1 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2 * incompatible change: the top level REPL now has only an ABORT
3 restart associated with it, not TOPLEVEL and ABORT as it used to.
4 TOP and TOPLEVEL are now available as debugger commands for
5 returning to the top level.
6 * incompatible change: forms evaluated in the REPL now use the
7 global optimization policy.
8 * incompatible change: user- and system-initialization files are
9 no longer processed with LOAD, but by READ and EVAL; hence the
10 global optimization policy, startup package, readtable, etc,
12 * internal entry point details and argument counts no longer appear
13 in backtraces unless explicitly requested by setting
14 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
15 * built-in and standard functions no longer have names like "top
16 level local call to FOO".
17 * fixed bug 32: functions defined in non-null lexical environments
18 now have more legible printed representation
19 * fixed bug 33: functions defined in non-null lexical environemnts
20 are now more amenable to inspection by INSPECT.
21 * workaround for bug 354: XEPs no longer appear in backtraces unless
23 * fixed bug: receiving the signal which results from attempting to
24 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
25 SIGBUS on other *BSDs) on architectures where the C stack is also
26 the Lisp stack (x86 and x86-64) from foreign code no longer leads
27 to debugger confusion or wild instability. (reported by Cheuksan
29 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
30 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
31 (reported by Lutz Euler)
32 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
33 compiled code. (reported by Paul Dietz)
34 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
35 specialized parameter to a method to something that is not TYPEP
36 the specializer is now possible.
37 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
38 face of package deletion.
39 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
40 pathname. (reported by David Tolpin on c.l.l)
41 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
42 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
43 STRING-STREAM and FILE-STREAM, respectively.
44 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
45 than previously; in addition, its test suite is now run on build.
46 (thanks to Luke Gorrie)
47 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
48 * optimization: added a immediate representation for single-floats
50 * fixed some bugs revealed by Paul Dietz' test suite:
51 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
52 ** attempting to create a package with a colliding nickname causes
53 correctable errors to be signalled.
55 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
56 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
57 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
59 * fixed loading of multiply forward-referenced layouts.
60 (thanks to Cheuksan Wang)
61 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
62 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
63 (see COMPILE-FILE documentation for details of :PRINT option)
64 * fixed bugs 19 and 317: fixed-format floating point printing is
65 more accurate. This also fixes a bug reported by Adam Warner
66 related to the ~@F format directive.
67 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
69 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
70 dispatch table if the from-char is a dispatch macro character.
71 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
72 by the wordsize no longer ignore the last word. (reported by Lutz
74 * fixed bug in type checking of dynamic-extent variables. (reported
76 * optimization: sequence traversal functions use their freedom to
77 coerce function designators to functions.
78 * optimization: code with many calls to CLOS methods specialized on
79 CLOS classes has had redundant type checks removed.
80 * optimization: added declarations to speed up operations that access
81 the internal character database (for example STRING-UPCASE)
82 * optimization: comparison operations between floats and sufficiently small
83 fixnums no longer create extra rationals
84 * fixed some bugs related to Unicode integration:
85 ** portions of multibyte characters at the end of buffers for
86 character-based file input are correctly transferred to the
87 start of the buffer at the next read.
88 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
89 passing it through to OPEN.
90 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
91 argument given to internal calls to OPEN.
92 ** the built-in comment readers (introduced by character sequences
93 ";" and "#|") are more forgiving to encoding errors; they will
94 STYLE-WARN and then attempt to resync the stream at a character
95 boundary. (thanks to Teemu Kalvas)
96 * fixed some bugs in the x86-64 port:
97 ** Negative short int return values from c-calls are sign-extended
99 ** The stack is aligned to 16-bytes for c-calls, as required by
100 the ABI. (thanks to Cheuksan Wang)
101 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
102 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
103 for enums. (thanks to Vincent Arkesteijn)
104 ** Multiple small optimizations and bugfixes for floating point
106 * fixed some bugs revealed by Paul Dietz' test suite:
107 ** Space, Tab, Linefeed, Return and Page have the invalid
108 secondary constituent character trait.
109 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
111 ** WITH-INPUT-FROM-STRING should only update the index place on
113 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
115 ** Bit-array operations (BIT-AND and similar) worked incorrectly
116 with one-dimensional arrays with fill pointers.
117 ** TYPE-OF failed on a complex with an integer realpart and a
119 ** compiler failure during type inference for the code of form
120 (IF (EQL X (THE ...)) ...) (MISC.535).
122 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
123 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
124 platform. The port supports 61-bit fixnums, large memory spaces
125 and reloading shared object files.
126 * enhancement: saving cores with foreign code loaded is now
127 supported on ppc/Darwin in addition to the previously supported
129 * enhancement: the statistical profiler now walks deeper into the
130 call stack for more meaningful call-graphs and accrued time
131 reports (x86/x86-64 only). It also now reports time spent in
133 * enhancement: it is now possible to trace most individual methods
134 of a generic function in addition to tracing the generic function
136 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
138 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
139 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
140 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
141 produces an error. (thanks to Vincent Arkesteijn)
142 * bug fix: NAMESTRING on pathnames with :WILD components in their
143 directories works correctly. (thanks to Artem V. Andreev)
144 * fixed bug 125: compiler preserves identity of closures. (reported
146 * bug fixed: functions with &REST arguments sometimes failed with
147 "Undefined function" when compiled with (DEBUG 3). (reported by
149 * bug fix: overflow during compiling of setting element of a bit
150 vector with constant index and value. (reported by Timmy Douglas)
151 * build fix: fixed the dependence on *LOAD-PATHNAME* and
152 *COMPILE-FILE-PATHNAME* being absolute pathnames.
153 * on x86 compiler partially supports stack allocation of dynamic-extent
155 * GO and RETURN-FROM do not check the extent of their exit points
156 when compiled with SAFETY 0.
157 * fixed some bugs related to Unicode integration:
158 ** encoding and decoding errors are now much more robustly
159 handled; it should now be possible to recover even from invalid
160 input or output to the terminal. (thanks to Teemu Kalvas)
161 ** provided a first cut at implementing STRING-TO-OCTETS and
162 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
163 ** altered the SB-MD5 contributed module slightly, changing the
164 interface just enough for it to be supportable for builds where
165 lisp characters are not eight bits.
166 * fixed some bugs revealed by Paul Dietz' test suite:
167 ** the FORMATTER-generated functions for ~V[ conditionals require
168 the correct number of arguments.
169 ** READ-FROM-STRING returns the mandated second value when applied
170 to displaced strings.
171 ** the #\Rubout and #\Backspace characters are treated as invalid
172 constituent characters by the tokenizer.
174 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
175 * new feature: reloading changed shared object files with
176 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
177 * new feature: references to foreign variables and functions
178 can now be compiled and loaded before the corresponding shared
179 object file is loaded, as long as the foreign definitions are
180 available at runtime.
181 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
182 supported. (thanks to Dan Debertin)
183 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
184 just keywords. (thanks to Vincent Arkesteijn)
185 * fixed bug #331: structure-class instances corresponding to
186 DEFSTRUCT forms are now created eagerly.
187 * fixed bug #345: backtraces from calls to undefined functions work
188 on x86 as well. Related bug #61 is now also partially fixed on x86
189 (backtraces from throws to unknown catch tags.)
190 * bug fix: lambda-list parsing is now stricter vrt. order and number
191 of lambda-list keywords.
192 * bug fix: as specified by AMOP, an error is signalled if a
193 class-option appears multiple times in a DEFCLASS form. (reported
195 * bug fix: RANDOM can be compiled when the compiler derives the type
196 of its numeric argument as a disjoint set of small integers.
197 (reported by Paul Dietz)
198 * bug fix: starting a core saved with shared objects loaded when
199 those objects are not available no longer causes threaded SBCL to
200 hang. (reported by Sean Ross)
201 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
203 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
204 argument, which must always be NIL. (reported by Kalle Niemitalo)
205 * bug fix: printing 1.0d+23 no longer results in an error.
206 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
207 * bug fix: structure slot setters preserve evaluation order. (thanks
209 * fixed some bugs related to Unicode integration:
210 ** RUN-PROGRAM can allow its child to take input from a Lisp
211 stream. (reported by Stefan Scholl)
212 ** ASDF-INSTALL successfully downloads .tar.gz files in all
213 locales. (reported by Ken Causey)
214 * fixed some bugs revealed by Paul Dietz' test suite:
215 ** INCF, DECF and REMF evaluate their place form as specified in
217 ** FORMATTER expands ~{ ~} iteration directives with V or #
218 parameters correctly.
219 ** FORMATTER deals with the ~@[ ~] conditional directive where the
220 consequent uses no arguments correctly.
221 ** the system has a partial understanding of the (COMPLEX RATIO)
223 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
226 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
227 * new feature: a build-time option (controlled by the :SB-UNICODE
228 keyword feature, enabled by default) for building the system with
229 support for the entire 21-bit character space defined by the
231 * new feature: the system now has rudimentary external-format
232 support; the primary user-visible change at this time is that
233 characters with the high bit set (such as lower-case-e-acute) will
234 print correctly to a terminal in a UTF-8 environment.
235 * minor incompatible change: BASE-CHAR no longer names a class;
236 however, CHARACTER continues to do so, as required by ANSI.
237 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
238 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
239 has been moved to the SB-EXT package (temporarily re-exported from
241 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
242 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
244 * fixed bug #351: better error-handlind and reporting for malformed
246 * fixed bug #350: bignum-printing is now more memory-efficient,
247 allowing printing of very large bignums, eg. (expt 2 10000000).
248 (reported by Bruno Haible)
249 * fixed bug #302: better primitive-type selection for intersection
251 * fixed bug #308: non-graphic characters now all have names, as
252 required. (reported by Bruno Haible)
253 * bug fix: redefining a class with different superclasses now correctly
254 removes it from the direct-subclasses of its previous superclasses.
255 (reported by David Morse)
256 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
257 new value now works. (reported by Bruno Haible)
258 * bug fix: correct canonicalization of multiple non-standard slot
259 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
260 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
261 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
262 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
263 now exists, an signals an error.
264 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
265 during macroexpansion for non-iteration variables as well. (reported
266 by Bruno Haible for CMUCL)
267 * bug fix: Cyclic structures and unprintable objects in compiler
268 messages no longer cause errors. (reported by Bruno Haible)
269 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
270 and READ-FROM-STRING all now return a primary value of NIL if
271 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
272 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
273 (reported by Bruno Haible for CMUCL)
274 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
275 do not designate a condition. (reported by Bruno Haible for
277 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
278 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
279 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
280 permitting resolution in favour of any of the conflicting symbols.
281 (reported by Bruno Haible for CMUCL)
282 * bug fix: EQUAL compiler optimizations is less aggressive on
283 strings which can potentially compare true despite having distinct
284 specialized array element types.
285 * bug fix: unit enumerations can be defined without dividing by
286 zero. (thanks to Vincent Arkesteijn)
287 * FORMAT compile-time argument count checking has been enhanced.
288 (report from Bruno Haible for CMUCL)
289 * a partial workaround for the bug 262: the compiler does not try to
290 inline-expand a local function doing RETURN-FROM from a deleted
291 BLOCK. (thanks to Peter Denno for the bug report and to David
292 Wragg for the simple test case)
293 * fixed some bugs revealed by Paul Dietz' test suite:
294 ** RENAME-PACKAGE allows all package designators as new package
296 ** constraint propagation and lambda variable substitution are
297 more cautious in dealing with partially deleted code.
298 ** compiler sometimes lost reoptimization passes.
299 ** CERROR, given a condition as condition designator, passes its
300 remaining arguments to the continue format control without
302 ** Case-altering FORMAT directives work correctly on non-ASCII
304 ** The REAL type specifier handles bounds outside the floating
305 point ranges without signalling FLOATING-POINT-OVERFLOW.
306 ** Functions with IR1-transformations can create intercomponent
307 references to global functions.
308 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
310 ** FORMAT ~R treats a nil value for its first parameter correctly.
312 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
313 * enhancement: saving cores with foreign code loaded is now
314 supported on x86/NetBSD and sparc/Linux in addition to the previously
316 * bug fix: on some platforms repeated installations caused multiple
317 copies of HTML documentation to be installed -- should not happen
318 any more. (reported by Stefan Scholl)
319 * bug fix: parsing self-recursive alien record types multiple times
320 no longer causes infinite recursion. (reported by Thomas F. Burdick,
321 original patch by Helmut Eller for CMUCL)
322 * bug fix: stack-exhaustion detection works now on NetBSD as well.
323 (thanks to Richard Kreuter)
324 * bug fix: defining classes whose accessors are methods on existing
325 generic functions in other (locked) packages no longer signals
326 bogus package lock violations. (reported by François-René Rideau)
327 * bug fix: special variables as DEFMETHOD parameters no longer have
328 associated bogus type declarations. (reported by David Wragg and
330 * bug fix: read-write consistency on streams of element-type
331 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
332 * bug fix: redefiniton of the only method of a generic function with
333 no DEFGENERIC no longer emits a full WARNING. In addition,
334 redefinition of generic functions with no DEFGENERIC to an
335 incompatible lambda list now signals an error. (thanks to Zach
337 * bug fix: DEFGENERIC now works even when there's a function of the
338 same name in an enclosing lexical environment. (thanks to Zach
340 * fixed compiler failure, caused by instrumenting code during
341 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
342 * optimization: added loop analysis and improved register allocation
343 * fixed some bugs revealed by Paul Dietz' test suite:
344 ** POSITION on displaced vectors with non-zero displacement
345 returns the right answer.
346 ** (SIMPLE-STRING) is a valid type specifier for sequence
348 ** *PRINT-LEVEL* handling for slotless structures is pedantically
350 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
351 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
353 ** The justification version of the FORMAT ~< directive treats
354 non-zero minpad parameter correctly.
356 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
357 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
358 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
359 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
360 the supported interface.
361 * new feature: Single-stepping of code compiled with DEBUG 2 or
362 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
363 * new feature: saving cores with foreign code loaded is now
364 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
365 Timothy Moore's work for CMUCL)
366 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
367 parameters to * instead of NIL if no initform is supplied.
368 (reported by Johan Bockgaard)
369 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
370 correctly when FUNCALL appears as the car of the form. Note:
371 despite this FUNCALL forms are not currently subject to
372 compiler-macro expansion. (port of Raymond Toy's fix for the
373 same from CMUCL, reported by Johan Bockgaard)
374 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
375 (thanks for Teemu Kalvas)
376 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
377 case instead of terminating the iteration (thanks for Julian
378 Squires, Sean Champ and Raymond Toy)
379 * bug fix: incorrect expansion of defgeneric that caused a style
380 warning. (thanks for Zach Beane)
381 * bug fix: slot accessor effective method computation works properly
382 for classes with multiple non-standard applicable methods on
383 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
384 * on x86 compiler supports stack allocation of results of LIST and
385 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
386 CMUCL implementation by Gerd Moellmann)
387 * fixed some bugs revealed by Paul Dietz' test suite:
388 ** FORMAT strings with both the ~<~:;~> form of the justification
389 directive and pretty-printing directives cause an error.
