is undefined under the ANSI spec. Note: This
2415 is specified by ANSI, but it's not very popular with programmers.
2416 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2417 macro in the SBCL sources for an example of a workaround which you
2418 might use to make portable ANSI-standard code which does what you want.
2419 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2420 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2421 when it's used as a non-toplevel form. (This is required in order
2422 to implement the DEFCONSTANT-EQX macro.)
2423 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2424 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2425 funny things, due to bugs in the implementation of EVAL-WHEN
2426 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2427 non-toplevel DEFCONSTANTs are uncommon.)
2428 * The core file version number and fasl file version number have been
2429 incremented, because the old noncompliant DEFCONSTANT behavior involved
2430 calling functions which no longer exist, and because I also took the
2431 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2432 * fixed bug 1 (error handling before read-eval-print loop starts), and
2433 redid debugger restarts and related debugger commands somewhat while
2435 ** The QUIT debugger command is gone, since it did something
2436 rather different than the SB-EXT:QUIT command, and since it never
2437 worked properly outside the main toplevel read/eval/print loop.
2438 Invoking the new TOPLEVEL restart provides the same functionality.
2439 ** The GO debugger command is also gone, since you can just invoke
2440 the CONTINUE restart directly instead.
2441 ** The TOP debugger command is also gone, since it's redundant with the
2442 FRAME 0 command, and since it interfered with abbreviations for the
2444 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2446 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2447 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2448 some porting work by Martin Atzmueller)
2449 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2450 with a boatload of patches which he ported to SBCL. Now that those
2452 ** The system tries to make sure that its low-priority messages
2453 are prefixed by semicolons, to help people who like to use
2454 syntax highlighting in their ILISP buffer. (This patch
2455 was originally due to Raymond Toy.)
2456 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2457 patches originally written by Raymond Toy.
2458 ** The compiler understands coercion between single-value and
2459 multiple-VALUES type expressions better, getting rid of some very
2460 weird behavior, thanks to patches originally by Robert MacLachlan
2461 and Douglas Crosher.
2462 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2463 lambda lists, thanks to a patch originally by Pierre Mai.
2464 ** The system no longer bogusly warns about "abbreviated type
2466 ** The compiler gets less confused by inlining and RETURN-FROM,
2467 thanks to some patches originally by Tim Moore.
2468 ** The system no longer hangs when dumping circular lists to fasl
2469 files, thanks to a patch originally from Douglas Crosher.
2470 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2471 undefined function error.
2472 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2473 in structure slot initforms) for now, documented workaround instead:-|
2474 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2475 * fixed bug 5: added stubs for various Gray stream functions called
2476 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2477 installed, at least appropriate type errors are generated
2478 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2479 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2480 consistently in DEFMETHOD forms.
2481 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2482 it doesn't seem to affect SBCL after all
2483 * The C runtime system now builds with better optimization and many
2484 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2486 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2488 * The system is now under CVS at SourceForge (instead of the
2489 CVS repository on my home machine).
2490 * The new signal handling code has been tweaked to treat register
2491 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2492 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2493 downstream, e.g. in debug-int.lisp, has implicit dependencies
2494 on the unsignedness of integer representation of machine words,
2495 and that caused the system to bomb out with infinite regress
2496 when trying to recover from type errors involving signed values,
2497 e.g. (BUTLAST '(1 2 3) -1).)
2498 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2499 (This was one of the bugs Peter Van Eynde reported back in July.)
2500 * The system now uses code inspired by Colin Walters' O(N)
2501 implementation of MAP (from the cmucl-imp@cons.org mailing
2502 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2503 inline the MAP operation, and there is more than one
2504 sequence argument to the MAP call (so that it can't just
2505 do ETYPECASE once and for all based on the type of the
2506 single sequence argument). (The old non-inline implementation
2507 of the general M-argument sequence-of-length-N case required
2508 O(M*N*N) time when any of the sequence arguments were LISTs.)
2509 * The QUIT :UNIX-CODE keyword argument has been renamed to
2510 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2511 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2513 * Raymond Wiker's patches to port dynamic loading from Linux to
2514 FreeBSD have been added.
2515 * The BUGS file is now more nearly up to date, thanks in large part
2516 to Martin Atzmueller's review of it.
2517 * The debugger now flushes standard output streams before it begins
2518 its output ("debugger invoked" and so forth).
