1 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2 * incompatible change: thread support for non-NPTL systems has
3 been removed - locking is buggy and unreliable. A threaded
4 SBCL build will now warn at startup and refuse to create
5 new threads, unless futex support is detected in the kernel
6 * incompatible change: the top level REPL now has only an ABORT
7 restart associated with it, not TOPLEVEL and ABORT as it used to.
8 TOP and TOPLEVEL are now available as debugger commands for
9 returning to the top level.
10 * incompatible change: forms evaluated in the REPL now use the
11 global optimization policy.
12 * incompatible change: user- and system-initialization files are
13 no longer processed with LOAD, but by READ and EVAL; hence the
14 global optimization policy, startup package, readtable, etc,
16 * The .fasl file format number has been incremented because of
17 various incompatible changes.
18 * internal entry point details and argument counts no longer appear
19 in backtraces unless explicitly requested by setting
20 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
21 * built-in and standard functions no longer have names like "top
22 level local call to FOO".
23 * fixed bug 32: functions defined in non-null lexical environments
24 now have more legible printed representation
25 * fixed bug 33: functions defined in non-null lexical environemnts
26 are now more amenable to inspection by INSPECT.
27 * workaround for bug 354: XEPs no longer appear in backtraces unless
29 * fixed bug: receiving the signal which results from attempting to
30 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
31 SIGBUS on other *BSDs) on architectures where the C stack is also
32 the Lisp stack (x86 and x86-64) from foreign code no longer leads
33 to debugger confusion or wild instability. (reported by Cheuksan
35 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
36 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
37 (reported by Lutz Euler)
38 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
39 compiled code. (reported by Paul Dietz)
40 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
41 specialized parameter to a method to something that is not TYPEP
42 the specializer is now possible.
43 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
44 face of package deletion.
45 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
46 pathname. (reported by David Tolpin on c.l.l)
47 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
48 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
49 STRING-STREAM and FILE-STREAM, respectively.
50 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
51 than previously; in addition, its test suite is now run on build.
52 (thanks to Luke Gorrie)
53 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
54 * optimization: added a immediate representation for single-floats
56 * fixed some bugs revealed by Paul Dietz' test suite:
57 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
58 ** attempting to create a package with a colliding nickname causes
59 correctable errors to be signalled.
60 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
61 ** several x86-64 backend bugs related to sign-extension of immediate
64 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
65 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
66 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
68 * fixed loading of multiply forward-referenced layouts.
69 (thanks to Cheuksan Wang)
70 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
71 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
72 (see COMPILE-FILE documentation for details of :PRINT option)
73 * fixed bugs 19 and 317: fixed-format floating point printing is
74 more accurate. This also fixes a bug reported by Adam Warner
75 related to the ~@F format directive.
76 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
78 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
79 dispatch table if the from-char is a dispatch macro character.
80 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
81 by the wordsize no longer ignore the last word. (reported by Lutz
83 * fixed bug in type checking of dynamic-extent variables. (reported
85 * optimization: sequence traversal functions use their freedom to
86 coerce function designators to functions.
87 * optimization: code with many calls to CLOS methods specialized on
88 CLOS classes has had redundant type checks removed.
89 * optimization: added declarations to speed up operations that access
90 the internal character database (for example STRING-UPCASE)
91 * optimization: comparison operations between floats and sufficiently small
92 fixnums no longer create extra rationals
93 * fixed some bugs related to Unicode integration:
94 ** portions of multibyte characters at the end of buffers for
95 character-based file input are correctly transferred to the
96 start of the buffer at the next read.
97 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
98 passing it through to OPEN.
99 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
100 argument given to internal calls to OPEN.
101 ** the built-in comment readers (introduced by character sequences
102 ";" and "#|") are more forgiving to encoding errors; they will
103 STYLE-WARN and then attempt to resync the stream at a character
104 boundary. (thanks to Teemu Kalvas)
105 * fixed some bugs in the x86-64 port:
106 ** Negative short int return values from c-calls are sign-extended
108 ** The stack is aligned to 16-bytes for c-calls, as required by
109 the ABI. (thanks to Cheuksan Wang)
110 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
111 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
112 for enums. (thanks to Vincent Arkesteijn)
113 ** Multiple small optimizations and bugfixes for floating point
115 * fixed some bugs revealed by Paul Dietz' test suite:
116 ** Space, Tab, Linefeed, Return and Page have the invalid
117 secondary constituent character trait.
118 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
120 ** WITH-INPUT-FROM-STRING should only update the index place on
122 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
124 ** Bit-array operations (BIT-AND and similar) worked incorrectly
125 with one-dimensional arrays with fill pointers.
126 ** TYPE-OF failed on a complex with an integer realpart and a
128 ** compiler failure during type inference for the code of form
129 (IF (EQL X (THE ...)) ...) (MISC.535).
131 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
132 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
133 platform. The port supports 61-bit fixnums, large memory spaces
134 and reloading shared object files.
135 * enhancement: saving cores with foreign code loaded is now
136 supported on ppc/Darwin in addition to the previously supported
138 * enhancement: the statistical profiler now walks deeper into the
139 call stack for more meaningful call-graphs and accrued time
140 reports (x86/x86-64 only). It also now reports time spent in
142 * enhancement: it is now possible to trace most individual methods
143 of a generic function in addition to tracing the generic function
145 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
147 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
148 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
149 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
150 produces an error. (thanks to Vincent Arkesteijn)
151 * bug fix: NAMESTRING on pathnames with :WILD components in their
152 directories works correctly. (thanks to Artem V. Andreev)
153 * fixed bug 125: compiler preserves identity of closures. (reported
155 * bug fixed: functions with &REST arguments sometimes failed with
156 "Undefined function" when compiled with (DEBUG 3). (reported by
158 * bug fix: overflow during compiling of setting element of a bit
159 vector with constant index and value. (reported by Timmy Douglas)
160 * build fix: fixed the dependence on *LOAD-PATHNAME* and
161 *COMPILE-FILE-PATHNAME* being absolute pathnames.
162 * on x86 compiler partially supports stack allocation of dynamic-extent
164 * GO and RETURN-FROM do not check the extent of their exit points
165 when compiled with SAFETY 0.
166 * fixed some bugs related to Unicode integration:
167 ** encoding and decoding errors are now much more robustly
168 handled; it should now be possible to recover even from invalid
169 input or output to the terminal. (thanks to Teemu Kalvas)
170 ** provided a first cut at implementing STRING-TO-OCTETS and
171 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
172 ** altered the SB-MD5 contributed module slightly, changing the
173 interface just enough for it to be supportable for builds where
174 lisp characters are not eight bits.
175 * fixed some bugs revealed by Paul Dietz' test suite:
176 ** the FORMATTER-generated functions for ~V[ conditionals require
177 the correct number of arguments.
178 ** READ-FROM-STRING returns the mandated second value when applied
179 to displaced strings.
180 ** the #\Rubout and #\Backspace characters are treated as invalid
181 constituent characters by the tokenizer.
183 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
184 * new feature: reloading changed shared object files with
185 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
186 * new feature: references to foreign variables and functions
187 can now be compiled and loaded before the corresponding shared
188 object file is loaded, as long as the foreign definitions are
189 available at runtime.
190 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
191 supported. (thanks to Dan Debertin)
192 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
193 just keywords. (thanks to Vincent Arkesteijn)
194 * fixed bug #331: structure-class instances corresponding to
195 DEFSTRUCT forms are now created eagerly.
196 * fixed bug #345: backtraces from calls to undefined functions work
197 on x86 as well. Related bug #61 is now also partially fixed on x86
198 (backtraces from throws to unknown catch tags.)
199 * bug fix: lambda-list parsing is now stricter vrt. order and number
200 of lambda-list keywords.
201 * bug fix: as specified by AMOP, an error is signalled if a
202 class-option appears multiple times in a DEFCLASS form. (reported
204 * bug fix: RANDOM can be compiled when the compiler derives the type
205 of its numeric argument as a disjoint set of small integers.
206 (reported by Paul Dietz)
207 * bug fix: starting a core saved with shared objects loaded when
208 those objects are not available no longer causes threaded SBCL to
209 hang. (reported by Sean Ross)
210 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
212 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
213 argument, which must always be NIL. (reported by Kalle Niemitalo)
214 * bug fix: printing 1.0d+23 no longer results in an error.
215 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
216 * bug fix: structure slot setters preserve evaluation order. (thanks
218 * fixed some bugs related to Unicode integration:
219 ** RUN-PROGRAM can allow its child to take input from a Lisp
220 stream. (reported by Stefan Scholl)
221 ** ASDF-INSTALL successfully downloads .tar.gz files in all
222 locales. (reported by Ken Causey)
223 * fixed some bugs revealed by Paul Dietz' test suite:
224 ** INCF, DECF and REMF evaluate their place form as specified in
226 ** FORMATTER expands ~{ ~} iteration directives with V or #
227 parameters correctly.
228 ** FORMATTER deals with the ~@[ ~] conditional directive where the
229 consequent uses no arguments correctly.
230 ** the system has a partial understanding of the (COMPLEX RATIO)
232 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
235 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
236 * new feature: a build-time option (controlled by the :SB-UNICODE
237 keyword feature, enabled by default) for building the system with
238 support for the entire 21-bit character space defined by the
240 * new feature: the system now has rudimentary external-format
241 support; the primary user-visible change at this time is that
242 characters with the high bit set (such as lower-case-e-acute) will
243 print correctly to a terminal in a UTF-8 environment.
244 * minor incompatible change: BASE-CHAR no longer names a class;
245 however, CHARACTER continues to do so, as required by ANSI.
246 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
247 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
248 has been moved to the SB-EXT package (temporarily re-exported from
250 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
251 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
253 * fixed bug #351: better error-handlind and reporting for malformed
255 * fixed bug #350: bignum-printing is now more memory-efficient,
256 allowing printing of very large bignums, eg. (expt 2 10000000).
257 (reported by Bruno Haible)
258 * fixed bug #302: better primitive-type selection for intersection
260 * fixed bug #308: non-graphic characters now all have names, as
261 required. (reported by Bruno Haible)
262 * bug fix: redefining a class with different superclasses now correctly
263 removes it from the direct-subclasses of its previous superclasses.
264 (reported by David Morse)
265 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
266 new value now works. (reported by Bruno Haible)
267 * bug fix: correct canonicalization of multiple non-standard slot
268 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
269 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
270 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
271 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
272 now exists, an signals an error.
273 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
274 during macroexpansion for non-iteration variables as well. (reported
275 by Bruno Haible for CMUCL)
276 * bug fix: Cyclic structures and unprintable objects in compiler
277 messages no longer cause errors. (reported by Bruno Haible)
278 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
279 and READ-FROM-STRING all now return a primary value of NIL if
280 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
281 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
282 (reported by Bruno Haible for CMUCL)
283 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
284 do not designate a condition. (reported by Bruno Haible for
286 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
287 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
288 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
289 permitting resolution in favour of any of the conflicting symbols.
290 (reported by Bruno Haible for CMUCL)
291 * bug fix: EQUAL compiler optimizations is less aggressive on
292 strings which can potentially compare true despite having distinct
293 specialized array element types.
294 * bug fix: unit enumerations can be defined without dividing by
295 zero. (thanks to Vincent Arkesteijn)
296 * FORMAT compile-time argument count checking has been enhanced.
297 (report from Bruno Haible for CMUCL)
298 * a partial workaround for the bug 262: the compiler does not try to
299 inline-expand a local function doing RETURN-FROM from a deleted
300 BLOCK. (thanks to Peter Denno for the bug report and to David
301 Wragg for the simple test case)
302 * fixed some bugs revealed by Paul Dietz' test suite:
303 ** RENAME-PACKAGE allows all package designators as new package
305 ** constraint propagation and lambda variable substitution are
306 more cautious in dealing with partially deleted code.
307 ** compiler sometimes lost reoptimization passes.
308 ** CERROR, given a condition as condition designator, passes its
309 remaining arguments to the continue format control without
311 ** Case-altering FORMAT directives work correctly on non-ASCII
313 ** The REAL type specifier handles bounds outside the floating
314 point ranges without signalling FLOATING-POINT-OVERFLOW.
315 ** Functions with IR1-transformations can create intercomponent
316 references to global functions.
317 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
319 ** FORMAT ~R treats a nil value for its first parameter correctly.
321 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
322 * enhancement: saving cores with foreign code loaded is now
323 supported on x86/NetBSD and sparc/Linux in addition to the previously
325 * bug fix: on some platforms repeated installations caused multiple
326 copies of HTML documentation to be installed -- should not happen
327 any more. (reported by Stefan Scholl)
328 * bug fix: parsing self-recursive alien record types multiple times
329 no longer causes infinite recursion. (reported by Thomas F. Burdick,
330 original patch by Helmut Eller for CMUCL)
331 * bug fix: stack-exhaustion detection works now on NetBSD as well.
332 (thanks to Richard Kreuter)
333 * bug fix: defining classes whose accessors are methods on existing
334 generic functions in other (locked) packages no longer signals
335 bogus package lock violations. (reported by François-René Rideau)
336 * bug fix: special variables as DEFMETHOD parameters no longer have
337 associated bogus type declarations. (reported by David Wragg and
339 * bug fix: read-write consistency on streams of element-type
340 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
341 * bug fix: redefiniton of the only method of a generic function with
342 no DEFGENERIC no longer emits a full WARNING. In addition,
343 redefinition of generic functions with no DEFGENERIC to an
344 incompatible lambda list now signals an error. (thanks to Zach
346 * bug fix: DEFGENERIC now works even when there's a function of the
347 same name in an enclosing lexical environment. (thanks to Zach
349 * fixed compiler failure, caused by instrumenting code during
350 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
351 * optimization: added loop analysis and improved register allocation
352 * fixed some bugs revealed by Paul Dietz' test suite:
353 ** POSITION on displaced vectors with non-zero displacement
354 returns the right answer.
355 ** (SIMPLE-STRING) is a valid type specifier for sequence
357 ** *PRINT-LEVEL* handling for slotless structures is pedantically
359 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
360 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
362 ** The justification version of the FORMAT ~< directive treats
363 non-zero minpad parameter correctly.
365 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
366 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
367 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
368 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
369 the supported interface.
370 * new feature: Single-stepping of code compiled with DEBUG 2 or
371 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
372 * new feature: saving cores with foreign code loaded is now
373 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
374 Timothy Moore's work for CMUCL)
375 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
376 parameters to * instead of NIL if no initform is supplied.
377 (reported by Johan Bockgaard)
378 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
379 correctly when FUNCALL appears as the car of the form. Note:
380 despite this FUNCALL forms are not currently subject to
381 compiler-macro expansion. (port of Raymond Toy's fix for the
382 same from CMUCL, reported by Johan Bockgaard)
383 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
384 (thanks for Teemu Kalvas)
385 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
386 case instead of terminating the iteration (thanks for Julian
387 Squires, Sean Champ and Raymond Toy)
388 * bug fix: incorrect expansion of defgeneric that caused a style
389 warning. (thanks for Zach Beane)
390 * bug fix: slot accessor effective method computation works properly
391 for classes with multiple non-standard applicable methods on
392 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
393 * on x86 compiler supports stack allocation of results of LIST and
394 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
395 CMUCL implementation by Gerd Moellmann)
396 * fixed some bugs revealed by Paul Dietz' test suite:
397 ** FORMAT strings with both the ~<~:;~> form of the justification
398 directive and pretty-printing directives cause an error.