391 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
392 * incompatible change: the internal functions
393 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
394 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
395 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
396 instead of the old functions.
397 * new feature: on platforms where "dladdr" is available foreign
398 function names now appear in backtraces. (based on Helmut Eller's
400 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
401 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
403 * documentation: ASDF manual is now distributed with SBCL.
404 * bug fix: SBCL can now load its contributed modules with REQUIRE
405 even if the system-provided entries have been removed from
406 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
408 * bug fix: backtraces involving undefined functions or assembly
409 routines are more informative. (thanks to Brian Downing)
410 * bug fix: mutually referent alien structures now work correctly.
411 (reported by Rick Taube)
412 * bug fix: structures defined by WITH-ALIEN can be referred to
413 within other definitions in the same WITH-ALIEN.
414 * bug fix: division operators (MOD, TRUNCATE and the like) with
415 constant zero divisors and integer dividends no longer generate
417 * bug fix: provide default methods for INPUT-STREAM-P and
418 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
419 * bug fix: improve the detection and resolution of MOP metacycles
420 (where effective method computation depends on the generic function
421 having its effective method computed). (reported by Bruno Haible)
422 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
423 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
424 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
425 * bug fix: pathnames with relative directory components can be
426 represented relative to default pathnames.
427 * optimization: in taking the GCD of bignums, reduce the two bignums
428 to approximately the same size (using Euclid's algorithm) before
429 applying the more sophisticated binary GCD. (thanks to Juho
431 * optimization: COUNT on bitvectors now operates word-at-a-time.
432 * optimization: ASH with a positive, but not necessarily constant,
433 (leftwards) shift, when in a modular context, is compiled to a
435 * fixed some bugs revealed by Paul Dietz' test suite:
436 ** FORMAT variable parameters ("~V<char>") are defaulted properly
437 if the corresponding argument is NIL.
438 ** FORMAT directives accepting a minpad parameter treat negative
440 ** literal commas as character parameters to FORMAT directives are
442 ** literal spaces directly after ~<Newline> directives within a
443 format-logical-block (~:< ~@:>) do not induce :FILL-style
444 conditional newlines.
445 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
446 as specified: it no longer includes conditional newlines).
447 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
449 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
450 :PER-LINE-PREFIX argument does not evaluate to a string.
452 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
453 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
454 the manual for details; for now, package locks can be disabled by
455 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
456 if no major problems are found then it is likely that they will be
457 compiled in unconditionally.
458 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
459 unsupported operators on all platforms. To load a shared library
460 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
461 object file, link it into a shared library outside of SBCL and
462 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
464 * bug workaround: ROOM T and the GC were not working together
465 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
466 reporting facility aren't preserved. That reporting has been
467 disabled, so now until and unless someone figures out how to make it
468 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
469 * minor incompatible change: as threatened around sbcl-0.8.0, the
470 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
471 The symbols are also exported from SB-PCL for backwards
472 compatibility, but more so than before SB-PCL should be treated as
473 an implementation-internal package.
474 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
476 * the SB-SPROF contrib now works on (most) non-x86 architectures.
477 It is known as of this release not to work on the Alpha, however.
478 * fixed bug #167: errors signalled due to illegal syntax in method
479 bodies are now more legible.
480 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
481 designators and can hence be used with TYPEP.
482 * fixed bug #333: CHECK-TYPE now ensures that the type error
483 signalled, if any, has the right object to be accessed by
484 TYPE-ERROR-DATUM. (reported by Tony Martinez)
485 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
486 5.1.2.3 for multiple-value place subforms. (reported by Kalle
488 * fixed bug #334: programmatic addition of slots using specialized
489 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
490 and :ALLOCATION :CLASS slots.
491 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
492 float exponents. (rereported by Peter Seibel)
493 * fixed a bug: #\Space (and other whitespace characters) are no
494 longer considered to be macro characters in standard syntax by
496 * fixed bug: initialization of condition class metaobjects no longer
497 causes an instance of the condition to be created. (reported by
499 * fixed bug: it is now possible to have slots such that
500 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
501 description is neither :INSTANCE nor :CLASS.
502 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
503 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
504 SLOT-VALUE-USING-CLASS), and no longer causes errors with
505 non-standard SLOT-DEFINITION-ALLOCATION values.
506 * fixed bugs: various race conditions handling exiting threads.
507 CL-PPCRE's thread torture test now passes.
508 * fixed arguably-a-bug: GC time proportional to number of threads in
509 system even when most of them are idle
510 * optimization: improved performance of BIT and SBIT on bit-vectors.
511 * .fasl file incompatibility: The fasl file version number has
512 been incremented because of changes associated with package locks.
514 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
515 * minor incompatible change: the system no longer provides
516 optimization hints (conditions which are TYPEP
517 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
518 mode; these hints are still emitted when higher SPEED optimization
520 * new contrib module: a sampling profiler (profiling by statistical
521 sampling, rather than by instrumenting functions) is available as
522 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
523 * the behaviour of the standard function ED is now customizeable by
524 third parties through a hook variable: see ED's documentation
525 string for information on the protocol.
526 * the compiler no longer emits efficiency notes for (FUNCALL X)
527 when the type of X is uncertain under default optimization
529 * fixed bug 276: mutating a binding of a specialized parameter to a
530 method to something that is not TYPEP the specializer is now
532 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
533 exhibit the required behaviour on the x86 platform. (thanks to
534 Peter van Eynde, Eric Marsden and Bruno Haible)
535 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
536 even for difficult arguments. (reported by Peter Graves)
537 * fixed bug 141a: the backquote printer now descends quoted
539 * fixed another bug in backquote printing: no more destructive
540 modification of the form's list structure. (reported by Brian
542 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
543 move between its address being taken and the call to
544 interrupt_thread, fixing a crashing race condition.
545 * the SB-POSIX contrib implementation has been adjusted so that it
546 no longer exhibits ridiculously poor performance when constructing
547 instances corresponding to C structs.
549 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
550 * minor incompatible change: the sb-grovel contrib now treats C
551 structures as alien (in the sense of SB-ALIEN) objects rather than
552 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
553 has implications for memory management of client code
554 (sb-grovel-returned objects must now be manually managed) and for
555 type safety (alien objects now have full types).
556 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
557 used to control emission of compiler diagnostics, rather than the
558 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
559 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
560 quality should be considered deprecated.
561 * install.sh now installs the user manual as well
562 * (not quite a new documentable feature, but worth considering in
563 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
564 beginnings of a semantically meaningful condition hierarchy is
565 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
567 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
568 designator as the defaults argument.
569 * fixed bug: Displaced arrays whose displaced-to array has become
570 too small now cause ARRAY-DIMENSION to signal an error, providing
571 for safer bounds-checking. (reported by Bruno Haible)
572 * fixed bug: DEFCLASS slot definitions with identical :READER and
573 :WRITER names now signal a reasonable error. (reported by Thomas
575 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
577 * fixed bug: Invalid slot specification errors now print correctly.
578 (thanks to Zach Beane)
579 * fixed bug 320: Shared to local slot value transfers in class
580 redefinitions now happen corrently from superclasses as
581 well. (reported by Bruno Haible)
582 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
584 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
585 improper lists correctly. (reported by Bruno Haible)
586 * fixed bug 313: source-transform for <fun-name> was erroneously
587 applied to a call of a value of a variable with name <fun-name>.
588 (reported by Antonio Menezes Leitao)
589 * fixed bug 307: The obsolete instance protocol ensures that
590 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
591 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
592 remove dead unknown-values globs from the middle of the stack.
593 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
594 easier to resume long computations after using *BREAK-ON-SIGNALS*
595 to diagnose and fix failures. (thanks to Nikodemus Siivola)
596 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
597 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
598 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
599 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
600 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
602 * fixed a (fairly theoretical) bug in string printing: if
603 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
604 does not have array-element-type equal to the most general string
606 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
607 function name. (thanks to Nikodemus Siivola)
608 * fixed bug: compile-time format string checker failed on
609 non-closed ~{. (reported by Thomas F Burdick)
610 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
611 don't warn on legal loop code involving more than one aggregate
612 boolean. (thanks to Nikodemus Siivola)
613 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
614 in keyword position to MAKE-PACKAGE was not being reported as
616 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
617 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
619 * fixed bugs 280 and 312: the checking for multiple definitions in a
620 file is less likely to become confused by uses of inline
622 * fixed bug: the #S reader macro performs the keyword coercion
623 specified for slot names. (reported by Kalle Niemitalo)
624 * fixed bug: lambda lists may contain symbols whose names start with
625 & but are not lambda-list-keywords; their occurrence triggers a
627 * fixed bug 321: define-method-combination argument lambda lists do
628 not cause invalid code to be generated when &optional or &aux
629 variables have default values. (reported by Bruno Haible)
630 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
631 have CLOS classes; this appears also to have fixed reported
632 instability in stack exhaustion detection.
633 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
634 MUFFLE-WARNING when no associated restart is present is now
636 * optimization: rearranged the expansion of various defining macros
637 so that each expands into only one top-level form in a
638 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
640 * optimization: used a previously little-used slot in symbols to
641 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
643 * fixed some bugs revealed by Paul Dietz' test suite:
644 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
646 ** Fixed an optimization bug related to inheritance of initforms
647 from local to shared slots.
648 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
649 ** CAST optimizer forgot to flush argument derived type.
650 ** print/read consistency on floats is now orders of magnitude
651 more likely. (thanks also to Bruno Haible for a similar report
653 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
654 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
655 for objects of type REAL. Make it so.
656 ** SXHASH is specified (infelicitously) to respect similarity,
657 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
658 it so. (thanks to Markus Ziegler)
659 ** on the Alpha, the compiler succeeds in compiling functions
660 returning a known number of arguments greater than 63.
661 ** fixed handling of invalid NIL arguments in keyword position
663 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
665 ** provided a readably-printable representation for RANDOM-STATE
667 ** ensured that pathnames, where they have a namestring, always
668 print using #P"..." syntax.
670 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
671 * Support for the forthcoming 2.0 version of the NetBSD kernel
672 running on x86 hardware has been added. (thanks to Perry
673 E. Metzger most immediately, and others for their past work)
674 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
675 current release version; SBCL's OpenBSD support had been broken
676 since about the time of OpenBSD's switch to ELF binary format.)
677 * [placeholder for DX summary]
678 ** user code with &REST lists declared dynamic-extent, under high
679 speed or space and low safety and debug optimization policy.
680 * The manual has been converted to Texinfo format and the debugger
681 chapter from the cmucl manual has been added.
682 * A facility has been added to extract documentation strings from
683 sbcl and store them as Texinfo-formatted snippets for inclusion in
684 the manual (via Texinfo's @include directive)
685 * bug fix: compiler emitted division in optimized DEREF. (thanks for
686 the test case to Dave Roberts)
687 * bug fix: multidimensional simple arrays loaded from FASLs had fill
688 pointers. (reported by Sean Ross)
689 * bug fix: PROFILE output is printed nicely even for large numerical
690 values. (thanks to Zach Beane)
691 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
692 greater than 32 handle EOF correctly.
693 * bug fix: on X86 an immediate argument of the IMUL instruction is
694 correctly printed in disassembly. (thanks to Lutz Euler)
695 * bug fix: class slots in redefined classes preserve their old
696 values. (thanks to Bruno Haible and Nikodemus Siivola)
697 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
698 succeeds. (reported by Marco Baringer)
699 * fixed some bugs revealed by Paul Dietz' test suite:
700 ** READ-SEQUENCE now works on ECHO-STREAMs.
701 ** RATIONALIZE works more according to its specification. (thanks
704 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
705 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
706 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
707 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
708 should matter to you only if you rebind the printer control
709 variables and then find you want different bindings in the
710 debugger than in the ordinary execution of your program.)
711 * The runtime build system has been tweaked to support building
712 (on SPARC/SunOS) using a C compiler which invokes Sun's own
713 assembler and linker. (thanks to Nikodemus Siivola)
714 * Unbound, undefined, undeclared variables now trigger full
715 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
716 more often programmer error than deliberate exploitation of undefined
718 * optimization: the hash algorithm for strings has changed to one
719 that is less vulnerable to spurious collisions. (thanks to Juho
721 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
722 needless bounds checking. (thanks to Juho Snellman)
723 * optimization: implemented multiplication as a modular
724 (UNSIGNED-BYTE 32) operation on the PPC backend.
725 * fixed some bugs revealed by Paul Dietz' test suite:
726 ** ADJUST-ARRAY now copies the datum in a zero rank array if
728 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
729 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
730 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
732 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
734 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
736 ** ECHO-STREAMs no longer attempt to echo the end of file value to
737 their output stream on EOF from read.
738 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
739 have been read to end-of-file.
740 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
742 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
743 description of determination of which consecutive characters
745 ** Printing the "Space" character with escaping on now yields "#\\ ",
746 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
747 ** Reading floating-point numbers with *READ-BASE* set to a number
748 less than 10 works correctly.
749 ** Reading floating-point numbers with *READ-BASE* set to a number
750 more than 10 works correctly.
751 ** Printing with *PRINT-READABLY* targets the standard readtable, not
752 the readtable currently in effect.
754 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
755 * minor incompatible change: parsing of namestrings on a physical
756 (Unix) host has changed; numbers after the final #\. in a
757 namestring are no longer interpreted as a version field. This is
758 intented to be largely invisible to the user, except that the
759 meaning of the namestring "*.*.*" has changed: it now refers to a
760 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
761 should usually be replaced by
762 (make-pathname :name :wild :type :wild :version :wild)
763 with the added benefit that this is more likely to be portable.
764 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
765 option to OPEN now signals an error if the file being opened
766 exists; this may have an impact on existing code.
767 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
768 signals received, on the PowerPC platforms (both Linux and
769 Darwin). (thanks to Pierre Mai for pointing out the location of
771 * several fixes on OS X: The system now builds and runs cleanly on
772 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
773 (thanks to Brian Mastenbrook)
774 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
775 second-resolution: integer multiples of 1/3600 between -24 and 24.
776 (thanks to Vincent Arkesteijn)
777 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
778 type when called with 1 argument; PEEK-CHAR checked type of
779 PEEK-TYPE only after having read first character from a
780 stream. (reported by Peter Graves)
781 * bug fix: the garbage collector now has much better locality
782 behaviour, and in particular no longer treats all memory as being
783 exhausted when many small objects point to each other in a deeply
785 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
786 recognized as being TYPEP their class.
787 * bug fix: the PUSHNEW documentation string has been corrected.
788 (thanks to Vincent Arkesteijn)
789 * bug fix: defaulting of the value for the last of an atypically
790 large number of multiple values being bound was not being
791 performed correctly on the Alpha or PPC platforms
792 * optimization: implemented multiplication as a modular
793 (UNSIGNED-BYTE 32) operation on the x86 backend.
794 * optimization: SEARCH on simple-base-strings can now be open-coded.
795 (see also contrib/compiler-extras.lisp for inspiration for
796 teaching the compiler about the Boyer-Moore algorithm).
797 * value, returned by MAX (and MIN) called with several EQUALP, but
798 not EQL, arguments now does not depend on compiler settings.
