1 changes in sbcl-0.8.22 relative to sbcl-0.8.21:
2 * incompatible change: the --noprogrammer option, deprecated since
3 version 0.7.5, has been removed. Please use the equivalent
4 --disable-debugger option instead.
5 * Null lexical environments are now printed as #<NULL-LEXENV>,
6 significantly reducing the amount of clutter in typical backtraces.
7 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
8 as other specialized array types, is much faster. SUBSEQ and
9 COPY-SEQ on such arrays have also been sped up.
10 * fixed bug: compiler error messages and summaries are now printed to
11 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
12 * fixed inference of the upper bound of an iteration variable.
13 (reported by Rajat Datta).
14 * fixed bug 376: CONJUGATE type deriver.
15 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
16 (reported by Baughn on #lisp)
17 * a cross-compiler bug on non-x86 platforms has been identified and
18 fixed. (thanks to Bruno Haible)
19 * fixed some bugs related to Unicode integration:
20 ** the restarts for recovering from input and output encoding
21 errors only appear when there is in fact such an error to
23 * fixed some bugs revealed by Paul Dietz' test suite:
24 ** MISC.549 and similar: late transformation of unsafe type
25 assertions into derived types caused unexpected code
27 ** SCALE-FLOAT type deriver is less wrong.
28 ** type derivers for EXP, LOG and similar functions compute result
29 types for complex arguments better.
30 ** (MISC.563) CONJUGATE type deriver works for very restricted
32 ** out-of-line type testers for character strings are available.
33 ** EQUAL compiler transform understands specialness of objects
35 ** accessing double-floats stored on the stack now works on x86-64.
36 ** debugger internals could sometimes create invalid lispobjs,
37 resulting in GC crashes.
39 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
40 * incompatible change: thread support for non-NPTL systems has
41 been removed - locking is buggy and unreliable. A threaded
42 SBCL build will now warn at startup and refuse to create
43 new threads, unless futex support is detected in the kernel
44 * incompatible change: the top level REPL now has only an ABORT
45 restart associated with it, not TOPLEVEL and ABORT as it used to.
46 TOP and TOPLEVEL are now available as debugger commands for
47 returning to the top level.
48 * incompatible change: forms evaluated in the REPL now use the
49 global optimization policy.
50 * incompatible change: user- and system-initialization files are
51 no longer processed with LOAD, but by READ and EVAL; hence the
52 global optimization policy, startup package, readtable, etc,
54 * The .fasl file format number has been incremented because of
55 various incompatible changes.
56 * internal entry point details and argument counts no longer appear
57 in backtraces unless explicitly requested by setting
58 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
59 * built-in and standard functions no longer have names like "top
60 level local call to FOO".
61 * fixed bug 32: functions defined in non-null lexical environments
62 now have more legible printed representation
63 * fixed bug 33: functions defined in non-null lexical environemnts
64 are now more amenable to inspection by INSPECT.
65 * workaround for bug 354: XEPs no longer appear in backtraces unless
67 * fixed bug: receiving the signal which results from attempting to
68 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
69 SIGBUS on other *BSDs) on architectures where the C stack is also
70 the Lisp stack (x86 and x86-64) from foreign code no longer leads
71 to debugger confusion or wild instability. (reported by Cheuksan
73 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
74 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
75 (reported by Lutz Euler)
76 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
77 compiled code. (reported by Paul Dietz)
78 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
79 specialized parameter to a method to something that is not TYPEP
80 the specializer is now possible.
81 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
82 face of package deletion.
83 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
84 pathname. (reported by David Tolpin on c.l.l)
85 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
86 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
87 STRING-STREAM and FILE-STREAM, respectively.
88 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
89 than previously; in addition, its test suite is now run on build.
90 (thanks to Luke Gorrie)
91 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
92 * optimization: added a immediate representation for single-floats
94 * fixed some bugs revealed by Paul Dietz' test suite:
95 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
96 ** attempting to create a package with a colliding nickname causes
97 correctable errors to be signalled.
98 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
99 ** several x86-64 backend bugs related to sign-extension of immediate
102 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
103 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
104 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
106 * fixed loading of multiply forward-referenced layouts.
107 (thanks to Cheuksan Wang)
108 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
109 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
110 (see COMPILE-FILE documentation for details of :PRINT option)
111 * fixed bugs 19 and 317: fixed-format floating point printing is
112 more accurate. This also fixes a bug reported by Adam Warner
113 related to the ~@F format directive.
114 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
116 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
117 dispatch table if the from-char is a dispatch macro character.
118 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
119 by the wordsize no longer ignore the last word. (reported by Lutz
121 * fixed bug in type checking of dynamic-extent variables. (reported
123 * optimization: sequence traversal functions use their freedom to
124 coerce function designators to functions.
125 * optimization: code with many calls to CLOS methods specialized on
126 CLOS classes has had redundant type checks removed.
127 * optimization: added declarations to speed up operations that access
128 the internal character database (for example STRING-UPCASE)
129 * optimization: comparison operations between floats and sufficiently small
130 fixnums no longer create extra rationals
131 * fixed some bugs related to Unicode integration:
132 ** portions of multibyte characters at the end of buffers for
133 character-based file input are correctly transferred to the
134 start of the buffer at the next read.
135 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
136 passing it through to OPEN.
137 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
138 argument given to internal calls to OPEN.
139 ** the built-in comment readers (introduced by character sequences
140 ";" and "#|") are more forgiving to encoding errors; they will
141 STYLE-WARN and then attempt to resync the stream at a character
142 boundary. (thanks to Teemu Kalvas)
143 * fixed some bugs in the x86-64 port:
144 ** Negative short int return values from c-calls are sign-extended
146 ** The stack is aligned to 16-bytes for c-calls, as required by
147 the ABI. (thanks to Cheuksan Wang)
148 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
149 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
150 for enums. (thanks to Vincent Arkesteijn)
151 ** Multiple small optimizations and bugfixes for floating point
153 * fixed some bugs revealed by Paul Dietz' test suite:
154 ** Space, Tab, Linefeed, Return and Page have the invalid
155 secondary constituent character trait.
156 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
158 ** WITH-INPUT-FROM-STRING should only update the index place on
160 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
162 ** Bit-array operations (BIT-AND and similar) worked incorrectly
163 with one-dimensional arrays with fill pointers.
164 ** TYPE-OF failed on a complex with an integer realpart and a
166 ** compiler failure during type inference for the code of form
167 (IF (EQL X (THE ...)) ...) (MISC.535).
169 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
170 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
171 platform. The port supports 61-bit fixnums, large memory spaces
172 and reloading shared object files.
173 * enhancement: saving cores with foreign code loaded is now
174 supported on ppc/Darwin in addition to the previously supported
176 * enhancement: the statistical profiler now walks deeper into the
177 call stack for more meaningful call-graphs and accrued time
178 reports (x86/x86-64 only). It also now reports time spent in
180 * enhancement: it is now possible to trace most individual methods
181 of a generic function in addition to tracing the generic function
183 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
185 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
186 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
187 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
188 produces an error. (thanks to Vincent Arkesteijn)
189 * bug fix: NAMESTRING on pathnames with :WILD components in their
190 directories works correctly. (thanks to Artem V. Andreev)
191 * fixed bug 125: compiler preserves identity of closures. (reported
193 * bug fixed: functions with &REST arguments sometimes failed with
194 "Undefined function" when compiled with (DEBUG 3). (reported by
196 * bug fix: overflow during compiling of setting element of a bit
197 vector with constant index and value. (reported by Timmy Douglas)
198 * build fix: fixed the dependence on *LOAD-PATHNAME* and
199 *COMPILE-FILE-PATHNAME* being absolute pathnames.
200 * on x86 compiler partially supports stack allocation of dynamic-extent
202 * GO and RETURN-FROM do not check the extent of their exit points
203 when compiled with SAFETY 0.
204 * fixed some bugs related to Unicode integration:
205 ** encoding and decoding errors are now much more robustly
206 handled; it should now be possible to recover even from invalid
207 input or output to the terminal. (thanks to Teemu Kalvas)
208 ** provided a first cut at implementing STRING-TO-OCTETS and
209 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
210 ** altered the SB-MD5 contributed module slightly, changing the
211 interface just enough for it to be supportable for builds where
212 lisp characters are not eight bits.
213 * fixed some bugs revealed by Paul Dietz' test suite:
214 ** the FORMATTER-generated functions for ~V[ conditionals require
215 the correct number of arguments.
216 ** READ-FROM-STRING returns the mandated second value when applied
217 to displaced strings.
218 ** the #\Rubout and #\Backspace characters are treated as invalid
219 constituent characters by the tokenizer.
221 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
222 * new feature: reloading changed shared object files with
223 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
224 * new feature: references to foreign variables and functions
225 can now be compiled and loaded before the corresponding shared
226 object file is loaded, as long as the foreign definitions are
227 available at runtime.
228 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
229 supported. (thanks to Dan Debertin)
230 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
231 just keywords. (thanks to Vincent Arkesteijn)
232 * fixed bug #331: structure-class instances corresponding to
233 DEFSTRUCT forms are now created eagerly.
234 * fixed bug #345: backtraces from calls to undefined functions work
235 on x86 as well. Related bug #61 is now also partially fixed on x86
236 (backtraces from throws to unknown catch tags.)
237 * bug fix: lambda-list parsing is now stricter vrt. order and number
238 of lambda-list keywords.
239 * bug fix: as specified by AMOP, an error is signalled if a
240 class-option appears multiple times in a DEFCLASS form. (reported
242 * bug fix: RANDOM can be compiled when the compiler derives the type
243 of its numeric argument as a disjoint set of small integers.
244 (reported by Paul Dietz)
245 * bug fix: starting a core saved with shared objects loaded when
246 those objects are not available no longer causes threaded SBCL to
247 hang. (reported by Sean Ross)
248 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
250 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
251 argument, which must always be NIL. (reported by Kalle Niemitalo)
252 * bug fix: printing 1.0d+23 no longer results in an error.
253 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
254 * bug fix: structure slot setters preserve evaluation order. (thanks
256 * fixed some bugs related to Unicode integration:
257 ** RUN-PROGRAM can allow its child to take input from a Lisp
258 stream. (reported by Stefan Scholl)
259 ** ASDF-INSTALL successfully downloads .tar.gz files in all
260 locales. (reported by Ken Causey)
261 * fixed some bugs revealed by Paul Dietz' test suite:
262 ** INCF, DECF and REMF evaluate their place form as specified in
264 ** FORMATTER expands ~{ ~} iteration directives with V or #
265 parameters correctly.
266 ** FORMATTER deals with the ~@[ ~] conditional directive where the
267 consequent uses no arguments correctly.
268 ** the system has a partial understanding of the (COMPLEX RATIO)
270 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
273 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
274 * new feature: a build-time option (controlled by the :SB-UNICODE
275 keyword feature, enabled by default) for building the system with
276 support for the entire 21-bit character space defined by the
278 * new feature: the system now has rudimentary external-format
279 support; the primary user-visible change at this time is that
280 characters with the high bit set (such as lower-case-e-acute) will
281 print correctly to a terminal in a UTF-8 environment.
282 * minor incompatible change: BASE-CHAR no longer names a class;
283 however, CHARACTER continues to do so, as required by ANSI.
284 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
285 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
286 has been moved to the SB-EXT package (temporarily re-exported from
288 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
289 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
291 * fixed bug #351: better error-handlind and reporting for malformed
293 * fixed bug #350: bignum-printing is now more memory-efficient,
294 allowing printing of very large bignums, eg. (expt 2 10000000).
295 (reported by Bruno Haible)
296 * fixed bug #302: better primitive-type selection for intersection
298 * fixed bug #308: non-graphic characters now all have names, as
299 required. (reported by Bruno Haible)
300 * bug fix: redefining a class with different superclasses now correctly
301 removes it from the direct-subclasses of its previous superclasses.
302 (reported by David Morse)
303 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
304 new value now works. (reported by Bruno Haible)
305 * bug fix: correct canonicalization of multiple non-standard slot
306 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
307 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
308 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
309 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
310 now exists, an signals an error.
311 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
312 during macroexpansion for non-iteration variables as well. (reported
313 by Bruno Haible for CMUCL)
314 * bug fix: Cyclic structures and unprintable objects in compiler
315 messages no longer cause errors. (reported by Bruno Haible)
316 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
317 and READ-FROM-STRING all now return a primary value of NIL if
318 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
319 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
320 (reported by Bruno Haible for CMUCL)
321 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
322 do not designate a condition. (reported by Bruno Haible for
324 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
325 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
326 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
327 permitting resolution in favour of any of the conflicting symbols.
328 (reported by Bruno Haible for CMUCL)
329 * bug fix: EQUAL compiler optimizations is less aggressive on
330 strings which can potentially compare true despite having distinct
331 specialized array element types.
332 * bug fix: unit enumerations can be defined without dividing by
333 zero. (thanks to Vincent Arkesteijn)
334 * FORMAT compile-time argument count checking has been enhanced.
335 (report from Bruno Haible for CMUCL)
336 * a partial workaround for the bug 262: the compiler does not try to
337 inline-expand a local function doing RETURN-FROM from a deleted
338 BLOCK. (thanks to Peter Denno for the bug report and to David
339 Wragg for the simple test case)
340 * fixed some bugs revealed by Paul Dietz' test suite:
341 ** RENAME-PACKAGE allows all package designators as new package
343 ** constraint propagation and lambda variable substitution are
344 more cautious in dealing with partially deleted code.
345 ** compiler sometimes lost reoptimization passes.
346 ** CERROR, given a condition as condition designator, passes its
347 remaining arguments to the continue format control without
349 ** Case-altering FORMAT directives work correctly on non-ASCII
351 ** The REAL type specifier handles bounds outside the floating
352 point ranges without signalling FLOATING-POINT-OVERFLOW.
353 ** Functions with IR1-transformations can create intercomponent
354 references to global functions.
355 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
357 ** FORMAT ~R treats a nil value for its first parameter correctly.
359 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
360 * enhancement: saving cores with foreign code loaded is now
361 supported on x86/NetBSD and sparc/Linux in addition to the previously
363 * bug fix: on some platforms repeated installations caused multiple
364 copies of HTML documentation to be installed -- should not happen
365 any more. (reported by Stefan Scholl)
366 * bug fix: parsing self-recursive alien record types multiple times
367 no longer causes infinite recursion. (reported by Thomas F. Burdick,
368 original patch by Helmut Eller for CMUCL)
369 * bug fix: stack-exhaustion detection works now on NetBSD as well.
370 (thanks to Richard Kreuter)
371 * bug fix: defining classes whose accessors are methods on existing
372 generic functions in other (locked) packages no longer signals
373 bogus package lock violations. (reported by François-René Rideau)
374 * bug fix: special variables as DEFMETHOD parameters no longer have
375 associated bogus type declarations. (reported by David Wragg and
377 * bug fix: read-write consistency on streams of element-type
378 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
379 * bug fix: redefiniton of the only method of a generic function with
380 no DEFGENERIC no longer emits a full WARNING. In addition,
381 redefinition of generic functions with no DEFGENERIC to an
382 incompatible lambda list now signals an error. (thanks to Zach
384 * bug fix: DEFGENERIC now works even when there's a function of the
385 same name in an enclosing lexical environment. (thanks to Zach
387 * fixed compiler failure, caused by instrumenting code during
388 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
389 * optimization: added loop analysis and improved register allocation
390 * fixed some bugs revealed by Paul Dietz' test suite:
391 ** POSITION on displaced vectors with non-zero displacement
392 returns the right answer.