400 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
401 * incompatible change: the internal functions
402 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
403 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
404 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
405 instead of the old functions.
406 * new feature: on platforms where "dladdr" is available foreign
407 function names now appear in backtraces. (based on Helmut Eller's
409 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
410 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
412 * documentation: ASDF manual is now distributed with SBCL.
413 * bug fix: SBCL can now load its contributed modules with REQUIRE
414 even if the system-provided entries have been removed from
415 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
417 * bug fix: backtraces involving undefined functions or assembly
418 routines are more informative. (thanks to Brian Downing)
419 * bug fix: mutually referent alien structures now work correctly.
420 (reported by Rick Taube)
421 * bug fix: structures defined by WITH-ALIEN can be referred to
422 within other definitions in the same WITH-ALIEN.
423 * bug fix: division operators (MOD, TRUNCATE and the like) with
424 constant zero divisors and integer dividends no longer generate
426 * bug fix: provide default methods for INPUT-STREAM-P and
427 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
428 * bug fix: improve the detection and resolution of MOP metacycles
429 (where effective method computation depends on the generic function
430 having its effective method computed). (reported by Bruno Haible)
431 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
432 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
433 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
434 * bug fix: pathnames with relative directory components can be
435 represented relative to default pathnames.
436 * optimization: in taking the GCD of bignums, reduce the two bignums
437 to approximately the same size (using Euclid's algorithm) before
438 applying the more sophisticated binary GCD. (thanks to Juho
440 * optimization: COUNT on bitvectors now operates word-at-a-time.
441 * optimization: ASH with a positive, but not necessarily constant,
442 (leftwards) shift, when in a modular context, is compiled to a
444 * fixed some bugs revealed by Paul Dietz' test suite:
445 ** FORMAT variable parameters ("~V<char>") are defaulted properly
446 if the corresponding argument is NIL.
447 ** FORMAT directives accepting a minpad parameter treat negative
449 ** literal commas as character parameters to FORMAT directives are
451 ** literal spaces directly after ~<Newline> directives within a
452 format-logical-block (~:< ~@:>) do not induce :FILL-style
453 conditional newlines.
454 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
455 as specified: it no longer includes conditional newlines).
456 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
458 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
459 :PER-LINE-PREFIX argument does not evaluate to a string.
461 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
462 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
463 the manual for details; for now, package locks can be disabled by
464 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
465 if no major problems are found then it is likely that they will be
466 compiled in unconditionally.
467 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
468 unsupported operators on all platforms. To load a shared library
469 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
470 object file, link it into a shared library outside of SBCL and
471 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
473 * bug workaround: ROOM T and the GC were not working together
474 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
475 reporting facility aren't preserved. That reporting has been
476 disabled, so now until and unless someone figures out how to make it
477 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
478 * minor incompatible change: as threatened around sbcl-0.8.0, the
479 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
480 The symbols are also exported from SB-PCL for backwards
481 compatibility, but more so than before SB-PCL should be treated as
482 an implementation-internal package.
483 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
485 * the SB-SPROF contrib now works on (most) non-x86 architectures.
486 It is known as of this release not to work on the Alpha, however.
487 * fixed bug #167: errors signalled due to illegal syntax in method
488 bodies are now more legible.
489 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
490 designators and can hence be used with TYPEP.
491 * fixed bug #333: CHECK-TYPE now ensures that the type error
492 signalled, if any, has the right object to be accessed by
493 TYPE-ERROR-DATUM. (reported by Tony Martinez)
494 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
495 5.1.2.3 for multiple-value place subforms. (reported by Kalle
497 * fixed bug #334: programmatic addition of slots using specialized
498 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
499 and :ALLOCATION :CLASS slots.
500 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
501 float exponents. (rereported by Peter Seibel)
502 * fixed a bug: #\Space (and other whitespace characters) are no
503 longer considered to be macro characters in standard syntax by
505 * fixed bug: initialization of condition class metaobjects no longer
506 causes an instance of the condition to be created. (reported by
508 * fixed bug: it is now possible to have slots such that
509 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
510 description is neither :INSTANCE nor :CLASS.
511 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
512 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
513 SLOT-VALUE-USING-CLASS), and no longer causes errors with
514 non-standard SLOT-DEFINITION-ALLOCATION values.
515 * fixed bugs: various race conditions handling exiting threads.
516 CL-PPCRE's thread torture test now passes.
517 * fixed arguably-a-bug: GC time proportional to number of threads in
518 system even when most of them are idle
519 * optimization: improved performance of BIT and SBIT on bit-vectors.
520 * .fasl file incompatibility: The fasl file version number has
521 been incremented because of changes associated with package locks.
523 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
524 * minor incompatible change: the system no longer provides
525 optimization hints (conditions which are TYPEP
526 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
527 mode; these hints are still emitted when higher SPEED optimization
529 * new contrib module: a sampling profiler (profiling by statistical
530 sampling, rather than by instrumenting functions) is available as
531 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
532 * the behaviour of the standard function ED is now customizeable by
533 third parties through a hook variable: see ED's documentation
534 string for information on the protocol.
535 * the compiler no longer emits efficiency notes for (FUNCALL X)
536 when the type of X is uncertain under default optimization
538 * fixed bug 276: mutating a binding of a specialized parameter to a
539 method to something that is not TYPEP the specializer is now
541 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
542 exhibit the required behaviour on the x86 platform. (thanks to
543 Peter van Eynde, Eric Marsden and Bruno Haible)
544 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
545 even for difficult arguments. (reported by Peter Graves)
546 * fixed bug 141a: the backquote printer now descends quoted
548 * fixed another bug in backquote printing: no more destructive
549 modification of the form's list structure. (reported by Brian
551 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
552 move between its address being taken and the call to
553 interrupt_thread, fixing a crashing race condition.
554 * the SB-POSIX contrib implementation has been adjusted so that it
555 no longer exhibits ridiculously poor performance when constructing
556 instances corresponding to C structs.
558 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
559 * minor incompatible change: the sb-grovel contrib now treats C
560 structures as alien (in the sense of SB-ALIEN) objects rather than
561 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
562 has implications for memory management of client code
563 (sb-grovel-returned objects must now be manually managed) and for
564 type safety (alien objects now have full types).
565 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
566 used to control emission of compiler diagnostics, rather than the
567 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
568 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
569 quality should be considered deprecated.
570 * install.sh now installs the user manual as well
571 * (not quite a new documentable feature, but worth considering in
572 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
573 beginnings of a semantically meaningful condition hierarchy is
574 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
576 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
577 designator as the defaults argument.
578 * fixed bug: Displaced arrays whose displaced-to array has become
579 too small now cause ARRAY-DIMENSION to signal an error, providing
580 for safer bounds-checking. (reported by Bruno Haible)
581 * fixed bug: DEFCLASS slot definitions with identical :READER and
582 :WRITER names now signal a reasonable error. (reported by Thomas
584 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
586 * fixed bug: Invalid slot specification errors now print correctly.
587 (thanks to Zach Beane)
588 * fixed bug 320: Shared to local slot value transfers in class
589 redefinitions now happen corrently from superclasses as
590 well. (reported by Bruno Haible)
591 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
593 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
594 improper lists correctly. (reported by Bruno Haible)
595 * fixed bug 313: source-transform for <fun-name> was erroneously
596 applied to a call of a value of a variable with name <fun-name>.
597 (reported by Antonio Menezes Leitao)
598 * fixed bug 307: The obsolete instance protocol ensures that
599 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
600 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
601 remove dead unknown-values globs from the middle of the stack.
602 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
603 easier to resume long computations after using *BREAK-ON-SIGNALS*
604 to diagnose and fix failures. (thanks to Nikodemus Siivola)
605 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
606 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
607 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
608 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
609 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
611 * fixed a (fairly theoretical) bug in string printing: if
612 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
613 does not have array-element-type equal to the most general string
615 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
616 function name. (thanks to Nikodemus Siivola)
617 * fixed bug: compile-time format string checker failed on
618 non-closed ~{. (reported by Thomas F Burdick)
619 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
620 don't warn on legal loop code involving more than one aggregate
621 boolean. (thanks to Nikodemus Siivola)
622 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
623 in keyword position to MAKE-PACKAGE was not being reported as
625 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
626 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
628 * fixed bugs 280 and 312: the checking for multiple definitions in a
629 file is less likely to become confused by uses of inline
631 * fixed bug: the #S reader macro performs the keyword coercion
632 specified for slot names. (reported by Kalle Niemitalo)
633 * fixed bug: lambda lists may contain symbols whose names start with
634 & but are not lambda-list-keywords; their occurrence triggers a
636 * fixed bug 321: define-method-combination argument lambda lists do
637 not cause invalid code to be generated when &optional or &aux
638 variables have default values. (reported by Bruno Haible)
639 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
640 have CLOS classes; this appears also to have fixed reported
641 instability in stack exhaustion detection.
642 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
643 MUFFLE-WARNING when no associated restart is present is now
645 * optimization: rearranged the expansion of various defining macros
646 so that each expands into only one top-level form in a
647 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
649 * optimization: used a previously little-used slot in symbols to
650 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
652 * fixed some bugs revealed by Paul Dietz' test suite:
653 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
655 ** Fixed an optimization bug related to inheritance of initforms
656 from local to shared slots.
657 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
658 ** CAST optimizer forgot to flush argument derived type.
659 ** print/read consistency on floats is now orders of magnitude
660 more likely. (thanks also to Bruno Haible for a similar report
662 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
663 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
664 for objects of type REAL. Make it so.
665 ** SXHASH is specified (infelicitously) to respect similarity,
666 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
667 it so. (thanks to Markus Ziegler)
668 ** on the Alpha, the compiler succeeds in compiling functions
669 returning a known number of arguments greater than 63.
670 ** fixed handling of invalid NIL arguments in keyword position
672 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
674 ** provided a readably-printable representation for RANDOM-STATE
676 ** ensured that pathnames, where they have a namestring, always
677 print using #P"..." syntax.
679 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
680 * Support for the forthcoming 2.0 version of the NetBSD kernel
681 running on x86 hardware has been added. (thanks to Perry
682 E. Metzger most immediately, and others for their past work)
683 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
684 current release version; SBCL's OpenBSD support had been broken
685 since about the time of OpenBSD's switch to ELF binary format.)
686 * [placeholder for DX summary]
687 ** user code with &REST lists declared dynamic-extent, under high
688 speed or space and low safety and debug optimization policy.
689 * The manual has been converted to Texinfo format and the debugger
690 chapter from the cmucl manual has been added.
691 * A facility has been added to extract documentation strings from
692 sbcl and store them as Texinfo-formatted snippets for inclusion in
693 the manual (via Texinfo's @include directive)
694 * bug fix: compiler emitted division in optimized DEREF. (thanks for
695 the test case to Dave Roberts)
696 * bug fix: multidimensional simple arrays loaded from FASLs had fill
697 pointers. (reported by Sean Ross)
698 * bug fix: PROFILE output is printed nicely even for large numerical
699 values. (thanks to Zach Beane)
700 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
701 greater than 32 handle EOF correctly.
702 * bug fix: on X86 an immediate argument of the IMUL instruction is
703 correctly printed in disassembly. (thanks to Lutz Euler)
704 * bug fix: class slots in redefined classes preserve their old
705 values. (thanks to Bruno Haible and Nikodemus Siivola)
706 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
707 succeeds. (reported by Marco Baringer)
708 * fixed some bugs revealed by Paul Dietz' test suite:
709 ** READ-SEQUENCE now works on ECHO-STREAMs.
710 ** RATIONALIZE works more according to its specification. (thanks
713 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
714 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
715 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
716 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
717 should matter to you only if you rebind the printer control
718 variables and then find you want different bindings in the
719 debugger than in the ordinary execution of your program.)
720 * The runtime build system has been tweaked to support building
721 (on SPARC/SunOS) using a C compiler which invokes Sun's own
722 assembler and linker. (thanks to Nikodemus Siivola)
723 * Unbound, undefined, undeclared variables now trigger full
724 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
725 more often programmer error than deliberate exploitation of undefined
727 * optimization: the hash algorithm for strings has changed to one
728 that is less vulnerable to spurious collisions. (thanks to Juho
730 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
731 needless bounds checking. (thanks to Juho Snellman)
732 * optimization: implemented multiplication as a modular
733 (UNSIGNED-BYTE 32) operation on the PPC backend.
734 * fixed some bugs revealed by Paul Dietz' test suite:
735 ** ADJUST-ARRAY now copies the datum in a zero rank array if
737 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
738 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
739 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
741 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
743 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
745 ** ECHO-STREAMs no longer attempt to echo the end of file value to
746 their output stream on EOF from read.
747 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
748 have been read to end-of-file.
749 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
751 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
752 description of determination of which consecutive characters
754 ** Printing the "Space" character with escaping on now yields "#\\ ",
755 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
756 ** Reading floating-point numbers with *READ-BASE* set to a number
757 less than 10 works correctly.
758 ** Reading floating-point numbers with *READ-BASE* set to a number
759 more than 10 works correctly.
760 ** Printing with *PRINT-READABLY* targets the standard readtable, not
761 the readtable currently in effect.
763 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
764 * minor incompatible change: parsing of namestrings on a physical
765 (Unix) host has changed; numbers after the final #\. in a
766 namestring are no longer interpreted as a version field. This is
767 intented to be largely invisible to the user, except that the
768 meaning of the namestring "*.*.*" has changed: it now refers to a
769 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
770 should usually be replaced by
771 (make-pathname :name :wild :type :wild :version :wild)
772 with the added benefit that this is more likely to be portable.
773 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
774 option to OPEN now signals an error if the file being opened
775 exists; this may have an impact on existing code.
776 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
777 signals received, on the PowerPC platforms (both Linux and
778 Darwin). (thanks to Pierre Mai for pointing out the location of
780 * several fixes on OS X: The system now builds and runs cleanly on
781 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
782 (thanks to Brian Mastenbrook)
783 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
784 second-resolution: integer multiples of 1/3600 between -24 and 24.
785 (thanks to Vincent Arkesteijn)
786 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
787 type when called with 1 argument; PEEK-CHAR checked type of
788 PEEK-TYPE only after having read first character from a
789 stream. (reported by Peter Graves)
790 * bug fix: the garbage collector now has much better locality
791 behaviour, and in particular no longer treats all memory as being
792 exhausted when many small objects point to each other in a deeply
794 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
795 recognized as being TYPEP their class.
796 * bug fix: the PUSHNEW documentation string has been corrected.
797 (thanks to Vincent Arkesteijn)
798 * bug fix: defaulting of the value for the last of an atypically
799 large number of multiple values being bound was not being
800 performed correctly on the Alpha or PPC platforms
801 * optimization: implemented multiplication as a modular
802 (UNSIGNED-BYTE 32) operation on the x86 backend.
803 * optimization: SEARCH on simple-base-strings can now be open-coded.