799 * fixed some bugs revealed by Paul Dietz' test suite:
800 ** in stack analysis liveness information is propagated from
801 non-local entry points.
802 ** pathwise CAST removing failed when the CAST node did not start
804 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
805 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
807 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
808 host is already defined.
809 ** RENAME-FILE works on streams instead of signalling an internal
811 ** PEEK-CHAR uses the current readtable when determining whether
812 or not a character is whitespace.
813 ** MERGE-PATHNAMES handles the case when the pathname does not
814 specify a name while the default-pathname specifies a version.
815 ** Pathnames now stand a better chance of respecting print/read
817 ** Attempting to use standardized file system operators with a
818 pathname with invalid :DIRECTORY components signals a
820 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
822 ** DIRECTORY on logical pathnames is more correct.
823 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
824 signal a TYPE-ERROR if their argument is not a stream.
825 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
826 designator argument does not designate a stream.
827 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
828 examining the synonym.
829 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
831 ** OPEN and WITH-OPEN-STREAM allow opening streams with
832 element-type larger than ([UN]SIGNED-BYTE 32).
834 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
835 * When built with the :SB-FUTEX feature, threaded builds now take
836 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
837 for faster/more reliable mutex and condition variable support.
838 * Incompatible change (but one you probably shouldn't have been using
839 anyway): the interface and code for arbitrating between multiple
840 threads in the same user session has been redesigned.
841 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
842 error when called without an explicit environment argument.
843 (thanks to Nikodemus Siivola)
844 * bug fix: buffered :DIRECTION :IO streams are less likely to become
845 confused about their position. (thanks to Adam Warner and Gerd
847 * bug fix: Pretty printing backquoted forms with unquotations in the
848 argument list position of various code constructs such as LAMBDA
849 now works correctly. (reported by Paul Dietz)
850 * bug fix: Pretty printing unquotations no longer loses all
851 stream position information.
852 * optimization: performance of string output streams is now less
853 poor for multiple small sequence writes.
854 * optimization: performance of CSUBTYPEP in the presence of complex
855 expressions involving CONS and NOT many times has been improved.
856 (reported by Paul Dietz)
857 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
859 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
860 streams for the REPL, for the debugger, and so on. (thanks to
862 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
863 * fixed some bugs revealed by Paul Dietz' test suite:
864 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
865 is ignored if it is too small, rather than propagating through
866 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
867 ** extremely complex negations of CONS types were not being
868 sufficiently canonicalized, leading to inconsistencies in
870 ** VALUES tranformer lost derived type.
872 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
873 * fixed a bootstrapping bug: the build process no longer assumes
874 that the various BOOLE-related constants have the same value in
875 host and target lisps. (noted by Paul Dietz' test suite on an
876 SBCL binary built from CLISP)
877 * The system can now be dynamically linked on the MIPS platform,
878 which enables dynamic loading of foreign code from Lisp. (thanks
879 to Ralf Baechle for discussions on the MIPS ABI)
880 * The system now records debugging information for its own source
881 files in a filesystem-position-independent manner, relative to
882 the "SYS" logical host.
883 * fixed a compiler bug: MV-LET convertion did not check references
884 to the "max args" entry point. (reported by Brian Downing)
885 * tweaked disassembly notes to be less confident about proclaiming
886 some instruction as an LRA. (thanks to Brian Downing)
887 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
888 now each have their own history, command character, and other
889 characteristics. (thanks to David Lichteblau)
890 * fixed some bugs revealed by Paul Dietz' test suite:
891 ** compiler failure in compiling LOGAND expressions including a
893 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
894 shift greater than 32.
895 ** FLUSH-DEST did not mark blocks for type check regeneration.
896 ** HANDLER-CASE failed to accept declarations in handler clauses
897 in some circumstances.
899 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
900 * New code in contrib/sb-introspect (still probably not entirely
901 stable yet) provides some support for smart Lisp development
902 environments like SLIME.
903 * The conditions signalled for errors occurring when loading .fasl
904 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
905 in a way which should help ASDF recover gracefully.
906 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
907 stuff has been cleaned up. If you code contrib/ stuff, this might
908 affect you, and you can look at contrib/README, contrib/STANDARDS,
909 and/or the 0.8.4.27 diff to check.
910 * In full calls the compiler now does not generate checks for declared
911 argument types for all arguments.
912 * various threading fixes
913 ** and some experimental patches which didn't make it into
914 the main tree for this release, but which are shipped in
915 contrib/experimental-thread.patch as a possible fix for some
916 failures (deadlock, spinning...) in GC-intensive multithreaded
918 * fixed PPC build problem (source code incompatibility of different
919 library versions): added offsetof() hackery which attempts to divine
920 where glibc maintainers put uc_mcontext today
921 * fixed bug 282: compiler does not trust type assertions while passing
922 arguments to a full call.
923 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
924 &OPTIONAL VALUES type parameter.
925 * fix bug 214: algorithm for noting rejected templates is now more
926 similar to that of template seletion. (also reported by rydis on
928 * fixed bug 141b: printing backquoted information readably and prettily
929 inserts a space where necessary.
930 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
931 (CHARACTER 10) are now reported as errors, rather than propagated
932 as unknown types. (reported by piso on #lisp)
933 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
934 with values NIL and :ERROR. (thanks to Milan Zamazal)
935 * fixed bug 191c: CLOS now does proper keyword argument checking as
936 described in CLHS 7.6.5 and 7.6.5.1.
937 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
938 longer raise an error; further, using a list as a for-as-arithmetic
939 counter now raises a meaningful error.
940 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
941 now understood by sequence creation functions such as MAKE-SEQUENCE
943 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
944 a STRING-INPUT-STREAM.
945 * compiler enhancement: SIGNUM is now better able to derive the type
947 * type declarations inside WITH-SLOTS are checked. (reported by
949 * fixed some bugs revealed by Paul Dietz' test suite:
950 ** incorrect optimization of TRUNCATE for a positive first
951 argument and negative second.
952 ** compiler failure in let-convertion during flushing dead code.
953 ** compiler failure while deriving type of TRUNCATE on an
954 interval, containing 0.
955 ** ASH of a negative bignum by a negative bignum count now returns
957 ** intersection of CONS types now canonicalizes properly, fixing
958 inconsistencies in SUBTYPEP.
960 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
961 * incompatible change: The --disable-debugger command line
962 option now clobbers the debugger at a more fundamental
963 level, by redefining #'INVOKE-DEBUGGER instead of by
964 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
965 is specified by ANSI to ignore *DEBUGGER-HOOK* and
966 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
967 enter the debugger REPL and then suffer recursive errors
968 because *DEBUG-IO* is also messed up in --disable-debugger mode;
969 while under the new system, BREAK in --disable-debugger mode
970 terminates the system just as an unhandled error would.
971 * fixed compiler performance when processing loops with a step >1;
972 * bug fix: DOCUMENTATION now retrieves generic function
973 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
974 support has been systematized, and now supports the methods
975 specified by ANSI, along with a default method and a method for
976 slot documentation. (reported by Nathan Froyd)
977 * bug fix: effective methods associated with a generic function are
978 no longer cached over a change of that generic function's method
979 combination. (reported by Andreas Fuchs)
980 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
981 names a non-existent directory. (thanks to Andreas Fuchs)
982 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
983 the PPC platform, lead to stack corruption; this has been fixed.
984 (reported by Rainer Joswig)
985 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
986 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
987 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
988 arithmetic in general, is now much more reliable on the PPC
990 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
991 platform now returns the right answer.
992 * optimization: restored some effective method precomputation in
993 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
994 precomputation is now tunable.
995 * optimization: compiler-internal data structure use has been
996 reviewed, and changes have been made that should improve the
997 performance of the compiler by about 20%.
998 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
999 simple-base-strings and simple-bit-vectors is improved.
1000 * optimization: the optimization of 32-bit logical and arithmetic
1001 functions introduced in version 0.8.3 on the x86 has been
1002 implemented on the mips, ppc and sparc platforms; an
1003 implementation of the same facility, but for 64-bit arithmetic,
1004 has been added for the alpha.
1005 * microoptimization: the compiler is better able to make use of the
1006 x86 LEA instruction for multiplication by constants.
1007 * bug fix: in some situations compiler did not report usage of
1008 generic arithmetic in (SPEED 3) policy.
1009 * bug 145b fix: compiler used wrong type specifier while converting
1010 MEMBER-types to numeric.
1011 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1013 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1014 index argument now works properly on the Alpha platform.
1015 * bug fix: floating point exception treatment on the Alpha platform
1017 * bug fix: FILE-POSITION works much better on string input and
1018 output streams. (thanks to Nikodemus Siivola)
1019 * bug fix: many threading/garbage collection symptoms sorted.
1020 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1021 might be pseudo-atomic.
1022 * internal change: Stopping for GC is now done with signals not
1023 ptrace. GC is now done in whichever thread wanted it, instead of
1025 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1027 * bug fix: result form in DO is not contained in the implicit
1029 * incompatible change: ICR structure is changed; the value part of
1030 CONTINUATION is now called LVAR; corresponding functions are
1031 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1032 * added type deriver for ISQRT (thanks to Robert E. Brown).
1033 * bug fix: better support for loading from the command line when an
1034 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1036 * fixed some bugs revealed by Paul Dietz' test suite:
1037 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1038 ** ROUND and FROUND now give the right answer when given very
1039 small float arguments.
1040 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1042 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1043 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1044 ** LCM should return a non-negative integer.
1045 ** PARSE-INTEGER returned the index of a terminator instead of the
1046 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1047 ** PARSE-INTEGER returned an incorrect index being applied to a
1049 ** LCM with two arguments of 0 returns 0 rather than signalling
1051 ** unsigned addition of a 32-bit constant with the high bit set no
1052 longer causes an internal compiler error on the x86.
1053 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1054 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1055 with negative last argument.
1056 ** byte specifiers with zero size and position no longer cause
1057 an error during type derivation.
1058 ** bignum multiplication on the Alpha platform now returns the
1060 * porting: The system now builds on SuSE AMD64, although it still
1061 generates a 32-bit binary.
1062 * .fasl file incompatibility: The fasl file version number has
1063 been incremented (because of the changes to internal compiler
1064 data structures referred to above).
1066 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1067 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1068 more accurately, on the Darwin kernel running on PowerPC hardware.
1069 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1070 * Compiler code deletion notes now signal a condition of type
1071 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1072 an associated MUFFLE-WARNING restart.
1073 * The compiler now performs limited argument count validation of
1074 constant format strings in FORMAT, and where appropriate in ERROR,
1075 CERROR and WARN. (thanks to Gerd Moellmann)
1076 * New ASDF-INSTALL contrib can be used for automatic download and
1077 installation of third-party Lisp code from CCLAN or other sites
1079 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1080 which forces another thread to execute a function supplied by the
1082 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1083 now accept and act upon their :ELEMENT-TYPE keyword argument.
1084 (reported by Martin Atzmueller, Edi Weitz)
1085 * bug fix: FILE-POSITION now accepts position designators up to
1086 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1087 is the greater. (thanks to Patrik Nordebo)
1088 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1089 Kalle Olavi Niemitalo)
1090 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1091 printable. (reported by Eric Marsden)
1092 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1093 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1094 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1095 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1096 (thanks to Stig Sandoe)
1097 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1098 UNREAD-CHAR now correctly includes the unread character in the
1099 target sequence. (thanks to Gerd Moellmann)
1100 * bug fix in threaded builds: the system can now be suspended and
1101 resumed by shell job control with minimal disruption.
1102 * bug fixes in times and timezones >2038AD
1103 * better handling of "where is GNU make?" problem in build scripts
1104 (thanks to Nikodemus Siivola)
1105 * new optimization: inside a named function any reference to a
1106 function with the same name is considered to be a self-reference;
1107 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1108 optimization quality.
1109 * new optimization on x86: logical functions and + now have
1110 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1111 used when the result is truncated to 32 bits.
1112 * VALUES declaration is partially enabled.
1113 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1114 * fixed some bugs revealed by Paul Dietz' test suite:
1115 ** The system now obeys the constraint imposed by
1116 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1117 types form a lattice under type intersection.
1118 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1119 ** ASSOC now ignores NIL elements in an alist.
1120 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1121 and (1+ MOST-POSITIVE-FIXNUM) answers.
1122 ** The addition of a method with invalid qualifiers to a generic
1123 function does not cause an error to be signalled immediately;
1124 a warning is signalled, and the error is generated only on
1125 calling the generic function.
1126 * changed .fasl file version number, in part to add type codes for
1127 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1128 obscure ANSI requirements
1130 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1131 * fixed bug 148: failure to inline-expand a local function left
1132 garbage, confusing the compiler.
1133 * fixed bugs 3cd: structure slot readers perform type check if the
1134 slot can have an invalid value (i.e. it is either not initialized
1135 or can be written with a less specific slot writer).
1136 * bug fix: the compiler now traps array references to elements off
1137 the end of an array; previously, the bounds checking in some
1138 circumstances could go off-by-one.
1139 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1141 * type declarations for array element types now obey the description
1142 on the CLHS page "Declaration TYPE", as per discussions on
1143 sbcl-help around 2003-05-08. This means that a declaration
1144 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1145 declaration, all references to BAR will be asserted or assumed
1146 (with THE, so dependent on compiler policy) to involve objects of
1147 type FOO. Note that no such declaration is implied in
1148 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1149 * declared types of functions from the "Conditions"
1150 chapter. (reported by Paul Dietz)
1151 * bug fix: CERROR accepts a function as its first argument.
1152 * bug fix: NTH an NTHCDR accept a bignum as index
1153 arguments. (reported by Adam Warner)
1154 * optimization: character compare routines now optimize comparing
1155 against a constant character. (reported by Gilbert Baumann)
1156 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1157 argument now works properly on the MIPS platform.
1158 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1159 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1160 declared to have a functional type, does not cause an error
1162 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1163 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1164 * fixed bug: if last continuation of a deleted block has a
1165 destination, this destination should be deleted too. (reported by
1167 * fixed a bug in the bootstrap process: the host compiler's values
1168 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1169 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1170 test case from Patrik Nordebo)
1171 * improved the ability of the disassembler on the PPC platform to
1172 provide helpful disassembly notes.
1173 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1174 the class in more cases than previously.
1175 * bug fix: FILE-POSITION now understands :START and :END for
1176 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1177 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1178 8) data. (thanks to David Lichteblau)
1179 * bug fix: it is possible to add a method to a generic function
1180 without lambda list.
1181 * bug fix: reader failed to signal END-OF-FILE inside an
1182 object representation. (reported by Nikodemus Siivola)
1183 * fixed some bugs revealed by Paul Dietz' test suite:
1184 ** LAST and [N]BUTLAST should accept a bignum.
1185 ** condition slot accessors are methods.
1186 ** (VECTOR NIL) is a subtype of STRING.
1188 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1189 * minor incompatible change: some nonsensical specialized lambda
1190 lists (used in DEFMETHOD) which were previously ignored now signal
1192 * minor incompatible change: the system is now aware of the types of
1193 variables in the COMMON-LISP package, and will signal errors for
1194 most violations of these type constraints (where previously they
1195 were silently accepted).