393 ** (SIMPLE-STRING) is a valid type specifier for sequence
395 ** *PRINT-LEVEL* handling for slotless structures is pedantically
397 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
398 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
400 ** The justification version of the FORMAT ~< directive treats
401 non-zero minpad parameter correctly.
403 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
404 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
405 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
406 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
407 the supported interface.
408 * new feature: Single-stepping of code compiled with DEBUG 2 or
409 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
410 * new feature: saving cores with foreign code loaded is now
411 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
412 Timothy Moore's work for CMUCL)
413 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
414 parameters to * instead of NIL if no initform is supplied.
415 (reported by Johan Bockgaard)
416 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
417 correctly when FUNCALL appears as the car of the form. Note:
418 despite this FUNCALL forms are not currently subject to
419 compiler-macro expansion. (port of Raymond Toy's fix for the
420 same from CMUCL, reported by Johan Bockgaard)
421 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
422 (thanks for Teemu Kalvas)
423 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
424 case instead of terminating the iteration (thanks for Julian
425 Squires, Sean Champ and Raymond Toy)
426 * bug fix: incorrect expansion of defgeneric that caused a style
427 warning. (thanks for Zach Beane)
428 * bug fix: slot accessor effective method computation works properly
429 for classes with multiple non-standard applicable methods on
430 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
431 * on x86 compiler supports stack allocation of results of LIST and
432 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
433 CMUCL implementation by Gerd Moellmann)
434 * fixed some bugs revealed by Paul Dietz' test suite:
435 ** FORMAT strings with both the ~<~:;~> form of the justification
436 directive and pretty-printing directives cause an error.
438 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
439 * incompatible change: the internal functions
440 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
441 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
442 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
443 instead of the old functions.
444 * new feature: on platforms where "dladdr" is available foreign
445 function names now appear in backtraces. (based on Helmut Eller's
447 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
448 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
450 * documentation: ASDF manual is now distributed with SBCL.
451 * bug fix: SBCL can now load its contributed modules with REQUIRE
452 even if the system-provided entries have been removed from
453 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
455 * bug fix: backtraces involving undefined functions or assembly
456 routines are more informative. (thanks to Brian Downing)
457 * bug fix: mutually referent alien structures now work correctly.
458 (reported by Rick Taube)
459 * bug fix: structures defined by WITH-ALIEN can be referred to
460 within other definitions in the same WITH-ALIEN.
461 * bug fix: division operators (MOD, TRUNCATE and the like) with
462 constant zero divisors and integer dividends no longer generate
464 * bug fix: provide default methods for INPUT-STREAM-P and
465 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
466 * bug fix: improve the detection and resolution of MOP metacycles
467 (where effective method computation depends on the generic function
468 having its effective method computed). (reported by Bruno Haible)
469 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
470 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
471 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
472 * bug fix: pathnames with relative directory components can be
473 represented relative to default pathnames.
474 * optimization: in taking the GCD of bignums, reduce the two bignums
475 to approximately the same size (using Euclid's algorithm) before
476 applying the more sophisticated binary GCD. (thanks to Juho
478 * optimization: COUNT on bitvectors now operates word-at-a-time.
479 * optimization: ASH with a positive, but not necessarily constant,
480 (leftwards) shift, when in a modular context, is compiled to a
482 * fixed some bugs revealed by Paul Dietz' test suite:
483 ** FORMAT variable parameters ("~V<char>") are defaulted properly
484 if the corresponding argument is NIL.
485 ** FORMAT directives accepting a minpad parameter treat negative
487 ** literal commas as character parameters to FORMAT directives are
489 ** literal spaces directly after ~<Newline> directives within a
490 format-logical-block (~:< ~@:>) do not induce :FILL-style
491 conditional newlines.
492 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
493 as specified: it no longer includes conditional newlines).
494 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
496 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
497 :PER-LINE-PREFIX argument does not evaluate to a string.
499 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
500 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
501 the manual for details; for now, package locks can be disabled by
502 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
503 if no major problems are found then it is likely that they will be
504 compiled in unconditionally.
505 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
506 unsupported operators on all platforms. To load a shared library
507 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
508 object file, link it into a shared library outside of SBCL and
509 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
511 * bug workaround: ROOM T and the GC were not working together
512 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
513 reporting facility aren't preserved. That reporting has been
514 disabled, so now until and unless someone figures out how to make it
515 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
516 * minor incompatible change: as threatened around sbcl-0.8.0, the
517 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
518 The symbols are also exported from SB-PCL for backwards
519 compatibility, but more so than before SB-PCL should be treated as
520 an implementation-internal package.
521 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
523 * the SB-SPROF contrib now works on (most) non-x86 architectures.
524 It is known as of this release not to work on the Alpha, however.
525 * fixed bug #167: errors signalled due to illegal syntax in method
526 bodies are now more legible.
527 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
528 designators and can hence be used with TYPEP.
529 * fixed bug #333: CHECK-TYPE now ensures that the type error
530 signalled, if any, has the right object to be accessed by
531 TYPE-ERROR-DATUM. (reported by Tony Martinez)
532 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
533 5.1.2.3 for multiple-value place subforms. (reported by Kalle
535 * fixed bug #334: programmatic addition of slots using specialized
536 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
537 and :ALLOCATION :CLASS slots.
538 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
539 float exponents. (rereported by Peter Seibel)
540 * fixed a bug: #\Space (and other whitespace characters) are no
541 longer considered to be macro characters in standard syntax by
543 * fixed bug: initialization of condition class metaobjects no longer
544 causes an instance of the condition to be created. (reported by
546 * fixed bug: it is now possible to have slots such that
547 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
548 description is neither :INSTANCE nor :CLASS.
549 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
550 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
551 SLOT-VALUE-USING-CLASS), and no longer causes errors with
552 non-standard SLOT-DEFINITION-ALLOCATION values.
553 * fixed bugs: various race conditions handling exiting threads.
554 CL-PPCRE's thread torture test now passes.
555 * fixed arguably-a-bug: GC time proportional to number of threads in
556 system even when most of them are idle
557 * optimization: improved performance of BIT and SBIT on bit-vectors.
558 * .fasl file incompatibility: The fasl file version number has
559 been incremented because of changes associated with package locks.
561 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
562 * minor incompatible change: the system no longer provides
563 optimization hints (conditions which are TYPEP
564 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
565 mode; these hints are still emitted when higher SPEED optimization
567 * new contrib module: a sampling profiler (profiling by statistical
568 sampling, rather than by instrumenting functions) is available as
569 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
570 * the behaviour of the standard function ED is now customizeable by
571 third parties through a hook variable: see ED's documentation
572 string for information on the protocol.
573 * the compiler no longer emits efficiency notes for (FUNCALL X)
574 when the type of X is uncertain under default optimization
576 * fixed bug 276: mutating a binding of a specialized parameter to a
577 method to something that is not TYPEP the specializer is now
579 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
580 exhibit the required behaviour on the x86 platform. (thanks to
581 Peter van Eynde, Eric Marsden and Bruno Haible)
582 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
583 even for difficult arguments. (reported by Peter Graves)
584 * fixed bug 141a: the backquote printer now descends quoted
586 * fixed another bug in backquote printing: no more destructive
587 modification of the form's list structure. (reported by Brian
589 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
590 move between its address being taken and the call to
591 interrupt_thread, fixing a crashing race condition.
592 * the SB-POSIX contrib implementation has been adjusted so that it
593 no longer exhibits ridiculously poor performance when constructing
594 instances corresponding to C structs.
596 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
597 * minor incompatible change: the sb-grovel contrib now treats C
598 structures as alien (in the sense of SB-ALIEN) objects rather than
599 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
600 has implications for memory management of client code
601 (sb-grovel-returned objects must now be manually managed) and for
602 type safety (alien objects now have full types).
603 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
604 used to control emission of compiler diagnostics, rather than the
605 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
606 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
607 quality should be considered deprecated.
608 * install.sh now installs the user manual as well
609 * (not quite a new documentable feature, but worth considering in
610 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
611 beginnings of a semantically meaningful condition hierarchy is
612 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
614 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
615 designator as the defaults argument.
616 * fixed bug: Displaced arrays whose displaced-to array has become
617 too small now cause ARRAY-DIMENSION to signal an error, providing
618 for safer bounds-checking. (reported by Bruno Haible)
619 * fixed bug: DEFCLASS slot definitions with identical :READER and
620 :WRITER names now signal a reasonable error. (reported by Thomas
622 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
624 * fixed bug: Invalid slot specification errors now print correctly.
625 (thanks to Zach Beane)
626 * fixed bug 320: Shared to local slot value transfers in class
627 redefinitions now happen corrently from superclasses as
628 well. (reported by Bruno Haible)
629 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
631 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
632 improper lists correctly. (reported by Bruno Haible)
633 * fixed bug 313: source-transform for <fun-name> was erroneously
634 applied to a call of a value of a variable with name <fun-name>.
635 (reported by Antonio Menezes Leitao)
636 * fixed bug 307: The obsolete instance protocol ensures that
637 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
638 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
639 remove dead unknown-values globs from the middle of the stack.
640 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
641 easier to resume long computations after using *BREAK-ON-SIGNALS*
642 to diagnose and fix failures. (thanks to Nikodemus Siivola)
643 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
644 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
645 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
646 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
647 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
649 * fixed a (fairly theoretical) bug in string printing: if
650 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
651 does not have array-element-type equal to the most general string
653 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
654 function name. (thanks to Nikodemus Siivola)
655 * fixed bug: compile-time format string checker failed on
656 non-closed ~{. (reported by Thomas F Burdick)
657 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
658 don't warn on legal loop code involving more than one aggregate
659 boolean. (thanks to Nikodemus Siivola)
660 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
661 in keyword position to MAKE-PACKAGE was not being reported as
663 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
664 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
666 * fixed bugs 280 and 312: the checking for multiple definitions in a
667 file is less likely to become confused by uses of inline
669 * fixed bug: the #S reader macro performs the keyword coercion
670 specified for slot names. (reported by Kalle Niemitalo)
671 * fixed bug: lambda lists may contain symbols whose names start with
672 & but are not lambda-list-keywords; their occurrence triggers a
674 * fixed bug 321: define-method-combination argument lambda lists do
675 not cause invalid code to be generated when &optional or &aux
676 variables have default values. (reported by Bruno Haible)
677 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
678 have CLOS classes; this appears also to have fixed reported
679 instability in stack exhaustion detection.
680 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
681 MUFFLE-WARNING when no associated restart is present is now
683 * optimization: rearranged the expansion of various defining macros
684 so that each expands into only one top-level form in a
685 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
687 * optimization: used a previously little-used slot in symbols to
688 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
690 * fixed some bugs revealed by Paul Dietz' test suite:
691 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
693 ** Fixed an optimization bug related to inheritance of initforms
694 from local to shared slots.
695 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
696 ** CAST optimizer forgot to flush argument derived type.
697 ** print/read consistency on floats is now orders of magnitude
698 more likely. (thanks also to Bruno Haible for a similar report
700 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
701 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
702 for objects of type REAL. Make it so.
703 ** SXHASH is specified (infelicitously) to respect similarity,
704 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
705 it so. (thanks to Markus Ziegler)
706 ** on the Alpha, the compiler succeeds in compiling functions
707 returning a known number of arguments greater than 63.
708 ** fixed handling of invalid NIL arguments in keyword position
710 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
712 ** provided a readably-printable representation for RANDOM-STATE
714 ** ensured that pathnames, where they have a namestring, always
715 print using #P"..." syntax.
717 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
718 * Support for the forthcoming 2.0 version of the NetBSD kernel
719 running on x86 hardware has been added. (thanks to Perry
720 E. Metzger most immediately, and others for their past work)
721 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
722 current release version; SBCL's OpenBSD support had been broken
723 since about the time of OpenBSD's switch to ELF binary format.)
724 * [placeholder for DX summary]
725 ** user code with &REST lists declared dynamic-extent, under high
726 speed or space and low safety and debug optimization policy.
727 * The manual has been converted to Texinfo format and the debugger
728 chapter from the cmucl manual has been added.
729 * A facility has been added to extract documentation strings from
730 sbcl and store them as Texinfo-formatted snippets for inclusion in
731 the manual (via Texinfo's @include directive)
732 * bug fix: compiler emitted division in optimized DEREF. (thanks for
733 the test case to Dave Roberts)
734 * bug fix: multidimensional simple arrays loaded from FASLs had fill
735 pointers. (reported by Sean Ross)
736 * bug fix: PROFILE output is printed nicely even for large numerical
737 values. (thanks to Zach Beane)
738 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
739 greater than 32 handle EOF correctly.
740 * bug fix: on X86 an immediate argument of the IMUL instruction is
741 correctly printed in disassembly. (thanks to Lutz Euler)
742 * bug fix: class slots in redefined classes preserve their old
743 values. (thanks to Bruno Haible and Nikodemus Siivola)
744 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
745 succeeds. (reported by Marco Baringer)
746 * fixed some bugs revealed by Paul Dietz' test suite:
747 ** READ-SEQUENCE now works on ECHO-STREAMs.
748 ** RATIONALIZE works more according to its specification. (thanks
751 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
752 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
753 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
754 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
755 should matter to you only if you rebind the printer control
756 variables and then find you want different bindings in the
757 debugger than in the ordinary execution of your program.)
758 * The runtime build system has been tweaked to support building
759 (on SPARC/SunOS) using a C compiler which invokes Sun's own
760 assembler and linker. (thanks to Nikodemus Siivola)
761 * Unbound, undefined, undeclared variables now trigger full
762 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
763 more often programmer error than deliberate exploitation of undefined
765 * optimization: the hash algorithm for strings has changed to one
766 that is less vulnerable to spurious collisions. (thanks to Juho
768 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
769 needless bounds checking. (thanks to Juho Snellman)
770 * optimization: implemented multiplication as a modular
771 (UNSIGNED-BYTE 32) operation on the PPC backend.
772 * fixed some bugs revealed by Paul Dietz' test suite:
773 ** ADJUST-ARRAY now copies the datum in a zero rank array if
775 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
776 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
777 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
779 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
781 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
783 ** ECHO-STREAMs no longer attempt to echo the end of file value to
784 their output stream on EOF from read.
785 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
786 have been read to end-of-file.
787 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
789 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
790 description of determination of which consecutive characters
792 ** Printing the "Space" character with escaping on now yields "#\\ ",
793 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
794 ** Reading floating-point numbers with *READ-BASE* set to a number
795 less than 10 works correctly.
796 ** Reading floating-point numbers with *READ-BASE* set to a number
797 more than 10 works correctly.
798 ** Printing with *PRINT-READABLY* targets the standard readtable, not
799 the readtable currently in effect.
801 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
802 * minor incompatible change: parsing of namestrings on a physical
803 (Unix) host has changed; numbers after the final #\. in a
804 namestring are no longer interpreted as a version field. This is
805 intented to be largely invisible to the user, except that the
806 meaning of the namestring "*.*.*" has changed: it now refers to a
807 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
808 should usually be replaced by
809 (make-pathname :name :wild :type :wild :version :wild)
810 with the added benefit that this is more likely to be portable.