804 (see also contrib/compiler-extras.lisp for inspiration for
805 teaching the compiler about the Boyer-Moore algorithm).
806 * value, returned by MAX (and MIN) called with several EQUALP, but
807 not EQL, arguments now does not depend on compiler settings.
808 * fixed some bugs revealed by Paul Dietz' test suite:
809 ** in stack analysis liveness information is propagated from
810 non-local entry points.
811 ** pathwise CAST removing failed when the CAST node did not start
813 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
814 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
816 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
817 host is already defined.
818 ** RENAME-FILE works on streams instead of signalling an internal
820 ** PEEK-CHAR uses the current readtable when determining whether
821 or not a character is whitespace.
822 ** MERGE-PATHNAMES handles the case when the pathname does not
823 specify a name while the default-pathname specifies a version.
824 ** Pathnames now stand a better chance of respecting print/read
826 ** Attempting to use standardized file system operators with a
827 pathname with invalid :DIRECTORY components signals a
829 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
831 ** DIRECTORY on logical pathnames is more correct.
832 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
833 signal a TYPE-ERROR if their argument is not a stream.
834 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
835 designator argument does not designate a stream.
836 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
837 examining the synonym.
838 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
840 ** OPEN and WITH-OPEN-STREAM allow opening streams with
841 element-type larger than ([UN]SIGNED-BYTE 32).
843 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
844 * When built with the :SB-FUTEX feature, threaded builds now take
845 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
846 for faster/more reliable mutex and condition variable support.
847 * Incompatible change (but one you probably shouldn't have been using
848 anyway): the interface and code for arbitrating between multiple
849 threads in the same user session has been redesigned.
850 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
851 error when called without an explicit environment argument.
852 (thanks to Nikodemus Siivola)
853 * bug fix: buffered :DIRECTION :IO streams are less likely to become
854 confused about their position. (thanks to Adam Warner and Gerd
856 * bug fix: Pretty printing backquoted forms with unquotations in the
857 argument list position of various code constructs such as LAMBDA
858 now works correctly. (reported by Paul Dietz)
859 * bug fix: Pretty printing unquotations no longer loses all
860 stream position information.
861 * optimization: performance of string output streams is now less
862 poor for multiple small sequence writes.
863 * optimization: performance of CSUBTYPEP in the presence of complex
864 expressions involving CONS and NOT many times has been improved.
865 (reported by Paul Dietz)
866 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
868 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
869 streams for the REPL, for the debugger, and so on. (thanks to
871 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
872 * fixed some bugs revealed by Paul Dietz' test suite:
873 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
874 is ignored if it is too small, rather than propagating through
875 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
876 ** extremely complex negations of CONS types were not being
877 sufficiently canonicalized, leading to inconsistencies in
879 ** VALUES tranformer lost derived type.
881 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
882 * fixed a bootstrapping bug: the build process no longer assumes
883 that the various BOOLE-related constants have the same value in
884 host and target lisps. (noted by Paul Dietz' test suite on an
885 SBCL binary built from CLISP)
886 * The system can now be dynamically linked on the MIPS platform,
887 which enables dynamic loading of foreign code from Lisp. (thanks
888 to Ralf Baechle for discussions on the MIPS ABI)
889 * The system now records debugging information for its own source
890 files in a filesystem-position-independent manner, relative to
891 the "SYS" logical host.
892 * fixed a compiler bug: MV-LET convertion did not check references
893 to the "max args" entry point. (reported by Brian Downing)
894 * tweaked disassembly notes to be less confident about proclaiming
895 some instruction as an LRA. (thanks to Brian Downing)
896 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
897 now each have their own history, command character, and other
898 characteristics. (thanks to David Lichteblau)
899 * fixed some bugs revealed by Paul Dietz' test suite:
900 ** compiler failure in compiling LOGAND expressions including a
902 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
903 shift greater than 32.
904 ** FLUSH-DEST did not mark blocks for type check regeneration.
905 ** HANDLER-CASE failed to accept declarations in handler clauses
906 in some circumstances.
908 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
909 * New code in contrib/sb-introspect (still probably not entirely
910 stable yet) provides some support for smart Lisp development
911 environments like SLIME.
912 * The conditions signalled for errors occurring when loading .fasl
913 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
914 in a way which should help ASDF recover gracefully.
915 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
916 stuff has been cleaned up. If you code contrib/ stuff, this might
917 affect you, and you can look at contrib/README, contrib/STANDARDS,
918 and/or the 0.8.4.27 diff to check.
919 * In full calls the compiler now does not generate checks for declared
920 argument types for all arguments.
921 * various threading fixes
922 ** and some experimental patches which didn't make it into
923 the main tree for this release, but which are shipped in
924 contrib/experimental-thread.patch as a possible fix for some
925 failures (deadlock, spinning...) in GC-intensive multithreaded
927 * fixed PPC build problem (source code incompatibility of different
928 library versions): added offsetof() hackery which attempts to divine
929 where glibc maintainers put uc_mcontext today
930 * fixed bug 282: compiler does not trust type assertions while passing
931 arguments to a full call.
932 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
933 &OPTIONAL VALUES type parameter.
934 * fix bug 214: algorithm for noting rejected templates is now more
935 similar to that of template seletion. (also reported by rydis on
937 * fixed bug 141b: printing backquoted information readably and prettily
938 inserts a space where necessary.
939 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
940 (CHARACTER 10) are now reported as errors, rather than propagated
941 as unknown types. (reported by piso on #lisp)
942 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
943 with values NIL and :ERROR. (thanks to Milan Zamazal)
944 * fixed bug 191c: CLOS now does proper keyword argument checking as
945 described in CLHS 7.6.5 and 7.6.5.1.
946 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
947 longer raise an error; further, using a list as a for-as-arithmetic
948 counter now raises a meaningful error.
949 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
950 now understood by sequence creation functions such as MAKE-SEQUENCE
952 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
953 a STRING-INPUT-STREAM.
954 * compiler enhancement: SIGNUM is now better able to derive the type
956 * type declarations inside WITH-SLOTS are checked. (reported by
958 * fixed some bugs revealed by Paul Dietz' test suite:
959 ** incorrect optimization of TRUNCATE for a positive first
960 argument and negative second.
961 ** compiler failure in let-convertion during flushing dead code.
962 ** compiler failure while deriving type of TRUNCATE on an
963 interval, containing 0.
964 ** ASH of a negative bignum by a negative bignum count now returns
966 ** intersection of CONS types now canonicalizes properly, fixing
967 inconsistencies in SUBTYPEP.
969 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
970 * incompatible change: The --disable-debugger command line
971 option now clobbers the debugger at a more fundamental
972 level, by redefining #'INVOKE-DEBUGGER instead of by
973 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
974 is specified by ANSI to ignore *DEBUGGER-HOOK* and
975 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
976 enter the debugger REPL and then suffer recursive errors
977 because *DEBUG-IO* is also messed up in --disable-debugger mode;
978 while under the new system, BREAK in --disable-debugger mode
979 terminates the system just as an unhandled error would.
980 * fixed compiler performance when processing loops with a step >1;
981 * bug fix: DOCUMENTATION now retrieves generic function
982 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
983 support has been systematized, and now supports the methods
984 specified by ANSI, along with a default method and a method for
985 slot documentation. (reported by Nathan Froyd)
986 * bug fix: effective methods associated with a generic function are
987 no longer cached over a change of that generic function's method
988 combination. (reported by Andreas Fuchs)
989 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
990 names a non-existent directory. (thanks to Andreas Fuchs)
991 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
992 the PPC platform, lead to stack corruption; this has been fixed.
993 (reported by Rainer Joswig)
994 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
995 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
996 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
997 arithmetic in general, is now much more reliable on the PPC
999 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1000 platform now returns the right answer.
1001 * optimization: restored some effective method precomputation in
1002 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1003 precomputation is now tunable.
1004 * optimization: compiler-internal data structure use has been
1005 reviewed, and changes have been made that should improve the
1006 performance of the compiler by about 20%.
1007 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1008 simple-base-strings and simple-bit-vectors is improved.
1009 * optimization: the optimization of 32-bit logical and arithmetic
1010 functions introduced in version 0.8.3 on the x86 has been
1011 implemented on the mips, ppc and sparc platforms; an
1012 implementation of the same facility, but for 64-bit arithmetic,
1013 has been added for the alpha.
1014 * microoptimization: the compiler is better able to make use of the
1015 x86 LEA instruction for multiplication by constants.
1016 * bug fix: in some situations compiler did not report usage of
1017 generic arithmetic in (SPEED 3) policy.
1018 * bug 145b fix: compiler used wrong type specifier while converting
1019 MEMBER-types to numeric.
1020 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1022 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1023 index argument now works properly on the Alpha platform.
1024 * bug fix: floating point exception treatment on the Alpha platform
1026 * bug fix: FILE-POSITION works much better on string input and
1027 output streams. (thanks to Nikodemus Siivola)
1028 * bug fix: many threading/garbage collection symptoms sorted.
1029 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1030 might be pseudo-atomic.
1031 * internal change: Stopping for GC is now done with signals not
1032 ptrace. GC is now done in whichever thread wanted it, instead of
1034 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1036 * bug fix: result form in DO is not contained in the implicit
1038 * incompatible change: ICR structure is changed; the value part of
1039 CONTINUATION is now called LVAR; corresponding functions are
1040 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1041 * added type deriver for ISQRT (thanks to Robert E. Brown).
1042 * bug fix: better support for loading from the command line when an
1043 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1045 * fixed some bugs revealed by Paul Dietz' test suite:
1046 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1047 ** ROUND and FROUND now give the right answer when given very
1048 small float arguments.
1049 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1051 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1052 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1053 ** LCM should return a non-negative integer.
1054 ** PARSE-INTEGER returned the index of a terminator instead of the
1055 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1056 ** PARSE-INTEGER returned an incorrect index being applied to a
1058 ** LCM with two arguments of 0 returns 0 rather than signalling
1060 ** unsigned addition of a 32-bit constant with the high bit set no
1061 longer causes an internal compiler error on the x86.
1062 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1063 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1064 with negative last argument.
1065 ** byte specifiers with zero size and position no longer cause
1066 an error during type derivation.
1067 ** bignum multiplication on the Alpha platform now returns the
1069 * porting: The system now builds on SuSE AMD64, although it still
1070 generates a 32-bit binary.
1071 * .fasl file incompatibility: The fasl file version number has
1072 been incremented (because of the changes to internal compiler
1073 data structures referred to above).
1075 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1076 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1077 more accurately, on the Darwin kernel running on PowerPC hardware.
1078 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1079 * Compiler code deletion notes now signal a condition of type
1080 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1081 an associated MUFFLE-WARNING restart.
1082 * The compiler now performs limited argument count validation of
1083 constant format strings in FORMAT, and where appropriate in ERROR,
1084 CERROR and WARN. (thanks to Gerd Moellmann)
1085 * New ASDF-INSTALL contrib can be used for automatic download and
1086 installation of third-party Lisp code from CCLAN or other sites
1088 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1089 which forces another thread to execute a function supplied by the
1091 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1092 now accept and act upon their :ELEMENT-TYPE keyword argument.
1093 (reported by Martin Atzmueller, Edi Weitz)
1094 * bug fix: FILE-POSITION now accepts position designators up to
1095 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1096 is the greater. (thanks to Patrik Nordebo)
1097 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1098 Kalle Olavi Niemitalo)
1099 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1100 printable. (reported by Eric Marsden)
1101 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1102 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1103 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1104 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1105 (thanks to Stig Sandoe)
1106 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1107 UNREAD-CHAR now correctly includes the unread character in the
1108 target sequence. (thanks to Gerd Moellmann)
1109 * bug fix in threaded builds: the system can now be suspended and
1110 resumed by shell job control with minimal disruption.
1111 * bug fixes in times and timezones >2038AD
1112 * better handling of "where is GNU make?" problem in build scripts
1113 (thanks to Nikodemus Siivola)
1114 * new optimization: inside a named function any reference to a
1115 function with the same name is considered to be a self-reference;
1116 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1117 optimization quality.
1118 * new optimization on x86: logical functions and + now have
1119 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1120 used when the result is truncated to 32 bits.
1121 * VALUES declaration is partially enabled.
1122 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1123 * fixed some bugs revealed by Paul Dietz' test suite:
1124 ** The system now obeys the constraint imposed by
1125 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1126 types form a lattice under type intersection.
1127 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1128 ** ASSOC now ignores NIL elements in an alist.
1129 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1130 and (1+ MOST-POSITIVE-FIXNUM) answers.
1131 ** The addition of a method with invalid qualifiers to a generic
1132 function does not cause an error to be signalled immediately;
1133 a warning is signalled, and the error is generated only on
1134 calling the generic function.
1135 * changed .fasl file version number, in part to add type codes for
1136 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1137 obscure ANSI requirements
1139 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1140 * fixed bug 148: failure to inline-expand a local function left
1141 garbage, confusing the compiler.
1142 * fixed bugs 3cd: structure slot readers perform type check if the
1143 slot can have an invalid value (i.e. it is either not initialized
1144 or can be written with a less specific slot writer).
1145 * bug fix: the compiler now traps array references to elements off
1146 the end of an array; previously, the bounds checking in some
1147 circumstances could go off-by-one.
1148 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1150 * type declarations for array element types now obey the description
1151 on the CLHS page "Declaration TYPE", as per discussions on
1152 sbcl-help around 2003-05-08. This means that a declaration
1153 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1154 declaration, all references to BAR will be asserted or assumed
1155 (with THE, so dependent on compiler policy) to involve objects of
1156 type FOO. Note that no such declaration is implied in
1157 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1158 * declared types of functions from the "Conditions"
1159 chapter. (reported by Paul Dietz)
1160 * bug fix: CERROR accepts a function as its first argument.
1161 * bug fix: NTH an NTHCDR accept a bignum as index
1162 arguments. (reported by Adam Warner)
1163 * optimization: character compare routines now optimize comparing
1164 against a constant character. (reported by Gilbert Baumann)
1165 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1166 argument now works properly on the MIPS platform.
1167 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1168 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1169 declared to have a functional type, does not cause an error
1171 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1172 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1173 * fixed bug: if last continuation of a deleted block has a
1174 destination, this destination should be deleted too. (reported by
1176 * fixed a bug in the bootstrap process: the host compiler's values
1177 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1178 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1179 test case from Patrik Nordebo)
1180 * improved the ability of the disassembler on the PPC platform to
1181 provide helpful disassembly notes.
1182 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1183 the class in more cases than previously.
1184 * bug fix: FILE-POSITION now understands :START and :END for
1185 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1186 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1187 8) data. (thanks to David Lichteblau)
1188 * bug fix: it is possible to add a method to a generic function
1189 without lambda list.
1190 * bug fix: reader failed to signal END-OF-FILE inside an
1191 object representation. (reported by Nikodemus Siivola)
1192 * fixed some bugs revealed by Paul Dietz' test suite:
1193 ** LAST and [N]BUTLAST should accept a bignum.
1194 ** condition slot accessors are methods.
1195 ** (VECTOR NIL) is a subtype of STRING.