1196 * minor incompatible change: COMPILE-FILE now uses the freedom
1197 afforded (ANSI 3.2.2.3) to use derived function types for
1198 functions defined in the same file. This also permits the system
1199 to warn on static type mismatches and function
1200 redefinition. (Currently it does not work with high DEBUG level.)
1201 * minor incompatible change: VALUES declaration is disabled.
1202 * When issuing notes, the compiler now signals a condition of type
1203 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1204 restart for use in user handlers. It is expected that the
1205 COMPILER-NOTE condition will eventually become a condition
1206 supertype to a hierarchy of note types, which will then be
1207 handleable in a similar fashion. However, at the moment, no such
1208 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1209 but it's an implementation detail, not a classification for the
1211 * Changes in type checking closed the following bugs:
1212 ** type checking of unused values (192b, 194d, 203);
1213 ** template selection based on unsafe type assertions (192c, 236);
1214 ** type checking in branches (194bc).
1215 * A short form of VALUES type specifier has ANSI meaning (it has
1216 increased the number of situations when SBCL cannot perform type
1218 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1219 keywords or constants is permissible.
1220 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1221 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1222 classes. (thanks to Antonio Martinez)
1223 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1224 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1225 argument to a pretty-print stream. Instead, it leaves any such
1226 operations to DESCRIBE-OBJECT methods.
1227 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1229 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1230 lambda lists. (thanks to David Lichteblau)
1231 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1232 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1233 (thanks to Gerd Moellmann)
1234 * fixed bug 246: increased compilation speed of long
1235 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1237 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1239 * DEFCONSTANT now throws a condition of type
1240 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1241 constant to a non-EQL value; CONTINUE and ABORT restarts
1242 respectively change and preserve the value.
1243 * fixed bug 63: The code walker, part of the implementation of CLOS,
1244 is now better at handling symbol macros.
1245 * bug fix: There is no longer an internal implementation type named
1246 CL:LENGTH. (reported by Raymond Toy)
1247 * bug fix: In macro-like defining macros/special operators the
1248 implicit block does not enclose the lambda list.
1249 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1250 enclosed in parentheses) are not suitable as type specifiers, and
1251 their use properly signals an error now.
1252 * bug fix: An argument count mismatch for a type specifier in code
1253 being compiled no longer causes an unhandled error at compile
1254 time, but signals a compile-time warning.
1255 * fixed simple vector readable printing
1256 * bug fix: DESCRIBE takes more care over whether the class
1257 precedence list slot of a class is bound before accessing it.
1258 (reported by Markus Krummenacker)
1259 * bug fix: FORMATTER can successfully compile pretty-printer format
1260 strings which use variants of the ~* directive inside.
1261 * bug fix: SEARCH now applies its TEST predicate to the elements of
1262 the arguments in the correct order. (thanks to Wolfhard Buss)
1263 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1264 in type inference. (thanks to Robert E. Brown)
1265 * bug fix: Reading in symbols with an explicit package name of ""
1266 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1267 * fixed some bugs revealed by Paul Dietz' test suite:
1268 ** NIL is now allowed as a structure slot name.
1269 ** Arbitrary numbers, not just REALs, are allowed in certain
1270 circumstances in LOOP for-as-arithmetic clauses.
1271 ** Multiple class redefinitions before slot access no longer
1272 causes a type error.
1273 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1274 association between the name and a class.
1275 ** Generic functions with non-standard method-combination and over
1276 five methods all of which return constants no longer return NIL
1277 after the first few invocations. (thanks to Gerd Moellmann)
1278 ** CALL-NEXT-METHOD with no arguments now passes the original
1279 values of the arguments, even in the presence of assignment.
1280 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1282 ** DISASSEMBLE works with closures and funcallable instances.
1283 ** ADD-METHOD now returns the generic function, not the new method.
1284 ** FIND-METHOD signals an error if the lengths of the specializers
1285 is incompatible with the generic function, even if the ERRORP
1287 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1288 which its argument is a member.
1289 ** DEFCLASS only redefines the class named by its class-name
1290 argument if that name is the proper name of the class;
1291 otherwise, it creates a new class.
1292 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1293 of the UNBOUND-SLOT condition to the name of the slot.
1294 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1295 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1296 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1297 SLOT-MAKUNBOUND in the specified fashion.
1299 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1300 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1301 cross-compilation host. As a consequence, we can now bootstrap our
1302 way up to SBCL starting with a bare gcc toolchain and human-readable
1303 source code (first the source to CLISP, then the source to SBCL).
1304 * A contributed module containing a partial implementation of the
1305 simple-streams interface has been included. (thanks to Rudi
1307 * A contributed module implementing the RFC1321 Message Digest
1308 Algorithm, known as MD5, has been included.
1309 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1310 no longer has any effect, as the code controlled by this feature
1311 has been deleted. (As far as we know, no-one has ever built using
1312 this feature, and its semantics were confused in any case).
1313 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1314 work on conditions (as required by the ANSI specification),
1315 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1316 expected behaviour on conditions. Users should note, however,
1317 that such behaviour is not required by the ANSI specification,
1318 and so use of this behaviour may render their code unportable.
1319 * fixed some bugs revealed by Paul Dietz' test suite:
1320 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1322 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1323 ** SXHASH on condition objects no longer returns NIL.
1324 ** :ALLOCATION :CLASS slots are better treated; their values are
1325 updated on class redefinition, and initforms inherited from
1326 superclasses are applied.
1327 ** REMOVE-METHOD returns its generic function argument even when
1328 no method was removed.
1329 ** SHARED-INITIALIZE now initializes the values of the requested
1330 slots, including those with :ALLOCATION :CLASS.
1331 ** ALLOCATE-INSTANCE now works on structure classes defined via
1332 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1334 ** SLOT-EXISTS-P now works on conditions, as well as structures
1336 ** MAKE-LOAD-FORM now has the required methods on
1337 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1338 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1339 keyword, and now implements the SLOT-NAMES argument.
1340 ** methods with &OPTIONAL arguments no longer allow too many
1341 arguments to be passed in the call without error.
1342 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1343 option is consistent with the required arguments of the generic
1344 function lambda list.
1345 * bug fix: REQUIRE accepts a string designator. (Thanks to
1347 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1348 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1349 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1350 * bug fix: The compiler now checks for duplicated variables in macro
1352 * bug fix: SETQ on globals returns the correct value.
1353 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1354 not a COMPILER-ERROR (followed by some other strange error on
1355 choosing the CONTINUE restart).
1356 * bug fix: make.sh and friends are now more consistent in the way that
1357 they look for GNU "make".
1359 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1360 * experimental native threads support (on x86 Linux >=2.4 only).
1361 This is not compiled in by default: you need to add :SB-THREAD to
1362 the target features. See the "Beyond ANSI" chapter of the manual
1364 * fix for longstanding nonANSIism: The old distinction between
1365 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1366 The return value from CL:FIND-CLASS is now a CLOS class, and
1367 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1368 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1369 * An interface to the MetaObject Protocol, as described in Kiczales,
1370 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1371 MIT Press, 1991, is available from the SB-MOP package.
1372 * incompatible change: the SB-PCL package should now be considered
1373 a private implementation detail, and no longer a semi-private MOP
1375 * minor incompatible change: due to rearrangement for threads, the
1376 control stack and binding stack are now allocated at arbitrary
1377 addresses instead of being hardcoded per-port. Users affected by
1378 this probably have to be doing advanced things with shared
1379 libraries, and will know who they are.
1380 * minor incompatible change: Previously, all --eval forms used were
1381 processed with READ before any of them were processed with EVAL.
1382 Now each --eval form is processed with both READ and EVAL before
1383 the next --eval form is processed. (Thus package operations like
1384 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1385 work as the user might reasonably expect.)
1386 * minor incompatible change: *STANDARD-INPUT* is now only an
1387 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1389 * minor incompatible change: Y-OR-N-P is now character-oriented,
1390 not line oriented. Also, YES-OR-NO-P now works without errors.
1391 (thanks to Antonio Martinez)
1392 * sb-aclrepl module improvements: an integrated inspector, added
1393 repl features, and a bug fix to :trace command.
1394 * Known functions, which cannot be open coded by the backend, are
1395 considered to be able to check types of their arguments. (fixing
1396 a bug report by Nathan J. Froyd)
1397 * fixed a bug in computing method discriminating functions: It is
1398 now possible to define methods specialized on classes which have
1399 forward-referenced superclasses. (thanks to Gerd Moellmann)
1400 * fixed evaluation order in optional entries (reported by Gilbert
1402 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1403 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1404 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1405 required-by-AMOP NAME argument, as well as CLASS and
1406 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1407 * fixed bug 20: DEFMETHOD can define methods using names that are
1408 not the proper names of classes to designate class specializers.
1409 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1410 Unix file descriptors, instead of blowing up. (thanks to Antonio
1412 * Garbage collection refactoring: user-visible change is that a
1413 call to the GC function during WITHOUT-GCING will not do garbage
1414 collection until the end of the WITHOUT-GCING. If you were doing
1415 this you were probably losing anyway.
1416 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1417 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1418 TYPEP the latter but not the former.
1419 * The compiler issues a full WARNING for calls to undefined functions
1420 with names from the CL package.
1421 * MAP-INTO for a vector destination is open coded. (reported by
1422 Brian Downing on c.l.l)
1423 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1424 documentation string.
1425 * fixed some bugs revealed by Paul Dietz' test suite:
1426 ** COPY-ALIST now signals an error if its argument is a dotted
1428 ** Condition slots are now accessed more correctly in the presence
1429 of multiple initargs for a given slot.
1430 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1431 exclude restarts of the same name associated with a different
1433 ** DEFCLASS of forward-referenced classes with another
1434 forward-referenced class in the superclasses list no longer
1436 ** Condition slots are now initialized once each, not multiple
1437 times. (thanks to Gerd Moellmann)
1438 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1440 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1441 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1442 arguments contain duplicated elements.
1443 ** RESTART-CASE understands local macros.
1444 ** RESTART-CASE associates exactly its own restarts with a condition.
1445 ** ENDP in safe mode checks its argument to be of type LIST.
1446 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1447 in question is unbound.
1448 ** Optimized MAKE-INSTANCE functions no longer cause internal
1449 assertion failures in the presence of duplicate initargs.
1450 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1451 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1452 propagated even in the ctor optimized implementation of
1454 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1456 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1457 * a better implementation of SXHASH on (simple) bit vectors,
1458 measured both in execution speed and in distribution of results
1459 over the positive fixnums, has been installed. Likewise, a better
1460 implementation of EQUAL for simple bit vectors is now available.
1461 * fixed CEILING optimization for a divisor of form 2^k.
1462 * fixed bug 240 (emitting extra style warnings "using the lexical
1463 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1464 by Antonio Martinez)
1465 * fixed SXHASH, giving different results for NIL depending on type
1466 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1467 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1468 variable. (found by Rolf Wester)
1469 * SBCL does not ignore type declarations for special
1470 variables. (reported by rif on c.l.l 2003-03-05)
1471 * some bug fixes in contrib/sb-aclrepl/
1472 * fixed some bugs revealed by Paul Dietz' test suite:
1473 ** a bug in the CONS type specifier, whereby the CAR and CDR
1474 types got intertwined, has been fixed;
1475 ** the type system is now able to reason about the interaction
1476 between INTEGER and RATIO types more completely;
1477 ** APPEND, [N]REVERSE and NRECONC check that those their
1478 arguments, which must be proper lists, are really so;
1479 ** An array specialized to be unable to hold elements has been
1480 implemented, as required -- yes, really -- by ANSI;
1481 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1482 on malformed property lists;
1484 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1485 * incompatible packaging change: in line with Unix convention,
1486 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1487 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1488 * REQUIRE and PROVIDE are now optionally capable of doing something
1489 useful. See the documentation string for REQUIRE.
1490 * infrastructure for a managed SBCL contrib system: contributed
1491 modules in this release include:
1492 ** the ASDF system definition facility;
1493 ** an interface to the BSD Sockets API;
1494 ** an ACL-like convenience interface to the repl;
1495 (thanks to Kevin Rosenberg)
1496 ** an implementation of ROTATE-BYTE, with efficient implementation
1498 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1499 gives a closer approximation to the right answer than previously.
1500 (thanks to Raymond Toy)
1501 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1502 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1503 optional environment arguments, as required by ANSI.
1504 * fixed bugs in other functions taking environment objects, allowing
1505 calls with an explicit NIL environment argument to be compiled
1507 * fixed bug 228: primary return values from
1508 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1509 COMPILE or FUNCTION.
1510 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1511 :TYPE will no longer signal errors on innocuous objects.
1512 * fixed bug 231b: SETQ is better at respecting type declarations in
1513 the lexical environment.
1514 * fixed a bug in DEFCLASS: classes named by symbols with no or
1515 unprintable packages can now be defined.
1516 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1517 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1518 * fixed compiler failure related to checking types of functions.
1519 (reported by Robert E. Brown)
1520 * the compiler is now much more consistent in its error-checking
1521 treatment of bounding index arguments to sequence functions: in
1522 (SAFETY 3) code, errors will be signalled in almost all cases if
1523 invalid sequence bounding indices are passed to functions defined
1524 by ANSI to operate on sequences.
1525 * fixed a bug in the build procedure: documentation of SBCL-specific
1526 packages is now preserved and available in the final Lisp image.
1527 * lifted FDEFINITION lookup out of loops in the implementation of
1528 many list operations. (thanks to Robert E. Brown)
1529 * fixed a bug in the reader: the #n# reader macro now works for
1530 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1531 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1532 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1533 E. Brown and Tony Martinez respectively)
1534 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1535 SLOT-BOUNDP: in particular, it is now able to optimize them much
1536 better, and is now not vulnerable to having packages renamed.
1537 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1538 always called when a slot is not present in an instance. (thanks
1540 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1541 longer signal an error on unreading a character following EOF on
1542 the previous constituent stream. (thanks to Tony Martinez)
1543 * fixed some bugs revealed by Paul Dietz' test suite:
1544 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1545 not just nonnegative fixnums;
1546 ** the logical bit-array operators such as BIT-AND now accept an
1547 explicit NIL for their "opt-arg" argument (to indicate a
1548 freshly-consed result bit-array);
1549 ** ELT now signals an error on an invalid sequence index in safe
1551 ** the type system is now cleverer about negations of numeric
1552 types, and consequently understands the BIGNUM and RATIO types
1554 ** the type system is now cleverer about the interaction between
1555 INTEGER and RATIO types: while bugs still remain, many more
1556 cases are accurately computed;
1557 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1558 if it is in the last clause;
1559 ** CONSTANTLY now correctly returns a side-effect-free function in
1561 ** DECLARE is no longer treated as a special-operator; in
1562 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1563 * incremented fasl file version number due to the change in the
1564 DEFSTRUCT-SLOT-DESCRIPTION structure.
1566 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1567 * minor incompatible change: code processed by the "interpreter" or
1568 EVAL now has a compilation optimization policy of (DEBUG 2)
1569 (changed from (DEBUG 1)) to improve debuggability of interactive
1570 development, and to allow the use of the debug RETURN command in
1572 * an experimental implementation of the RETURN command for the
1573 debugger has been included. (thanks to Frederik Kuivinen)
1574 * fixed bug 62: constraints were not propagated into a loop.