811 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
812 option to OPEN now signals an error if the file being opened
813 exists; this may have an impact on existing code.
814 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
815 signals received, on the PowerPC platforms (both Linux and
816 Darwin). (thanks to Pierre Mai for pointing out the location of
818 * several fixes on OS X: The system now builds and runs cleanly on
819 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
820 (thanks to Brian Mastenbrook)
821 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
822 second-resolution: integer multiples of 1/3600 between -24 and 24.
823 (thanks to Vincent Arkesteijn)
824 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
825 type when called with 1 argument; PEEK-CHAR checked type of
826 PEEK-TYPE only after having read first character from a
827 stream. (reported by Peter Graves)
828 * bug fix: the garbage collector now has much better locality
829 behaviour, and in particular no longer treats all memory as being
830 exhausted when many small objects point to each other in a deeply
832 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
833 recognized as being TYPEP their class.
834 * bug fix: the PUSHNEW documentation string has been corrected.
835 (thanks to Vincent Arkesteijn)
836 * bug fix: defaulting of the value for the last of an atypically
837 large number of multiple values being bound was not being
838 performed correctly on the Alpha or PPC platforms
839 * optimization: implemented multiplication as a modular
840 (UNSIGNED-BYTE 32) operation on the x86 backend.
841 * optimization: SEARCH on simple-base-strings can now be open-coded.
842 (see also contrib/compiler-extras.lisp for inspiration for
843 teaching the compiler about the Boyer-Moore algorithm).
844 * value, returned by MAX (and MIN) called with several EQUALP, but
845 not EQL, arguments now does not depend on compiler settings.
846 * fixed some bugs revealed by Paul Dietz' test suite:
847 ** in stack analysis liveness information is propagated from
848 non-local entry points.
849 ** pathwise CAST removing failed when the CAST node did not start
851 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
852 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
854 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
855 host is already defined.
856 ** RENAME-FILE works on streams instead of signalling an internal
858 ** PEEK-CHAR uses the current readtable when determining whether
859 or not a character is whitespace.
860 ** MERGE-PATHNAMES handles the case when the pathname does not
861 specify a name while the default-pathname specifies a version.
862 ** Pathnames now stand a better chance of respecting print/read
864 ** Attempting to use standardized file system operators with a
865 pathname with invalid :DIRECTORY components signals a
867 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
869 ** DIRECTORY on logical pathnames is more correct.
870 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
871 signal a TYPE-ERROR if their argument is not a stream.
872 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
873 designator argument does not designate a stream.
874 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
875 examining the synonym.
876 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
878 ** OPEN and WITH-OPEN-STREAM allow opening streams with
879 element-type larger than ([UN]SIGNED-BYTE 32).
881 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
882 * When built with the :SB-FUTEX feature, threaded builds now take
883 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
884 for faster/more reliable mutex and condition variable support.
885 * Incompatible change (but one you probably shouldn't have been using
886 anyway): the interface and code for arbitrating between multiple
887 threads in the same user session has been redesigned.
888 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
889 error when called without an explicit environment argument.
890 (thanks to Nikodemus Siivola)
891 * bug fix: buffered :DIRECTION :IO streams are less likely to become
892 confused about their position. (thanks to Adam Warner and Gerd
894 * bug fix: Pretty printing backquoted forms with unquotations in the
895 argument list position of various code constructs such as LAMBDA
896 now works correctly. (reported by Paul Dietz)
897 * bug fix: Pretty printing unquotations no longer loses all
898 stream position information.
899 * optimization: performance of string output streams is now less
900 poor for multiple small sequence writes.
901 * optimization: performance of CSUBTYPEP in the presence of complex
902 expressions involving CONS and NOT many times has been improved.
903 (reported by Paul Dietz)
904 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
906 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
907 streams for the REPL, for the debugger, and so on. (thanks to
909 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
910 * fixed some bugs revealed by Paul Dietz' test suite:
911 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
912 is ignored if it is too small, rather than propagating through
913 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
914 ** extremely complex negations of CONS types were not being
915 sufficiently canonicalized, leading to inconsistencies in
917 ** VALUES tranformer lost derived type.
919 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
920 * fixed a bootstrapping bug: the build process no longer assumes
921 that the various BOOLE-related constants have the same value in
922 host and target lisps. (noted by Paul Dietz' test suite on an
923 SBCL binary built from CLISP)
924 * The system can now be dynamically linked on the MIPS platform,
925 which enables dynamic loading of foreign code from Lisp. (thanks
926 to Ralf Baechle for discussions on the MIPS ABI)
927 * The system now records debugging information for its own source
928 files in a filesystem-position-independent manner, relative to
929 the "SYS" logical host.
930 * fixed a compiler bug: MV-LET convertion did not check references
931 to the "max args" entry point. (reported by Brian Downing)
932 * tweaked disassembly notes to be less confident about proclaiming
933 some instruction as an LRA. (thanks to Brian Downing)
934 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
935 now each have their own history, command character, and other
936 characteristics. (thanks to David Lichteblau)
937 * fixed some bugs revealed by Paul Dietz' test suite:
938 ** compiler failure in compiling LOGAND expressions including a
940 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
941 shift greater than 32.
942 ** FLUSH-DEST did not mark blocks for type check regeneration.
943 ** HANDLER-CASE failed to accept declarations in handler clauses
944 in some circumstances.
946 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
947 * New code in contrib/sb-introspect (still probably not entirely
948 stable yet) provides some support for smart Lisp development
949 environments like SLIME.
950 * The conditions signalled for errors occurring when loading .fasl
951 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
952 in a way which should help ASDF recover gracefully.
953 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
954 stuff has been cleaned up. If you code contrib/ stuff, this might
955 affect you, and you can look at contrib/README, contrib/STANDARDS,
956 and/or the 0.8.4.27 diff to check.
957 * In full calls the compiler now does not generate checks for declared
958 argument types for all arguments.
959 * various threading fixes
960 ** and some experimental patches which didn't make it into
961 the main tree for this release, but which are shipped in
962 contrib/experimental-thread.patch as a possible fix for some
963 failures (deadlock, spinning...) in GC-intensive multithreaded
965 * fixed PPC build problem (source code incompatibility of different
966 library versions): added offsetof() hackery which attempts to divine
967 where glibc maintainers put uc_mcontext today
968 * fixed bug 282: compiler does not trust type assertions while passing
969 arguments to a full call.
970 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
971 &OPTIONAL VALUES type parameter.
972 * fix bug 214: algorithm for noting rejected templates is now more
973 similar to that of template seletion. (also reported by rydis on
975 * fixed bug 141b: printing backquoted information readably and prettily
976 inserts a space where necessary.
977 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
978 (CHARACTER 10) are now reported as errors, rather than propagated
979 as unknown types. (reported by piso on #lisp)
980 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
981 with values NIL and :ERROR. (thanks to Milan Zamazal)
982 * fixed bug 191c: CLOS now does proper keyword argument checking as
983 described in CLHS 7.6.5 and 7.6.5.1.
984 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
985 longer raise an error; further, using a list as a for-as-arithmetic
986 counter now raises a meaningful error.
987 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
988 now understood by sequence creation functions such as MAKE-SEQUENCE
990 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
991 a STRING-INPUT-STREAM.
992 * compiler enhancement: SIGNUM is now better able to derive the type
994 * type declarations inside WITH-SLOTS are checked. (reported by
996 * fixed some bugs revealed by Paul Dietz' test suite:
997 ** incorrect optimization of TRUNCATE for a positive first
998 argument and negative second.
999 ** compiler failure in let-convertion during flushing dead code.
1000 ** compiler failure while deriving type of TRUNCATE on an
1001 interval, containing 0.
1002 ** ASH of a negative bignum by a negative bignum count now returns
1004 ** intersection of CONS types now canonicalizes properly, fixing
1005 inconsistencies in SUBTYPEP.
1007 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1008 * incompatible change: The --disable-debugger command line
1009 option now clobbers the debugger at a more fundamental
1010 level, by redefining #'INVOKE-DEBUGGER instead of by
1011 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1012 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1013 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1014 enter the debugger REPL and then suffer recursive errors
1015 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1016 while under the new system, BREAK in --disable-debugger mode
1017 terminates the system just as an unhandled error would.
1018 * fixed compiler performance when processing loops with a step >1;
1019 * bug fix: DOCUMENTATION now retrieves generic function
1020 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1021 support has been systematized, and now supports the methods
1022 specified by ANSI, along with a default method and a method for
1023 slot documentation. (reported by Nathan Froyd)
1024 * bug fix: effective methods associated with a generic function are
1025 no longer cached over a change of that generic function's method
1026 combination. (reported by Andreas Fuchs)
1027 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1028 names a non-existent directory. (thanks to Andreas Fuchs)
1029 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1030 the PPC platform, lead to stack corruption; this has been fixed.
1031 (reported by Rainer Joswig)
1032 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1033 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1034 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1035 arithmetic in general, is now much more reliable on the PPC
1037 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1038 platform now returns the right answer.
1039 * optimization: restored some effective method precomputation in
1040 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1041 precomputation is now tunable.
1042 * optimization: compiler-internal data structure use has been
1043 reviewed, and changes have been made that should improve the
1044 performance of the compiler by about 20%.
1045 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1046 simple-base-strings and simple-bit-vectors is improved.
1047 * optimization: the optimization of 32-bit logical and arithmetic
1048 functions introduced in version 0.8.3 on the x86 has been
1049 implemented on the mips, ppc and sparc platforms; an
1050 implementation of the same facility, but for 64-bit arithmetic,
1051 has been added for the alpha.
1052 * microoptimization: the compiler is better able to make use of the
1053 x86 LEA instruction for multiplication by constants.
1054 * bug fix: in some situations compiler did not report usage of
1055 generic arithmetic in (SPEED 3) policy.
1056 * bug 145b fix: compiler used wrong type specifier while converting
1057 MEMBER-types to numeric.
1058 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1060 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1061 index argument now works properly on the Alpha platform.
1062 * bug fix: floating point exception treatment on the Alpha platform
1064 * bug fix: FILE-POSITION works much better on string input and
1065 output streams. (thanks to Nikodemus Siivola)
1066 * bug fix: many threading/garbage collection symptoms sorted.
1067 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1068 might be pseudo-atomic.
1069 * internal change: Stopping for GC is now done with signals not
1070 ptrace. GC is now done in whichever thread wanted it, instead of
1072 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1074 * bug fix: result form in DO is not contained in the implicit
1076 * incompatible change: ICR structure is changed; the value part of
1077 CONTINUATION is now called LVAR; corresponding functions are
1078 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1079 * added type deriver for ISQRT (thanks to Robert E. Brown).
1080 * bug fix: better support for loading from the command line when an
1081 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1083 * fixed some bugs revealed by Paul Dietz' test suite:
1084 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1085 ** ROUND and FROUND now give the right answer when given very
1086 small float arguments.
1087 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1089 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1090 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1091 ** LCM should return a non-negative integer.
1092 ** PARSE-INTEGER returned the index of a terminator instead of the
1093 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1094 ** PARSE-INTEGER returned an incorrect index being applied to a
1096 ** LCM with two arguments of 0 returns 0 rather than signalling
1098 ** unsigned addition of a 32-bit constant with the high bit set no
1099 longer causes an internal compiler error on the x86.
1100 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1101 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1102 with negative last argument.
1103 ** byte specifiers with zero size and position no longer cause
1104 an error during type derivation.
1105 ** bignum multiplication on the Alpha platform now returns the
1107 * porting: The system now builds on SuSE AMD64, although it still
1108 generates a 32-bit binary.
1109 * .fasl file incompatibility: The fasl file version number has
1110 been incremented (because of the changes to internal compiler
1111 data structures referred to above).
1113 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1114 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1115 more accurately, on the Darwin kernel running on PowerPC hardware.
1116 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1117 * Compiler code deletion notes now signal a condition of type
1118 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1119 an associated MUFFLE-WARNING restart.
1120 * The compiler now performs limited argument count validation of
1121 constant format strings in FORMAT, and where appropriate in ERROR,
1122 CERROR and WARN. (thanks to Gerd Moellmann)
1123 * New ASDF-INSTALL contrib can be used for automatic download and
1124 installation of third-party Lisp code from CCLAN or other sites
1126 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1127 which forces another thread to execute a function supplied by the
1129 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1130 now accept and act upon their :ELEMENT-TYPE keyword argument.
1131 (reported by Martin Atzmueller, Edi Weitz)
1132 * bug fix: FILE-POSITION now accepts position designators up to
1133 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1134 is the greater. (thanks to Patrik Nordebo)
1135 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1136 Kalle Olavi Niemitalo)
1137 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1138 printable. (reported by Eric Marsden)
1139 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1140 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1141 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1142 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1143 (thanks to Stig Sandoe)
1144 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1145 UNREAD-CHAR now correctly includes the unread character in the
1146 target sequence. (thanks to Gerd Moellmann)
1147 * bug fix in threaded builds: the system can now be suspended and
1148 resumed by shell job control with minimal disruption.
1149 * bug fixes in times and timezones >2038AD
1150 * better handling of "where is GNU make?" problem in build scripts
1151 (thanks to Nikodemus Siivola)
1152 * new optimization: inside a named function any reference to a
1153 function with the same name is considered to be a self-reference;
1154 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1155 optimization quality.
1156 * new optimization on x86: logical functions and + now have
1157 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1158 used when the result is truncated to 32 bits.
1159 * VALUES declaration is partially enabled.
1160 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1161 * fixed some bugs revealed by Paul Dietz' test suite:
1162 ** The system now obeys the constraint imposed by
1163 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1164 types form a lattice under type intersection.
1165 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1166 ** ASSOC now ignores NIL elements in an alist.
1167 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1168 and (1+ MOST-POSITIVE-FIXNUM) answers.
1169 ** The addition of a method with invalid qualifiers to a generic
1170 function does not cause an error to be signalled immediately;
1171 a warning is signalled, and the error is generated only on
1172 calling the generic function.
1173 * changed .fasl file version number, in part to add type codes for
1174 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1175 obscure ANSI requirements
1177 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1178 * fixed bug 148: failure to inline-expand a local function left
1179 garbage, confusing the compiler.
1180 * fixed bugs 3cd: structure slot readers perform type check if the
1181 slot can have an invalid value (i.e. it is either not initialized
1182 or can be written with a less specific slot writer).
1183 * bug fix: the compiler now traps array references to elements off
1184 the end of an array; previously, the bounds checking in some
1185 circumstances could go off-by-one.
1186 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1188 * type declarations for array element types now obey the description
1189 on the CLHS page "Declaration TYPE", as per discussions on
1190 sbcl-help around 2003-05-08. This means that a declaration
1191 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1192 declaration, all references to BAR will be asserted or assumed
1193 (with THE, so dependent on compiler policy) to involve objects of
1194 type FOO. Note that no such declaration is implied in
1195 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1196 * declared types of functions from the "Conditions"
1197 chapter. (reported by Paul Dietz)
1198 * bug fix: CERROR accepts a function as its first argument.
1199 * bug fix: NTH an NTHCDR accept a bignum as index
1200 arguments. (reported by Adam Warner)
1201 * optimization: character compare routines now optimize comparing
1202 against a constant character. (reported by Gilbert Baumann)
1203 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1204 argument now works properly on the MIPS platform.