2519 * The core version number and fasl file version number have both
2520 been incremented, because of incompatible changes in the layout
2522 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2523 used to not be called for a saved Lisp image.)
2524 * Martin Atzmueller's version of a patch to fix a compiler crash,
2525 as posted on sbcl-devel 13 September 2000, has been installed.
2526 * Instead of installing Martin Atzmueller's patch for the
2527 compiler transform for SUBSEQ, I deleted the compiler transform,
2528 and transforms for some similar consing operations.
2529 * A bug in signal handling which kept TRACE from working on OpenBSD
2531 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2532 away by constant folding
2533 * The system now defines its address space constants in one place
2534 (in the Lisp sources), and propagates them automatically elsewhere
2535 (through GENESIS and the sbcl.h file). Therefore, patching the
2536 address map is less unnecessarily tedious and error-prone. The
2537 Lisp names of address space constants have also been systematized.
2538 * CVS tags like dollar-Header-dollar have been removed from
2539 the sources, because they have never saved me trouble and
2540 they've been source of trouble working with patches and other
2541 diff-related operations.
2542 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2543 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2545 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2547 * The system has been ported to OpenBSD.
2548 * The system now compiles with a simple "sh make.sh" on the systems
2549 that it's supported on. I.e., now you no longer need to tweak
2550 text in the target-features.lisp-expr and symlinks in src/runtime/
2551 by hand, the make.sh takes care of it for you.
2552 * The system is no longer so grossly inefficient when compiling code
2553 involving vectors implemented as general (not simple) vectors (VECTOR T),
2554 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2555 which dares to use the various sequence functions on non-simple
2556 vectors, takes less of a performance hit.
2557 * There is now a primitive type predicate VECTOR-T-P
2558 to test for the (VECTOR T) type, so that e.g.
2559 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2560 can now be compiled with some semblance of efficiency. (The old code
2561 turned the type declaration into a full call to %TYPEP at runtime!)
2562 * AREF on (VECTOR T) is still not fast, since it's still compiled
2563 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2564 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2565 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2566 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2567 without making full calls to SUBTYPEP at runtime.
2568 (Some analogous efficiency issues for non-simple vectors specialized to
2569 element types other than T, or for non-simple multidimensional arrays,
2570 have not been addressed. They could almost certainly be handled the
2571 same way if anyone is motivated to do so.)
2572 * The changes in array handling break binary compatibility, so
2573 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2574 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2575 instead of (VALUES T T).
2576 * By following the instructions that Dan Barlow posted to sbcl-devel
2577 on 2 July 2000, I was able to enable primitive dynamic object
2578 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2579 functionality is not implemented (since it calls ld to resolve
2580 library references automatically, requiring RUN-PROGRAM for its
2581 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2582 try to resolve library references) is now supported.
2583 * The system now flushes the standard output streams when it terminates,
2584 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2585 them at several other probably-convenient times, e.g. in each pass of
2586 the toplevel read-eval-print loop, and after evaluating a form given
2587 as an "--eval" command-line option. (These changes were motivated by a
2588 discussion of stream flushing issues on cmucl-imp in August 2000.)
2589 * The source transform for TYPEP of array types no longer assumes
2590 that an array whose element type is a not-yet-defined type
2591 is implemented as an array of T, but instead punts, so that the
2592 type will be interpreted at runtime.
2593 * There is now some support for cross-compiling in make.sh: each of
2594 the phases of make.sh has its own script. (This should be transparent
2595 to people doing ordinary, non-cross-compile builds.)
2596 * Since my laptop doesn't have hundreds of megabytes of memory like
2597 my desktop machine, I became more motivated to do some items on
2598 my to-do list in order to reduce the size of the system a little:
2599 ** Arrange for various needed-only-at-cold-init things to be
2600 uninterned after cold init. To support this, those things have
2601 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2602 symbols with such names are now uninterned after cold init).
2603 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2604 fasl files for cold load.
2605 ** Remove the old compiler structure pooling code (which used to
2606 be conditional on the target feature :SB-ALLOC) completely.
2607 ** Redo the representation of some data in cold init to be more compact.
2608 (I also looked into supporting byte compiled code at bootstrap time,
2609 which would probably reduce the size of the system a lot, but that
2610 looked too complicated, so I punted for now.)
2611 * The maximum signal nesting depth in the src/runtime/ support code has
2612 been reduced from 4096 to 256. (I don't know any reason for the very
2613 large old value. If the new smaller value turns out to break something,
2614 I'll probably just bump it back up.)
2615 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2617 * Many, many bugs reported by Peter Van Eynde have been added to
2618 the BUGS list; some have even been fixed.