1575 * fixed bug in embedded calls of SORT (reported and investigated by
1577 * fixed some bugs revealed by Paul Dietz' test suite:
1578 ** printing and reading of arrays with some dimensions having
1579 length 0 (thanks to Gerd Moellmann);
1580 ** BOA constructor with &AUX argument without a default value does
1581 not cause a type error;
1582 ** CONSTANTP now returns true for all self-evaluating objects.
1584 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1585 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1586 accessors that are related by inheritance, as specified in the
1587 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1588 Valtteri Vuorikoski)
1589 * The compiler is now able to inline functions that were defined in
1590 a complex lexical environment (e.g. inside a MACROLET).
1591 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1593 * Support for the upcoming FreeBSD-5.0 release has been included.
1594 (thanks to Dag-Erling Smorgrav)
1595 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1596 effect when it is not in a toplevel context.
1597 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1598 stand a better chance of being correct. (thanks to Gerd
1600 * fixed bug in COERCE, which now signals an error on coercing a
1601 rational to a bounded real type which excludes the expected
1603 * The compiler is now able to derive types more accurately from the
1604 COERCE and COMPILE functions.
1605 * fixed bug 223: functional binding is considered to be constant
1606 only for symbols in the CL package.
1607 * fixed bug 231: SETQ did not check the type of a variable being set
1608 (reported by Robert E. Brown)
1609 * A new optimization for MAKE-INSTANCE has been included, fixing
1610 various bugs (including relating to :ALLOCATION :CLASS slots and
1611 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1613 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1614 ** As required by ANSI, LOOP now disallows anonymous collection
1615 clauses such as COLLECT I in conjunction with aggregate boolean
1616 clauses such as THEREIS (= I 1);
1617 ** LOOP now signals an error when any variable is reused in the
1618 same loop (including the potentially useful construct analogous
1619 to WITH A = 1 WITH A = (1+ A);
1620 ** IT is only a special loop symbol within the first clause of a
1621 conditional loop clause;
1622 ** LOOP with a typed iteration variable over a hashtable now
1623 signals a type error iff it should.
1624 * fixed some other bugs revealed by Paul Dietz' test suite:
1625 ** FILE-STREAM now names the class previously known as FD-STREAM;
1626 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1627 argument) no longer signals an error;
1628 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1629 of the slot symbol, rather than using the current package
1630 ((:CONC-NAME "") continues to intern the slot's name in the
1632 * incremented fasl file version number, because of the incompatible
1633 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1634 of the new implementation of DEFINE-COMPILER-MACRO.
1636 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1637 * Support for building SBCL for MIPS platforms running in
1638 little-endian mode has now been checked in, and basic
1639 functionality on said platforms verified.
1640 * minor incompatible change: PCL now records the pathname of a file
1641 in which methods and the like are defined, rather than its
1643 * minor incompatible change: TRUENAME now considers the truename of
1644 a file naming a directory to be the pathname with :DIRECTORY
1645 component indicating that directory.
1646 * minor incompatible change: a NAMED clause in the extended form of
1647 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1648 reason for the previous behaviour is unclear.
1649 * more systematization and improvement of CLOS and MOP conformance
1650 in PCL (thanks to Gerd Moellman and Pierre Mai):
1651 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1653 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1654 primary methods with no specializers;
1655 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1657 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1658 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1659 FORWARD-REFERENCED-CLASSes; error reporting on
1660 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1662 ** SXHASH on CLOS instances now uses a slot internal to the
1663 instance to return different numbers on distinct instances,
1664 while preserving the same return value through invocations of
1666 ** DEFMETHOD signals errors when methods with longer incongruent
1667 lambda lists are added to generic functions;
1668 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1669 CLASS, as specified in AMOP;
1670 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1671 on the order returned by the primary method for classes of
1672 class STANDARD-CLASS;
1673 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1674 * fixed some bugs shown by Paul Dietz' test suite:
1675 ** DOLIST puts its body in TAGBODY;
1676 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1678 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1679 value producing form;
1680 ** if more variables are given to PROGV than values, extra
1681 variables are bound and made to have no value;
1682 ** NSUBSTITUTE on list arguments gets the right answer with
1684 ** ELT signals an error of type TYPE-ERROR when the index argument
1685 is not a valid sequence index;
1686 ** LOOP signals (at macroexpansion time) an error of type
1687 PROGRAM-ERROR when duplicate variable names are found;
1688 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1689 ** FUNCALL of special-operators now cause an error of type
1691 ** PSETQ now works as required in the presence of side-effecting
1692 symbol-macro places;
1693 ** NCONC accepts any object as its last argument;
1694 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1696 ** loop-for-as-package does not require a package to be explicitely
1698 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1700 * fixed bug 166: compiler preserves "there is a way to go"
1701 invariant when deleting code.
1702 * fixed bug 172: macro lambda lists with required arguments after
1703 &REST arguments now cause an error to be signalled. (thanks to
1705 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1706 bug. (thanks to Matthew Danish)
1707 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1709 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1710 arguments are lexically rebound. (thanks to Gerd Moellmann and
1712 * fixed bug 194: error messages are now more informative when there
1713 is no primary method applicable in a call to a generic
1714 function. (thanks to Gerd Moellmann)
1715 * fixed bug in command line argument checking (thanks to Julian
1717 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1718 (thanks to Matthew Danish)
1719 * incremented fasl file version number, because of the
1720 SXHASH-related changes in the layout of CLOS data structures
1722 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1723 * minor incompatible change: The runtime (the Unix executable named
1724 "sbcl") is now much pickier about the .core files it will load.
1725 Essentially it now requires .core files to descend from the same
1726 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1727 as the runtime does. (The intent is to prevent the crashes which
1728 can occur, and which can even be reported as mysterious failures,
1729 when people patch the sources or change the build parameters
1730 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1731 sbcl and .core files.)
1732 * fixed bug: VALUES-LIST is no longer optimized away.
1733 * fixed bug 142: The FFI conversion of C string values to Lisp
1734 string values no longer conses excessively. (thanks to Nathan
1735 Froyd porting Raymond Toy's fix to CMU CL)
1736 * began to systematize and improve MOP conformance in PCL (thanks to
1737 Nathan Froyd, Gerd Moellman and Pierre Mai):
1738 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1740 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1741 ** FINALIZE-INHERITANCE is now called on class finalization;
1742 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1743 argument precedence order.
1744 * fixed bug 202: The compiler no longer fails on functions whose
1745 derived types contradict their declared type.
1746 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1747 so it can be non-toplevel.
1748 * The fasl file version number has changed (because of the new
1749 implementation of DEFMACRO).
1750 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1751 safe code, that any length requirement by their type-specifier
1752 argument is valid. The exceptions to this are described in bug
1754 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1755 functions now check the types of their inputs as required by ANSI.
1756 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1757 introduced symbol is DECLAREd to be SPECIAL.
1758 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1759 * fixed bug: :COUNT argument to sequence functions may be negative
1760 * fixed bug: body of DO-SYMBOLS may contain declarations
1761 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1762 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1763 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1764 symbol macro only once
1765 * fixed printing of call frame when argument list is unavailable
1766 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1767 * compiler no longer signals WARNING on unknown keyword
1770 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1771 * A beta-quality port to the mips architecture running Linux,
1772 based on the old CMUCL backend, has been made. It has been tested
1773 on a big-endian kernel, and works sufficiently well to be able to
1774 rebuild itself; it has not been tested in little-endian mode.
1775 * fixed an inconsistency between gencgc.c and purify.c which made
1776 dumping/loading .core files unreliable
1777 * fixed bug 120a: The compiler now deals correctly with IFs where
1778 the consequent is the same as the alternative, instead of
1779 misderiving the return type. (thanks to Alexey Dejneka)
1780 * fixed bug 113: Logical pathnames are now dumpable (the logical
1781 host is resolved at load-time, throwing an error if it is not
1783 * fixed bug 174: FORMAT's error message is slightly clearer when a
1784 non-printing character is used in a format directive.
1785 * fixed several bugs in compiler checking of type declarations, i.e.
1786 violations of the Python "declarations are assertions" principle
1787 (thanks to Alexey Dejneka)
1788 * fixed several bugs in PCL's error checking (thanks to Gerd
1790 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1792 * fixed bug in compilation of functions as first class values
1793 (thanks to Antonio Martinez-Shotton)
1794 * The compiler's handling TYPE-ERRORs which it can prove will
1795 inevitably happen at runtime has been cleaned up and corrected
1796 in several ways. (thanks to Alexey Dejneka)
1797 * improved argument type checking for various basic arithmetic
1798 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1799 lavished on them in the past that they can be compiled in many
1800 ways in different special cases
1801 * fixed bug 181: compiler checks validity of user supplied type
1803 * cleaned up code flushing in optimization: Function calls which
1804 should signal errors for safety purposes (e.g. which ANSI says
1805 should signal errors when their arguments are of incorrect type)
1806 are no longer optimized away.
1807 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1808 * incremented fasl file version number, because changes in the
1809 implementation of sequence functions like COERCE caused
1810 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1811 in old inline expansions) to become undefined. (Actually these
1812 changes were later undone, so we might very well be binary
1813 compatible with 0.7.7 after all, but leaving the version number
1814 incremented seemed like the simplest and most conservative
1817 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1818 * An alpha-quality port to the parisc architecture running Linux,
1819 based on the old CMUCL backend, has been made. This, even more so
1820 than the other backends, should be considered still a work in
1821 progress; known problems include that the Linux kernel in 64-bit
1822 mode does not propagate the correct sigcontext structure to
1823 userspace, and consequently SBCL on a parisc64 kernel will not
1825 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1826 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1827 suppressed.) Also now that the compiler is looking at declarations
1828 in the environment, it checks optimization declarations as well,
1829 and suppresses inlining when (> DEBUG SPEED).
1830 * More fixes have been made to treatment of floating point exception
1831 treatment and other Unix signals. In particular, floating point
1832 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1833 * The detection and handling of control stack exhaustion (infinite
1834 or very deeply nested recursion) has changed. Stack exhaustion
1835 detection is now done by write-protecting pages at the OS level
1836 and applies at all optimization settings; when found, a
1837 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1838 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1839 be caught using IGNORE-ERRORS.
1840 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1841 correctly when transforming recursive LABELS functions to LETs.
1842 (thanks to Alexey Dejneka)
1843 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1844 that are names of constants or global variables.
1845 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1846 alien routines with docstrings.
1847 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1848 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1850 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1851 (thanks to Eric Marsden)
1852 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1853 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1854 * Bug fix: The SPARC backend can now compile functions involving
1855 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1856 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1857 to a functions expecting a PATHNAME-DESIGNATOR.
1858 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1859 lambda lists. (thanks to Alexey Dejneka)
1860 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1861 functions so that EXPORTing the name of the class doesn't cause
1862 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1863 to Antonio Martinez for reporting this)
1864 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1865 constant-folded without failing an assertion. (thanks to Einar
1866 Floystad Dorum for reporting this)
1867 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1868 types are now dealt with more correctly by the compiler.
1869 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1870 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1871 behaviour analogous to RENAME-FILE. This puts its behaviour more
1872 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1874 * The fasl file version number has changed again. (because of the
1875 bug fix involving the names of PCL MAKE-INSTANCE functions)
1877 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1878 * bug fix: Floating point exceptions are treated much more
1879 consistently on the x86/Linux and PPC/Linux platforms.
1880 * Array initialization with :INITIAL-ELEMENT is now much faster for
1881 cases when the compiler cannot open code the array creation, but
1882 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1883 array accesses have also seen a speed increase.
1884 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1885 specified. (This was at the root of some bad interactions between
1886 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1888 * bug fix: Internal error arguments for undefined functions are now
1889 computed correctly on the PPC/Linux platform.
1890 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1891 Raymond Toy's patch for CMU CL)
1892 * Support for the Solaris 9 operating environment has been included
1893 (thanks to Daniel Merritt)
1894 * A very ugly but hopefully complete draft of the missing FFI chapter
1895 of the manual has been created by reformatting the corresponding
1896 CMU CL manual chapter into (currently very ugly and incoherent)
1897 DocBook and bringing it up to date for SBCL behavior. Thus, the
1898 manual is now essentially complete, at least by my extreme
1899 once-and-only-once standards, whereby it's acceptable to refer to
1900 the doc strings of SB-EXT functions as the primary documentation.
1901 * The fasl file version number has changed again, due to cleanup of
1902 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1903 is no longer a static symbol.)
1905 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1906 * SBCL now builds with OpenMCL (version 0.12) as the
1907 cross-compilation host; also, more progress has been made toward
1908 bootstrapping under CLISP.
1909 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1911 * bug 158 fixed: The compiler can now deal with integer loop
1912 increments different from 1; fixing this turned out also to fix
1914 * bug 169 fixed: no more bogus warnings about using lexical bindings
1915 despite the presence of perfectly good SPECIAL declarations (thanks
1916 to David Lichteblau)
1917 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1918 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1919 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1921 * bug 180 fixed: Method combination specifications no longer ignore
1922 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1923 * bug fix: Structure type predicate functions now check their argument
1924 count as they should.
1925 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1926 correctly. (thanks to Pierre Mai)
1927 * minor incompatible change: The --noprogrammer option is deprecated
1928 in favor of the new --disable-debugger option, which is very similar.
1929 (The major difference is that it takes effect at a slightly different
1930 time at startup, causing handling of errors in --sysinit and
1931 --userinit files will be affected differently.) The
1932 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1933 been added to allow this functionality to be controlled from ordinary
1934 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1935 maintainers, who might want to run non-interactive scripts to
1936 build SBCL cores which will later be used interactively.)
1937 * minor incompatible change: The LOAD function no longer, when given
1938 a wild pathname to load, loads all files matching that pathname.
1939 Instead, an error of type FILE-ERROR is signalled.
1941 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1942 * bug 147 fixed: The compiler preserves its block link/count
1943 invariants more correctly now so that it doesn't crash. (thanks
1945 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1947 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1948 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1949 no-value-for-this-character as NIL (as specified by ANSI).
1950 * HOST-NAMESTRING on physical pathnames now returns a string that is
1951 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1952 (thanks to Christophe Rhodes)
1953 * The Alpha port handles icache flushing more correctly. (thanks to
1955 * More progress has been made toward bootstrapping under CLISP. (thanks
1956 to Christophe Rhodes)
1957 * The fasl file format has changed again, because dynamic loading
1958 on OpenBSD (which has non-ELF object files) motivated some cleanups
1959 in the way that foreign symbols are transformed and passed around.
1960 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1961 is no longer treated as whitespace by the reader, but instead as
1962 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1963 |A
\7fB|, instead of A as it used to.