1205 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1206 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1207 declared to have a functional type, does not cause an error
1209 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1210 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1211 * fixed bug: if last continuation of a deleted block has a
1212 destination, this destination should be deleted too. (reported by
1214 * fixed a bug in the bootstrap process: the host compiler's values
1215 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1216 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1217 test case from Patrik Nordebo)
1218 * improved the ability of the disassembler on the PPC platform to
1219 provide helpful disassembly notes.
1220 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1221 the class in more cases than previously.
1222 * bug fix: FILE-POSITION now understands :START and :END for
1223 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1224 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1225 8) data. (thanks to David Lichteblau)
1226 * bug fix: it is possible to add a method to a generic function
1227 without lambda list.
1228 * bug fix: reader failed to signal END-OF-FILE inside an
1229 object representation. (reported by Nikodemus Siivola)
1230 * fixed some bugs revealed by Paul Dietz' test suite:
1231 ** LAST and [N]BUTLAST should accept a bignum.
1232 ** condition slot accessors are methods.
1233 ** (VECTOR NIL) is a subtype of STRING.
1235 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1236 * minor incompatible change: some nonsensical specialized lambda
1237 lists (used in DEFMETHOD) which were previously ignored now signal
1239 * minor incompatible change: the system is now aware of the types of
1240 variables in the COMMON-LISP package, and will signal errors for
1241 most violations of these type constraints (where previously they
1242 were silently accepted).
1243 * minor incompatible change: COMPILE-FILE now uses the freedom
1244 afforded (ANSI 3.2.2.3) to use derived function types for
1245 functions defined in the same file. This also permits the system
1246 to warn on static type mismatches and function
1247 redefinition. (Currently it does not work with high DEBUG level.)
1248 * minor incompatible change: VALUES declaration is disabled.
1249 * When issuing notes, the compiler now signals a condition of type
1250 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1251 restart for use in user handlers. It is expected that the
1252 COMPILER-NOTE condition will eventually become a condition
1253 supertype to a hierarchy of note types, which will then be
1254 handleable in a similar fashion. However, at the moment, no such
1255 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1256 but it's an implementation detail, not a classification for the
1258 * Changes in type checking closed the following bugs:
1259 ** type checking of unused values (192b, 194d, 203);
1260 ** template selection based on unsafe type assertions (192c, 236);
1261 ** type checking in branches (194bc).
1262 * A short form of VALUES type specifier has ANSI meaning (it has
1263 increased the number of situations when SBCL cannot perform type
1265 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1266 keywords or constants is permissible.
1267 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1268 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1269 classes. (thanks to Antonio Martinez)
1270 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1271 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1272 argument to a pretty-print stream. Instead, it leaves any such
1273 operations to DESCRIBE-OBJECT methods.
1274 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1276 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1277 lambda lists. (thanks to David Lichteblau)
1278 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1279 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1280 (thanks to Gerd Moellmann)
1281 * fixed bug 246: increased compilation speed of long
1282 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1284 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1286 * DEFCONSTANT now throws a condition of type
1287 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1288 constant to a non-EQL value; CONTINUE and ABORT restarts
1289 respectively change and preserve the value.
1290 * fixed bug 63: The code walker, part of the implementation of CLOS,
1291 is now better at handling symbol macros.
1292 * bug fix: There is no longer an internal implementation type named
1293 CL:LENGTH. (reported by Raymond Toy)
1294 * bug fix: In macro-like defining macros/special operators the
1295 implicit block does not enclose the lambda list.
1296 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1297 enclosed in parentheses) are not suitable as type specifiers, and
1298 their use properly signals an error now.
1299 * bug fix: An argument count mismatch for a type specifier in code
1300 being compiled no longer causes an unhandled error at compile
1301 time, but signals a compile-time warning.
1302 * fixed simple vector readable printing
1303 * bug fix: DESCRIBE takes more care over whether the class
1304 precedence list slot of a class is bound before accessing it.
1305 (reported by Markus Krummenacker)
1306 * bug fix: FORMATTER can successfully compile pretty-printer format
1307 strings which use variants of the ~* directive inside.
1308 * bug fix: SEARCH now applies its TEST predicate to the elements of
1309 the arguments in the correct order. (thanks to Wolfhard Buss)
1310 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1311 in type inference. (thanks to Robert E. Brown)
1312 * bug fix: Reading in symbols with an explicit package name of ""
1313 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1314 * fixed some bugs revealed by Paul Dietz' test suite:
1315 ** NIL is now allowed as a structure slot name.
1316 ** Arbitrary numbers, not just REALs, are allowed in certain
1317 circumstances in LOOP for-as-arithmetic clauses.
1318 ** Multiple class redefinitions before slot access no longer
1319 causes a type error.
1320 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1321 association between the name and a class.
1322 ** Generic functions with non-standard method-combination and over
1323 five methods all of which return constants no longer return NIL
1324 after the first few invocations. (thanks to Gerd Moellmann)
1325 ** CALL-NEXT-METHOD with no arguments now passes the original
1326 values of the arguments, even in the presence of assignment.
1327 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1329 ** DISASSEMBLE works with closures and funcallable instances.
1330 ** ADD-METHOD now returns the generic function, not the new method.
1331 ** FIND-METHOD signals an error if the lengths of the specializers
1332 is incompatible with the generic function, even if the ERRORP
1334 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1335 which its argument is a member.
1336 ** DEFCLASS only redefines the class named by its class-name
1337 argument if that name is the proper name of the class;
1338 otherwise, it creates a new class.
1339 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1340 of the UNBOUND-SLOT condition to the name of the slot.
1341 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1342 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1343 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1344 SLOT-MAKUNBOUND in the specified fashion.
1346 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1347 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1348 cross-compilation host. As a consequence, we can now bootstrap our
1349 way up to SBCL starting with a bare gcc toolchain and human-readable
1350 source code (first the source to CLISP, then the source to SBCL).
1351 * A contributed module containing a partial implementation of the
1352 simple-streams interface has been included. (thanks to Rudi
1354 * A contributed module implementing the RFC1321 Message Digest
1355 Algorithm, known as MD5, has been included.
1356 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1357 no longer has any effect, as the code controlled by this feature
1358 has been deleted. (As far as we know, no-one has ever built using
1359 this feature, and its semantics were confused in any case).
1360 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1361 work on conditions (as required by the ANSI specification),
1362 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1363 expected behaviour on conditions. Users should note, however,
1364 that such behaviour is not required by the ANSI specification,
1365 and so use of this behaviour may render their code unportable.
1366 * fixed some bugs revealed by Paul Dietz' test suite:
1367 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1369 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1370 ** SXHASH on condition objects no longer returns NIL.
1371 ** :ALLOCATION :CLASS slots are better treated; their values are
1372 updated on class redefinition, and initforms inherited from
1373 superclasses are applied.
1374 ** REMOVE-METHOD returns its generic function argument even when
1375 no method was removed.
1376 ** SHARED-INITIALIZE now initializes the values of the requested
1377 slots, including those with :ALLOCATION :CLASS.
1378 ** ALLOCATE-INSTANCE now works on structure classes defined via
1379 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1381 ** SLOT-EXISTS-P now works on conditions, as well as structures
1383 ** MAKE-LOAD-FORM now has the required methods on
1384 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1385 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1386 keyword, and now implements the SLOT-NAMES argument.
1387 ** methods with &OPTIONAL arguments no longer allow too many
1388 arguments to be passed in the call without error.
1389 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1390 option is consistent with the required arguments of the generic
1391 function lambda list.
1392 * bug fix: REQUIRE accepts a string designator. (Thanks to
1394 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1395 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1396 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1397 * bug fix: The compiler now checks for duplicated variables in macro
1399 * bug fix: SETQ on globals returns the correct value.
1400 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1401 not a COMPILER-ERROR (followed by some other strange error on
1402 choosing the CONTINUE restart).
1403 * bug fix: make.sh and friends are now more consistent in the way that
1404 they look for GNU "make".
1406 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1407 * experimental native threads support (on x86 Linux >=2.4 only).
1408 This is not compiled in by default: you need to add :SB-THREAD to
1409 the target features. See the "Beyond ANSI" chapter of the manual
1411 * fix for longstanding nonANSIism: The old distinction between
1412 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1413 The return value from CL:FIND-CLASS is now a CLOS class, and
1414 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1415 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1416 * An interface to the MetaObject Protocol, as described in Kiczales,
1417 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1418 MIT Press, 1991, is available from the SB-MOP package.
1419 * incompatible change: the SB-PCL package should now be considered
1420 a private implementation detail, and no longer a semi-private MOP
1422 * minor incompatible change: due to rearrangement for threads, the
1423 control stack and binding stack are now allocated at arbitrary
1424 addresses instead of being hardcoded per-port. Users affected by
1425 this probably have to be doing advanced things with shared
1426 libraries, and will know who they are.
1427 * minor incompatible change: Previously, all --eval forms used were
1428 processed with READ before any of them were processed with EVAL.
1429 Now each --eval form is processed with both READ and EVAL before
1430 the next --eval form is processed. (Thus package operations like
1431 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1432 work as the user might reasonably expect.)
1433 * minor incompatible change: *STANDARD-INPUT* is now only an
1434 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1436 * minor incompatible change: Y-OR-N-P is now character-oriented,
1437 not line oriented. Also, YES-OR-NO-P now works without errors.
1438 (thanks to Antonio Martinez)
1439 * sb-aclrepl module improvements: an integrated inspector, added
1440 repl features, and a bug fix to :trace command.
1441 * Known functions, which cannot be open coded by the backend, are
1442 considered to be able to check types of their arguments. (fixing
1443 a bug report by Nathan J. Froyd)
1444 * fixed a bug in computing method discriminating functions: It is
1445 now possible to define methods specialized on classes which have
1446 forward-referenced superclasses. (thanks to Gerd Moellmann)
1447 * fixed evaluation order in optional entries (reported by Gilbert
1449 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1450 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1451 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1452 required-by-AMOP NAME argument, as well as CLASS and
1453 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1454 * fixed bug 20: DEFMETHOD can define methods using names that are
1455 not the proper names of classes to designate class specializers.
1456 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1457 Unix file descriptors, instead of blowing up. (thanks to Antonio
1459 * Garbage collection refactoring: user-visible change is that a
1460 call to the GC function during WITHOUT-GCING will not do garbage
1461 collection until the end of the WITHOUT-GCING. If you were doing
1462 this you were probably losing anyway.
1463 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1464 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1465 TYPEP the latter but not the former.
1466 * The compiler issues a full WARNING for calls to undefined functions
1467 with names from the CL package.
1468 * MAP-INTO for a vector destination is open coded. (reported by
1469 Brian Downing on c.l.l)
1470 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1471 documentation string.
1472 * fixed some bugs revealed by Paul Dietz' test suite:
1473 ** COPY-ALIST now signals an error if its argument is a dotted
1475 ** Condition slots are now accessed more correctly in the presence
1476 of multiple initargs for a given slot.
1477 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1478 exclude restarts of the same name associated with a different
1480 ** DEFCLASS of forward-referenced classes with another
1481 forward-referenced class in the superclasses list no longer
1483 ** Condition slots are now initialized once each, not multiple
1484 times. (thanks to Gerd Moellmann)
1485 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1487 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1488 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1489 arguments contain duplicated elements.
1490 ** RESTART-CASE understands local macros.
1491 ** RESTART-CASE associates exactly its own restarts with a condition.
1492 ** ENDP in safe mode checks its argument to be of type LIST.
1493 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1494 in question is unbound.
1495 ** Optimized MAKE-INSTANCE functions no longer cause internal
1496 assertion failures in the presence of duplicate initargs.
1497 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1498 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1499 propagated even in the ctor optimized implementation of
1501 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1503 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1504 * a better implementation of SXHASH on (simple) bit vectors,
1505 measured both in execution speed and in distribution of results
1506 over the positive fixnums, has been installed. Likewise, a better
1507 implementation of EQUAL for simple bit vectors is now available.
1508 * fixed CEILING optimization for a divisor of form 2^k.
1509 * fixed bug 240 (emitting extra style warnings "using the lexical
1510 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1511 by Antonio Martinez)
1512 * fixed SXHASH, giving different results for NIL depending on type
1513 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1514 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1515 variable. (found by Rolf Wester)
1516 * SBCL does not ignore type declarations for special
1517 variables. (reported by rif on c.l.l 2003-03-05)
1518 * some bug fixes in contrib/sb-aclrepl/
1519 * fixed some bugs revealed by Paul Dietz' test suite:
1520 ** a bug in the CONS type specifier, whereby the CAR and CDR
1521 types got intertwined, has been fixed;
1522 ** the type system is now able to reason about the interaction
1523 between INTEGER and RATIO types more completely;
1524 ** APPEND, [N]REVERSE and NRECONC check that those their
1525 arguments, which must be proper lists, are really so;
1526 ** An array specialized to be unable to hold elements has been
1527 implemented, as required -- yes, really -- by ANSI;
1528 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1529 on malformed property lists;
1531 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1532 * incompatible packaging change: in line with Unix convention,
1533 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1534 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1535 * REQUIRE and PROVIDE are now optionally capable of doing something
1536 useful. See the documentation string for REQUIRE.
1537 * infrastructure for a managed SBCL contrib system: contributed
1538 modules in this release include:
1539 ** the ASDF system definition facility;
1540 ** an interface to the BSD Sockets API;
1541 ** an ACL-like convenience interface to the repl;
1542 (thanks to Kevin Rosenberg)
1543 ** an implementation of ROTATE-BYTE, with efficient implementation
1545 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1546 gives a closer approximation to the right answer than previously.
1547 (thanks to Raymond Toy)
1548 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1549 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1550 optional environment arguments, as required by ANSI.
1551 * fixed bugs in other functions taking environment objects, allowing
1552 calls with an explicit NIL environment argument to be compiled
1554 * fixed bug 228: primary return values from
1555 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1556 COMPILE or FUNCTION.
1557 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1558 :TYPE will no longer signal errors on innocuous objects.
1559 * fixed bug 231b: SETQ is better at respecting type declarations in
1560 the lexical environment.
1561 * fixed a bug in DEFCLASS: classes named by symbols with no or
1562 unprintable packages can now be defined.
1563 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1564 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1565 * fixed compiler failure related to checking types of functions.
1566 (reported by Robert E. Brown)
1567 * the compiler is now much more consistent in its error-checking
1568 treatment of bounding index arguments to sequence functions: in
1569 (SAFETY 3) code, errors will be signalled in almost all cases if
1570 invalid sequence bounding indices are passed to functions defined
1571 by ANSI to operate on sequences.
1572 * fixed a bug in the build procedure: documentation of SBCL-specific
1573 packages is now preserved and available in the final Lisp image.
1574 * lifted FDEFINITION lookup out of loops in the implementation of
1575 many list operations. (thanks to Robert E. Brown)
1576 * fixed a bug in the reader: the #n# reader macro now works for
1577 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1578 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1579 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1580 E. Brown and Tony Martinez respectively)
1581 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1582 SLOT-BOUNDP: in particular, it is now able to optimize them much
1583 better, and is now not vulnerable to having packages renamed.