2619 * While enabling dynamic object file loading, I tried to make the
2620 code easier to understand, renaming various functions and variables
2621 with less ambiguous names, and changing some function calling
2622 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2623 * While trying to figure out how to do the OpenBSD port, I tried to
2624 clean up some of the code in src/runtime/. In particular, I dropped
2625 support for non-POSIX signal handling, added various comments,
2626 tweaked the code to reduce the number of compilation warnings, and
2627 renamed some files to increase consistency.
2628 * To support the new automatic configuration functionality in make.sh,
2629 the source file target-features.lisp-expr has been replaced with the
2630 source file base-target-features.lisp-expr and the machine-generated
2631 file local-target-features.lisp-expr.
2632 * fixed a stupid quoting error in make.sh so that using CMU CL
2633 "lisp -batch" as cross-compilation host works again
2635 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2637 * DESCRIBE no longer tries to call itself recursively to describe
2638 bound/fbound values, so that it no longer fails on symbols which are
2639 bound to themselves (like keywords, T, and NIL).
2640 * DESCRIBE now works on generic functions.
2641 * The printer now prints less-screwed-up representations of closures
2642 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2643 * A private symbol is used instead of the :EMPTY keyword previously
2644 used to mark empty slots in hash tables. Thus
2645 (DEFVAR *HT* (MAKE-HASH-TABLE))
2646 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2647 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2648 now does what ANSI says that it should. (You can still get
2649 similar noncompliant behavior if bang on the hash table
2650 implementation with all the symbols you get back from
2651 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2652 This breaks binary compatibility, since tests for equality to
2653 :EMPTY are wired into things like the macroexpansion of
2654 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2656 * There's now a minimal placeholder implementation for CL:STEP,
2657 as required by ANSI.
2658 * An obscure bug in the interaction of the normal compiler, the byte
2659 compiler, inlining, and structure predicates has been patched
2660 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2661 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2663 * Missing ordinary arguments in a macro call are now detected even
2664 when the macro lambda list contains &KEY or &REST.
2665 * The debugger no longer complains about encountering the top of the
2666 stack when you type "FRAME 0" to explicitly instruct it to go to
2667 the top of the stack. And it now prints the frame you request even
2668 if it's the current frame (instead of saying "You are here.").
2669 * As specified by ANSI, the system now always prints keywords
2670 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2672 * The default initial SIZE of HASH-TABLEs is now smaller.
2673 * Type information from CLOS class dispatch is now propagated
2674 into DEFMETHOD bodies, so that e.g.
2675 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2677 is now basically equivalent to
2678 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2679 (DECLARE (TYPE SINGLE-FLOAT X))
2681 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2682 operation, without having to do run-time type dispatch.
2683 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2684 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2685 * Since I don't seem to be making big file reorganizations very often
2686 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2687 is overflowing my ability to conveniently back them up), I've finally
2688 checked the system into CVS. (The CVS repository is on my home system,
2689 not at SourceForge -- putting it on SourceForge might come later.)
2690 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2691 high-level GC-NOTIFY-FOO functions send their output. (There's
2692 still very little control of where low-level verbose GC functions
2693 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2694 less than it used to -- the GC-NOTIFY-FOO functions are now under
2695 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2696 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2697 in only one place in the source code, and propagates it automatically
2698 everywhere that it's needed. Thus e.g. when I bump the version from
2699 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2700 * The C source files now include boilerplate legalese and documentation
2701 at the head of each file (just as the Lisp source files already did).
2702 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2703 to his page will be replaced with a link to his new CLiki service.
2705 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2707 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2708 * The build process now looks for GNU make under the default name "gmake",
2709 instead of "make" as it used to. If GNU make is not available as "gmake"
2710 on your system, you can change this default behavior by setting the
2711 GNUMAKE environment variable.
2712 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2713 can build without error under CMU CL.
2715 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2717 * There is now a partial SBCL user manual (with some new text and some
2718 text cribbed from the CMU CL manual).
2719 * The beginnings of a profiler have been added (starting with the
2720 CMU CL profiler and simplifying and cleaning up). Eventually the
2721 main interface should be through the TRACE macro, but for now,
2722 it's still accessed through vaguely CMU-CL-style functions and macros
2723 exported from the package SB-PROFILE.