1965 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1966 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1967 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1968 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1969 this as of 0.7.1.45, but was left out of the previous news section
1970 (thanks to Dan Barlow)
1971 * SBCL now runs on the Solaris operating system on SPARC architectures
1972 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1973 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1974 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1975 * SPARC backend cleanups, allowing builds of cores optimized for V8
1976 and V9 SPARCS, and also emission of code targeted to a particular
1977 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1979 * SBCL is closer to bootstrapping under CLISP, thanks to various
1980 fixes by Christophe Rhodes.
1981 * The fasl file format has changed again, to allow the compiler's
1982 INFO database to support symbol macros.
1983 * The user manual (in doc/) is formatted into HTML more nicely.
1984 (thanks to coreythomas)
1985 * The system is smarter about SUBTYPEP relationships, especially
1986 those involving NOT types (including types such as ATOM which are
1987 represented internally using NOT types). Thus SUBTYPEP is less
1988 likely to return (VALUES NIL NIL) in general, and in particular
1989 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1991 * The fasl file format has changed again, because the internal
1992 representation of types now includes a new slot to support the new
1993 SUBTYPEP-of-NOT-types logic.
1994 * (not a change in the main branch of SBCL, but a related prototype
1995 which can hopefully be merged into the main branch of SBCL in the
1996 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1997 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1998 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1999 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2000 dispatch macro character combinations. (thanks to Alexey Dejneka)
2001 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2002 that superficially look like logical namestrings correctly.
2003 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2005 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2006 default supplied-p) for &optional and &key arguments. (thanks to
2009 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2010 * incompatible change: The compiler is now less aggressive about
2011 tail call optimization, doing it only when (> SPACE DEBUG) or
2012 (> SPEED DEBUG). (This is an incompatible change because there are
2013 programs which relied on the old CMU-CL-style behavior to optimize
2014 away their unbounded recursion which will now die of stack overflow.)
2015 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2016 for non-GENCGC systems has been increased to 20M (since that
2017 seems much closer to the likely performance optimum for modern
2018 systems than the old 4M value was)
2019 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2020 and *DEBUG-PRINT-LEVEL*
2021 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2022 of CMU CL's support for SPARC, and various endianness and other
2023 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2024 * new syntactic sugar for the Unix command line: --load foo.bar is now
2025 an alternate notation for --eval '(load "foo.bar")'.
2027 ** The system now detects stack overflow and handles it gracefully,
2028 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2029 optimization settings. (This is a good thing in general, and
2030 its introduction in this version should be particularly timely
2031 for anyone whose code fails because of suppression of tail
2033 ** The system now hunts for the C variable "environ" in a more
2034 devious way, to avoid segfaults when the C library version differs
2035 between compile time and run time. (thanks to Christophe Rhodes)
2036 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2037 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2038 ** The compiler no longer issues bogus style warnings for undefined
2039 classes in the same source file as the DEFCLASSes which defined
2040 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2042 ** fixes in CONDITION class precedence list for undefined function
2043 errors (thanks to Alexei Dejneka)
2044 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2045 correctly. (thanks to Dan Barlow)
2046 ** portability fixes aiming at bootstrapping under CLISP (thanks
2047 to Dave McDonald and Christophe Rhodes)
2048 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2049 ** fixes in type translation and and type inference (thanks to
2051 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2052 ** various fixes in the new ports (thanks to Dan Barlow)
2053 * several changes related to debugging:
2054 ** suppression of tail recursion, as noted above
2055 ** stack overflow detection, as noted above
2056 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2057 is now the default. (For some time encapsulation has been more
2058 reliable than the breakpoint-based :ENCAPSULATE NIL
2059 implementation, at least on X86 systems; and I just noticed that
2060 encapsulation also seems closer to the spirit of the ANSI
2063 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2065 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2066 up properly again. (There was a packaging bug in 0.7.0 which
2067 left their definitions in SB-SYS::LOAD-FOREIGN and
2068 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2069 vital for most things which interface to C-level interfaces,
2070 like extensions working with sockets or databases or
2071 Perl-compatible regexes or whatever, and the need to fix
2072 this bug is the main reason that 0.7.1 was released so
2074 ** DEFGENERIC is now choosier about the methods it redefines, so that
2075 reLOADing a previously-LOADed file containing DEFGENERICs does
2076 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2077 cycle now works as it should. (thanks to Alexey Dejneka)
2078 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2079 fixed by Christophe Rhodes. (He actually submitted this patch
2080 months ago, and I delayed until after 0.7.0.)
2081 ** Bug 111 (internal compiler confusion about runtime checks on
2082 FUNCTION types) was fixed by Alexey Dejneka.
2083 * Some internal cleanups (getting rid of variables which aren't
2084 needed now that the byte interpreter is gone) caused the fasl
2085 file format number to change again.
2087 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2088 * major incompatible change: The default fasl file extension, i.e. the
2089 default extension for files produced by COMPILE-FILE, has changed
2090 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2092 ** There are many changes in the implementation of the compiler.
2093 SBCL is now essentially a compiler-only implementation of ANSI
2094 Common Lisp. EVAL still "interprets" a few special cases, but
2095 almost all the interesting cases are handled by creating
2096 a LAMBDA expression, calling COMPILE on it, then calling
2097 FUNCALL on the result.
2098 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2099 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2100 Since the code is newer, there might still be some new bugs
2101 (though not as many as before Martin Atzmueller's fixes:-). But
2102 the new code is substantially simpler and clearer, and hopefully
2103 any remaining bugs will be simpler, less fundamental, and more
2104 fixable then the bugs in the old code.
2105 ** The revised compiler is still a little unsteady on its feet.
2107 *** The debugging information it produces (particularly the names
2108 of FUNCTION objects) is sometimes much less useful than what
2109 the old compiler produced.
2110 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2111 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2112 within a MACROLET) has been temporarily weakened.
2113 ** There are new compiler optimizations for various functions:
2114 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2115 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2116 *** the math functions TRUNCATE, FLOOR, and CEILING
2117 *** the function-of-all-trades COERCE
2118 Mostly these should be transparent, but there's one
2119 potentially-annoying problem (bug 117): when the compiler
2120 inline-expands a function and does type analysis on the result,
2121 it can create control paths which have type mismatches, and
2122 when it can't prove that those control paths aren't taken,
2123 it will issue WARNINGs about the type mismatches. This is
2124 a particular problem in practice for the new sequence functions.
2125 It's not clear how this should be fixed, and for now, a
2126 workaround is given in the entry for 117 in the BUGS file.
2127 ** (Because of the interaction between the two previous items --
2128 occasional inlining problems and new inline expansions -- some
2129 of the new sequence function optimizations won't really kick in
2130 completely until debugging information, and then inlining, are
2131 straightened out in some future version.)
2132 * minor incompatible changes:
2133 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2134 DIRECTORY no longer implicitly promotes NIL slots of its
2135 pathname argument to :WILD. In particular, when you ask for the
2136 contents of a directory (which you used to be able to do without
2137 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2138 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2139 ** changes in behavior that ANSI explicitly defines to be
2140 implementation dependent:
2141 *** The new compiler-only implementation still conforms with ANSI,
2142 but acts a little different than before. Besides the obvious
2143 changes in performance tradeoffs (that the cost per form passed
2144 to EVAL has gone up, and the cost per form executed by EVAL
2145 has gone down), the behavior of the system changes a little
2146 because there are no longer any interpreted function objects.
2147 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2148 e.g. doing COMPILE on the output of interactive DEFUN is
2150 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2152 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2153 has changed from (:CL) to NIL.
2154 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2155 e.g. #\Newline and #\Tab, don't have names specified in the
2156 ANSI Common Lisp standard, is now based on their ASCII symbolic
2157 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2158 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2159 are no longer used for output.
2160 ** changes in internal implementation constants:
2161 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2162 4 million. (If your application spends a lot of time GCing and
2163 you have a lot of RAM, you might want to experiment with
2164 increasing it even more.)
2165 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2166 However, almost all old code should still continue to work without
2167 immediate update, as SB-C-CALL is now a (deprecated) nickname
2169 ** Old operator names in the style DEF-FOO are now deprecated in
2170 favor of new corresponding names DEFINE-FOO, for consistency with
2171 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2172 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2173 internal symbols, but a few supported extensions like
2174 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2175 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2176 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2177 etc. as you get deeper into recursive calls to the debugger
2178 command loop, instead of the old "5]", "5]]", "5]]]"
2179 sequence. (I was motivated to do this when squabbles between
2180 ILISP and SBCL left me very deeply nested in the debugger. In the
2181 short term, this change will probably provoke more ILISP/SBCL
2182 squabbles, but hopefully it will be an improvement in the long run.)
2183 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2184 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2185 function, since declaiming return types involving aliens is
2186 (1) annoyingly messy to do by hand and (2) vital to efficient
2187 compilation of code which calls such functions.
2188 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2189 longer reexported by the SB-EXT package. They're solely useful
2190 for alien code, so it seems more logical that you should get
2191 them from the SB-ALIEN package, not in SB-EXT.
2192 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2193 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2194 features. Instead, the code that they used to control is always
2195 built into the system.
2196 * many other bug fixes
2197 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2198 advantage of the new EVAL-WHEN stuff and to clean them up in
2199 general, and they are now more ANSI-compliant in a number of
2200 ways. Martin Atzmueller is responsible for a lot of this.
2201 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2203 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2204 *** correct ERROR type for various file operations
2205 *** some fixes for Lisp streams
2206 *** DEFMETHOD syntax checking
2207 *** changing old weird representation of debug information as
2208 strings (which, among their other deficiencies, don't transform
2209 correctly when you rename packages, and don't change their
2210 print representation when you change things like *PACKAGE*
2211 and *PRINT-LENGTH*) to symbols and lists of symbols
2212 He also made several improvements and fixed several bugs in DESCRIBE.
2213 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2215 *** misbehavior of WRITE-STRING/WRITE-LINE
2216 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2217 and several other LOOP problems as well
2218 *** DIRECTORY when similar filenames are present
2219 *** DEFGENERIC with :METHOD options
2220 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2221 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2222 *** argument ordering in FIND with :TEST option
2223 *** mishandled package designator argument in APROPOS-LIST
2224 *** various problems in the backquote readmacro
2225 *** a bug in APROPOS
2226 *** probably some others that I'm not describing very well here,
2227 since the CVS log documents them by reference to sbcl-devel
2228 messages, and the SourceForge archives aren't working well.:-(
2229 ** Dan Barlow improved the Alpha port (and is making progress on the
2230 PPC port, for those of you who think different).
2231 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2232 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2233 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2234 continues to work on the SPARC port (for those of you in a position
2235 to look down upon our little PC-compatible boxes from a great height).
2236 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2237 and uses that copy, rather than the current dynamic value, when
2238 it's trying to decide whether to truncate output. Thus e.g.
2239 (let ((*print-lines* 50))
2240 (pprint-logical-block (stream nil)
2242 (let ((*print-lines* 8))
2243 (print (aref possiblybigthings i) stream)))))
2244 should now truncate the logical block only at 50 lines, instead of
2245 often truncating it at 8 lines, as it did before.
2246 * The doc/cmucl/ directory, containing old CMU CL documentation
2247 from the time of the fork, is no longer part of the base system.
2248 SourceForge has shut down its anonymous FTP service, and with it
2249 my original plan for distributing the old CMU CL documentation
2250 there. For now, if you need these files you can download an old
2251 SBCL source release and extract them from it.
2252 * The fasl file version number changed again, for dozens of reasons,
2253 some of which are apparent above.
2255 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2256 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2257 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2258 patch, so that DISASSEMBLE output is much nicer.
2259 * The code in the SB-PROFILE package now seems reasonably stable.
2260 I still haven't decided what the final interface should look like
2261 (I'd like PROFILE to interact cleanly with TRACE, since both
2262 facilities use function encapsulation) but if you have a need
2263 for profiling now, you can probably use it successfully with
2264 the current CMU-CL-style interface.
2265 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2266 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2267 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2268 link now returns the dangling link itself, and for similar
2269 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2270 link itself. (In these cases the old code signalled an error and
2271 looped endlessly, respectively.) Thus, DIRECTORY now works even
2272 in the presence of dangling and cyclic symbolic links.
2273 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2274 is now a supported extension again, since the consensus on
2275 sbcl-devel was that it can be useful for ordinary development
2276 work, not just for debugging SBCL itself.
2277 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2278 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2279 that currently-defined functions might be redefined later with
2280 different return types.
2281 * Hash tables can be printed readably, as inspired by CMU CL code
2282 of Eric Marsden and SBCL code of Martin Atzmueller.
2283 * better error handling in CLOS method combination, thanks to
2284 Martin Atzmueller porting Pierre Mai's CMU CL patches
2285 * more overflow fixes for >16Mbyte I/O buffers
2286 * A bug in READ has been fixed, so that now a single Ctrl-D
2287 character suffices to cause end-of-file on character streams.
2288 In particular, now you only need one Ctrl-D at the command
2289 line (not two) to exit SBCL.
2290 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2292 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2293 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2294 does the right thing.
2295 * The implementation of some type tests, especially for CONDITION
2296 types, is now tidier and maybe faster, due to CMU CL code
2297 originally by Douglas Crosher, ported by Martin Atzmueller.
2298 * Some math functions have been fixed, and there are new
2299 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2300 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2301 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2302 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2303 likely to be built into sbcl-0.7.0.)
2304 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2305 CL, ported to SBCL by Martin Atzmueller)
2306 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2307 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2308 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2309 * There's a new slam.sh hack to shorten the edit/compile/debug
2310 cycle for low-level changes to SBCL itself, and a new
2311 :SB-AFTER-XC-CORE target feature to control the generation of
2312 the after-xc.core file needed by slam.sh.
2313 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2314 COMPILE-FILE is no longer supported, so that now every function
2315 gets an entry point, so that block compilation looks a little
2316 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2317 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2318 returns the number of bytes consed since the system started,
2319 rather than the number consed since the first time the function
2320 was called. (The new definition parallels ANSI functions like
2321 CL:GET-INTERNAL-RUN-TIME.)
2322 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2323 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2324 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2325 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2326 * Fasl file version numbers are now independent of the target CPU,
2327 since historically most system changes which required version
2328 number changes have affected all CPUs equally. Similarly,
2329 the byte fasl file version is now equal to the ordinary
2332 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2333 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2334 is no longer recognized. I apologize for this, because it was
2335 listed in SB-EXT as a supported extension, but I found that
2336 its existing behavior was poorly specified, as well as incorrectly
2337 specified, and it looked like too much of a mess to straighten it
2338 out. I have enough on my hands trying to get ANSI stuff to work..
2339 * many patches ported from CMU CL by Martin Atzmueller, with
2340 half a dozen bug fixes in pretty-printing and the debugger, and
2341 half a dozen others elsewhere
2342 * fixed bug 13: Floating point infinities are now supported again.
2343 They might still be a little bit flaky, but thanks to bug reports
2344 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2345 as flaky as they were.
2346 * The --noprogrammer command line option is now supported. (Its
2347 behavior is slightly different in detail from what the old man
2348 page claimed it would do, but it's still appropriate under the
2349 same circumstances that the man page talks about.)