1584 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1585 always called when a slot is not present in an instance. (thanks
1587 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1588 longer signal an error on unreading a character following EOF on
1589 the previous constituent stream. (thanks to Tony Martinez)
1590 * fixed some bugs revealed by Paul Dietz' test suite:
1591 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1592 not just nonnegative fixnums;
1593 ** the logical bit-array operators such as BIT-AND now accept an
1594 explicit NIL for their "opt-arg" argument (to indicate a
1595 freshly-consed result bit-array);
1596 ** ELT now signals an error on an invalid sequence index in safe
1598 ** the type system is now cleverer about negations of numeric
1599 types, and consequently understands the BIGNUM and RATIO types
1601 ** the type system is now cleverer about the interaction between
1602 INTEGER and RATIO types: while bugs still remain, many more
1603 cases are accurately computed;
1604 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1605 if it is in the last clause;
1606 ** CONSTANTLY now correctly returns a side-effect-free function in
1608 ** DECLARE is no longer treated as a special-operator; in
1609 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1610 * incremented fasl file version number due to the change in the
1611 DEFSTRUCT-SLOT-DESCRIPTION structure.
1613 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1614 * minor incompatible change: code processed by the "interpreter" or
1615 EVAL now has a compilation optimization policy of (DEBUG 2)
1616 (changed from (DEBUG 1)) to improve debuggability of interactive
1617 development, and to allow the use of the debug RETURN command in
1619 * an experimental implementation of the RETURN command for the
1620 debugger has been included. (thanks to Frederik Kuivinen)
1621 * fixed bug 62: constraints were not propagated into a loop.
1622 * fixed bug in embedded calls of SORT (reported and investigated by
1624 * fixed some bugs revealed by Paul Dietz' test suite:
1625 ** printing and reading of arrays with some dimensions having
1626 length 0 (thanks to Gerd Moellmann);
1627 ** BOA constructor with &AUX argument without a default value does
1628 not cause a type error;
1629 ** CONSTANTP now returns true for all self-evaluating objects.
1631 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1632 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1633 accessors that are related by inheritance, as specified in the
1634 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1635 Valtteri Vuorikoski)
1636 * The compiler is now able to inline functions that were defined in
1637 a complex lexical environment (e.g. inside a MACROLET).
1638 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1640 * Support for the upcoming FreeBSD-5.0 release has been included.
1641 (thanks to Dag-Erling Smorgrav)
1642 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1643 effect when it is not in a toplevel context.
1644 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1645 stand a better chance of being correct. (thanks to Gerd
1647 * fixed bug in COERCE, which now signals an error on coercing a
1648 rational to a bounded real type which excludes the expected
1650 * The compiler is now able to derive types more accurately from the
1651 COERCE and COMPILE functions.
1652 * fixed bug 223: functional binding is considered to be constant
1653 only for symbols in the CL package.
1654 * fixed bug 231: SETQ did not check the type of a variable being set
1655 (reported by Robert E. Brown)
1656 * A new optimization for MAKE-INSTANCE has been included, fixing
1657 various bugs (including relating to :ALLOCATION :CLASS slots and
1658 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1660 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1661 ** As required by ANSI, LOOP now disallows anonymous collection
1662 clauses such as COLLECT I in conjunction with aggregate boolean
1663 clauses such as THEREIS (= I 1);
1664 ** LOOP now signals an error when any variable is reused in the
1665 same loop (including the potentially useful construct analogous
1666 to WITH A = 1 WITH A = (1+ A);
1667 ** IT is only a special loop symbol within the first clause of a
1668 conditional loop clause;
1669 ** LOOP with a typed iteration variable over a hashtable now
1670 signals a type error iff it should.
1671 * fixed some other bugs revealed by Paul Dietz' test suite:
1672 ** FILE-STREAM now names the class previously known as FD-STREAM;
1673 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1674 argument) no longer signals an error;
1675 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1676 of the slot symbol, rather than using the current package
1677 ((:CONC-NAME "") continues to intern the slot's name in the
1679 * incremented fasl file version number, because of the incompatible
1680 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1681 of the new implementation of DEFINE-COMPILER-MACRO.
1683 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1684 * Support for building SBCL for MIPS platforms running in
1685 little-endian mode has now been checked in, and basic
1686 functionality on said platforms verified.
1687 * minor incompatible change: PCL now records the pathname of a file
1688 in which methods and the like are defined, rather than its
1690 * minor incompatible change: TRUENAME now considers the truename of
1691 a file naming a directory to be the pathname with :DIRECTORY
1692 component indicating that directory.
1693 * minor incompatible change: a NAMED clause in the extended form of
1694 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1695 reason for the previous behaviour is unclear.
1696 * more systematization and improvement of CLOS and MOP conformance
1697 in PCL (thanks to Gerd Moellman and Pierre Mai):
1698 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1700 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1701 primary methods with no specializers;
1702 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1704 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1705 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1706 FORWARD-REFERENCED-CLASSes; error reporting on
1707 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1709 ** SXHASH on CLOS instances now uses a slot internal to the
1710 instance to return different numbers on distinct instances,
1711 while preserving the same return value through invocations of
1713 ** DEFMETHOD signals errors when methods with longer incongruent
1714 lambda lists are added to generic functions;
1715 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1716 CLASS, as specified in AMOP;
1717 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1718 on the order returned by the primary method for classes of
1719 class STANDARD-CLASS;
1720 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1721 * fixed some bugs shown by Paul Dietz' test suite:
1722 ** DOLIST puts its body in TAGBODY;
1723 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1725 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1726 value producing form;
1727 ** if more variables are given to PROGV than values, extra
1728 variables are bound and made to have no value;
1729 ** NSUBSTITUTE on list arguments gets the right answer with
1731 ** ELT signals an error of type TYPE-ERROR when the index argument
1732 is not a valid sequence index;
1733 ** LOOP signals (at macroexpansion time) an error of type
1734 PROGRAM-ERROR when duplicate variable names are found;
1735 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1736 ** FUNCALL of special-operators now cause an error of type
1738 ** PSETQ now works as required in the presence of side-effecting
1739 symbol-macro places;
1740 ** NCONC accepts any object as its last argument;
1741 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1743 ** loop-for-as-package does not require a package to be explicitely
1745 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1747 * fixed bug 166: compiler preserves "there is a way to go"
1748 invariant when deleting code.
1749 * fixed bug 172: macro lambda lists with required arguments after
1750 &REST arguments now cause an error to be signalled. (thanks to
1752 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1753 bug. (thanks to Matthew Danish)
1754 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1756 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1757 arguments are lexically rebound. (thanks to Gerd Moellmann and
1759 * fixed bug 194: error messages are now more informative when there
1760 is no primary method applicable in a call to a generic
1761 function. (thanks to Gerd Moellmann)
1762 * fixed bug in command line argument checking (thanks to Julian
1764 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1765 (thanks to Matthew Danish)
1766 * incremented fasl file version number, because of the
1767 SXHASH-related changes in the layout of CLOS data structures
1769 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1770 * minor incompatible change: The runtime (the Unix executable named
1771 "sbcl") is now much pickier about the .core files it will load.
1772 Essentially it now requires .core files to descend from the same
1773 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1774 as the runtime does. (The intent is to prevent the crashes which
1775 can occur, and which can even be reported as mysterious failures,
1776 when people patch the sources or change the build parameters
1777 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1778 sbcl and .core files.)
1779 * fixed bug: VALUES-LIST is no longer optimized away.
1780 * fixed bug 142: The FFI conversion of C string values to Lisp
1781 string values no longer conses excessively. (thanks to Nathan
1782 Froyd porting Raymond Toy's fix to CMU CL)
1783 * began to systematize and improve MOP conformance in PCL (thanks to
1784 Nathan Froyd, Gerd Moellman and Pierre Mai):
1785 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1787 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1788 ** FINALIZE-INHERITANCE is now called on class finalization;
1789 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1790 argument precedence order.
1791 * fixed bug 202: The compiler no longer fails on functions whose
1792 derived types contradict their declared type.
1793 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1794 so it can be non-toplevel.
1795 * The fasl file version number has changed (because of the new
1796 implementation of DEFMACRO).
1797 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1798 safe code, that any length requirement by their type-specifier
1799 argument is valid. The exceptions to this are described in bug
1801 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1802 functions now check the types of their inputs as required by ANSI.
1803 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1804 introduced symbol is DECLAREd to be SPECIAL.
1805 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1806 * fixed bug: :COUNT argument to sequence functions may be negative
1807 * fixed bug: body of DO-SYMBOLS may contain declarations
1808 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1809 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1810 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1811 symbol macro only once
1812 * fixed printing of call frame when argument list is unavailable
1813 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1814 * compiler no longer signals WARNING on unknown keyword
1817 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1818 * A beta-quality port to the mips architecture running Linux,
1819 based on the old CMUCL backend, has been made. It has been tested
1820 on a big-endian kernel, and works sufficiently well to be able to
1821 rebuild itself; it has not been tested in little-endian mode.
1822 * fixed an inconsistency between gencgc.c and purify.c which made
1823 dumping/loading .core files unreliable
1824 * fixed bug 120a: The compiler now deals correctly with IFs where
1825 the consequent is the same as the alternative, instead of
1826 misderiving the return type. (thanks to Alexey Dejneka)
1827 * fixed bug 113: Logical pathnames are now dumpable (the logical
1828 host is resolved at load-time, throwing an error if it is not
1830 * fixed bug 174: FORMAT's error message is slightly clearer when a
1831 non-printing character is used in a format directive.
1832 * fixed several bugs in compiler checking of type declarations, i.e.
1833 violations of the Python "declarations are assertions" principle
1834 (thanks to Alexey Dejneka)
1835 * fixed several bugs in PCL's error checking (thanks to Gerd
1837 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1839 * fixed bug in compilation of functions as first class values
1840 (thanks to Antonio Martinez-Shotton)
1841 * The compiler's handling TYPE-ERRORs which it can prove will
1842 inevitably happen at runtime has been cleaned up and corrected
1843 in several ways. (thanks to Alexey Dejneka)
1844 * improved argument type checking for various basic arithmetic
1845 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1846 lavished on them in the past that they can be compiled in many
1847 ways in different special cases
1848 * fixed bug 181: compiler checks validity of user supplied type
1850 * cleaned up code flushing in optimization: Function calls which
1851 should signal errors for safety purposes (e.g. which ANSI says
1852 should signal errors when their arguments are of incorrect type)
1853 are no longer optimized away.
1854 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1855 * incremented fasl file version number, because changes in the
1856 implementation of sequence functions like COERCE caused
1857 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1858 in old inline expansions) to become undefined. (Actually these
1859 changes were later undone, so we might very well be binary
1860 compatible with 0.7.7 after all, but leaving the version number
1861 incremented seemed like the simplest and most conservative
1864 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1865 * An alpha-quality port to the parisc architecture running Linux,
1866 based on the old CMUCL backend, has been made. This, even more so
1867 than the other backends, should be considered still a work in
1868 progress; known problems include that the Linux kernel in 64-bit
1869 mode does not propagate the correct sigcontext structure to
1870 userspace, and consequently SBCL on a parisc64 kernel will not
1872 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1873 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1874 suppressed.) Also now that the compiler is looking at declarations
1875 in the environment, it checks optimization declarations as well,
1876 and suppresses inlining when (> DEBUG SPEED).
1877 * More fixes have been made to treatment of floating point exception
1878 treatment and other Unix signals. In particular, floating point
1879 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1880 * The detection and handling of control stack exhaustion (infinite
1881 or very deeply nested recursion) has changed. Stack exhaustion
1882 detection is now done by write-protecting pages at the OS level
1883 and applies at all optimization settings; when found, a
1884 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1885 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1886 be caught using IGNORE-ERRORS.
1887 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1888 correctly when transforming recursive LABELS functions to LETs.
1889 (thanks to Alexey Dejneka)
1890 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1891 that are names of constants or global variables.
1892 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1893 alien routines with docstrings.
1894 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1895 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1897 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1898 (thanks to Eric Marsden)
1899 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1900 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1901 * Bug fix: The SPARC backend can now compile functions involving
1902 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1903 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1904 to a functions expecting a PATHNAME-DESIGNATOR.
1905 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1906 lambda lists. (thanks to Alexey Dejneka)
1907 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1908 functions so that EXPORTing the name of the class doesn't cause
1909 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1910 to Antonio Martinez for reporting this)
1911 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1912 constant-folded without failing an assertion. (thanks to Einar
1913 Floystad Dorum for reporting this)
1914 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1915 types are now dealt with more correctly by the compiler.
1916 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1917 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1918 behaviour analogous to RENAME-FILE. This puts its behaviour more
1919 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1921 * The fasl file version number has changed again. (because of the
1922 bug fix involving the names of PCL MAKE-INSTANCE functions)
1924 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1925 * bug fix: Floating point exceptions are treated much more
1926 consistently on the x86/Linux and PPC/Linux platforms.
1927 * Array initialization with :INITIAL-ELEMENT is now much faster for
1928 cases when the compiler cannot open code the array creation, but
1929 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1930 array accesses have also seen a speed increase.
1931 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1932 specified. (This was at the root of some bad interactions between
1933 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1935 * bug fix: Internal error arguments for undefined functions are now
1936 computed correctly on the PPC/Linux platform.
1937 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1938 Raymond Toy's patch for CMU CL)
1939 * Support for the Solaris 9 operating environment has been included
1940 (thanks to Daniel Merritt)
1941 * A very ugly but hopefully complete draft of the missing FFI chapter
1942 of the manual has been created by reformatting the corresponding
1943 CMU CL manual chapter into (currently very ugly and incoherent)
1944 DocBook and bringing it up to date for SBCL behavior. Thus, the
1945 manual is now essentially complete, at least by my extreme
1946 once-and-only-once standards, whereby it's acceptable to refer to
1947 the doc strings of SB-EXT functions as the primary documentation.
1948 * The fasl file version number has changed again, due to cleanup of
1949 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1950 is no longer a static symbol.)
1952 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1953 * SBCL now builds with OpenMCL (version 0.12) as the
1954 cross-compilation host; also, more progress has been made toward
1955 bootstrapping under CLISP.
1956 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1958 * bug 158 fixed: The compiler can now deal with integer loop
1959 increments different from 1; fixing this turned out also to fix
1961 * bug 169 fixed: no more bogus warnings about using lexical bindings
1962 despite the presence of perfectly good SPECIAL declarations (thanks
1963 to David Lichteblau)
1964 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1965 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1966 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1968 * bug 180 fixed: Method combination specifications no longer ignore
1969 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1970 * bug fix: Structure type predicate functions now check their argument
1971 count as they should.
1972 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1973 correctly. (thanks to Pierre Mai)
1974 * minor incompatible change: The --noprogrammer option is deprecated
1975 in favor of the new --disable-debugger option, which is very similar.
1976 (The major difference is that it takes effect at a slightly different
1977 time at startup, causing handling of errors in --sysinit and
1978 --userinit files will be affected differently.) The
1979 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1980 been added to allow this functionality to be controlled from ordinary
1981 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1982 maintainers, who might want to run non-interactive scripts to
1983 build SBCL cores which will later be used interactively.)
1984 * minor incompatible change: The LOAD function no longer, when given
1985 a wild pathname to load, loads all files matching that pathname.
1986 Instead, an error of type FILE-ERROR is signalled.
1988 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1989 * bug 147 fixed: The compiler preserves its block link/count
1990 invariants more correctly now so that it doesn't crash. (thanks
1992 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1994 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1995 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1996 no-value-for-this-character as NIL (as specified by ANSI).