2724 * Some problems left over from porting CMU CL to the new
2725 cross-compilation bootstrap process have been cleaned up:
2726 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2727 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2728 part when getting rid of the compiler *BACKEND* stuff.)
2729 ** The value of *NULL-TYPE* was screwed up, because it was
2730 being initialized before the type system knew the final
2731 definition of the 'NULL type. This screwed up several key
2732 optimizations in the compiler, causing inefficiency in all sorts
2733 of places. (I found it because I wanted to understand why
2734 GET-INTERNAL-RUN-TIME was consing.)
2735 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2736 PROCLAIM FTYPE information. Unfortunately this broke binary
2737 compatibility again, since now the forms output by DEFGENERIC
2738 to refer to functions which didn't exist in 0.6.3.
2739 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2740 can use the new (as of 0.6.3) transform for SOME into MAP into
2742 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2743 values to (INTEGER 0 1000000), so that the time code will no longer
2744 occasionally get blown up by Linux returning 1000000 microseconds
2745 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2746 its output conform to the ANSI spec. (Alas, this makes its output
2747 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2748 * A full call to MAP NIL with a single sequence argument no longer conses.
2749 * fixes to problems pointed out by Martin Atzmueller:
2750 * The manual page no longer talks about multiprocessing as though
2751 it were currently supported.
2752 * The ILISP support patches have been removed from the distribution,
2753 because as of version 5.10.1, ILISP now supports SBCL without us
2754 having to maintain patches.
2755 * added a modified version of Raymond Toy's recent CMU CL patch for
2756 EQUALP comparison of HASH-TABLE
2758 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2760 * The system still can't cross-compile itself with
2761 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2762 entails), but at least it can compile more of itself that way
2763 than it used to be able to, and various buglets which were uncovered
2764 by trying to cross-compile itself that way have now been fixed.
2765 * This release breaks binary compatibility again. This time
2766 at least I've incremented the FASL file format version to 2, so that the
2767 problem can be detected reliably instead of just causing weird errors.
2768 * various new style warnings:
2769 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2770 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2771 ** using the lexical binding of a variable named in the *FOO* style
2772 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2773 as specified by ANSI.
2774 * *RANDOM-STATE* is no longer automatically initialized from
2775 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2776 default behavior of the system is to repeat its behavior every time
2777 it's run. If you'd like to change this behavior, you can always
2778 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2779 old convention there was no comparably easy way to get the system to
2780 repeat its behavior every time it was run.
2781 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2782 FTYPE declarations has been removed, in favor of their ANSI
2783 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2784 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2785 the types of their sequence arguments can be determined at compile time.
2786 This is done through a new open code expansion for MAP which eliminates
2787 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2788 argument types can be determined at compile time.
2789 * The optimizer now transforms COERCE into an identity operation when it
2790 can prove that the coerced object is already of the correct type. (This
2791 can be a win for machine generated code, including the output of other
2792 optimization transforms, such as the MAP transform above.)
2793 * Credit information has been moved from source file headers into CREDITS.
2794 * Source file headers have been made more standard.
2795 * The CASE macro now compiles without complaining even when it has
2798 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2800 * (Note that the way that the PCL macroexpansions were rewritten
2801 to accommodate the change in DEFGENERIC below breaks binary
2802 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2803 not run under sbcl-0.6.2. Once we get out of alpha releases,
2804 i.e. hit release 1.0.0, we'll probably try to maintain binary
2805 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2806 and sbcl-1.4.4. Until then, however, it might be fairly common
2807 for maintenance releases to break binary compatibility.)
2808 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2810 * The reporting of unhandled signals has been changed to print some
2811 explanatory text as well as the report form. (Previously only
2812 the report form was printed.)
2813 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2814 it defines, so that the compiler no longer issues undefined function
2815 warnings for compiled-but-not-yet-loaded generic functions.
2816 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2817 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2818 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2819 section of the ANSI spec are supported.
2820 * The "" nickname for the "KEYWORD" package has been removed.
2821 The reader still handles symbol tokens which begin with a package marker
2822 as keywords, but it doesn't expose its mechanism for doing so in the
2823 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2824 * The system now issues STYLE-WARNINGs for contradictory TYPE
2825 proclamations. (Warnings for contradictory FTYPE proclamations would
2826 be nice too, but those can't be done usefully unless the type system
2827 is made smarter about FUNCTION types.)