1197 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1198 * minor incompatible change: some nonsensical specialized lambda
1199 lists (used in DEFMETHOD) which were previously ignored now signal
1201 * minor incompatible change: the system is now aware of the types of
1202 variables in the COMMON-LISP package, and will signal errors for
1203 most violations of these type constraints (where previously they
1204 were silently accepted).
1205 * minor incompatible change: COMPILE-FILE now uses the freedom
1206 afforded (ANSI 3.2.2.3) to use derived function types for
1207 functions defined in the same file. This also permits the system
1208 to warn on static type mismatches and function
1209 redefinition. (Currently it does not work with high DEBUG level.)
1210 * minor incompatible change: VALUES declaration is disabled.
1211 * When issuing notes, the compiler now signals a condition of type
1212 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1213 restart for use in user handlers. It is expected that the
1214 COMPILER-NOTE condition will eventually become a condition
1215 supertype to a hierarchy of note types, which will then be
1216 handleable in a similar fashion. However, at the moment, no such
1217 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1218 but it's an implementation detail, not a classification for the
1220 * Changes in type checking closed the following bugs:
1221 ** type checking of unused values (192b, 194d, 203);
1222 ** template selection based on unsafe type assertions (192c, 236);
1223 ** type checking in branches (194bc).
1224 * A short form of VALUES type specifier has ANSI meaning (it has
1225 increased the number of situations when SBCL cannot perform type
1227 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1228 keywords or constants is permissible.
1229 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1230 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1231 classes. (thanks to Antonio Martinez)
1232 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1233 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1234 argument to a pretty-print stream. Instead, it leaves any such
1235 operations to DESCRIBE-OBJECT methods.
1236 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1238 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1239 lambda lists. (thanks to David Lichteblau)
1240 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1241 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1242 (thanks to Gerd Moellmann)
1243 * fixed bug 246: increased compilation speed of long
1244 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1246 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1248 * DEFCONSTANT now throws a condition of type
1249 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1250 constant to a non-EQL value; CONTINUE and ABORT restarts
1251 respectively change and preserve the value.
1252 * fixed bug 63: The code walker, part of the implementation of CLOS,
1253 is now better at handling symbol macros.
1254 * bug fix: There is no longer an internal implementation type named
1255 CL:LENGTH. (reported by Raymond Toy)
1256 * bug fix: In macro-like defining macros/special operators the
1257 implicit block does not enclose the lambda list.
1258 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1259 enclosed in parentheses) are not suitable as type specifiers, and
1260 their use properly signals an error now.
1261 * bug fix: An argument count mismatch for a type specifier in code
1262 being compiled no longer causes an unhandled error at compile
1263 time, but signals a compile-time warning.
1264 * fixed simple vector readable printing
1265 * bug fix: DESCRIBE takes more care over whether the class
1266 precedence list slot of a class is bound before accessing it.
1267 (reported by Markus Krummenacker)
1268 * bug fix: FORMATTER can successfully compile pretty-printer format
1269 strings which use variants of the ~* directive inside.
1270 * bug fix: SEARCH now applies its TEST predicate to the elements of
1271 the arguments in the correct order. (thanks to Wolfhard Buss)
1272 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1273 in type inference. (thanks to Robert E. Brown)
1274 * bug fix: Reading in symbols with an explicit package name of ""
1275 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1276 * fixed some bugs revealed by Paul Dietz' test suite:
1277 ** NIL is now allowed as a structure slot name.
1278 ** Arbitrary numbers, not just REALs, are allowed in certain
1279 circumstances in LOOP for-as-arithmetic clauses.
1280 ** Multiple class redefinitions before slot access no longer
1281 causes a type error.
1282 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1283 association between the name and a class.
1284 ** Generic functions with non-standard method-combination and over
1285 five methods all of which return constants no longer return NIL
1286 after the first few invocations. (thanks to Gerd Moellmann)
1287 ** CALL-NEXT-METHOD with no arguments now passes the original
1288 values of the arguments, even in the presence of assignment.
1289 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1291 ** DISASSEMBLE works with closures and funcallable instances.
1292 ** ADD-METHOD now returns the generic function, not the new method.
1293 ** FIND-METHOD signals an error if the lengths of the specializers
1294 is incompatible with the generic function, even if the ERRORP
1296 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1297 which its argument is a member.
1298 ** DEFCLASS only redefines the class named by its class-name
1299 argument if that name is the proper name of the class;
1300 otherwise, it creates a new class.
1301 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1302 of the UNBOUND-SLOT condition to the name of the slot.
1303 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1304 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1305 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1306 SLOT-MAKUNBOUND in the specified fashion.
1308 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1309 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1310 cross-compilation host. As a consequence, we can now bootstrap our
1311 way up to SBCL starting with a bare gcc toolchain and human-readable
1312 source code (first the source to CLISP, then the source to SBCL).
1313 * A contributed module containing a partial implementation of the
1314 simple-streams interface has been included. (thanks to Rudi
1316 * A contributed module implementing the RFC1321 Message Digest
1317 Algorithm, known as MD5, has been included.
1318 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1319 no longer has any effect, as the code controlled by this feature
1320 has been deleted. (As far as we know, no-one has ever built using
1321 this feature, and its semantics were confused in any case).
1322 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1323 work on conditions (as required by the ANSI specification),
1324 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1325 expected behaviour on conditions. Users should note, however,
1326 that such behaviour is not required by the ANSI specification,
1327 and so use of this behaviour may render their code unportable.
1328 * fixed some bugs revealed by Paul Dietz' test suite:
1329 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1331 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1332 ** SXHASH on condition objects no longer returns NIL.
1333 ** :ALLOCATION :CLASS slots are better treated; their values are
1334 updated on class redefinition, and initforms inherited from
1335 superclasses are applied.
1336 ** REMOVE-METHOD returns its generic function argument even when
1337 no method was removed.
1338 ** SHARED-INITIALIZE now initializes the values of the requested
1339 slots, including those with :ALLOCATION :CLASS.
1340 ** ALLOCATE-INSTANCE now works on structure classes defined via
1341 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1343 ** SLOT-EXISTS-P now works on conditions, as well as structures
1345 ** MAKE-LOAD-FORM now has the required methods on
1346 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1347 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1348 keyword, and now implements the SLOT-NAMES argument.
1349 ** methods with &OPTIONAL arguments no longer allow too many
1350 arguments to be passed in the call without error.
1351 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1352 option is consistent with the required arguments of the generic
1353 function lambda list.
1354 * bug fix: REQUIRE accepts a string designator. (Thanks to
1356 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1357 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1358 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1359 * bug fix: The compiler now checks for duplicated variables in macro
1361 * bug fix: SETQ on globals returns the correct value.
1362 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1363 not a COMPILER-ERROR (followed by some other strange error on
1364 choosing the CONTINUE restart).
1365 * bug fix: make.sh and friends are now more consistent in the way that
1366 they look for GNU "make".
1368 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1369 * experimental native threads support (on x86 Linux >=2.4 only).
1370 This is not compiled in by default: you need to add :SB-THREAD to
1371 the target features. See the "Beyond ANSI" chapter of the manual
1373 * fix for longstanding nonANSIism: The old distinction between
1374 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1375 The return value from CL:FIND-CLASS is now a CLOS class, and
1376 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1377 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1378 * An interface to the MetaObject Protocol, as described in Kiczales,
1379 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1380 MIT Press, 1991, is available from the SB-MOP package.
1381 * incompatible change: the SB-PCL package should now be considered
1382 a private implementation detail, and no longer a semi-private MOP
1384 * minor incompatible change: due to rearrangement for threads, the
1385 control stack and binding stack are now allocated at arbitrary
1386 addresses instead of being hardcoded per-port. Users affected by
1387 this probably have to be doing advanced things with shared
1388 libraries, and will know who they are.
1389 * minor incompatible change: Previously, all --eval forms used were
1390 processed with READ before any of them were processed with EVAL.
1391 Now each --eval form is processed with both READ and EVAL before
1392 the next --eval form is processed. (Thus package operations like
1393 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1394 work as the user might reasonably expect.)
1395 * minor incompatible change: *STANDARD-INPUT* is now only an
1396 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1398 * minor incompatible change: Y-OR-N-P is now character-oriented,
1399 not line oriented. Also, YES-OR-NO-P now works without errors.
1400 (thanks to Antonio Martinez)
1401 * sb-aclrepl module improvements: an integrated inspector, added
1402 repl features, and a bug fix to :trace command.
1403 * Known functions, which cannot be open coded by the backend, are
1404 considered to be able to check types of their arguments. (fixing
1405 a bug report by Nathan J. Froyd)
1406 * fixed a bug in computing method discriminating functions: It is
1407 now possible to define methods specialized on classes which have
1408 forward-referenced superclasses. (thanks to Gerd Moellmann)
1409 * fixed evaluation order in optional entries (reported by Gilbert
1411 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1412 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1413 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1414 required-by-AMOP NAME argument, as well as CLASS and
1415 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1416 * fixed bug 20: DEFMETHOD can define methods using names that are
1417 not the proper names of classes to designate class specializers.
1418 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1419 Unix file descriptors, instead of blowing up. (thanks to Antonio
1421 * Garbage collection refactoring: user-visible change is that a
1422 call to the GC function during WITHOUT-GCING will not do garbage
1423 collection until the end of the WITHOUT-GCING. If you were doing
1424 this you were probably losing anyway.
1425 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1426 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1427 TYPEP the latter but not the former.
1428 * The compiler issues a full WARNING for calls to undefined functions
1429 with names from the CL package.
1430 * MAP-INTO for a vector destination is open coded. (reported by
1431 Brian Downing on c.l.l)
1432 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1433 documentation string.
1434 * fixed some bugs revealed by Paul Dietz' test suite:
1435 ** COPY-ALIST now signals an error if its argument is a dotted
1437 ** Condition slots are now accessed more correctly in the presence
1438 of multiple initargs for a given slot.
1439 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1440 exclude restarts of the same name associated with a different
1442 ** DEFCLASS of forward-referenced classes with another
1443 forward-referenced class in the superclasses list no longer
1445 ** Condition slots are now initialized once each, not multiple
1446 times. (thanks to Gerd Moellmann)
1447 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1449 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1450 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1451 arguments contain duplicated elements.
1452 ** RESTART-CASE understands local macros.
1453 ** RESTART-CASE associates exactly its own restarts with a condition.
1454 ** ENDP in safe mode checks its argument to be of type LIST.
1455 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1456 in question is unbound.
1457 ** Optimized MAKE-INSTANCE functions no longer cause internal
1458 assertion failures in the presence of duplicate initargs.
1459 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1460 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1461 propagated even in the ctor optimized implementation of
1463 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1465 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1466 * a better implementation of SXHASH on (simple) bit vectors,
1467 measured both in execution speed and in distribution of results
1468 over the positive fixnums, has been installed. Likewise, a better
1469 implementation of EQUAL for simple bit vectors is now available.
1470 * fixed CEILING optimization for a divisor of form 2^k.
1471 * fixed bug 240 (emitting extra style warnings "using the lexical
1472 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1473 by Antonio Martinez)
1474 * fixed SXHASH, giving different results for NIL depending on type
1475 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1476 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1477 variable. (found by Rolf Wester)
1478 * SBCL does not ignore type declarations for special
1479 variables. (reported by rif on c.l.l 2003-03-05)
1480 * some bug fixes in contrib/sb-aclrepl/
1481 * fixed some bugs revealed by Paul Dietz' test suite:
1482 ** a bug in the CONS type specifier, whereby the CAR and CDR
1483 types got intertwined, has been fixed;
1484 ** the type system is now able to reason about the interaction
1485 between INTEGER and RATIO types more completely;
1486 ** APPEND, [N]REVERSE and NRECONC check that those their
1487 arguments, which must be proper lists, are really so;
1488 ** An array specialized to be unable to hold elements has been
1489 implemented, as required -- yes, really -- by ANSI;
1490 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1491 on malformed property lists;
1493 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1494 * incompatible packaging change: in line with Unix convention,
1495 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1496 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1497 * REQUIRE and PROVIDE are now optionally capable of doing something
1498 useful. See the documentation string for REQUIRE.
1499 * infrastructure for a managed SBCL contrib system: contributed
1500 modules in this release include:
1501 ** the ASDF system definition facility;
1502 ** an interface to the BSD Sockets API;
1503 ** an ACL-like convenience interface to the repl;
1504 (thanks to Kevin Rosenberg)
1505 ** an implementation of ROTATE-BYTE, with efficient implementation
1507 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1508 gives a closer approximation to the right answer than previously.
1509 (thanks to Raymond Toy)
1510 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1511 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1512 optional environment arguments, as required by ANSI.
1513 * fixed bugs in other functions taking environment objects, allowing
1514 calls with an explicit NIL environment argument to be compiled
1516 * fixed bug 228: primary return values from
1517 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1518 COMPILE or FUNCTION.
1519 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1520 :TYPE will no longer signal errors on innocuous objects.
1521 * fixed bug 231b: SETQ is better at respecting type declarations in
1522 the lexical environment.
1523 * fixed a bug in DEFCLASS: classes named by symbols with no or
1524 unprintable packages can now be defined.
1525 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1526 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1527 * fixed compiler failure related to checking types of functions.
1528 (reported by Robert E. Brown)
1529 * the compiler is now much more consistent in its error-checking
1530 treatment of bounding index arguments to sequence functions: in
1531 (SAFETY 3) code, errors will be signalled in almost all cases if
1532 invalid sequence bounding indices are passed to functions defined
1533 by ANSI to operate on sequences.
1534 * fixed a bug in the build procedure: documentation of SBCL-specific
1535 packages is now preserved and available in the final Lisp image.
1536 * lifted FDEFINITION lookup out of loops in the implementation of
1537 many list operations. (thanks to Robert E. Brown)
1538 * fixed a bug in the reader: the #n# reader macro now works for
1539 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1540 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1541 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1542 E. Brown and Tony Martinez respectively)
1543 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1544 SLOT-BOUNDP: in particular, it is now able to optimize them much
1545 better, and is now not vulnerable to having packages renamed.
1546 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1547 always called when a slot is not present in an instance. (thanks
1549 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1550 longer signal an error on unreading a character following EOF on
1551 the previous constituent stream. (thanks to Tony Martinez)
1552 * fixed some bugs revealed by Paul Dietz' test suite:
1553 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1554 not just nonnegative fixnums;
1555 ** the logical bit-array operators such as BIT-AND now accept an
1556 explicit NIL for their "opt-arg" argument (to indicate a
1557 freshly-consed result bit-array);
1558 ** ELT now signals an error on an invalid sequence index in safe
1560 ** the type system is now cleverer about negations of numeric
1561 types, and consequently understands the BIGNUM and RATIO types
1563 ** the type system is now cleverer about the interaction between
1564 INTEGER and RATIO types: while bugs still remain, many more
1565 cases are accurately computed;
1566 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1567 if it is in the last clause;
1568 ** CONSTANTLY now correctly returns a side-effect-free function in
1570 ** DECLARE is no longer treated as a special-operator; in
1571 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1572 * incremented fasl file version number due to the change in the
1573 DEFSTRUCT-SLOT-DESCRIPTION structure.
1575 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1576 * minor incompatible change: code processed by the "interpreter" or
1577 EVAL now has a compilation optimization policy of (DEBUG 2)
1578 (changed from (DEBUG 1)) to improve debuggability of interactive
1579 development, and to allow the use of the debug RETURN command in
1581 * an experimental implementation of the RETURN command for the
1582 debugger has been included. (thanks to Frederik Kuivinen)
1583 * fixed bug 62: constraints were not propagated into a loop.