1997 * HOST-NAMESTRING on physical pathnames now returns a string that is
1998 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1999 (thanks to Christophe Rhodes)
2000 * The Alpha port handles icache flushing more correctly. (thanks to
2002 * More progress has been made toward bootstrapping under CLISP. (thanks
2003 to Christophe Rhodes)
2004 * The fasl file format has changed again, because dynamic loading
2005 on OpenBSD (which has non-ELF object files) motivated some cleanups
2006 in the way that foreign symbols are transformed and passed around.
2007 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2008 is no longer treated as whitespace by the reader, but instead as
2009 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2010 |A
\7fB|, instead of A as it used to.
2012 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2013 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2014 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2015 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2016 this as of 0.7.1.45, but was left out of the previous news section
2017 (thanks to Dan Barlow)
2018 * SBCL now runs on the Solaris operating system on SPARC architectures
2019 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2020 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2021 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2022 * SPARC backend cleanups, allowing builds of cores optimized for V8
2023 and V9 SPARCS, and also emission of code targeted to a particular
2024 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2026 * SBCL is closer to bootstrapping under CLISP, thanks to various
2027 fixes by Christophe Rhodes.
2028 * The fasl file format has changed again, to allow the compiler's
2029 INFO database to support symbol macros.
2030 * The user manual (in doc/) is formatted into HTML more nicely.
2031 (thanks to coreythomas)
2032 * The system is smarter about SUBTYPEP relationships, especially
2033 those involving NOT types (including types such as ATOM which are
2034 represented internally using NOT types). Thus SUBTYPEP is less
2035 likely to return (VALUES NIL NIL) in general, and in particular
2036 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2038 * The fasl file format has changed again, because the internal
2039 representation of types now includes a new slot to support the new
2040 SUBTYPEP-of-NOT-types logic.
2041 * (not a change in the main branch of SBCL, but a related prototype
2042 which can hopefully be merged into the main branch of SBCL in the
2043 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2044 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2045 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2046 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2047 dispatch macro character combinations. (thanks to Alexey Dejneka)
2048 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2049 that superficially look like logical namestrings correctly.
2050 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2052 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2053 default supplied-p) for &optional and &key arguments. (thanks to
2056 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2057 * incompatible change: The compiler is now less aggressive about
2058 tail call optimization, doing it only when (> SPACE DEBUG) or
2059 (> SPEED DEBUG). (This is an incompatible change because there are
2060 programs which relied on the old CMU-CL-style behavior to optimize
2061 away their unbounded recursion which will now die of stack overflow.)
2062 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2063 for non-GENCGC systems has been increased to 20M (since that
2064 seems much closer to the likely performance optimum for modern
2065 systems than the old 4M value was)
2066 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2067 and *DEBUG-PRINT-LEVEL*
2068 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2069 of CMU CL's support for SPARC, and various endianness and other
2070 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2071 * new syntactic sugar for the Unix command line: --load foo.bar is now
2072 an alternate notation for --eval '(load "foo.bar")'.
2074 ** The system now detects stack overflow and handles it gracefully,
2075 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2076 optimization settings. (This is a good thing in general, and
2077 its introduction in this version should be particularly timely
2078 for anyone whose code fails because of suppression of tail
2080 ** The system now hunts for the C variable "environ" in a more
2081 devious way, to avoid segfaults when the C library version differs
2082 between compile time and run time. (thanks to Christophe Rhodes)
2083 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2084 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2085 ** The compiler no longer issues bogus style warnings for undefined
2086 classes in the same source file as the DEFCLASSes which defined
2087 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2089 ** fixes in CONDITION class precedence list for undefined function
2090 errors (thanks to Alexei Dejneka)
2091 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2092 correctly. (thanks to Dan Barlow)
2093 ** portability fixes aiming at bootstrapping under CLISP (thanks
2094 to Dave McDonald and Christophe Rhodes)
2095 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2096 ** fixes in type translation and and type inference (thanks to
2098 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2099 ** various fixes in the new ports (thanks to Dan Barlow)
2100 * several changes related to debugging:
2101 ** suppression of tail recursion, as noted above
2102 ** stack overflow detection, as noted above
2103 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2104 is now the default. (For some time encapsulation has been more
2105 reliable than the breakpoint-based :ENCAPSULATE NIL
2106 implementation, at least on X86 systems; and I just noticed that
2107 encapsulation also seems closer to the spirit of the ANSI
2110 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2112 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2113 up properly again. (There was a packaging bug in 0.7.0 which
2114 left their definitions in SB-SYS::LOAD-FOREIGN and
2115 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2116 vital for most things which interface to C-level interfaces,
2117 like extensions working with sockets or databases or
2118 Perl-compatible regexes or whatever, and the need to fix
2119 this bug is the main reason that 0.7.1 was released so
2121 ** DEFGENERIC is now choosier about the methods it redefines, so that
2122 reLOADing a previously-LOADed file containing DEFGENERICs does
2123 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2124 cycle now works as it should. (thanks to Alexey Dejneka)
2125 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2126 fixed by Christophe Rhodes. (He actually submitted this patch
2127 months ago, and I delayed until after 0.7.0.)
2128 ** Bug 111 (internal compiler confusion about runtime checks on
2129 FUNCTION types) was fixed by Alexey Dejneka.
2130 * Some internal cleanups (getting rid of variables which aren't
2131 needed now that the byte interpreter is gone) caused the fasl
2132 file format number to change again.
2134 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2135 * major incompatible change: The default fasl file extension, i.e. the
2136 default extension for files produced by COMPILE-FILE, has changed
2137 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2139 ** There are many changes in the implementation of the compiler.
2140 SBCL is now essentially a compiler-only implementation of ANSI
2141 Common Lisp. EVAL still "interprets" a few special cases, but
2142 almost all the interesting cases are handled by creating
2143 a LAMBDA expression, calling COMPILE on it, then calling
2144 FUNCALL on the result.
2145 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2146 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2147 Since the code is newer, there might still be some new bugs
2148 (though not as many as before Martin Atzmueller's fixes:-). But
2149 the new code is substantially simpler and clearer, and hopefully
2150 any remaining bugs will be simpler, less fundamental, and more
2151 fixable then the bugs in the old code.
2152 ** The revised compiler is still a little unsteady on its feet.
2154 *** The debugging information it produces (particularly the names
2155 of FUNCTION objects) is sometimes much less useful than what
2156 the old compiler produced.
2157 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2158 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2159 within a MACROLET) has been temporarily weakened.
2160 ** There are new compiler optimizations for various functions:
2161 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2162 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2163 *** the math functions TRUNCATE, FLOOR, and CEILING
2164 *** the function-of-all-trades COERCE
2165 Mostly these should be transparent, but there's one
2166 potentially-annoying problem (bug 117): when the compiler
2167 inline-expands a function and does type analysis on the result,
2168 it can create control paths which have type mismatches, and
2169 when it can't prove that those control paths aren't taken,
2170 it will issue WARNINGs about the type mismatches. This is
2171 a particular problem in practice for the new sequence functions.
2172 It's not clear how this should be fixed, and for now, a
2173 workaround is given in the entry for 117 in the BUGS file.
2174 ** (Because of the interaction between the two previous items --
2175 occasional inlining problems and new inline expansions -- some
2176 of the new sequence function optimizations won't really kick in
2177 completely until debugging information, and then inlining, are
2178 straightened out in some future version.)
2179 * minor incompatible changes:
2180 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2181 DIRECTORY no longer implicitly promotes NIL slots of its
2182 pathname argument to :WILD. In particular, when you ask for the
2183 contents of a directory (which you used to be able to do without
2184 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2185 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2186 ** changes in behavior that ANSI explicitly defines to be
2187 implementation dependent:
2188 *** The new compiler-only implementation still conforms with ANSI,
2189 but acts a little different than before. Besides the obvious
2190 changes in performance tradeoffs (that the cost per form passed
2191 to EVAL has gone up, and the cost per form executed by EVAL
2192 has gone down), the behavior of the system changes a little
2193 because there are no longer any interpreted function objects.
2194 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2195 e.g. doing COMPILE on the output of interactive DEFUN is
2197 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2199 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2200 has changed from (:CL) to NIL.
2201 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2202 e.g. #\Newline and #\Tab, don't have names specified in the
2203 ANSI Common Lisp standard, is now based on their ASCII symbolic
2204 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2205 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2206 are no longer used for output.
2207 ** changes in internal implementation constants:
2208 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2209 4 million. (If your application spends a lot of time GCing and
2210 you have a lot of RAM, you might want to experiment with
2211 increasing it even more.)
2212 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2213 However, almost all old code should still continue to work without
2214 immediate update, as SB-C-CALL is now a (deprecated) nickname
2216 ** Old operator names in the style DEF-FOO are now deprecated in
2217 favor of new corresponding names DEFINE-FOO, for consistency with
2218 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2219 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2220 internal symbols, but a few supported extensions like
2221 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2222 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2223 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2224 etc. as you get deeper into recursive calls to the debugger
2225 command loop, instead of the old "5]", "5]]", "5]]]"
2226 sequence. (I was motivated to do this when squabbles between
2227 ILISP and SBCL left me very deeply nested in the debugger. In the
2228 short term, this change will probably provoke more ILISP/SBCL
2229 squabbles, but hopefully it will be an improvement in the long run.)
2230 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2231 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2232 function, since declaiming return types involving aliens is
2233 (1) annoyingly messy to do by hand and (2) vital to efficient
2234 compilation of code which calls such functions.
2235 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2236 longer reexported by the SB-EXT package. They're solely useful
2237 for alien code, so it seems more logical that you should get
2238 them from the SB-ALIEN package, not in SB-EXT.
2239 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2240 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2241 features. Instead, the code that they used to control is always
2242 built into the system.
2243 * many other bug fixes
2244 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2245 advantage of the new EVAL-WHEN stuff and to clean them up in
2246 general, and they are now more ANSI-compliant in a number of
2247 ways. Martin Atzmueller is responsible for a lot of this.
2248 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2250 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2251 *** correct ERROR type for various file operations
2252 *** some fixes for Lisp streams
2253 *** DEFMETHOD syntax checking
2254 *** changing old weird representation of debug information as
2255 strings (which, among their other deficiencies, don't transform
2256 correctly when you rename packages, and don't change their
2257 print representation when you change things like *PACKAGE*
2258 and *PRINT-LENGTH*) to symbols and lists of symbols
2259 He also made several improvements and fixed several bugs in DESCRIBE.
2260 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2262 *** misbehavior of WRITE-STRING/WRITE-LINE
2263 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2264 and several other LOOP problems as well
2265 *** DIRECTORY when similar filenames are present
2266 *** DEFGENERIC with :METHOD options
2267 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2268 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2269 *** argument ordering in FIND with :TEST option
2270 *** mishandled package designator argument in APROPOS-LIST
2271 *** various problems in the backquote readmacro
2272 *** a bug in APROPOS
2273 *** probably some others that I'm not describing very well here,
2274 since the CVS log documents them by reference to sbcl-devel
2275 messages, and the SourceForge archives aren't working well.:-(
2276 ** Dan Barlow improved the Alpha port (and is making progress on the
2277 PPC port, for those of you who think different).
2278 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2279 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2280 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2281 continues to work on the SPARC port (for those of you in a position
2282 to look down upon our little PC-compatible boxes from a great height).
2283 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2284 and uses that copy, rather than the current dynamic value, when
2285 it's trying to decide whether to truncate output. Thus e.g.
2286 (let ((*print-lines* 50))
2287 (pprint-logical-block (stream nil)
2289 (let ((*print-lines* 8))
2290 (print (aref possiblybigthings i) stream)))))
2291 should now truncate the logical block only at 50 lines, instead of
2292 often truncating it at 8 lines, as it did before.
2293 * The doc/cmucl/ directory, containing old CMU CL documentation
2294 from the time of the fork, is no longer part of the base system.
2295 SourceForge has shut down its anonymous FTP service, and with it
2296 my original plan for distributing the old CMU CL documentation
2297 there. For now, if you need these files you can download an old
2298 SBCL source release and extract them from it.
2299 * The fasl file version number changed again, for dozens of reasons,
2300 some of which are apparent above.
2302 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2303 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2304 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2305 patch, so that DISASSEMBLE output is much nicer.
2306 * The code in the SB-PROFILE package now seems reasonably stable.
2307 I still haven't decided what the final interface should look like
2308 (I'd like PROFILE to interact cleanly with TRACE, since both
2309 facilities use function encapsulation) but if you have a need
2310 for profiling now, you can probably use it successfully with
2311 the current CMU-CL-style interface.
2312 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2313 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2314 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2315 link now returns the dangling link itself, and for similar
2316 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2317 link itself. (In these cases the old code signalled an error and
2318 looped endlessly, respectively.) Thus, DIRECTORY now works even
2319 in the presence of dangling and cyclic symbolic links.
2320 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2321 is now a supported extension again, since the consensus on
2322 sbcl-devel was that it can be useful for ordinary development
2323 work, not just for debugging SBCL itself.
2324 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2325 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2326 that currently-defined functions might be redefined later with
2327 different return types.
2328 * Hash tables can be printed readably, as inspired by CMU CL code
2329 of Eric Marsden and SBCL code of Martin Atzmueller.
2330 * better error handling in CLOS method combination, thanks to
2331 Martin Atzmueller porting Pierre Mai's CMU CL patches
2332 * more overflow fixes for >16Mbyte I/O buffers
2333 * A bug in READ has been fixed, so that now a single Ctrl-D
2334 character suffices to cause end-of-file on character streams.
2335 In particular, now you only need one Ctrl-D at the command
2336 line (not two) to exit SBCL.
2337 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2339 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2340 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2341 does the right thing.
2342 * The implementation of some type tests, especially for CONDITION
2343 types, is now tidier and maybe faster, due to CMU CL code
2344 originally by Douglas Crosher, ported by Martin Atzmueller.
2345 * Some math functions have been fixed, and there are new
2346 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2347 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2348 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2349 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2350 likely to be built into sbcl-0.7.0.)
2351 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2352 CL, ported to SBCL by Martin Atzmueller)
2353 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2354 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2355 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2356 * There's a new slam.sh hack to shorten the edit/compile/debug
2357 cycle for low-level changes to SBCL itself, and a new
2358 :SB-AFTER-XC-CORE target feature to control the generation of
2359 the after-xc.core file needed by slam.sh.
2360 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2361 COMPILE-FILE is no longer supported, so that now every function
2362 gets an entry point, so that block compilation looks a little
2363 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2364 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2365 returns the number of bytes consed since the system started,
2366 rather than the number consed since the first time the function
2367 was called. (The new definition parallels ANSI functions like
2368 CL:GET-INTERNAL-RUN-TIME.)
2369 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2370 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2371 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2372 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2373 * Fasl file version numbers are now independent of the target CPU,
2374 since historically most system changes which required version
2375 number changes have affected all CPUs equally. Similarly,
2376 the byte fasl file version is now equal to the ordinary
2379 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2380 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2381 is no longer recognized. I apologize for this, because it was
2382 listed in SB-EXT as a supported extension, but I found that
2383 its existing behavior was poorly specified, as well as incorrectly
2384 specified, and it looked like too much of a mess to straighten it
2385 out. I have enough on my hands trying to get ANSI stuff to work..
2386 * many patches ported from CMU CL by Martin Atzmueller, with
2387 half a dozen bug fixes in pretty-printing and the debugger, and
2388 half a dozen others elsewhere
2389 * fixed bug 13: Floating point infinities are now supported again.