2828 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2829 systematized, so that "*target-*.lisp is supposed to exist only on the
2830 target and imply that there's a related file which exists on the
2831 host, and *host-*.lisp is supposed to exist only on the host and imply
2832 that there's a related file which exists on the target. This involves a
2833 lot of renaming. Hopefully the acute confusion caused by the renaming
2834 will be justified by the reduction in chronic confusion..
2835 ** runtime-type.lisp -> early-target-type.lisp
2836 ** target-type.lisp -> late-target-type.lisp
2837 ** early-host-format.lisp -> early-format.lisp
2838 ** late-host-format.lisp -> late-format.lisp
2839 ** host-error.lisp -> misc-error.lisp
2840 ** early-error.lisp -> early-target-error.lisp
2841 ** late-error.lisp -> late-target-error.lisp
2842 ** host-defboot.lisp -> early-defboot.lisp
2843 ** code/misc.lisp -> code/target-misc.lisp
2844 ** code/host-misc.lisp -> code/misc.lisp
2845 ** code/numbers.lisp -> code/target-numbers.lisp
2846 ** code/early-numbers.lisp -> numbers.lisp
2847 ** early-host-type.lisp -> early-type.lisp
2848 ** late-host-type.lisp -> late-type.lisp
2849 ** host-typep.lisp -> typep.lisp
2850 ** load.lisp -> target-load.lisp
2851 ** host-load.lisp -> load.lisp
2852 ** host-disassem.lisp -> disassem.lisp
2853 ** host-insts.lisp -> insts.lisp
2854 ** byte-comp.lisp -> target-byte-comp.lisp
2855 ** host-byte-comp.lisp -> byte-comp.lisp
2856 ** host-signal.lisp -> signal.lisp
2857 ** host-defstruct.lisp -> defstruct.lisp
2858 ** late-target-type.lisp -> deftypes-for-target.lisp
2859 Furthermore, several other previously target-only files foo.lisp (e.g.
2860 hash-table.lisp and random.lisp) have been split into a target-and-host
2861 foo.lisp file and a target-only target-foo.lisp file, with their key type
2862 definitions in the target-and-host part, so that the cross-compiler will
2863 know more about target types.
2864 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2865 gone away. In their place are various *BACKEND-FOO* variables
2866 corresponding to the slots of the old structure.
2867 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2868 into the target SBCL has been fixed.
2869 * The chill.lisp system for loading cold code into a running SBCL
2871 * Support for the CMU CL "scavenger hook" extension has been removed.
2872 (It was undocumented and unused in the CMU CL sources that SBCL was
2873 derived from, and stale in sbcl-0.6.1.)
2874 * Various errors in the cross-compiler type system were detected
2875 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2876 (enabling various consistency checks). Many of them were fixed,
2877 but some hard problems remain, so the compiler is back to
2878 running without *TYPE-SYSTEM-INITIALIZED* for now.
2879 * As part of the cross-compiler type system cleanup, I implemented
2880 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2881 * I have started adding UNCROSS calls throughout the type system
2882 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2883 on cross-compiler input files will be able to go away, and various
2885 * CONSTANTP now returns true for quoted forms (as explicitly required
2888 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2890 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2891 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2892 (MAKE-LIST -1). In the longer term, it ought to become true
2893 that declarations are assertions even at SAFETY 1. For now, it's not
2894 quite true even at SAFETY 3, but it's at least more nearly true..
2895 (Note that this change seems to increases the size of the system by
2896 O(5%) and to decrease the speed of the compiler by 20% or more.)
2897 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2898 for the problem of printing dozens of lines of distracting information
2899 about low-level system machinery as part of the top stack frame
2900 on entry to the debugger when an undefined function was called.
2901 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2902 is not reset to COMMON-LISP-USER.
2903 * Compilation of stuff related to dyncount.lisp has been made conditional
2904 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2905 smaller. The various dyncount-related symbols have been moved into
2906 a new "SB-DYNCOUNT" package.
2907 * tty-inspect.lisp has been renamed to inspect.lisp.
2908 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2909 feature has gone away. (When we eventually port to other flavors of
2910 libc and/or Unix, we'll try to make the differences between flavors
2911 invisible at the user level.)
2912 * Various other *FEATURES* tags, and/or their associated conditionals,
2913 have been removed if obsolescent, or given better documentation, or
2914 sometimes given more-mnemonic names.