2350 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2351 are now supported, and enabled by default. Thus, the compiler can
2352 handle many floating point and complex operations much less
2353 inefficiently. (Thus e.g. you can implement a complex FFT
2355 * The compiler now detects type mismatches between DECLAIM FTYPE
2356 and DEFUN better, and implements CHECK-TYPE more correctly, and
2357 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2358 thanks to patches from Martin Atzmueller.
2359 * various fixes to make the cross-compiler more portable to
2360 ANSI-conforming-but-different cross-compilation hosts (notably
2361 Lispworks for Windows, following bug reports from Arthur Lemmens)
2362 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2363 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2364 fixed thanks to Pierre Mai's CMU CL patches.
2365 * fixes to keep the system from overflowing internal counters when
2366 it tries to use i/o buffers larger than 16M bytes
2367 * fixed bug 45a: Various internal functions required to support
2368 complex special functions have been merged from CMU CL sources.
2369 (When I was first setting up SBCL, I misunderstood a compile-time
2370 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2371 * improved support for type intersection and union, fixing bug 12
2372 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2373 more obscure bugs as well
2374 * some steps toward byte-compiling non-performance-critical
2375 parts of the system, courtesy of patches from Martin Atzmueller
2376 * Christophe Rhodes has made some debian packages of sbcl at
2377 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2378 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2379 stable, but are nonetheless usable. When he's ready, I'd be happy
2380 to add them to the SourceForge "File Releases" section. (And if
2381 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2382 * new fasl file format version number (because of changes in
2383 internal representation of (OR ..) types to accommodate the new
2384 support for (AND ..) types, among other things)
2386 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2387 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2389 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2391 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2392 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2393 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2394 * DESCRIBE now works on structure objects again.
2395 * Most function call argument type mismatches are now handled as
2396 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2397 know whether the function will be redefined before the call is
2398 executed. (The compiler could flag local calls with full WARNINGs,
2399 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2400 it doesn't keep track of enough information to know whether calls
2401 are local in this sense.)
2402 * Compiler output is now more verbose, with messages truncated
2403 later than before. (There should be some supported way for users
2404 to override the default verbosity, but I haven't decided how to
2405 provide it yet, so this behavior is still controlled by the internal
2406 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2407 src/compiler/ir1util.lisp.)
2408 * Fasl file format version numbers have increased again, because
2409 support for the Gray streams extension changes the layout of the
2410 system's STREAM objects.
2411 * The Gray subclassable streams extension now works, thanks to a
2412 patch from Martin Atzmueller.
2413 * The full LOAD-FOREIGN extension (not just the primitive
2414 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2415 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2416 but like most other programs, it defaults to copying the Unix
2417 environment from the original process instead of starting the
2418 new process in an empty environment.
2419 * Extensions which manipulate the Unix environment now support
2420 an :ENVIRONMENT keyword option which doesn't smash case or
2421 do other bad things. The CMU-CL-style :ENV option is retained
2422 for porting convenience.
2423 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2424 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2426 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2428 * A patch from Martin Atzmueller seems to have solved the SIGINT
2429 problem, and as far as we know, signal-handling now works cleanly.
2430 (If you find any new bugs, please report them!)
2431 * The system no longer defaults Lisp source file names to types
2432 ".l", ".cl", or ".lsp", but only to ".lisp".
2433 * The compiler no longer uses special default file extensions for
2434 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2435 seems to expect a single default extension for all compiled code,
2436 and there's no compelling reason to try to stretch the standard
2437 to allow two different extensions.) Instead, byte-compiled files
2438 default to the same extension as native-compiled files.
2439 * Fasl file format version numbers have increased again, because
2440 a rearrangement of internal implementation packages made some
2441 dumped symbols in old fasl files unreadable in new cores.
2442 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2443 many fewer weird special cases.
2444 * Bug #17 (differing COMPILE-FILE behavior between logical and
2445 physical pathnames) has been fixed, and some related misbehavior too,
2446 thanks to a patch from Martin Atzmueller.
2447 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2448 by Tim Moore, ported to SBCL by Martin Atzmueller.
2449 * Martin Atzmueller fixed several filesystem-related problems,
2450 including bug #36, in part by porting CMU CL patches, which were
2451 written in part by Paul Werkowski.
2452 * More compiler warnings in src/runtime/ are gone, thanks to
2453 more patches from Martin Atzmueller.
2454 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2457 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2459 * DESCRIBE now works on CONDITION objects.
2460 * The debugger now handles errors which arise when trying to print
2461 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2463 * The build system now uses an additional file, customize-target-features.lisp,
2464 to allow local modifications to the target *FEATURES* list. (The point of
2465 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2466 debugging features enabled, without having to worry about propagating it
2467 into everyone's system when I do a "cvs update".) When no
2468 customize-target-features.lisp file exists, the target *FEATURES* list
2469 should be constructed the same way as before.
2470 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2471 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2472 warn about behavior which is undefined under the ANSI spec. Note: This
2473 is specified by ANSI, but it's not very popular with programmers.
2474 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2475 macro in the SBCL sources for an example of a workaround which you
2476 might use to make portable ANSI-standard code which does what you want.
2477 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2478 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2479 when it's used as a non-toplevel form. (This is required in order
2480 to implement the DEFCONSTANT-EQX macro.)
2481 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2482 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2483 funny things, due to bugs in the implementation of EVAL-WHEN
2484 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2485 non-toplevel DEFCONSTANTs are uncommon.)
2486 * The core file version number and fasl file version number have been
2487 incremented, because the old noncompliant DEFCONSTANT behavior involved
2488 calling functions which no longer exist, and because I also took the
2489 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2490 * fixed bug 1 (error handling before read-eval-print loop starts), and
2491 redid debugger restarts and related debugger commands somewhat while
2493 ** The QUIT debugger command is gone, since it did something
2494 rather different than the SB-EXT:QUIT command, and since it never
2495 worked properly outside the main toplevel read/eval/print loop.
2496 Invoking the new TOPLEVEL restart provides the same functionality.
2497 ** The GO debugger command is also gone, since you can just invoke
2498 the CONTINUE restart directly instead.
2499 ** The TOP debugger command is also gone, since it's redundant with the
2500 FRAME 0 command, and since it interfered with abbreviations for the
2502 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2504 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2505 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2506 some porting work by Martin Atzmueller)
2507 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2508 with a boatload of patches which he ported to SBCL. Now that those
2510 ** The system tries to make sure that its low-priority messages
2511 are prefixed by semicolons, to help people who like to use
2512 syntax highlighting in their ILISP buffer. (This patch
2513 was originally due to Raymond Toy.)
2514 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2515 patches originally written by Raymond Toy.
2516 ** The compiler understands coercion between single-value and
2517 multiple-VALUES type expressions better, getting rid of some very
2518 weird behavior, thanks to patches originally by Robert MacLachlan
2519 and Douglas Crosher.
2520 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2521 lambda lists, thanks to a patch originally by Pierre Mai.
2522 ** The system no longer bogusly warns about "abbreviated type
2524 ** The compiler gets less confused by inlining and RETURN-FROM,
2525 thanks to some patches originally by Tim Moore.
2526 ** The system no longer hangs when dumping circular lists to fasl
2527 files, thanks to a patch originally from Douglas Crosher.
2528 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2529 undefined function error.
2530 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2531 in structure slot initforms) for now, documented workaround instead:-|
2532 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2533 * fixed bug 5: added stubs for various Gray stream functions called
2534 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2535 installed, at least appropriate type errors are generated
2536 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2537 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2538 consistently in DEFMETHOD forms.
2539 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2540 it doesn't seem to affect SBCL after all
2541 * The C runtime system now builds with better optimization and many
2542 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2544 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2546 * The system is now under CVS at SourceForge (instead of the
2547 CVS repository on my home machine).
2548 * The new signal handling code has been tweaked to treat register
2549 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2550 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2551 downstream, e.g. in debug-int.lisp, has implicit dependencies
2552 on the unsignedness of integer representation of machine words,
2553 and that caused the system to bomb out with infinite regress
2554 when trying to recover from type errors involving signed values,
2555 e.g. (BUTLAST '(1 2 3) -1).)
2556 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2557 (This was one of the bugs Peter Van Eynde reported back in July.)
2558 * The system now uses code inspired by Colin Walters' O(N)
2559 implementation of MAP (from the cmucl-imp@cons.org mailing
2560 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2561 inline the MAP operation, and there is more than one
2562 sequence argument to the MAP call (so that it can't just
2563 do ETYPECASE once and for all based on the type of the
2564 single sequence argument). (The old non-inline implementation
2565 of the general M-argument sequence-of-length-N case required
2566 O(M*N*N) time when any of the sequence arguments were LISTs.)
2567 * The QUIT :UNIX-CODE keyword argument has been renamed to
2568 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2569 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2571 * Raymond Wiker's patches to port dynamic loading from Linux to
2572 FreeBSD have been added.
2573 * The BUGS file is now more nearly up to date, thanks in large part
2574 to Martin Atzmueller's review of it.
2575 * The debugger now flushes standard output streams before it begins
2576 its output ("debugger invoked" and so forth).
2577 * The core version number and fasl file version number have both
2578 been incremented, because of incompatible changes in the layout
2580 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2581 used to not be called for a saved Lisp image.)
2582 * Martin Atzmueller's version of a patch to fix a compiler crash,
2583 as posted on sbcl-devel 13 September 2000, has been installed.
2584 * Instead of installing Martin Atzmueller's patch for the
2585 compiler transform for SUBSEQ, I deleted the compiler transform,
2586 and transforms for some similar consing operations.
2587 * A bug in signal handling which kept TRACE from working on OpenBSD
2589 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2590 away by constant folding
2591 * The system now defines its address space constants in one place
2592 (in the Lisp sources), and propagates them automatically elsewhere
2593 (through GENESIS and the sbcl.h file). Therefore, patching the
2594 address map is less unnecessarily tedious and error-prone. The
2595 Lisp names of address space constants have also been systematized.
2596 * CVS tags like dollar-Header-dollar have been removed from
2597 the sources, because they have never saved me trouble and
2598 they've been source of trouble working with patches and other
2599 diff-related operations.
2600 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2601 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2603 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2605 * The system has been ported to OpenBSD.
2606 * The system now compiles with a simple "sh make.sh" on the systems
2607 that it's supported on. I.e., now you no longer need to tweak
2608 text in the target-features.lisp-expr and symlinks in src/runtime/
2609 by hand, the make.sh takes care of it for you.
2610 * The system is no longer so grossly inefficient when compiling code
2611 involving vectors implemented as general (not simple) vectors (VECTOR T),
2612 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2613 which dares to use the various sequence functions on non-simple
2614 vectors, takes less of a performance hit.
2615 * There is now a primitive type predicate VECTOR-T-P
2616 to test for the (VECTOR T) type, so that e.g.
2617 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2618 can now be compiled with some semblance of efficiency. (The old code
2619 turned the type declaration into a full call to %TYPEP at runtime!)
2620 * AREF on (VECTOR T) is still not fast, since it's still compiled
2621 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2622 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2623 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2624 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2625 without making full calls to SUBTYPEP at runtime.
2626 (Some analogous efficiency issues for non-simple vectors specialized to
2627 element types other than T, or for non-simple multidimensional arrays,
2628 have not been addressed. They could almost certainly be handled the
2629 same way if anyone is motivated to do so.)
2630 * The changes in array handling break binary compatibility, so
2631 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2632 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2633 instead of (VALUES T T).
2634 * By following the instructions that Dan Barlow posted to sbcl-devel
2635 on 2 July 2000, I was able to enable primitive dynamic object
2636 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2637 functionality is not implemented (since it calls ld to resolve
2638 library references automatically, requiring RUN-PROGRAM for its
2639 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2640 try to resolve library references) is now supported.
2641 * The system now flushes the standard output streams when it terminates,
2642 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2643 them at several other probably-convenient times, e.g. in each pass of
2644 the toplevel read-eval-print loop, and after evaluating a form given
2645 as an "--eval" command-line option. (These changes were motivated by a
2646 discussion of stream flushing issues on cmucl-imp in August 2000.)
2647 * The source transform for TYPEP of array types no longer assumes
2648 that an array whose element type is a not-yet-defined type
2649 is implemented as an array of T, but instead punts, so that the
2650 type will be interpreted at runtime.
2651 * There is now some support for cross-compiling in make.sh: each of
2652 the phases of make.sh has its own script. (This should be transparent
2653 to people doing ordinary, non-cross-compile builds.)
2654 * Since my laptop doesn't have hundreds of megabytes of memory like
2655 my desktop machine, I became more motivated to do some items on
2656 my to-do list in order to reduce the size of the system a little:
2657 ** Arrange for various needed-only-at-cold-init things to be
2658 uninterned after cold init. To support this, those things have
2659 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2660 symbols with such names are now uninterned after cold init).
2661 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2662 fasl files for cold load.
2663 ** Remove the old compiler structure pooling code (which used to
2664 be conditional on the target feature :SB-ALLOC) completely.
2665 ** Redo the representation of some data in cold init to be more compact.
2666 (I also looked into supporting byte compiled code at bootstrap time,
2667 which would probably reduce the size of the system a lot, but that
2668 looked too complicated, so I punted for now.)
2669 * The maximum signal nesting depth in the src/runtime/ support code has
2670 been reduced from 4096 to 256. (I don't know any reason for the very
2671 large old value. If the new smaller value turns out to break something,
2672 I'll probably just bump it back up.)
2673 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2675 * Many, many bugs reported by Peter Van Eynde have been added to
2676 the BUGS list; some have even been fixed.
2677 * While enabling dynamic object file loading, I tried to make the
2678 code easier to understand, renaming various functions and variables
2679 with less ambiguous names, and changing some function calling
2680 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2681 * While trying to figure out how to do the OpenBSD port, I tried to
2682 clean up some of the code in src/runtime/. In particular, I dropped
2683 support for non-POSIX signal handling, added various comments,
2684 tweaked the code to reduce the number of compilation warnings, and
2685 renamed some files to increase consistency.
2686 * To support the new automatic configuration functionality in make.sh,
2687 the source file target-features.lisp-expr has been replaced with the
2688 source file base-target-features.lisp-expr and the machine-generated
2689 file local-target-features.lisp-expr.
2690 * fixed a stupid quoting error in make.sh so that using CMU CL
2691 "lisp -batch" as cross-compilation host works again
2693 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2695 * DESCRIBE no longer tries to call itself recursively to describe
2696 bound/fbound values, so that it no longer fails on symbols which are
2697 bound to themselves (like keywords, T, and NIL).
2698 * DESCRIBE now works on generic functions.
2699 * The printer now prints less-screwed-up representations of closures
2700 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2701 * A private symbol is used instead of the :EMPTY keyword previously
2702 used to mark empty slots in hash tables. Thus
2703 (DEFVAR *HT* (MAKE-HASH-TABLE))
2704 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2705 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2706 now does what ANSI says that it should. (You can still get
2707 similar noncompliant behavior if bang on the hash table
2708 implementation with all the symbols you get back from
2709 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2710 This breaks binary compatibility, since tests for equality to
2711 :EMPTY are wired into things like the macroexpansion of
2712 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2714 * There's now a minimal placeholder implementation for CL:STEP,
2715 as required by ANSI.