1584 * fixed bug in embedded calls of SORT (reported and investigated by
1586 * fixed some bugs revealed by Paul Dietz' test suite:
1587 ** printing and reading of arrays with some dimensions having
1588 length 0 (thanks to Gerd Moellmann);
1589 ** BOA constructor with &AUX argument without a default value does
1590 not cause a type error;
1591 ** CONSTANTP now returns true for all self-evaluating objects.
1593 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1594 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1595 accessors that are related by inheritance, as specified in the
1596 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1597 Valtteri Vuorikoski)
1598 * The compiler is now able to inline functions that were defined in
1599 a complex lexical environment (e.g. inside a MACROLET).
1600 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1602 * Support for the upcoming FreeBSD-5.0 release has been included.
1603 (thanks to Dag-Erling Smorgrav)
1604 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1605 effect when it is not in a toplevel context.
1606 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1607 stand a better chance of being correct. (thanks to Gerd
1609 * fixed bug in COERCE, which now signals an error on coercing a
1610 rational to a bounded real type which excludes the expected
1612 * The compiler is now able to derive types more accurately from the
1613 COERCE and COMPILE functions.
1614 * fixed bug 223: functional binding is considered to be constant
1615 only for symbols in the CL package.
1616 * fixed bug 231: SETQ did not check the type of a variable being set
1617 (reported by Robert E. Brown)
1618 * A new optimization for MAKE-INSTANCE has been included, fixing
1619 various bugs (including relating to :ALLOCATION :CLASS slots and
1620 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1622 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1623 ** As required by ANSI, LOOP now disallows anonymous collection
1624 clauses such as COLLECT I in conjunction with aggregate boolean
1625 clauses such as THEREIS (= I 1);
1626 ** LOOP now signals an error when any variable is reused in the
1627 same loop (including the potentially useful construct analogous
1628 to WITH A = 1 WITH A = (1+ A);
1629 ** IT is only a special loop symbol within the first clause of a
1630 conditional loop clause;
1631 ** LOOP with a typed iteration variable over a hashtable now
1632 signals a type error iff it should.
1633 * fixed some other bugs revealed by Paul Dietz' test suite:
1634 ** FILE-STREAM now names the class previously known as FD-STREAM;
1635 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1636 argument) no longer signals an error;
1637 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1638 of the slot symbol, rather than using the current package
1639 ((:CONC-NAME "") continues to intern the slot's name in the
1641 * incremented fasl file version number, because of the incompatible
1642 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1643 of the new implementation of DEFINE-COMPILER-MACRO.
1645 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1646 * Support for building SBCL for MIPS platforms running in
1647 little-endian mode has now been checked in, and basic
1648 functionality on said platforms verified.
1649 * minor incompatible change: PCL now records the pathname of a file
1650 in which methods and the like are defined, rather than its
1652 * minor incompatible change: TRUENAME now considers the truename of
1653 a file naming a directory to be the pathname with :DIRECTORY
1654 component indicating that directory.
1655 * minor incompatible change: a NAMED clause in the extended form of
1656 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1657 reason for the previous behaviour is unclear.
1658 * more systematization and improvement of CLOS and MOP conformance
1659 in PCL (thanks to Gerd Moellman and Pierre Mai):
1660 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1662 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1663 primary methods with no specializers;
1664 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1666 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1667 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1668 FORWARD-REFERENCED-CLASSes; error reporting on
1669 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1671 ** SXHASH on CLOS instances now uses a slot internal to the
1672 instance to return different numbers on distinct instances,
1673 while preserving the same return value through invocations of
1675 ** DEFMETHOD signals errors when methods with longer incongruent
1676 lambda lists are added to generic functions;
1677 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1678 CLASS, as specified in AMOP;
1679 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1680 on the order returned by the primary method for classes of
1681 class STANDARD-CLASS;
1682 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1683 * fixed some bugs shown by Paul Dietz' test suite:
1684 ** DOLIST puts its body in TAGBODY;
1685 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1687 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1688 value producing form;
1689 ** if more variables are given to PROGV than values, extra
1690 variables are bound and made to have no value;
1691 ** NSUBSTITUTE on list arguments gets the right answer with
1693 ** ELT signals an error of type TYPE-ERROR when the index argument
1694 is not a valid sequence index;
1695 ** LOOP signals (at macroexpansion time) an error of type
1696 PROGRAM-ERROR when duplicate variable names are found;
1697 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1698 ** FUNCALL of special-operators now cause an error of type
1700 ** PSETQ now works as required in the presence of side-effecting
1701 symbol-macro places;
1702 ** NCONC accepts any object as its last argument;
1703 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1705 ** loop-for-as-package does not require a package to be explicitely
1707 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1709 * fixed bug 166: compiler preserves "there is a way to go"
1710 invariant when deleting code.
1711 * fixed bug 172: macro lambda lists with required arguments after
1712 &REST arguments now cause an error to be signalled. (thanks to
1714 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1715 bug. (thanks to Matthew Danish)
1716 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1718 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1719 arguments are lexically rebound. (thanks to Gerd Moellmann and
1721 * fixed bug 194: error messages are now more informative when there
1722 is no primary method applicable in a call to a generic
1723 function. (thanks to Gerd Moellmann)
1724 * fixed bug in command line argument checking (thanks to Julian
1726 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1727 (thanks to Matthew Danish)
1728 * incremented fasl file version number, because of the
1729 SXHASH-related changes in the layout of CLOS data structures
1731 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1732 * minor incompatible change: The runtime (the Unix executable named
1733 "sbcl") is now much pickier about the .core files it will load.
1734 Essentially it now requires .core files to descend from the same
1735 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1736 as the runtime does. (The intent is to prevent the crashes which
1737 can occur, and which can even be reported as mysterious failures,
1738 when people patch the sources or change the build parameters
1739 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1740 sbcl and .core files.)
1741 * fixed bug: VALUES-LIST is no longer optimized away.
1742 * fixed bug 142: The FFI conversion of C string values to Lisp
1743 string values no longer conses excessively. (thanks to Nathan
1744 Froyd porting Raymond Toy's fix to CMU CL)
1745 * began to systematize and improve MOP conformance in PCL (thanks to
1746 Nathan Froyd, Gerd Moellman and Pierre Mai):
1747 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1749 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1750 ** FINALIZE-INHERITANCE is now called on class finalization;
1751 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1752 argument precedence order.
1753 * fixed bug 202: The compiler no longer fails on functions whose
1754 derived types contradict their declared type.
1755 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1756 so it can be non-toplevel.
1757 * The fasl file version number has changed (because of the new
1758 implementation of DEFMACRO).
1759 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1760 safe code, that any length requirement by their type-specifier
1761 argument is valid. The exceptions to this are described in bug
1763 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1764 functions now check the types of their inputs as required by ANSI.
1765 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1766 introduced symbol is DECLAREd to be SPECIAL.
1767 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1768 * fixed bug: :COUNT argument to sequence functions may be negative
1769 * fixed bug: body of DO-SYMBOLS may contain declarations
1770 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1771 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1772 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1773 symbol macro only once
1774 * fixed printing of call frame when argument list is unavailable
1775 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1776 * compiler no longer signals WARNING on unknown keyword
1779 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1780 * A beta-quality port to the mips architecture running Linux,
1781 based on the old CMUCL backend, has been made. It has been tested
1782 on a big-endian kernel, and works sufficiently well to be able to
1783 rebuild itself; it has not been tested in little-endian mode.
1784 * fixed an inconsistency between gencgc.c and purify.c which made
1785 dumping/loading .core files unreliable
1786 * fixed bug 120a: The compiler now deals correctly with IFs where
1787 the consequent is the same as the alternative, instead of
1788 misderiving the return type. (thanks to Alexey Dejneka)
1789 * fixed bug 113: Logical pathnames are now dumpable (the logical
1790 host is resolved at load-time, throwing an error if it is not
1792 * fixed bug 174: FORMAT's error message is slightly clearer when a
1793 non-printing character is used in a format directive.
1794 * fixed several bugs in compiler checking of type declarations, i.e.
1795 violations of the Python "declarations are assertions" principle
1796 (thanks to Alexey Dejneka)
1797 * fixed several bugs in PCL's error checking (thanks to Gerd
1799 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1801 * fixed bug in compilation of functions as first class values
1802 (thanks to Antonio Martinez-Shotton)
1803 * The compiler's handling TYPE-ERRORs which it can prove will
1804 inevitably happen at runtime has been cleaned up and corrected
1805 in several ways. (thanks to Alexey Dejneka)
1806 * improved argument type checking for various basic arithmetic
1807 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1808 lavished on them in the past that they can be compiled in many
1809 ways in different special cases
1810 * fixed bug 181: compiler checks validity of user supplied type
1812 * cleaned up code flushing in optimization: Function calls which
1813 should signal errors for safety purposes (e.g. which ANSI says
1814 should signal errors when their arguments are of incorrect type)
1815 are no longer optimized away.
1816 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1817 * incremented fasl file version number, because changes in the
1818 implementation of sequence functions like COERCE caused
1819 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1820 in old inline expansions) to become undefined. (Actually these
1821 changes were later undone, so we might very well be binary
1822 compatible with 0.7.7 after all, but leaving the version number
1823 incremented seemed like the simplest and most conservative
1826 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1827 * An alpha-quality port to the parisc architecture running Linux,
1828 based on the old CMUCL backend, has been made. This, even more so
1829 than the other backends, should be considered still a work in
1830 progress; known problems include that the Linux kernel in 64-bit
1831 mode does not propagate the correct sigcontext structure to
1832 userspace, and consequently SBCL on a parisc64 kernel will not
1834 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1835 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1836 suppressed.) Also now that the compiler is looking at declarations
1837 in the environment, it checks optimization declarations as well,
1838 and suppresses inlining when (> DEBUG SPEED).
1839 * More fixes have been made to treatment of floating point exception
1840 treatment and other Unix signals. In particular, floating point
1841 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1842 * The detection and handling of control stack exhaustion (infinite
1843 or very deeply nested recursion) has changed. Stack exhaustion
1844 detection is now done by write-protecting pages at the OS level
1845 and applies at all optimization settings; when found, a
1846 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1847 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1848 be caught using IGNORE-ERRORS.
1849 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1850 correctly when transforming recursive LABELS functions to LETs.
1851 (thanks to Alexey Dejneka)
1852 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1853 that are names of constants or global variables.
1854 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1855 alien routines with docstrings.
1856 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1857 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1859 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1860 (thanks to Eric Marsden)
1861 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1862 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1863 * Bug fix: The SPARC backend can now compile functions involving
1864 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1865 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1866 to a functions expecting a PATHNAME-DESIGNATOR.
1867 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1868 lambda lists. (thanks to Alexey Dejneka)
1869 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1870 functions so that EXPORTing the name of the class doesn't cause
1871 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1872 to Antonio Martinez for reporting this)
1873 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1874 constant-folded without failing an assertion. (thanks to Einar
1875 Floystad Dorum for reporting this)
1876 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1877 types are now dealt with more correctly by the compiler.
1878 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1879 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1880 behaviour analogous to RENAME-FILE. This puts its behaviour more
1881 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1883 * The fasl file version number has changed again. (because of the
1884 bug fix involving the names of PCL MAKE-INSTANCE functions)
1886 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1887 * bug fix: Floating point exceptions are treated much more
1888 consistently on the x86/Linux and PPC/Linux platforms.
1889 * Array initialization with :INITIAL-ELEMENT is now much faster for
1890 cases when the compiler cannot open code the array creation, but
1891 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1892 array accesses have also seen a speed increase.
1893 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1894 specified. (This was at the root of some bad interactions between
1895 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1897 * bug fix: Internal error arguments for undefined functions are now
1898 computed correctly on the PPC/Linux platform.
1899 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1900 Raymond Toy's patch for CMU CL)
1901 * Support for the Solaris 9 operating environment has been included
1902 (thanks to Daniel Merritt)
1903 * A very ugly but hopefully complete draft of the missing FFI chapter
1904 of the manual has been created by reformatting the corresponding
1905 CMU CL manual chapter into (currently very ugly and incoherent)
1906 DocBook and bringing it up to date for SBCL behavior. Thus, the
1907 manual is now essentially complete, at least by my extreme
1908 once-and-only-once standards, whereby it's acceptable to refer to
1909 the doc strings of SB-EXT functions as the primary documentation.
1910 * The fasl file version number has changed again, due to cleanup of
1911 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1912 is no longer a static symbol.)
1914 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1915 * SBCL now builds with OpenMCL (version 0.12) as the
1916 cross-compilation host; also, more progress has been made toward
1917 bootstrapping under CLISP.
1918 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1920 * bug 158 fixed: The compiler can now deal with integer loop
1921 increments different from 1; fixing this turned out also to fix
1923 * bug 169 fixed: no more bogus warnings about using lexical bindings
1924 despite the presence of perfectly good SPECIAL declarations (thanks
1925 to David Lichteblau)
1926 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1927 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1928 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1930 * bug 180 fixed: Method combination specifications no longer ignore
1931 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1932 * bug fix: Structure type predicate functions now check their argument
1933 count as they should.
1934 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1935 correctly. (thanks to Pierre Mai)
1936 * minor incompatible change: The --noprogrammer option is deprecated
1937 in favor of the new --disable-debugger option, which is very similar.
1938 (The major difference is that it takes effect at a slightly different
1939 time at startup, causing handling of errors in --sysinit and
1940 --userinit files will be affected differently.) The
1941 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1942 been added to allow this functionality to be controlled from ordinary
1943 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1944 maintainers, who might want to run non-interactive scripts to
1945 build SBCL cores which will later be used interactively.)
1946 * minor incompatible change: The LOAD function no longer, when given
1947 a wild pathname to load, loads all files matching that pathname.
1948 Instead, an error of type FILE-ERROR is signalled.
1950 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1951 * bug 147 fixed: The compiler preserves its block link/count
1952 invariants more correctly now so that it doesn't crash. (thanks
1954 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1956 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1957 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1958 no-value-for-this-character as NIL (as specified by ANSI).
1959 * HOST-NAMESTRING on physical pathnames now returns a string that is
1960 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1961 (thanks to Christophe Rhodes)
1962 * The Alpha port handles icache flushing more correctly. (thanks to
1964 * More progress has been made toward bootstrapping under CLISP. (thanks
1965 to Christophe Rhodes)
1966 * The fasl file format has changed again, because dynamic loading
1967 on OpenBSD (which has non-ELF object files) motivated some cleanups
1968 in the way that foreign symbols are transformed and passed around.
1969 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1970 is no longer treated as whitespace by the reader, but instead as
1971 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1972 |A
\7fB|, instead of A as it used to.