2390 They might still be a little bit flaky, but thanks to bug reports
2391 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2392 as flaky as they were.
2393 * The --noprogrammer command line option is now supported. (Its
2394 behavior is slightly different in detail from what the old man
2395 page claimed it would do, but it's still appropriate under the
2396 same circumstances that the man page talks about.)
2397 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2398 are now supported, and enabled by default. Thus, the compiler can
2399 handle many floating point and complex operations much less
2400 inefficiently. (Thus e.g. you can implement a complex FFT
2402 * The compiler now detects type mismatches between DECLAIM FTYPE
2403 and DEFUN better, and implements CHECK-TYPE more correctly, and
2404 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2405 thanks to patches from Martin Atzmueller.
2406 * various fixes to make the cross-compiler more portable to
2407 ANSI-conforming-but-different cross-compilation hosts (notably
2408 Lispworks for Windows, following bug reports from Arthur Lemmens)
2409 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2410 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2411 fixed thanks to Pierre Mai's CMU CL patches.
2412 * fixes to keep the system from overflowing internal counters when
2413 it tries to use i/o buffers larger than 16M bytes
2414 * fixed bug 45a: Various internal functions required to support
2415 complex special functions have been merged from CMU CL sources.
2416 (When I was first setting up SBCL, I misunderstood a compile-time
2417 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2418 * improved support for type intersection and union, fixing bug 12
2419 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2420 more obscure bugs as well
2421 * some steps toward byte-compiling non-performance-critical
2422 parts of the system, courtesy of patches from Martin Atzmueller
2423 * Christophe Rhodes has made some debian packages of sbcl at
2424 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2425 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2426 stable, but are nonetheless usable. When he's ready, I'd be happy
2427 to add them to the SourceForge "File Releases" section. (And if
2428 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2429 * new fasl file format version number (because of changes in
2430 internal representation of (OR ..) types to accommodate the new
2431 support for (AND ..) types, among other things)
2433 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2434 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2436 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2438 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2439 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2440 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2441 * DESCRIBE now works on structure objects again.
2442 * Most function call argument type mismatches are now handled as
2443 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2444 know whether the function will be redefined before the call is
2445 executed. (The compiler could flag local calls with full WARNINGs,
2446 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2447 it doesn't keep track of enough information to know whether calls
2448 are local in this sense.)
2449 * Compiler output is now more verbose, with messages truncated
2450 later than before. (There should be some supported way for users
2451 to override the default verbosity, but I haven't decided how to
2452 provide it yet, so this behavior is still controlled by the internal
2453 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2454 src/compiler/ir1util.lisp.)
2455 * Fasl file format version numbers have increased again, because
2456 support for the Gray streams extension changes the layout of the
2457 system's STREAM objects.
2458 * The Gray subclassable streams extension now works, thanks to a
2459 patch from Martin Atzmueller.
2460 * The full LOAD-FOREIGN extension (not just the primitive
2461 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2462 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2463 but like most other programs, it defaults to copying the Unix
2464 environment from the original process instead of starting the
2465 new process in an empty environment.
2466 * Extensions which manipulate the Unix environment now support
2467 an :ENVIRONMENT keyword option which doesn't smash case or
2468 do other bad things. The CMU-CL-style :ENV option is retained
2469 for porting convenience.
2470 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2471 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2473 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2475 * A patch from Martin Atzmueller seems to have solved the SIGINT
2476 problem, and as far as we know, signal-handling now works cleanly.
2477 (If you find any new bugs, please report them!)
2478 * The system no longer defaults Lisp source file names to types
2479 ".l", ".cl", or ".lsp", but only to ".lisp".
2480 * The compiler no longer uses special default file extensions for
2481 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2482 seems to expect a single default extension for all compiled code,
2483 and there's no compelling reason to try to stretch the standard
2484 to allow two different extensions.) Instead, byte-compiled files
2485 default to the same extension as native-compiled files.
2486 * Fasl file format version numbers have increased again, because
2487 a rearrangement of internal implementation packages made some
2488 dumped symbols in old fasl files unreadable in new cores.
2489 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2490 many fewer weird special cases.
2491 * Bug #17 (differing COMPILE-FILE behavior between logical and
2492 physical pathnames) has been fixed, and some related misbehavior too,
2493 thanks to a patch from Martin Atzmueller.
2494 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2495 by Tim Moore, ported to SBCL by Martin Atzmueller.
2496 * Martin Atzmueller fixed several filesystem-related problems,
2497 including bug #36, in part by porting CMU CL patches, which were
2498 written in part by Paul Werkowski.
2499 * More compiler warnings in src/runtime/ are gone, thanks to
2500 more patches from Martin Atzmueller.
2501 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2504 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2506 * DESCRIBE now works on CONDITION objects.
2507 * The debugger now handles errors which arise when trying to print
2508 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2510 * The build system now uses an additional file, customize-target-features.lisp,
2511 to allow local modifications to the target *FEATURES* list. (The point of
2512 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2513 debugging features enabled, without having to worry about propagating it
2514 into everyone's system when I do a "cvs update".) When no
2515 customize-target-features.lisp file exists, the target *FEATURES* list
2516 should be constructed the same way as before.
2517 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2518 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2519 warn about behavior which is undefined under the ANSI spec. Note: This
2520 is specified by ANSI, but it's not very popular with programmers.
2521 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2522 macro in the SBCL sources for an example of a workaround which you
2523 might use to make portable ANSI-standard code which does what you want.
2524 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2525 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2526 when it's used as a non-toplevel form. (This is required in order
2527 to implement the DEFCONSTANT-EQX macro.)
2528 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2529 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2530 funny things, due to bugs in the implementation of EVAL-WHEN
2531 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2532 non-toplevel DEFCONSTANTs are uncommon.)
2533 * The core file version number and fasl file version number have been
2534 incremented, because the old noncompliant DEFCONSTANT behavior involved
2535 calling functions which no longer exist, and because I also took the
2536 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2537 * fixed bug 1 (error handling before read-eval-print loop starts), and
2538 redid debugger restarts and related debugger commands somewhat while
2540 ** The QUIT debugger command is gone, since it did something
2541 rather different than the SB-EXT:QUIT command, and since it never
2542 worked properly outside the main toplevel read/eval/print loop.
2543 Invoking the new TOPLEVEL restart provides the same functionality.
2544 ** The GO debugger command is also gone, since you can just invoke
2545 the CONTINUE restart directly instead.
2546 ** The TOP debugger command is also gone, since it's redundant with the
2547 FRAME 0 command, and since it interfered with abbreviations for the
2549 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2551 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2552 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2553 some porting work by Martin Atzmueller)
2554 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2555 with a boatload of patches which he ported to SBCL. Now that those
2557 ** The system tries to make sure that its low-priority messages
2558 are prefixed by semicolons, to help people who like to use
2559 syntax highlighting in their ILISP buffer. (This patch
2560 was originally due to Raymond Toy.)
2561 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2562 patches originally written by Raymond Toy.
2563 ** The compiler understands coercion between single-value and
2564 multiple-VALUES type expressions better, getting rid of some very
2565 weird behavior, thanks to patches originally by Robert MacLachlan
2566 and Douglas Crosher.
2567 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2568 lambda lists, thanks to a patch originally by Pierre Mai.
2569 ** The system no longer bogusly warns about "abbreviated type
2571 ** The compiler gets less confused by inlining and RETURN-FROM,
2572 thanks to some patches originally by Tim Moore.
2573 ** The system no longer hangs when dumping circular lists to fasl
2574 files, thanks to a patch originally from Douglas Crosher.
2575 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2576 undefined function error.
2577 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2578 in structure slot initforms) for now, documented workaround instead:-|
2579 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2580 * fixed bug 5: added stubs for various Gray stream functions called
2581 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2582 installed, at least appropriate type errors are generated
2583 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2584 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2585 consistently in DEFMETHOD forms.
2586 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2587 it doesn't seem to affect SBCL after all
2588 * The C runtime system now builds with better optimization and many
2589 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2591 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2593 * The system is now under CVS at SourceForge (instead of the
2594 CVS repository on my home machine).
2595 * The new signal handling code has been tweaked to treat register
2596 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2597 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2598 downstream, e.g. in debug-int.lisp, has implicit dependencies
2599 on the unsignedness of integer representation of machine words,
2600 and that caused the system to bomb out with infinite regress
2601 when trying to recover from type errors involving signed values,
2602 e.g. (BUTLAST '(1 2 3) -1).)
2603 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2604 (This was one of the bugs Peter Van Eynde reported back in July.)
2605 * The system now uses code inspired by Colin Walters' O(N)
2606 implementation of MAP (from the cmucl-imp@cons.org mailing
2607 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2608 inline the MAP operation, and there is more than one
2609 sequence argument to the MAP call (so that it can't just
2610 do ETYPECASE once and for all based on the type of the
2611 single sequence argument). (The old non-inline implementation
2612 of the general M-argument sequence-of-length-N case required
2613 O(M*N*N) time when any of the sequence arguments were LISTs.)
2614 * The QUIT :UNIX-CODE keyword argument has been renamed to
2615 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2616 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2618 * Raymond Wiker's patches to port dynamic loading from Linux to
2619 FreeBSD have been added.
2620 * The BUGS file is now more nearly up to date, thanks in large part
2621 to Martin Atzmueller's review of it.
2622 * The debugger now flushes standard output streams before it begins
2623 its output ("debugger invoked" and so forth).
2624 * The core version number and fasl file version number have both
2625 been incremented, because of incompatible changes in the layout
2627 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2628 used to not be called for a saved Lisp image.)
2629 * Martin Atzmueller's version of a patch to fix a compiler crash,
2630 as posted on sbcl-devel 13 September 2000, has been installed.
2631 * Instead of installing Martin Atzmueller's patch for the
2632 compiler transform for SUBSEQ, I deleted the compiler transform,
2633 and transforms for some similar consing operations.
2634 * A bug in signal handling which kept TRACE from working on OpenBSD
2636 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2637 away by constant folding
2638 * The system now defines its address space constants in one place
2639 (in the Lisp sources), and propagates them automatically elsewhere
2640 (through GENESIS and the sbcl.h file). Therefore, patching the
2641 address map is less unnecessarily tedious and error-prone. The
2642 Lisp names of address space constants have also been systematized.
2643 * CVS tags like dollar-Header-dollar have been removed from
2644 the sources, because they have never saved me trouble and
2645 they've been source of trouble working with patches and other
2646 diff-related operations.
2647 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2648 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2650 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2652 * The system has been ported to OpenBSD.
2653 * The system now compiles with a simple "sh make.sh" on the systems
2654 that it's supported on. I.e., now you no longer need to tweak
2655 text in the target-features.lisp-expr and symlinks in src/runtime/
2656 by hand, the make.sh takes care of it for you.
2657 * The system is no longer so grossly inefficient when compiling code
2658 involving vectors implemented as general (not simple) vectors (VECTOR T),
2659 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2660 which dares to use the various sequence functions on non-simple
2661 vectors, takes less of a performance hit.
2662 * There is now a primitive type predicate VECTOR-T-P
2663 to test for the (VECTOR T) type, so that e.g.
2664 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2665 can now be compiled with some semblance of efficiency. (The old code
2666 turned the type declaration into a full call to %TYPEP at runtime!)
2667 * AREF on (VECTOR T) is still not fast, since it's still compiled
2668 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2669 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2670 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2671 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2672 without making full calls to SUBTYPEP at runtime.
2673 (Some analogous efficiency issues for non-simple vectors specialized to
2674 element types other than T, or for non-simple multidimensional arrays,
2675 have not been addressed. They could almost certainly be handled the
2676 same way if anyone is motivated to do so.)
2677 * The changes in array handling break binary compatibility, so
2678 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2679 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2680 instead of (VALUES T T).
2681 * By following the instructions that Dan Barlow posted to sbcl-devel
2682 on 2 July 2000, I was able to enable primitive dynamic object
2683 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2684 functionality is not implemented (since it calls ld to resolve
2685 library references automatically, requiring RUN-PROGRAM for its
2686 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2687 try to resolve library references) is now supported.
2688 * The system now flushes the standard output streams when it terminates,
2689 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2690 them at several other probably-convenient times, e.g. in each pass of
2691 the toplevel read-eval-print loop, and after evaluating a form given
2692 as an "--eval" command-line option. (These changes were motivated by a
2693 discussion of stream flushing issues on cmucl-imp in August 2000.)
2694 * The source transform for TYPEP of array types no longer assumes
2695 that an array whose element type is a not-yet-defined type
2696 is implemented as an array of T, but instead punts, so that the
2697 type will be interpreted at runtime.
2698 * There is now some support for cross-compiling in make.sh: each of
2699 the phases of make.sh has its own script. (This should be transparent
2700 to people doing ordinary, non-cross-compile builds.)
2701 * Since my laptop doesn't have hundreds of megabytes of memory like
2702 my desktop machine, I became more motivated to do some items on
2703 my to-do list in order to reduce the size of the system a little:
2704 ** Arrange for various needed-only-at-cold-init things to be
2705 uninterned after cold init. To support this, those things have
2706 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2707 symbols with such names are now uninterned after cold init).
2708 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2709 fasl files for cold load.
2710 ** Remove the old compiler structure pooling code (which used to
2711 be conditional on the target feature :SB-ALLOC) completely.
2712 ** Redo the representation of some data in cold init to be more compact.
2713 (I also looked into supporting byte compiled code at bootstrap time,
2714 which would probably reduce the size of the system a lot, but that
2715 looked too complicated, so I punted for now.)
2716 * The maximum signal nesting depth in the src/runtime/ support code has
2717 been reduced from 4096 to 256. (I don't know any reason for the very
2718 large old value. If the new smaller value turns out to break something,
2719 I'll probably just bump it back up.)
2720 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2722 * Many, many bugs reported by Peter Van Eynde have been added to
2723 the BUGS list; some have even been fixed.
2724 * While enabling dynamic object file loading, I tried to make the
2725 code easier to understand, renaming various functions and variables
2726 with less ambiguous names, and changing some function calling
2727 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2728 * While trying to figure out how to do the OpenBSD port, I tried to
2729 clean up some of the code in src/runtime/. In particular, I dropped
2730 support for non-POSIX signal handling, added various comments,
2731 tweaked the code to reduce the number of compilation warnings, and
2732 renamed some files to increase consistency.
2733 * To support the new automatic configuration functionality in make.sh,
2734 the source file target-features.lisp-expr has been replaced with the
2735 source file base-target-features.lisp-expr and the machine-generated
2736 file local-target-features.lisp-expr.
2737 * fixed a stupid quoting error in make.sh so that using CMU CL
2738 "lisp -batch" as cross-compilation host works again
2740 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2742 * DESCRIBE no longer tries to call itself recursively to describe
2743 bound/fbound values, so that it no longer fails on symbols which are
2744 bound to themselves (like keywords, T, and NIL).
2745 * DESCRIBE now works on generic functions.
2746 * The printer now prints less-screwed-up representations of closures
2747 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2748 * A private symbol is used instead of the :EMPTY keyword previously
2749 used to mark empty slots in hash tables. Thus
2750 (DEFVAR *HT* (MAKE-HASH-TABLE))
2751 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2752 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2753 now does what ANSI says that it should. (You can still get
2754 similar noncompliant behavior if bang on the hash table
2755 implementation with all the symbols you get back from
2756 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2757 This breaks binary compatibility, since tests for equality to
2758 :EMPTY are wired into things like the macroexpansion of
2759 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2761 * There's now a minimal placeholder implementation for CL:STEP,
2762 as required by ANSI.