2716 * An obscure bug in the interaction of the normal compiler, the byte
2717 compiler, inlining, and structure predicates has been patched
2718 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2719 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2721 * Missing ordinary arguments in a macro call are now detected even
2722 when the macro lambda list contains &KEY or &REST.
2723 * The debugger no longer complains about encountering the top of the
2724 stack when you type "FRAME 0" to explicitly instruct it to go to
2725 the top of the stack. And it now prints the frame you request even
2726 if it's the current frame (instead of saying "You are here.").
2727 * As specified by ANSI, the system now always prints keywords
2728 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2730 * The default initial SIZE of HASH-TABLEs is now smaller.
2731 * Type information from CLOS class dispatch is now propagated
2732 into DEFMETHOD bodies, so that e.g.
2733 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2735 is now basically equivalent to
2736 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2737 (DECLARE (TYPE SINGLE-FLOAT X))
2739 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2740 operation, without having to do run-time type dispatch.
2741 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2742 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2743 * Since I don't seem to be making big file reorganizations very often
2744 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2745 is overflowing my ability to conveniently back them up), I've finally
2746 checked the system into CVS. (The CVS repository is on my home system,
2747 not at SourceForge -- putting it on SourceForge might come later.)
2748 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2749 high-level GC-NOTIFY-FOO functions send their output. (There's
2750 still very little control of where low-level verbose GC functions
2751 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2752 less than it used to -- the GC-NOTIFY-FOO functions are now under
2753 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2754 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2755 in only one place in the source code, and propagates it automatically
2756 everywhere that it's needed. Thus e.g. when I bump the version from
2757 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2758 * The C source files now include boilerplate legalese and documentation
2759 at the head of each file (just as the Lisp source files already did).
2760 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2761 to his page will be replaced with a link to his new CLiki service.
2763 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2765 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2766 * The build process now looks for GNU make under the default name "gmake",
2767 instead of "make" as it used to. If GNU make is not available as "gmake"
2768 on your system, you can change this default behavior by setting the
2769 GNUMAKE environment variable.
2770 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2771 can build without error under CMU CL.
2773 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2775 * There is now a partial SBCL user manual (with some new text and some
2776 text cribbed from the CMU CL manual).
2777 * The beginnings of a profiler have been added (starting with the
2778 CMU CL profiler and simplifying and cleaning up). Eventually the
2779 main interface should be through the TRACE macro, but for now,
2780 it's still accessed through vaguely CMU-CL-style functions and macros
2781 exported from the package SB-PROFILE.
2782 * Some problems left over from porting CMU CL to the new
2783 cross-compilation bootstrap process have been cleaned up:
2784 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2785 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2786 part when getting rid of the compiler *BACKEND* stuff.)
2787 ** The value of *NULL-TYPE* was screwed up, because it was
2788 being initialized before the type system knew the final
2789 definition of the 'NULL type. This screwed up several key
2790 optimizations in the compiler, causing inefficiency in all sorts
2791 of places. (I found it because I wanted to understand why
2792 GET-INTERNAL-RUN-TIME was consing.)
2793 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2794 PROCLAIM FTYPE information. Unfortunately this broke binary
2795 compatibility again, since now the forms output by DEFGENERIC
2796 to refer to functions which didn't exist in 0.6.3.
2797 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2798 can use the new (as of 0.6.3) transform for SOME into MAP into
2800 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2801 values to (INTEGER 0 1000000), so that the time code will no longer
2802 occasionally get blown up by Linux returning 1000000 microseconds
2803 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2804 its output conform to the ANSI spec. (Alas, this makes its output
2805 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2806 * A full call to MAP NIL with a single sequence argument no longer conses.
2807 * fixes to problems pointed out by Martin Atzmueller:
2808 * The manual page no longer talks about multiprocessing as though
2809 it were currently supported.
2810 * The ILISP support patches have been removed from the distribution,
2811 because as of version 5.10.1, ILISP now supports SBCL without us
2812 having to maintain patches.
2813 * added a modified version of Raymond Toy's recent CMU CL patch for
2814 EQUALP comparison of HASH-TABLE
2816 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2818 * The system still can't cross-compile itself with
2819 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2820 entails), but at least it can compile more of itself that way
2821 than it used to be able to, and various buglets which were uncovered
2822 by trying to cross-compile itself that way have now been fixed.
2823 * This release breaks binary compatibility again. This time
2824 at least I've incremented the FASL file format version to 2, so that the
2825 problem can be detected reliably instead of just causing weird errors.
2826 * various new style warnings:
2827 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2828 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2829 ** using the lexical binding of a variable named in the *FOO* style
2830 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2831 as specified by ANSI.
2832 * *RANDOM-STATE* is no longer automatically initialized from
2833 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2834 default behavior of the system is to repeat its behavior every time
2835 it's run. If you'd like to change this behavior, you can always
2836 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2837 old convention there was no comparably easy way to get the system to
2838 repeat its behavior every time it was run.
2839 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2840 FTYPE declarations has been removed, in favor of their ANSI
2841 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2842 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2843 the types of their sequence arguments can be determined at compile time.
2844 This is done through a new open code expansion for MAP which eliminates
2845 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2846 argument types can be determined at compile time.
2847 * The optimizer now transforms COERCE into an identity operation when it
2848 can prove that the coerced object is already of the correct type. (This
2849 can be a win for machine generated code, including the output of other
2850 optimization transforms, such as the MAP transform above.)
2851 * Credit information has been moved from source file headers into CREDITS.
2852 * Source file headers have been made more standard.
2853 * The CASE macro now compiles without complaining even when it has
2856 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2858 * (Note that the way that the PCL macroexpansions were rewritten
2859 to accommodate the change in DEFGENERIC below breaks binary
2860 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2861 not run under sbcl-0.6.2. Once we get out of alpha releases,
2862 i.e. hit release 1.0.0, we'll probably try to maintain binary
2863 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2864 and sbcl-1.4.4. Until then, however, it might be fairly common
2865 for maintenance releases to break binary compatibility.)
2866 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2868 * The reporting of unhandled signals has been changed to print some
2869 explanatory text as well as the report form. (Previously only
2870 the report form was printed.)
2871 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2872 it defines, so that the compiler no longer issues undefined function
2873 warnings for compiled-but-not-yet-loaded generic functions.
2874 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2875 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2876 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2877 section of the ANSI spec are supported.
2878 * The "" nickname for the "KEYWORD" package has been removed.
2879 The reader still handles symbol tokens which begin with a package marker
2880 as keywords, but it doesn't expose its mechanism for doing so in the
2881 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2882 * The system now issues STYLE-WARNINGs for contradictory TYPE
2883 proclamations. (Warnings for contradictory FTYPE proclamations would
2884 be nice too, but those can't be done usefully unless the type system
2885 is made smarter about FUNCTION types.)
2886 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2887 systematized, so that "*target-*.lisp is supposed to exist only on the
2888 target and imply that there's a related file which exists on the
2889 host, and *host-*.lisp is supposed to exist only on the host and imply
2890 that there's a related file which exists on the target. This involves a
2891 lot of renaming. Hopefully the acute confusion caused by the renaming
2892 will be justified by the reduction in chronic confusion..
2893 ** runtime-type.lisp -> early-target-type.lisp
2894 ** target-type.lisp -> late-target-type.lisp
2895 ** early-host-format.lisp -> early-format.lisp
2896 ** late-host-format.lisp -> late-format.lisp
2897 ** host-error.lisp -> misc-error.lisp
2898 ** early-error.lisp -> early-target-error.lisp
2899 ** late-error.lisp -> late-target-error.lisp
2900 ** host-defboot.lisp -> early-defboot.lisp
2901 ** code/misc.lisp -> code/target-misc.lisp
2902 ** code/host-misc.lisp -> code/misc.lisp
2903 ** code/numbers.lisp -> code/target-numbers.lisp
2904 ** code/early-numbers.lisp -> numbers.lisp
2905 ** early-host-type.lisp -> early-type.lisp
2906 ** late-host-type.lisp -> late-type.lisp
2907 ** host-typep.lisp -> typep.lisp
2908 ** load.lisp -> target-load.lisp
2909 ** host-load.lisp -> load.lisp
2910 ** host-disassem.lisp -> disassem.lisp
2911 ** host-insts.lisp -> insts.lisp
2912 ** byte-comp.lisp -> target-byte-comp.lisp
2913 ** host-byte-comp.lisp -> byte-comp.lisp
2914 ** host-signal.lisp -> signal.lisp
2915 ** host-defstruct.lisp -> defstruct.lisp
2916 ** late-target-type.lisp -> deftypes-for-target.lisp
2917 Furthermore, several other previously target-only files foo.lisp (e.g.
2918 hash-table.lisp and random.lisp) have been split into a target-and-host
2919 foo.lisp file and a target-only target-foo.lisp file, with their key type
2920 definitions in the target-and-host part, so that the cross-compiler will
2921 know more about target types.
2922 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2923 gone away. In their place are various *BACKEND-FOO* variables
2924 corresponding to the slots of the old structure.
2925 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2926 into the target SBCL has been fixed.
2927 * The chill.lisp system for loading cold code into a running SBCL
2929 * Support for the CMU CL "scavenger hook" extension has been removed.
2930 (It was undocumented and unused in the CMU CL sources that SBCL was
2931 derived from, and stale in sbcl-0.6.1.)
2932 * Various errors in the cross-compiler type system were detected
2933 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2934 (enabling various consistency checks). Many of them were fixed,
2935 but some hard problems remain, so the compiler is back to
2936 running without *TYPE-SYSTEM-INITIALIZED* for now.
2937 * As part of the cross-compiler type system cleanup, I implemented
2938 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2939 * I have started adding UNCROSS calls throughout the type system
2940 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2941 on cross-compiler input files will be able to go away, and various
2943 * CONSTANTP now returns true for quoted forms (as explicitly required
2946 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2948 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2949 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2950 (MAKE-LIST -1). In the longer term, it ought to become true
2951 that declarations are assertions even at SAFETY 1. For now, it's not
2952 quite true even at SAFETY 3, but it's at least more nearly true..
2953 (Note that this change seems to increases the size of the system by
2954 O(5%) and to decrease the speed of the compiler by 20% or more.)
2955 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2956 for the problem of printing dozens of lines of distracting information
2957 about low-level system machinery as part of the top stack frame
2958 on entry to the debugger when an undefined function was called.
2959 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2960 is not reset to COMMON-LISP-USER.
2961 * Compilation of stuff related to dyncount.lisp has been made conditional
2962 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2963 smaller. The various dyncount-related symbols have been moved into
2964 a new "SB-DYNCOUNT" package.
2965 * tty-inspect.lisp has been renamed to inspect.lisp.
2966 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2967 feature has gone away. (When we eventually port to other flavors of
2968 libc and/or Unix, we'll try to make the differences between flavors
2969 invisible at the user level.)
2970 * Various other *FEATURES* tags, and/or their associated conditionals,
2971 have been removed if obsolescent, or given better documentation, or
2972 sometimes given more-mnemonic names.
2974 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2976 * tidied up "make.sh" script
2977 * tidied up system directory structure
2978 * better "clean.sh" behavior
2979 * added doc/FOR-CMUCL-DEVELOPERS
2980 * many many small tweaks to output format, e.g. removing possibly-confusing
2981 trailing #\. character in DESCRIBE-INSTANCE
2982 * (EQUALP #\A 'A) no longer signals an error.
2983 * new hashing code, including EQUALP hashing
2984 * tidied up Lisp initialization and toplevel
2985 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2986 * command line argument processing
2987 * added POSIX-GETENV function to deal with Unix-ish environment variables
2988 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2989 terminating SBCL on EOF
2990 * non-verbose GC by default
2991 * There is no more "sbcl" shell script; the sbcl file is now the C
2992 runtime executable (just like CMU CL).
2993 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2995 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2996 symbol and package information as Lisp native symbol and package objects
2997 instead of strings naming symbols and strings naming packages. This way,
2998 whenever packages are renamed (as in warm init), debug information is
2999 transformed along with everything else.
3000 * tweaked the optimization policy declarations which control the building
3001 of SBCL itself. Now, among other things, the system no longer saves
3002 source location debugging information. (This helps two problems at once
3003 by reducing SBCL size and by keeping SBCL from trying to look for its
3004 sources -- which may not exist -- when reporting errors.)
3005 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3006 debugging and testing purposes
3007 * cleaned up printing, making the printer call PRINT-OBJECT for
3008 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3009 methods, giving nearly-ANSI behavior
3010 * converted almost all special variables to use *FOO* naming convention
3011 * deleted PARSE-TIME functionality, since it can be done portably
3012 * moved some files out of cold init into warm init
3013 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3015 * regularized formatting of source files
3016 * added an install.sh script
3017 * fixed ridiculous memory usage of cross-compiler by making
3018 compiler/alloc.lisp not try to do pooling unless it can hook
3019 itself into the GC of the cross-compilation host. Now the system
3020 builds nicely on my old laptop.
3021 * added :SB-ALLOC in target-features.lisp-expr
3022 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3023 was not implemented)
3024 * re-did condition handling and note reporting in the compiler. Notes
3025 are no longer handled by signalling conditions. Style warnings
3026 and warnings are handled more correctly and reported in such a way
3027 that it's easy to find one or the other in your output (so that you
3028 can e.g. figure out which of many problems caused COMPILE-FILE to
3030 * changed the severity of several compiler warnings from full WARNING
3031 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3032 compiler note reporting so that it doesn't use the condition system
3033 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3034 COMPILE-FILE command)
3035 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3036 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3037 no longer supported, since their implementation was deeply intertwingled
3038 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3039 * removed bogus "support" for compiler macros named (SETF FOO), and
3040 removed the compiler macro for SETF INFO (but only after making a fool
3041 of myself on the cmucl-imp mailing list by posting a bogus patch for
3042 DEFINE-COMPILER-MACRO..)
3043 * Compiled files containing forms which have side effects on the Lisp
3044 reader (such as DEFPACKAGE forms) are now handled more correctly.
3045 (Compiler queuing of top level lambdas has been suppressed by setting
3046 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3047 * deleted various currently-unused source files, e.g. gengc.lisp. They
3048 may be added back at some point e.g. when porting to other architectures,
3049 but until they are it's distracting to distribute them and to try to
3051 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3052 there were so many of them they're just distractions, and UNCROSS is
3053 known to be able to handle the current sources
3054 * moved PROFILE functionality into TRACE, so that it will be clear
3055 how the wrapping and unwrapping of functions when you profile them
3056 interacts with the wrapping and unwrapping of functions when you
3057 trace them. (Actually, the functionality isn't there yet, but at least
3058 the interface specification is there. Hopefully, the functionality will
3059 arrive with some future maintenance release.)
3060 * removed host-oops.lisp
3061 * changed signature of QUIT function to allow UNIX-CODE argument
3062 * fixed READ-SEQUENCE bug
3063 * tweaked verbose GC output so that it looks more like the progress
3064 output that ANSI specifies for functions like LOAD
3065 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3066 * added <http://sbcl.sourceforge.com> to the banner information printed by