1974 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1975 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1976 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1977 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1978 this as of 0.7.1.45, but was left out of the previous news section
1979 (thanks to Dan Barlow)
1980 * SBCL now runs on the Solaris operating system on SPARC architectures
1981 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1982 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1983 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1984 * SPARC backend cleanups, allowing builds of cores optimized for V8
1985 and V9 SPARCS, and also emission of code targeted to a particular
1986 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1988 * SBCL is closer to bootstrapping under CLISP, thanks to various
1989 fixes by Christophe Rhodes.
1990 * The fasl file format has changed again, to allow the compiler's
1991 INFO database to support symbol macros.
1992 * The user manual (in doc/) is formatted into HTML more nicely.
1993 (thanks to coreythomas)
1994 * The system is smarter about SUBTYPEP relationships, especially
1995 those involving NOT types (including types such as ATOM which are
1996 represented internally using NOT types). Thus SUBTYPEP is less
1997 likely to return (VALUES NIL NIL) in general, and in particular
1998 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2000 * The fasl file format has changed again, because the internal
2001 representation of types now includes a new slot to support the new
2002 SUBTYPEP-of-NOT-types logic.
2003 * (not a change in the main branch of SBCL, but a related prototype
2004 which can hopefully be merged into the main branch of SBCL in the
2005 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2006 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2007 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2008 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2009 dispatch macro character combinations. (thanks to Alexey Dejneka)
2010 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2011 that superficially look like logical namestrings correctly.
2012 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2014 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2015 default supplied-p) for &optional and &key arguments. (thanks to
2018 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2019 * incompatible change: The compiler is now less aggressive about
2020 tail call optimization, doing it only when (> SPACE DEBUG) or
2021 (> SPEED DEBUG). (This is an incompatible change because there are
2022 programs which relied on the old CMU-CL-style behavior to optimize
2023 away their unbounded recursion which will now die of stack overflow.)
2024 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2025 for non-GENCGC systems has been increased to 20M (since that
2026 seems much closer to the likely performance optimum for modern
2027 systems than the old 4M value was)
2028 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2029 and *DEBUG-PRINT-LEVEL*
2030 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2031 of CMU CL's support for SPARC, and various endianness and other
2032 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2033 * new syntactic sugar for the Unix command line: --load foo.bar is now
2034 an alternate notation for --eval '(load "foo.bar")'.
2036 ** The system now detects stack overflow and handles it gracefully,
2037 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2038 optimization settings. (This is a good thing in general, and
2039 its introduction in this version should be particularly timely
2040 for anyone whose code fails because of suppression of tail
2042 ** The system now hunts for the C variable "environ" in a more
2043 devious way, to avoid segfaults when the C library version differs
2044 between compile time and run time. (thanks to Christophe Rhodes)
2045 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2046 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2047 ** The compiler no longer issues bogus style warnings for undefined
2048 classes in the same source file as the DEFCLASSes which defined
2049 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2051 ** fixes in CONDITION class precedence list for undefined function
2052 errors (thanks to Alexei Dejneka)
2053 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2054 correctly. (thanks to Dan Barlow)
2055 ** portability fixes aiming at bootstrapping under CLISP (thanks
2056 to Dave McDonald and Christophe Rhodes)
2057 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2058 ** fixes in type translation and and type inference (thanks to
2060 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2061 ** various fixes in the new ports (thanks to Dan Barlow)
2062 * several changes related to debugging:
2063 ** suppression of tail recursion, as noted above
2064 ** stack overflow detection, as noted above
2065 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2066 is now the default. (For some time encapsulation has been more
2067 reliable than the breakpoint-based :ENCAPSULATE NIL
2068 implementation, at least on X86 systems; and I just noticed that
2069 encapsulation also seems closer to the spirit of the ANSI
2072 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2074 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2075 up properly again. (There was a packaging bug in 0.7.0 which
2076 left their definitions in SB-SYS::LOAD-FOREIGN and
2077 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2078 vital for most things which interface to C-level interfaces,
2079 like extensions working with sockets or databases or
2080 Perl-compatible regexes or whatever, and the need to fix
2081 this bug is the main reason that 0.7.1 was released so
2083 ** DEFGENERIC is now choosier about the methods it redefines, so that
2084 reLOADing a previously-LOADed file containing DEFGENERICs does
2085 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2086 cycle now works as it should. (thanks to Alexey Dejneka)
2087 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2088 fixed by Christophe Rhodes. (He actually submitted this patch
2089 months ago, and I delayed until after 0.7.0.)
2090 ** Bug 111 (internal compiler confusion about runtime checks on
2091 FUNCTION types) was fixed by Alexey Dejneka.
2092 * Some internal cleanups (getting rid of variables which aren't
2093 needed now that the byte interpreter is gone) caused the fasl
2094 file format number to change again.
2096 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2097 * major incompatible change: The default fasl file extension, i.e. the
2098 default extension for files produced by COMPILE-FILE, has changed
2099 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2101 ** There are many changes in the implementation of the compiler.
2102 SBCL is now essentially a compiler-only implementation of ANSI
2103 Common Lisp. EVAL still "interprets" a few special cases, but
2104 almost all the interesting cases are handled by creating
2105 a LAMBDA expression, calling COMPILE on it, then calling
2106 FUNCALL on the result.
2107 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2108 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2109 Since the code is newer, there might still be some new bugs
2110 (though not as many as before Martin Atzmueller's fixes:-). But
2111 the new code is substantially simpler and clearer, and hopefully
2112 any remaining bugs will be simpler, less fundamental, and more
2113 fixable then the bugs in the old code.
2114 ** The revised compiler is still a little unsteady on its feet.
2116 *** The debugging information it produces (particularly the names
2117 of FUNCTION objects) is sometimes much less useful than what
2118 the old compiler produced.
2119 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2120 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2121 within a MACROLET) has been temporarily weakened.
2122 ** There are new compiler optimizations for various functions:
2123 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2124 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2125 *** the math functions TRUNCATE, FLOOR, and CEILING
2126 *** the function-of-all-trades COERCE
2127 Mostly these should be transparent, but there's one
2128 potentially-annoying problem (bug 117): when the compiler
2129 inline-expands a function and does type analysis on the result,
2130 it can create control paths which have type mismatches, and
2131 when it can't prove that those control paths aren't taken,
2132 it will issue WARNINGs about the type mismatches. This is
2133 a particular problem in practice for the new sequence functions.
2134 It's not clear how this should be fixed, and for now, a
2135 workaround is given in the entry for 117 in the BUGS file.
2136 ** (Because of the interaction between the two previous items --
2137 occasional inlining problems and new inline expansions -- some
2138 of the new sequence function optimizations won't really kick in
2139 completely until debugging information, and then inlining, are
2140 straightened out in some future version.)
2141 * minor incompatible changes:
2142 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2143 DIRECTORY no longer implicitly promotes NIL slots of its
2144 pathname argument to :WILD. In particular, when you ask for the
2145 contents of a directory (which you used to be able to do without
2146 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2147 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2148 ** changes in behavior that ANSI explicitly defines to be
2149 implementation dependent:
2150 *** The new compiler-only implementation still conforms with ANSI,
2151 but acts a little different than before. Besides the obvious
2152 changes in performance tradeoffs (that the cost per form passed
2153 to EVAL has gone up, and the cost per form executed by EVAL
2154 has gone down), the behavior of the system changes a little
2155 because there are no longer any interpreted function objects.
2156 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2157 e.g. doing COMPILE on the output of interactive DEFUN is
2159 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2161 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2162 has changed from (:CL) to NIL.
2163 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2164 e.g. #\Newline and #\Tab, don't have names specified in the
2165 ANSI Common Lisp standard, is now based on their ASCII symbolic
2166 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2167 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2168 are no longer used for output.
2169 ** changes in internal implementation constants:
2170 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2171 4 million. (If your application spends a lot of time GCing and
2172 you have a lot of RAM, you might want to experiment with
2173 increasing it even more.)
2174 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2175 However, almost all old code should still continue to work without
2176 immediate update, as SB-C-CALL is now a (deprecated) nickname
2178 ** Old operator names in the style DEF-FOO are now deprecated in
2179 favor of new corresponding names DEFINE-FOO, for consistency with
2180 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2181 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2182 internal symbols, but a few supported extensions like
2183 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2184 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2185 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2186 etc. as you get deeper into recursive calls to the debugger
2187 command loop, instead of the old "5]", "5]]", "5]]]"
2188 sequence. (I was motivated to do this when squabbles between
2189 ILISP and SBCL left me very deeply nested in the debugger. In the
2190 short term, this change will probably provoke more ILISP/SBCL
2191 squabbles, but hopefully it will be an improvement in the long run.)
2192 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2193 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2194 function, since declaiming return types involving aliens is
2195 (1) annoyingly messy to do by hand and (2) vital to efficient
2196 compilation of code which calls such functions.
2197 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2198 longer reexported by the SB-EXT package. They're solely useful
2199 for alien code, so it seems more logical that you should get
2200 them from the SB-ALIEN package, not in SB-EXT.
2201 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2202 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2203 features. Instead, the code that they used to control is always
2204 built into the system.
2205 * many other bug fixes
2206 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2207 advantage of the new EVAL-WHEN stuff and to clean them up in
2208 general, and they are now more ANSI-compliant in a number of
2209 ways. Martin Atzmueller is responsible for a lot of this.
2210 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2212 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2213 *** correct ERROR type for various file operations
2214 *** some fixes for Lisp streams
2215 *** DEFMETHOD syntax checking
2216 *** changing old weird representation of debug information as
2217 strings (which, among their other deficiencies, don't transform
2218 correctly when you rename packages, and don't change their
2219 print representation when you change things like *PACKAGE*
2220 and *PRINT-LENGTH*) to symbols and lists of symbols
2221 He also made several improvements and fixed several bugs in DESCRIBE.
2222 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2224 *** misbehavior of WRITE-STRING/WRITE-LINE
2225 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2226 and several other LOOP problems as well
2227 *** DIRECTORY when similar filenames are present
2228 *** DEFGENERIC with :METHOD options
2229 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2230 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2231 *** argument ordering in FIND with :TEST option
2232 *** mishandled package designator argument in APROPOS-LIST
2233 *** various problems in the backquote readmacro
2234 *** a bug in APROPOS
2235 *** probably some others that I'm not describing very well here,
2236 since the CVS log documents them by reference to sbcl-devel
2237 messages, and the SourceForge archives aren't working well.:-(
2238 ** Dan Barlow improved the Alpha port (and is making progress on the
2239 PPC port, for those of you who think different).
2240 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2241 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2242 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2243 continues to work on the SPARC port (for those of you in a position
2244 to look down upon our little PC-compatible boxes from a great height).
2245 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2246 and uses that copy, rather than the current dynamic value, when
2247 it's trying to decide whether to truncate output. Thus e.g.
2248 (let ((*print-lines* 50))
2249 (pprint-logical-block (stream nil)
2251 (let ((*print-lines* 8))
2252 (print (aref possiblybigthings i) stream)))))
2253 should now truncate the logical block only at 50 lines, instead of
2254 often truncating it at 8 lines, as it did before.
2255 * The doc/cmucl/ directory, containing old CMU CL documentation
2256 from the time of the fork, is no longer part of the base system.
2257 SourceForge has shut down its anonymous FTP service, and with it
2258 my original plan for distributing the old CMU CL documentation
2259 there. For now, if you need these files you can download an old
2260 SBCL source release and extract them from it.
2261 * The fasl file version number changed again, for dozens of reasons,
2262 some of which are apparent above.
2264 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2265 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2266 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2267 patch, so that DISASSEMBLE output is much nicer.
2268 * The code in the SB-PROFILE package now seems reasonably stable.
2269 I still haven't decided what the final interface should look like
2270 (I'd like PROFILE to interact cleanly with TRACE, since both
2271 facilities use function encapsulation) but if you have a need
2272 for profiling now, you can probably use it successfully with
2273 the current CMU-CL-style interface.
2274 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2275 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2276 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2277 link now returns the dangling link itself, and for similar
2278 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2279 link itself. (In these cases the old code signalled an error and
2280 looped endlessly, respectively.) Thus, DIRECTORY now works even
2281 in the presence of dangling and cyclic symbolic links.
2282 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2283 is now a supported extension again, since the consensus on
2284 sbcl-devel was that it can be useful for ordinary development
2285 work, not just for debugging SBCL itself.
2286 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2287 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2288 that currently-defined functions might be redefined later with
2289 different return types.
2290 * Hash tables can be printed readably, as inspired by CMU CL code
2291 of Eric Marsden and SBCL code of Martin Atzmueller.
2292 * better error handling in CLOS method combination, thanks to
2293 Martin Atzmueller porting Pierre Mai's CMU CL patches
2294 * more overflow fixes for >16Mbyte I/O buffers
2295 * A bug in READ has been fixed, so that now a single Ctrl-D
2296 character suffices to cause end-of-file on character streams.
2297 In particular, now you only need one Ctrl-D at the command
2298 line (not two) to exit SBCL.
2299 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2301 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2302 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2303 does the right thing.
2304 * The implementation of some type tests, especially for CONDITION
2305 types, is now tidier and maybe faster, due to CMU CL code
2306 originally by Douglas Crosher, ported by Martin Atzmueller.
2307 * Some math functions have been fixed, and there are new
2308 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2309 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2310 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2311 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2312 likely to be built into sbcl-0.7.0.)
2313 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2314 CL, ported to SBCL by Martin Atzmueller)
2315 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2316 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2317 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2318 * There's a new slam.sh hack to shorten the edit/compile/debug
2319 cycle for low-level changes to SBCL itself, and a new
2320 :SB-AFTER-XC-CORE target feature to control the generation of
2321 the after-xc.core file needed by slam.sh.
2322 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2323 COMPILE-FILE is no longer supported, so that now every function
2324 gets an entry point, so that block compilation looks a little
2325 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2326 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2327 returns the number of bytes consed since the system started,
2328 rather than the number consed since the first time the function
2329 was called. (The new definition parallels ANSI functions like
2330 CL:GET-INTERNAL-RUN-TIME.)
2331 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2332 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2333 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2334 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2335 * Fasl file version numbers are now independent of the target CPU,
2336 since historically most system changes which required version
2337 number changes have affected all CPUs equally. Similarly,
2338 the byte fasl file version is now equal to the ordinary
2341 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2342 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2343 is no longer recognized. I apologize for this, because it was
2344 listed in SB-EXT as a supported extension, but I found that
2345 its existing behavior was poorly specified, as well as incorrectly
2346 specified, and it looked like too much of a mess to straighten it
2347 out. I have enough on my hands trying to get ANSI stuff to work..
2348 * many patches ported from CMU CL by Martin Atzmueller, with
2349 half a dozen bug fixes in pretty-printing and the debugger, and
2350 half a dozen others elsewhere
2351 * fixed bug 13: Floating point infinities are now supported again.
2352 They might still be a little bit flaky, but thanks to bug reports
2353 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2354 as flaky as they were.
2355 * The --noprogrammer command line option is now supported. (Its
2356 behavior is slightly different in detail from what the old man
2357 page claimed it would do, but it's still appropriate under the
2358 same circumstances that the man page talks about.)