2763 * An obscure bug in the interaction of the normal compiler, the byte
2764 compiler, inlining, and structure predicates has been patched
2765 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2766 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2768 * Missing ordinary arguments in a macro call are now detected even
2769 when the macro lambda list contains &KEY or &REST.
2770 * The debugger no longer complains about encountering the top of the
2771 stack when you type "FRAME 0" to explicitly instruct it to go to
2772 the top of the stack. And it now prints the frame you request even
2773 if it's the current frame (instead of saying "You are here.").
2774 * As specified by ANSI, the system now always prints keywords
2775 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2777 * The default initial SIZE of HASH-TABLEs is now smaller.
2778 * Type information from CLOS class dispatch is now propagated
2779 into DEFMETHOD bodies, so that e.g.
2780 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2782 is now basically equivalent to
2783 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2784 (DECLARE (TYPE SINGLE-FLOAT X))
2786 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2787 operation, without having to do run-time type dispatch.
2788 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2789 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2790 * Since I don't seem to be making big file reorganizations very often
2791 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2792 is overflowing my ability to conveniently back them up), I've finally
2793 checked the system into CVS. (The CVS repository is on my home system,
2794 not at SourceForge -- putting it on SourceForge might come later.)
2795 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2796 high-level GC-NOTIFY-FOO functions send their output. (There's
2797 still very little control of where low-level verbose GC functions
2798 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2799 less than it used to -- the GC-NOTIFY-FOO functions are now under
2800 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2801 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2802 in only one place in the source code, and propagates it automatically
2803 everywhere that it's needed. Thus e.g. when I bump the version from
2804 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2805 * The C source files now include boilerplate legalese and documentation
2806 at the head of each file (just as the Lisp source files already did).
2807 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2808 to his page will be replaced with a link to his new CLiki service.
2810 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2812 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2813 * The build process now looks for GNU make under the default name "gmake",
2814 instead of "make" as it used to. If GNU make is not available as "gmake"
2815 on your system, you can change this default behavior by setting the
2816 GNUMAKE environment variable.
2817 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2818 can build without error under CMU CL.
2820 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2822 * There is now a partial SBCL user manual (with some new text and some
2823 text cribbed from the CMU CL manual).
2824 * The beginnings of a profiler have been added (starting with the
2825 CMU CL profiler and simplifying and cleaning up). Eventually the
2826 main interface should be through the TRACE macro, but for now,
2827 it's still accessed through vaguely CMU-CL-style functions and macros
2828 exported from the package SB-PROFILE.
2829 * Some problems left over from porting CMU CL to the new
2830 cross-compilation bootstrap process have been cleaned up:
2831 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2832 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2833 part when getting rid of the compiler *BACKEND* stuff.)
2834 ** The value of *NULL-TYPE* was screwed up, because it was
2835 being initialized before the type system knew the final
2836 definition of the 'NULL type. This screwed up several key
2837 optimizations in the compiler, causing inefficiency in all sorts
2838 of places. (I found it because I wanted to understand why
2839 GET-INTERNAL-RUN-TIME was consing.)
2840 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2841 PROCLAIM FTYPE information. Unfortunately this broke binary
2842 compatibility again, since now the forms output by DEFGENERIC
2843 to refer to functions which didn't exist in 0.6.3.
2844 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2845 can use the new (as of 0.6.3) transform for SOME into MAP into
2847 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2848 values to (INTEGER 0 1000000), so that the time code will no longer
2849 occasionally get blown up by Linux returning 1000000 microseconds
2850 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2851 its output conform to the ANSI spec. (Alas, this makes its output
2852 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2853 * A full call to MAP NIL with a single sequence argument no longer conses.
2854 * fixes to problems pointed out by Martin Atzmueller:
2855 * The manual page no longer talks about multiprocessing as though
2856 it were currently supported.
2857 * The ILISP support patches have been removed from the distribution,
2858 because as of version 5.10.1, ILISP now supports SBCL without us
2859 having to maintain patches.
2860 * added a modified version of Raymond Toy's recent CMU CL patch for
2861 EQUALP comparison of HASH-TABLE
2863 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2865 * The system still can't cross-compile itself with
2866 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2867 entails), but at least it can compile more of itself that way
2868 than it used to be able to, and various buglets which were uncovered
2869 by trying to cross-compile itself that way have now been fixed.
2870 * This release breaks binary compatibility again. This time
2871 at least I've incremented the FASL file format version to 2, so that the
2872 problem can be detected reliably instead of just causing weird errors.
2873 * various new style warnings:
2874 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2875 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2876 ** using the lexical binding of a variable named in the *FOO* style
2877 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2878 as specified by ANSI.
2879 * *RANDOM-STATE* is no longer automatically initialized from
2880 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2881 default behavior of the system is to repeat its behavior every time
2882 it's run. If you'd like to change this behavior, you can always
2883 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2884 old convention there was no comparably easy way to get the system to
2885 repeat its behavior every time it was run.
2886 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2887 FTYPE declarations has been removed, in favor of their ANSI
2888 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2889 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2890 the types of their sequence arguments can be determined at compile time.
2891 This is done through a new open code expansion for MAP which eliminates
2892 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2893 argument types can be determined at compile time.
2894 * The optimizer now transforms COERCE into an identity operation when it
2895 can prove that the coerced object is already of the correct type. (This
2896 can be a win for machine generated code, including the output of other
2897 optimization transforms, such as the MAP transform above.)
2898 * Credit information has been moved from source file headers into CREDITS.
2899 * Source file headers have been made more standard.
2900 * The CASE macro now compiles without complaining even when it has
2903 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2905 * (Note that the way that the PCL macroexpansions were rewritten
2906 to accommodate the change in DEFGENERIC below breaks binary
2907 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2908 not run under sbcl-0.6.2. Once we get out of alpha releases,
2909 i.e. hit release 1.0.0, we'll probably try to maintain binary
2910 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2911 and sbcl-1.4.4. Until then, however, it might be fairly common
2912 for maintenance releases to break binary compatibility.)
2913 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2915 * The reporting of unhandled signals has been changed to print some
2916 explanatory text as well as the report form. (Previously only
2917 the report form was printed.)
2918 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2919 it defines, so that the compiler no longer issues undefined function
2920 warnings for compiled-but-not-yet-loaded generic functions.
2921 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2922 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2923 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2924 section of the ANSI spec are supported.
2925 * The "" nickname for the "KEYWORD" package has been removed.
2926 The reader still handles symbol tokens which begin with a package marker
2927 as keywords, but it doesn't expose its mechanism for doing so in the
2928 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2929 * The system now issues STYLE-WARNINGs for contradictory TYPE
2930 proclamations. (Warnings for contradictory FTYPE proclamations would
2931 be nice too, but those can't be done usefully unless the type system
2932 is made smarter about FUNCTION types.)
2933 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2934 systematized, so that "*target-*.lisp is supposed to exist only on the
2935 target and imply that there's a related file which exists on the
2936 host, and *host-*.lisp is supposed to exist only on the host and imply
2937 that there's a related file which exists on the target. This involves a
2938 lot of renaming. Hopefully the acute confusion caused by the renaming
2939 will be justified by the reduction in chronic confusion..
2940 ** runtime-type.lisp -> early-target-type.lisp
2941 ** target-type.lisp -> late-target-type.lisp
2942 ** early-host-format.lisp -> early-format.lisp
2943 ** late-host-format.lisp -> late-format.lisp
2944 ** host-error.lisp -> misc-error.lisp
2945 ** early-error.lisp -> early-target-error.lisp
2946 ** late-error.lisp -> late-target-error.lisp
2947 ** host-defboot.lisp -> early-defboot.lisp
2948 ** code/misc.lisp -> code/target-misc.lisp
2949 ** code/host-misc.lisp -> code/misc.lisp
2950 ** code/numbers.lisp -> code/target-numbers.lisp
2951 ** code/early-numbers.lisp -> numbers.lisp
2952 ** early-host-type.lisp -> early-type.lisp
2953 ** late-host-type.lisp -> late-type.lisp
2954 ** host-typep.lisp -> typep.lisp
2955 ** load.lisp -> target-load.lisp
2956 ** host-load.lisp -> load.lisp
2957 ** host-disassem.lisp -> disassem.lisp
2958 ** host-insts.lisp -> insts.lisp
2959 ** byte-comp.lisp -> target-byte-comp.lisp
2960 ** host-byte-comp.lisp -> byte-comp.lisp
2961 ** host-signal.lisp -> signal.lisp
2962 ** host-defstruct.lisp -> defstruct.lisp
2963 ** late-target-type.lisp -> deftypes-for-target.lisp
2964 Furthermore, several other previously target-only files foo.lisp (e.g.
2965 hash-table.lisp and random.lisp) have been split into a target-and-host
2966 foo.lisp file and a target-only target-foo.lisp file, with their key type
2967 definitions in the target-and-host part, so that the cross-compiler will
2968 know more about target types.
2969 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2970 gone away. In their place are various *BACKEND-FOO* variables
2971 corresponding to the slots of the old structure.
2972 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2973 into the target SBCL has been fixed.
2974 * The chill.lisp system for loading cold code into a running SBCL
2976 * Support for the CMU CL "scavenger hook" extension has been removed.
2977 (It was undocumented and unused in the CMU CL sources that SBCL was
2978 derived from, and stale in sbcl-0.6.1.)
2979 * Various errors in the cross-compiler type system were detected
2980 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2981 (enabling various consistency checks). Many of them were fixed,
2982 but some hard problems remain, so the compiler is back to
2983 running without *TYPE-SYSTEM-INITIALIZED* for now.
2984 * As part of the cross-compiler type system cleanup, I implemented
2985 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2986 * I have started adding UNCROSS calls throughout the type system
2987 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2988 on cross-compiler input files will be able to go away, and various
2990 * CONSTANTP now returns true for quoted forms (as explicitly required
2993 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2995 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2996 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2997 (MAKE-LIST -1). In the longer term, it ought to become true
2998 that declarations are assertions even at SAFETY 1. For now, it's not
2999 quite true even at SAFETY 3, but it's at least more nearly true..
3000 (Note that this change seems to increases the size of the system by
3001 O(5%) and to decrease the speed of the compiler by 20% or more.)
3002 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3003 for the problem of printing dozens of lines of distracting information
3004 about low-level system machinery as part of the top stack frame
3005 on entry to the debugger when an undefined function was called.
3006 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3007 is not reset to COMMON-LISP-USER.
3008 * Compilation of stuff related to dyncount.lisp has been made conditional
3009 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3010 smaller. The various dyncount-related symbols have been moved into
3011 a new "SB-DYNCOUNT" package.
3012 * tty-inspect.lisp has been renamed to inspect.lisp.
3013 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3014 feature has gone away. (When we eventually port to other flavors of
3015 libc and/or Unix, we'll try to make the differences between flavors
3016 invisible at the user level.)
3017 * Various other *FEATURES* tags, and/or their associated conditionals,
3018 have been removed if obsolescent, or given better documentation, or
3019 sometimes given more-mnemonic names.
3021 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3023 * tidied up "make.sh" script
3024 * tidied up system directory structure
3025 * better "clean.sh" behavior
3026 * added doc/FOR-CMUCL-DEVELOPERS
3027 * many many small tweaks to output format, e.g. removing possibly-confusing
3028 trailing #\. character in DESCRIBE-INSTANCE
3029 * (EQUALP #\A 'A) no longer signals an error.
3030 * new hashing code, including EQUALP hashing
3031 * tidied up Lisp initialization and toplevel
3032 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3033 * command line argument processing
3034 * added POSIX-GETENV function to deal with Unix-ish environment variables
3035 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3036 terminating SBCL on EOF
3037 * non-verbose GC by default
3038 * There is no more "sbcl" shell script; the sbcl file is now the C
3039 runtime executable (just like CMU CL).
3040 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3042 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3043 symbol and package information as Lisp native symbol and package objects
3044 instead of strings naming symbols and strings naming packages. This way,
3045 whenever packages are renamed (as in warm init), debug information is
3046 transformed along with everything else.
3047 * tweaked the optimization policy declarations which control the building
3048 of SBCL itself. Now, among other things, the system no longer saves
3049 source location debugging information. (This helps two problems at once
3050 by reducing SBCL size and by keeping SBCL from trying to look for its
3051 sources -- which may not exist -- when reporting errors.)
3052 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3053 debugging and testing purposes
3054 * cleaned up printing, making the printer call PRINT-OBJECT for
3055 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3056 methods, giving nearly-ANSI behavior
3057 * converted almost all special variables to use *FOO* naming convention
3058 * deleted PARSE-TIME functionality, since it can be done portably
3059 * moved some files out of cold init into warm init
3060 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3062 * regularized formatting of source files
3063 * added an install.sh script
3064 * fixed ridiculous memory usage of cross-compiler by making
3065 compiler/alloc.lisp not try to do pooling unless it can hook
3066 itself into the GC of the cross-compilation host. Now the system
3067 builds nicely on my old laptop.
3068 * added :SB-ALLOC in target-features.lisp-expr
3069 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3070 was not implemented)
3071 * re-did condition handling and note reporting in the compiler. Notes
3072 are no longer handled by signalling conditions. Style warnings
3073 and warnings are handled more correctly and reported in such a way
3074 that it's easy to find one or the other in your output (so that you
3075 can e.g. figure out which of many problems caused COMPILE-FILE to
3077 * changed the severity of several compiler warnings from full WARNING
3078 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3079 compiler note reporting so that it doesn't use the condition system
3080 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3081 COMPILE-FILE command)
3082 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3083 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3084 no longer supported, since their implementation was deeply intertwingled
3085 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3086 * removed bogus "support" for compiler macros named (SETF FOO), and
3087 removed the compiler macro for SETF INFO (but only after making a fool
3088 of myself on the cmucl-imp mailing list by posting a bogus patch for
3089 DEFINE-COMPILER-MACRO..)
3090 * Compiled files containing forms which have side effects on the Lisp
3091 reader (such as DEFPACKAGE forms) are now handled more correctly.
3092 (Compiler queuing of top level lambdas has been suppressed by setting
3093 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3094 * deleted various currently-unused source files, e.g. gengc.lisp. They
3095 may be added back at some point e.g. when porting to other architectures,
3096 but until they are it's distracting to distribute them and to try to
3098 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3099 there were so many of them they're just distractions, and UNCROSS is
3100 known to be able to handle the current sources
3101 * moved PROFILE functionality into TRACE, so that it will be clear
3102 how the wrapping and unwrapping of functions when you profile them
3103 interacts with the wrapping and unwrapping of functions when you
3104 trace them. (Actually, the functionality isn't there yet, but at least
3105 the interface specification is there. Hopefully, the functionality will
3106 arrive with some future maintenance release.)
3107 * removed host-oops.lisp
3108 * changed signature of QUIT function to allow UNIX-CODE argument
3109 * fixed READ-SEQUENCE bug
3110 * tweaked verbose GC output so that it looks more like the progress
3111 output that ANSI specifies for functions like LOAD
3112 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3113 * added <http://sbcl.sourceforge.com> to the banner information printed by