2916 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2918 * tidied up "make.sh" script
2919 * tidied up system directory structure
2920 * better "clean.sh" behavior
2921 * added doc/FOR-CMUCL-DEVELOPERS
2922 * many many small tweaks to output format, e.g. removing possibly-confusing
2923 trailing #\. character in DESCRIBE-INSTANCE
2924 * (EQUALP #\A 'A) no longer signals an error.
2925 * new hashing code, including EQUALP hashing
2926 * tidied up Lisp initialization and toplevel
2927 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2928 * command line argument processing
2929 * added POSIX-GETENV function to deal with Unix-ish environment variables
2930 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2931 terminating SBCL on EOF
2932 * non-verbose GC by default
2933 * There is no more "sbcl" shell script; the sbcl file is now the C
2934 runtime executable (just like CMU CL).
2935 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2937 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2938 symbol and package information as Lisp native symbol and package objects
2939 instead of strings naming symbols and strings naming packages. This way,
2940 whenever packages are renamed (as in warm init), debug information is
2941 transformed along with everything else.
2942 * tweaked the optimization policy declarations which control the building
2943 of SBCL itself. Now, among other things, the system no longer saves
2944 source location debugging information. (This helps two problems at once
2945 by reducing SBCL size and by keeping SBCL from trying to look for its
2946 sources -- which may not exist -- when reporting errors.)
2947 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2948 debugging and testing purposes
2949 * cleaned up printing, making the printer call PRINT-OBJECT for
2950 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2951 methods, giving nearly-ANSI behavior
2952 * converted almost all special variables to use *FOO* naming convention
2953 * deleted PARSE-TIME functionality, since it can be done portably
2954 * moved some files out of cold init into warm init
2955 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2957 * regularized formatting of source files
2958 * added an install.sh script
2959 * fixed ridiculous memory usage of cross-compiler by making
2960 compiler/alloc.lisp not try to do pooling unless it can hook
2961 itself into the GC of the cross-compilation host. Now the system
2962 builds nicely on my old laptop.
2963 * added :SB-ALLOC in target-features.lisp-expr
2964 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
2965 was not implemented)
2966 * re-did condition handling and note reporting in the compiler. Notes
2967 are no longer handled by signalling conditions. Style warnings
2968 and warnings are handled more correctly and reported in such a way
2969 that it's easy to find one or the other in your output (so that you
2970 can e.g. figure out which of many problems caused COMPILE-FILE to
2972 * changed the severity of several compiler warnings from full WARNING
2973 to STYLE-WARNING in order to conform with the ANSI spec; also changed
2974 compiler note reporting so that it doesn't use the condition system
2975 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
2976 COMPILE-FILE command)
2977 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
2978 function. As a consequence, START-BLOCK and END-BLOCK declarations are
2979 no longer supported, since their implementation was deeply intertwingled
2980 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
2981 * removed bogus "support" for compiler macros named (SETF FOO), and
2982 removed the compiler macro for SETF INFO (but only after making a fool
2983 of myself on the cmucl-imp mailing list by posting a bogus patch for
2984 DEFINE-COMPILER-MACRO..)
2985 * Compiled files containing forms which have side effects on the Lisp
2986 reader (such as DEFPACKAGE forms) are now handled more correctly.
2987 (Compiler queuing of top level lambdas has been suppressed by setting
2988 *TOP-LEVEL-LAMBDA-MAX* to 0. )
2989 * deleted various currently-unused source files, e.g. gengc.lisp. They
2990 may be added back at some point e.g. when porting to other architectures,
2991 but until they are it's distracting to distribute them and to try to
2993 * deleted "UNCROSS couldn't recurse through.." style warnings, since
2994 there were so many of them they're just distractions, and UNCROSS is
2995 known to be able to handle the current sources
2996 * moved PROFILE functionality into TRACE, so that it will be clear
2997 how the wrapping and unwrapping of functions when you profile them
2998 interacts with the wrapping and unwrapping of functions when you
2999 trace them. (Actually, the functionality isn't there yet, but at least
3000 the interface specification is there. Hopefully, the functionality will
3001 arrive with some future maintenance release.)
3002 * removed host-oops.lisp
3003 * changed signature of QUIT function to allow UNIX-CODE argument
3004 * fixed READ-SEQUENCE bug
3005 * tweaked verbose GC output so that it looks more like the progress
3006 output that ANSI specifies for functions like LOAD
3007 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3008 * added <http://sbcl.sourceforge.com> to the banner information printed by