2359 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2360 are now supported, and enabled by default. Thus, the compiler can
2361 handle many floating point and complex operations much less
2362 inefficiently. (Thus e.g. you can implement a complex FFT
2364 * The compiler now detects type mismatches between DECLAIM FTYPE
2365 and DEFUN better, and implements CHECK-TYPE more correctly, and
2366 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2367 thanks to patches from Martin Atzmueller.
2368 * various fixes to make the cross-compiler more portable to
2369 ANSI-conforming-but-different cross-compilation hosts (notably
2370 Lispworks for Windows, following bug reports from Arthur Lemmens)
2371 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2372 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2373 fixed thanks to Pierre Mai's CMU CL patches.
2374 * fixes to keep the system from overflowing internal counters when
2375 it tries to use i/o buffers larger than 16M bytes
2376 * fixed bug 45a: Various internal functions required to support
2377 complex special functions have been merged from CMU CL sources.
2378 (When I was first setting up SBCL, I misunderstood a compile-time
2379 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2380 * improved support for type intersection and union, fixing bug 12
2381 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2382 more obscure bugs as well
2383 * some steps toward byte-compiling non-performance-critical
2384 parts of the system, courtesy of patches from Martin Atzmueller
2385 * Christophe Rhodes has made some debian packages of sbcl at
2386 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2387 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2388 stable, but are nonetheless usable. When he's ready, I'd be happy
2389 to add them to the SourceForge "File Releases" section. (And if
2390 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2391 * new fasl file format version number (because of changes in
2392 internal representation of (OR ..) types to accommodate the new
2393 support for (AND ..) types, among other things)
2395 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2396 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2398 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2400 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2401 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2402 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2403 * DESCRIBE now works on structure objects again.
2404 * Most function call argument type mismatches are now handled as
2405 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2406 know whether the function will be redefined before the call is
2407 executed. (The compiler could flag local calls with full WARNINGs,
2408 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2409 it doesn't keep track of enough information to know whether calls
2410 are local in this sense.)
2411 * Compiler output is now more verbose, with messages truncated
2412 later than before. (There should be some supported way for users
2413 to override the default verbosity, but I haven't decided how to
2414 provide it yet, so this behavior is still controlled by the internal
2415 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2416 src/compiler/ir1util.lisp.)
2417 * Fasl file format version numbers have increased again, because
2418 support for the Gray streams extension changes the layout of the
2419 system's STREAM objects.
2420 * The Gray subclassable streams extension now works, thanks to a
2421 patch from Martin Atzmueller.
2422 * The full LOAD-FOREIGN extension (not just the primitive
2423 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2424 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2425 but like most other programs, it defaults to copying the Unix
2426 environment from the original process instead of starting the
2427 new process in an empty environment.
2428 * Extensions which manipulate the Unix environment now support
2429 an :ENVIRONMENT keyword option which doesn't smash case or
2430 do other bad things. The CMU-CL-style :ENV option is retained
2431 for porting convenience.
2432 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2433 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2435 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2437 * A patch from Martin Atzmueller seems to have solved the SIGINT
2438 problem, and as far as we know, signal-handling now works cleanly.
2439 (If you find any new bugs, please report them!)
2440 * The system no longer defaults Lisp source file names to types
2441 ".l", ".cl", or ".lsp", but only to ".lisp".
2442 * The compiler no longer uses special default file extensions for
2443 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2444 seems to expect a single default extension for all compiled code,
2445 and there's no compelling reason to try to stretch the standard
2446 to allow two different extensions.) Instead, byte-compiled files
2447 default to the same extension as native-compiled files.
2448 * Fasl file format version numbers have increased again, because
2449 a rearrangement of internal implementation packages made some
2450 dumped symbols in old fasl files unreadable in new cores.
2451 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2452 many fewer weird special cases.
2453 * Bug #17 (differing COMPILE-FILE behavior between logical and
2454 physical pathnames) has been fixed, and some related misbehavior too,
2455 thanks to a patch from Martin Atzmueller.
2456 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2457 by Tim Moore, ported to SBCL by Martin Atzmueller.
2458 * Martin Atzmueller fixed several filesystem-related problems,
2459 including bug #36, in part by porting CMU CL patches, which were
2460 written in part by Paul Werkowski.
2461 * More compiler warnings in src/runtime/ are gone, thanks to
2462 more patches from Martin Atzmueller.
2463 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2466 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2468 * DESCRIBE now works on CONDITION objects.
2469 * The debugger now handles errors which arise when trying to print
2470 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2472 * The build system now uses an additional file, customize-target-features.lisp,
2473 to allow local modifications to the target *FEATURES* list. (The point of
2474 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2475 debugging features enabled, without having to worry about propagating it
2476 into everyone's system when I do a "cvs update".) When no
2477 customize-target-features.lisp file exists, the target *FEATURES* list
2478 should be constructed the same way as before.
2479 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2480 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2481 warn about behavior which is undefined under the ANSI spec. Note: This
2482 is specified by ANSI, but it's not very popular with programmers.
2483 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2484 macro in the SBCL sources for an example of a workaround which you
2485 might use to make portable ANSI-standard code which does what you want.
2486 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2487 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2488 when it's used as a non-toplevel form. (This is required in order
2489 to implement the DEFCONSTANT-EQX macro.)
2490 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2491 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2492 funny things, due to bugs in the implementation of EVAL-WHEN
2493 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2494 non-toplevel DEFCONSTANTs are uncommon.)
2495 * The core file version number and fasl file version number have been
2496 incremented, because the old noncompliant DEFCONSTANT behavior involved
2497 calling functions which no longer exist, and because I also took the
2498 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2499 * fixed bug 1 (error handling before read-eval-print loop starts), and
2500 redid debugger restarts and related debugger commands somewhat while
2502 ** The QUIT debugger command is gone, since it did something
2503 rather different than the SB-EXT:QUIT command, and since it never
2504 worked properly outside the main toplevel read/eval/print loop.
2505 Invoking the new TOPLEVEL restart provides the same functionality.
2506 ** The GO debugger command is also gone, since you can just invoke
2507 the CONTINUE restart directly instead.
2508 ** The TOP debugger command is also gone, since it's redundant with the
2509 FRAME 0 command, and since it interfered with abbreviations for the
2511 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2513 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2514 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2515 some porting work by Martin Atzmueller)
2516 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2517 with a boatload of patches which he ported to SBCL. Now that those
2519 ** The system tries to make sure that its low-priority messages
2520 are prefixed by semicolons, to help people who like to use
2521 syntax highlighting in their ILISP buffer. (This patch
2522 was originally due to Raymond Toy.)
2523 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2524 patches originally written by Raymond Toy.
2525 ** The compiler understands coercion between single-value and
2526 multiple-VALUES type expressions better, getting rid of some very
2527 weird behavior, thanks to patches originally by Robert MacLachlan
2528 and Douglas Crosher.
2529 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2530 lambda lists, thanks to a patch originally by Pierre Mai.
2531 ** The system no longer bogusly warns about "abbreviated type
2533 ** The compiler gets less confused by inlining and RETURN-FROM,
2534 thanks to some patches originally by Tim Moore.
2535 ** The system no longer hangs when dumping circular lists to fasl
2536 files, thanks to a patch originally from Douglas Crosher.
2537 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2538 undefined function error.
2539 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2540 in structure slot initforms) for now, documented workaround instead:-|
2541 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2542 * fixed bug 5: added stubs for various Gray stream functions called
2543 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2544 installed, at least appropriate type errors are generated
2545 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2546 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2547 consistently in DEFMETHOD forms.
2548 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2549 it doesn't seem to affect SBCL after all
2550 * The C runtime system now builds with better optimization and many
2551 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2553 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2555 * The system is now under CVS at SourceForge (instead of the
2556 CVS repository on my home machine).
2557 * The new signal handling code has been tweaked to treat register
2558 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2559 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2560 downstream, e.g. in debug-int.lisp, has implicit dependencies
2561 on the unsignedness of integer representation of machine words,
2562 and that caused the system to bomb out with infinite regress
2563 when trying to recover from type errors involving signed values,
2564 e.g. (BUTLAST '(1 2 3) -1).)
2565 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2566 (This was one of the bugs Peter Van Eynde reported back in July.)
2567 * The system now uses code inspired by Colin Walters' O(N)
2568 implementation of MAP (from the cmucl-imp@cons.org mailing
2569 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2570 inline the MAP operation, and there is more than one
2571 sequence argument to the MAP call (so that it can't just
2572 do ETYPECASE once and for all based on the type of the
2573 single sequence argument). (The old non-inline implementation
2574 of the general M-argument sequence-of-length-N case required
2575 O(M*N*N) time when any of the sequence arguments were LISTs.)
2576 * The QUIT :UNIX-CODE keyword argument has been renamed to
2577 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2578 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2580 * Raymond Wiker's patches to port dynamic loading from Linux to
2581 FreeBSD have been added.
2582 * The BUGS file is now more nearly up to date, thanks in large part
2583 to Martin Atzmueller's review of it.
2584 * The debugger now flushes standard output streams before it begins
2585 its output ("debugger invoked" and so forth).
2586 * The core version number and fasl file version number have both
2587 been incremented, because of incompatible changes in the layout
2589 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2590 used to not be called for a saved Lisp image.)
2591 * Martin Atzmueller's version of a patch to fix a compiler crash,
2592 as posted on sbcl-devel 13 September 2000, has been installed.
2593 * Instead of installing Martin Atzmueller's patch for the
2594 compiler transform for SUBSEQ, I deleted the compiler transform,
2595 and transforms for some similar consing operations.
2596 * A bug in signal handling which kept TRACE from working on OpenBSD
2598 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2599 away by constant folding
2600 * The system now defines its address space constants in one place
2601 (in the Lisp sources), and propagates them automatically elsewhere
2602 (through GENESIS and the sbcl.h file). Therefore, patching the
2603 address map is less unnecessarily tedious and error-prone. The
2604 Lisp names of address space constants have also been systematized.
2605 * CVS tags like dollar-Header-dollar have been removed from
2606 the sources, because they have never saved me trouble and
2607 they've been source of trouble working with patches and other
2608 diff-related operations.
2609 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2610 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2612 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2614 * The system has been ported to OpenBSD.
2615 * The system now compiles with a simple "sh make.sh" on the systems
2616 that it's supported on. I.e., now you no longer need to tweak
2617 text in the target-features.lisp-expr and symlinks in src/runtime/
2618 by hand, the make.sh takes care of it for you.
2619 * The system is no longer so grossly inefficient when compiling code
2620 involving vectors implemented as general (not simple) vectors (VECTOR T),
2621 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2622 which dares to use the various sequence functions on non-simple
2623 vectors, takes less of a performance hit.
2624 * There is now a primitive type predicate VECTOR-T-P
2625 to test for the (VECTOR T) type, so that e.g.
2626 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2627 can now be compiled with some semblance of efficiency. (The old code
2628 turned the type declaration into a full call to %TYPEP at runtime!)
2629 * AREF on (VECTOR T) is still not fast, since it's still compiled
2630 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2631 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2632 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2633 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2634 without making full calls to SUBTYPEP at runtime.
2635 (Some analogous efficiency issues for non-simple vectors specialized to
2636 element types other than T, or for non-simple multidimensional arrays,
2637 have not been addressed. They could almost certainly be handled the
2638 same way if anyone is motivated to do so.)
2639 * The changes in array handling break binary compatibility, so
2640 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2641 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2642 instead of (VALUES T T).
2643 * By following the instructions that Dan Barlow posted to sbcl-devel
2644 on 2 July 2000, I was able to enable primitive dynamic object
2645 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2646 functionality is not implemented (since it calls ld to resolve
2647 library references automatically, requiring RUN-PROGRAM for its
2648 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2649 try to resolve library references) is now supported.
2650 * The system now flushes the standard output streams when it terminates,
2651 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2652 them at several other probably-convenient times, e.g. in each pass of
2653 the toplevel read-eval-print loop, and after evaluating a form given
2654 as an "--eval" command-line option. (These changes were motivated by a
2655 discussion of stream flushing issues on cmucl-imp in August 2000.)
2656 * The source transform for TYPEP of array types no longer assumes
2657 that an array whose element type is a not-yet-defined type
2658 is implemented as an array of T, but instead punts, so that the
2659 type will be interpreted at runtime.
2660 * There is now some support for cross-compiling in make.sh: each of
2661 the phases of make.sh has its own script. (This should be transparent
2662 to people doing ordinary, non-cross-compile builds.)
2663 * Since my laptop doesn't have hundreds of megabytes of memory like
2664 my desktop machine, I became more motivated to do some items on
2665 my to-do list in order to reduce the size of the system a little:
2666 ** Arrange for various needed-only-at-cold-init things to be
2667 uninterned after cold init. To support this, those things have
2668 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2669 symbols with such names are now uninterned after cold init).
2670 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2671 fasl files for cold load.
2672 ** Remove the old compiler structure pooling code (which used to
2673 be conditional on the target feature :SB-ALLOC) completely.
2674 ** Redo the representation of some data in cold init to be more compact.
2675 (I also looked into supporting byte compiled code at bootstrap time,
2676 which would probably reduce the size of the system a lot, but that
2677 looked too complicated, so I punted for now.)
2678 * The maximum signal nesting depth in the src/runtime/ support code has
2679 been reduced from 4096 to 256. (I don't know any reason for the very
2680 large old value. If the new smaller value turns out to break something,
2681 I'll probably just bump it back up.)
2682 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2684 * Many, many bugs reported by Peter Van Eynde have been added to
2685 the BUGS list; some have even been fixed.
2686 * While enabling dynamic object file loading, I tried to make the
2687 code easier to understand, renaming various functions and variables
2688 with less ambiguous names, and changing some function calling
2689 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2690 * While trying to figure out how to do the OpenBSD port, I tried to
2691 clean up some of the code in src/runtime/. In particular, I dropped
2692 support for non-POSIX signal handling, added various comments,
2693 tweaked the code to reduce the number of compilation warnings, and
2694 renamed some files to increase consistency.
2695 * To support the new automatic configuration functionality in make.sh,
2696 the source file target-features.lisp-expr has been replaced with the
2697 source file base-target-features.lisp-expr and the machine-generated
2698 file local-target-features.lisp-expr.
2699 * fixed a stupid quoting error in make.sh so that using CMU CL
2700 "lisp -batch" as cross-compilation host works again
2702 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2704 * DESCRIBE no longer tries to call itself recursively to describe
2705 bound/fbound values, so that it no longer fails on symbols which are
2706 bound to themselves (like keywords, T, and NIL).
2707 * DESCRIBE now works on generic functions.
2708 * The printer now prints less-screwed-up representations of closures
2709 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2710 * A private symbol is used instead of the :EMPTY keyword previously
2711 used to mark empty slots in hash tables. Thus
2712 (DEFVAR *HT* (MAKE-HASH-TABLE))
2713 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2714 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2715 now does what ANSI says that it should. (You can still get
2716 similar noncompliant behavior if bang on the hash table
2717 implementation with all the symbols you get back from
2718 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2719 This breaks binary compatibility, since tests for equality to
2720 :EMPTY are wired into things like the macroexpansion of
2721 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2723 * There's now a minimal placeholder implementation for CL:STEP,
2724 as required by ANSI.
2725 * An obscure bug in the interaction of the normal compiler, the byte
2726 compiler, inlining, and structure predicates has been patched
2727 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2728 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2730 * Missing ordinary arguments in a macro call are now detected even
2731 when the macro lambda list contains &KEY or &REST.
2732 * The debugger no longer complains about encountering the top of the
2733 stack when you type "FRAME 0" to explicitly instruct it to go to
2734 the top of the stack. And it now prints the frame you request even
2735 if it's the current frame (instead of saying "You are here.").
2736 * As specified by ANSI, the system now always prints keywords
2737 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2739 * The default initial SIZE of HASH-TABLEs is now smaller.
2740 * Type information from CLOS class dispatch is now propagated
2741 into DEFMETHOD bodies, so that e.g.
2742 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2744 is now basically equivalent to
2745 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2746 (DECLARE (TYPE SINGLE-FLOAT X))
2748 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2749 operation, without having to do run-time type dispatch.
2750 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2751 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2752 * Since I don't seem to be making big file reorganizations very often
2753 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2754 is overflowing my ability to conveniently back them up), I've finally
2755 checked the system into CVS. (The CVS repository is on my home system,
2756 not at SourceForge -- putting it on SourceForge might come later.)
2757 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2758 high-level GC-NOTIFY-FOO functions send their output. (There's
2759 still very little control of where low-level verbose GC functions
2760 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2761 less than it used to -- the GC-NOTIFY-FOO functions are now under
2762 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2763 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2764 in only one place in the source code, and propagates it automatically
2765 everywhere that it's needed. Thus e.g. when I bump the version from
2766 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2767 * The C source files now include boilerplate legalese and documentation
2768 at the head of each file (just as the Lisp source files already did).
2769 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2770 to his page will be replaced with a link to his new CLiki service.
2772 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2774 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2775 * The build process now looks for GNU make under the default name "gmake",
2776 instead of "make" as it used to. If GNU make is not available as "gmake"
2777 on your system, you can change this default behavior by setting the
2778 GNUMAKE environment variable.
2779 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2780 can build without error under CMU CL.
2782 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2784 * There is now a partial SBCL user manual (with some new text and some
2785 text cribbed from the CMU CL manual).
2786 * The beginnings of a profiler have been added (starting with the
2787 CMU CL profiler and simplifying and cleaning up). Eventually the
2788 main interface should be through the TRACE macro, but for now,
2789 it's still accessed through vaguely CMU-CL-style functions and macros
2790 exported from the package SB-PROFILE.
2791 * Some problems left over from porting CMU CL to the new
2792 cross-compilation bootstrap process have been cleaned up:
2793 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2794 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2795 part when getting rid of the compiler *BACKEND* stuff.)
2796 ** The value of *NULL-TYPE* was screwed up, because it was
2797 being initialized before the type system knew the final
2798 definition of the 'NULL type. This screwed up several key
2799 optimizations in the compiler, causing inefficiency in all sorts
2800 of places. (I found it because I wanted to understand why
2801 GET-INTERNAL-RUN-TIME was consing.)
2802 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2803 PROCLAIM FTYPE information. Unfortunately this broke binary
2804 compatibility again, since now the forms output by DEFGENERIC
2805 to refer to functions which didn't exist in 0.6.3.
2806 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2807 can use the new (as of 0.6.3) transform for SOME into MAP into
2809 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2810 values to (INTEGER 0 1000000), so that the time code will no longer
2811 occasionally get blown up by Linux returning 1000000 microseconds
2812 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2813 its output conform to the ANSI spec. (Alas, this makes its output
2814 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2815 * A full call to MAP NIL with a single sequence argument no longer conses.
2816 * fixes to problems pointed out by Martin Atzmueller:
2817 * The manual page no longer talks about multiprocessing as though
2818 it were currently supported.
2819 * The ILISP support patches have been removed from the distribution,
2820 because as of version 5.10.1, ILISP now supports SBCL without us
2821 having to maintain patches.
2822 * added a modified version of Raymond Toy's recent CMU CL patch for
2823 EQUALP comparison of HASH-TABLE
2825 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2827 * The system still can't cross-compile itself with
2828 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2829 entails), but at least it can compile more of itself that way
2830 than it used to be able to, and various buglets which were uncovered
2831 by trying to cross-compile itself that way have now been fixed.
2832 * This release breaks binary compatibility again. This time
2833 at least I've incremented the FASL file format version to 2, so that the
2834 problem can be detected reliably instead of just causing weird errors.
2835 * various new style warnings:
2836 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2837 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2838 ** using the lexical binding of a variable named in the *FOO* style
2839 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2840 as specified by ANSI.
2841 * *RANDOM-STATE* is no longer automatically initialized from
2842 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2843 default behavior of the system is to repeat its behavior every time
2844 it's run. If you'd like to change this behavior, you can always
2845 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2846 old convention there was no comparably easy way to get the system to
2847 repeat its behavior every time it was run.
2848 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2849 FTYPE declarations has been removed, in favor of their ANSI
2850 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2851 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2852 the types of their sequence arguments can be determined at compile time.
2853 This is done through a new open code expansion for MAP which eliminates
2854 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2855 argument types can be determined at compile time.
2856 * The optimizer now transforms COERCE into an identity operation when it
2857 can prove that the coerced object is already of the correct type. (This
2858 can be a win for machine generated code, including the output of other
2859 optimization transforms, such as the MAP transform above.)
2860 * Credit information has been moved from source file headers into CREDITS.
2861 * Source file headers have been made more standard.
2862 * The CASE macro now compiles without complaining even when it has
2865 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2867 * (Note that the way that the PCL macroexpansions were rewritten
2868 to accommodate the change in DEFGENERIC below breaks binary
2869 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2870 not run under sbcl-0.6.2. Once we get out of alpha releases,
2871 i.e. hit release 1.0.0, we'll probably try to maintain binary
2872 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2873 and sbcl-1.4.4. Until then, however, it might be fairly common
2874 for maintenance releases to break binary compatibility.)
2875 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2877 * The reporting of unhandled signals has been changed to print some
2878 explanatory text as well as the report form. (Previously only
2879 the report form was printed.)
2880 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2881 it defines, so that the compiler no longer issues undefined function
2882 warnings for compiled-but-not-yet-loaded generic functions.
2883 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2884 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2885 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2886 section of the ANSI spec are supported.
2887 * The "" nickname for the "KEYWORD" package has been removed.
2888 The reader still handles symbol tokens which begin with a package marker
2889 as keywords, but it doesn't expose its mechanism for doing so in the
2890 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2891 * The system now issues STYLE-WARNINGs for contradictory TYPE
2892 proclamations. (Warnings for contradictory FTYPE proclamations would
2893 be nice too, but those can't be done usefully unless the type system
2894 is made smarter about FUNCTION types.)
2895 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2896 systematized, so that "*target-*.lisp is supposed to exist only on the
2897 target and imply that there's a related file which exists on the
2898 host, and *host-*.lisp is supposed to exist only on the host and imply
2899 that there's a related file which exists on the target. This involves a
2900 lot of renaming. Hopefully the acute confusion caused by the renaming
2901 will be justified by the reduction in chronic confusion..
2902 ** runtime-type.lisp -> early-target-type.lisp
2903 ** target-type.lisp -> late-target-type.lisp
2904 ** early-host-format.lisp -> early-format.lisp
2905 ** late-host-format.lisp -> late-format.lisp
2906 ** host-error.lisp -> misc-error.lisp
2907 ** early-error.lisp -> early-target-error.lisp
2908 ** late-error.lisp -> late-target-error.lisp
2909 ** host-defboot.lisp -> early-defboot.lisp
2910 ** code/misc.lisp -> code/target-misc.lisp
2911 ** code/host-misc.lisp -> code/misc.lisp
2912 ** code/numbers.lisp -> code/target-numbers.lisp
2913 ** code/early-numbers.lisp -> numbers.lisp
2914 ** early-host-type.lisp -> early-type.lisp
2915 ** late-host-type.lisp -> late-type.lisp
2916 ** host-typep.lisp -> typep.lisp
2917 ** load.lisp -> target-load.lisp
2918 ** host-load.lisp -> load.lisp
2919 ** host-disassem.lisp -> disassem.lisp
2920 ** host-insts.lisp -> insts.lisp
2921 ** byte-comp.lisp -> target-byte-comp.lisp
2922 ** host-byte-comp.lisp -> byte-comp.lisp
2923 ** host-signal.lisp -> signal.lisp
2924 ** host-defstruct.lisp -> defstruct.lisp
2925 ** late-target-type.lisp -> deftypes-for-target.lisp
2926 Furthermore, several other previously target-only files foo.lisp (e.g.
2927 hash-table.lisp and random.lisp) have been split into a target-and-host
2928 foo.lisp file and a target-only target-foo.lisp file, with their key type
2929 definitions in the target-and-host part, so that the cross-compiler will
2930 know more about target types.
2931 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2932 gone away. In their place are various *BACKEND-FOO* variables
2933 corresponding to the slots of the old structure.
2934 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2935 into the target SBCL has been fixed.
2936 * The chill.lisp system for loading cold code into a running SBCL
2938 * Support for the CMU CL "scavenger hook" extension has been removed.
2939 (It was undocumented and unused in the CMU CL sources that SBCL was
2940 derived from, and stale in sbcl-0.6.1.)
2941 * Various errors in the cross-compiler type system were detected
2942 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2943 (enabling various consistency checks). Many of them were fixed,
2944 but some hard problems remain, so the compiler is back to
2945 running without *TYPE-SYSTEM-INITIALIZED* for now.
2946 * As part of the cross-compiler type system cleanup, I implemented
2947 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2948 * I have started adding UNCROSS calls throughout the type system
2949 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2950 on cross-compiler input files will be able to go away, and various
2952 * CONSTANTP now returns true for quoted forms (as explicitly required
2955 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2957 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2958 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2959 (MAKE-LIST -1). In the longer term, it ought to become true
2960 that declarations are assertions even at SAFETY 1. For now, it's not
2961 quite true even at SAFETY 3, but it's at least more nearly true..
2962 (Note that this change seems to increases the size of the system by
2963 O(5%) and to decrease the speed of the compiler by 20% or more.)
2964 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2965 for the problem of printing dozens of lines of distracting information
2966 about low-level system machinery as part of the top stack frame
2967 on entry to the debugger when an undefined function was called.
2968 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2969 is not reset to COMMON-LISP-USER.
2970 * Compilation of stuff related to dyncount.lisp has been made conditional
2971 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2972 smaller. The various dyncount-related symbols have been moved into
2973 a new "SB-DYNCOUNT" package.
2974 * tty-inspect.lisp has been renamed to inspect.lisp.
2975 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2976 feature has gone away. (When we eventually port to other flavors of
2977 libc and/or Unix, we'll try to make the differences between flavors
2978 invisible at the user level.)
2979 * Various other *FEATURES* tags, and/or their associated conditionals,
2980 have been removed if obsolescent, or given better documentation, or
2981 sometimes given more-mnemonic names.
2983 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2985 * tidied up "make.sh" script
2986 * tidied up system directory structure
2987 * better "clean.sh" behavior
2988 * added doc/FOR-CMUCL-DEVELOPERS
2989 * many many small tweaks to output format, e.g. removing possibly-confusing
2990 trailing #\. character in DESCRIBE-INSTANCE
2991 * (EQUALP #\A 'A) no longer signals an error.
2992 * new hashing code, including EQUALP hashing
2993 * tidied up Lisp initialization and toplevel
2994 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2995 * command line argument processing
2996 * added POSIX-GETENV function to deal with Unix-ish environment variables
2997 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2998 terminating SBCL on EOF
2999 * non-verbose GC by default
3000 * There is no more "sbcl" shell script; the sbcl file is now the C
3001 runtime executable (just like CMU CL).
3002 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3004 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3005 symbol and package information as Lisp native symbol and package objects
3006 instead of strings naming symbols and strings naming packages. This way,
3007 whenever packages are renamed (as in warm init), debug information is
3008 transformed along with everything else.
3009 * tweaked the optimization policy declarations which control the building
3010 of SBCL itself. Now, among other things, the system no longer saves
3011 source location debugging information. (This helps two problems at once
3012 by reducing SBCL size and by keeping SBCL from trying to look for its
3013 sources -- which may not exist -- when reporting errors.)
3014 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3015 debugging and testing purposes
3016 * cleaned up printing, making the printer call PRINT-OBJECT for
3017 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3018 methods, giving nearly-ANSI behavior
3019 * converted almost all special variables to use *FOO* naming convention
3020 * deleted PARSE-TIME functionality, since it can be done portably
3021 * moved some files out of cold init into warm init
3022 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3024 * regularized formatting of source files
3025 * added an install.sh script
3026 * fixed ridiculous memory usage of cross-compiler by making
3027 compiler/alloc.lisp not try to do pooling unless it can hook
3028 itself into the GC of the cross-compilation host. Now the system
3029 builds nicely on my old laptop.
3030 * added :SB-ALLOC in target-features.lisp-expr
3031 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3032 was not implemented)
3033 * re-did condition handling and note reporting in the compiler. Notes
3034 are no longer handled by signalling conditions. Style warnings
3035 and warnings are handled more correctly and reported in such a way
3036 that it's easy to find one or the other in your output (so that you
3037 can e.g. figure out which of many problems caused COMPILE-FILE to
3039 * changed the severity of several compiler warnings from full WARNING
3040 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3041 compiler note reporting so that it doesn't use the condition system
3042 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3043 COMPILE-FILE command)
3044 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3045 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3046 no longer supported, since their implementation was deeply intertwingled
3047 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3048 * removed bogus "support" for compiler macros named (SETF FOO), and
3049 removed the compiler macro for SETF INFO (but only after making a fool
3050 of myself on the cmucl-imp mailing list by posting a bogus patch for
3051 DEFINE-COMPILER-MACRO..)
3052 * Compiled files containing forms which have side effects on the Lisp
3053 reader (such as DEFPACKAGE forms) are now handled more correctly.
3054 (Compiler queuing of top level lambdas has been suppressed by setting
3055 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3056 * deleted various currently-unused source files, e.g. gengc.lisp. They
3057 may be added back at some point e.g. when porting to other architectures,
3058 but until they are it's distracting to distribute them and to try to
3060 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3061 there were so many of them they're just distractions, and UNCROSS is
3062 known to be able to handle the current sources
3063 * moved PROFILE functionality into TRACE, so that it will be clear
3064 how the wrapping and unwrapping of functions when you profile them
3065 interacts with the wrapping and unwrapping of functions when you
3066 trace them. (Actually, the functionality isn't there yet, but at least
3067 the interface specification is there. Hopefully, the functionality will
3068 arrive with some future maintenance release.)
3069 * removed host-oops.lisp
3070 * changed signature of QUIT function to allow UNIX-CODE argument
3071 * fixed READ-SEQUENCE bug
3072 * tweaked verbose GC output so that it looks more like the progress
3073 output that ANSI specifies for functions like LOAD
3074 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3075 * added <http://sbcl.sourceforge.com> to the banner information printed by