1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
3 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
4 space for its heap. (thanks to Thiemo Seufer)
5 * minor incompatible change: eof selects abort in the debugger.
6 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
7 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
9 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
10 deprecated, and will go away in a future revision of SBCL.
11 * bug fix: discriminating functions for generic function classes
12 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
13 make invalid assumptions about method precedence order. (reported
15 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
16 * bug fix: correct pathname printing with printer escaping is on.
17 (thanks to Kevin Reid)
18 * bug fix: complex VOP definitions in "user-space" no longer trigger
19 package locks. (reported by Zach Beane)
20 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
21 for funcallable-instance functions, and is no different from
23 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
24 characters in the current readtable. (reported by Nicholas Neuss)
25 * bug fix: SUBTYPEP on various CONS types returns more a more
26 accurate acknowledgment of its certainty.
27 * optimizations: REMOVE-DUPLICATES now runs in linear time on
28 lists in some cases. This partially fixes bug 384.
29 * flush all standard streams before prompting in the REPL and the
32 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
33 there is only one thread in the session
34 ** bug fix: memory leak for streams created in one thread and
37 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
38 * New feature: Experimental support for bivalent streams: streams
39 opened with :element-type :default now allow character and binary
41 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
42 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
44 * bug fix: no more highly sporadic "couldn't check whether ~S is
45 readable" when reading a stream and an interrupt hits in the middle
46 of a select system call
47 * compiler better recognizes complex arrays (reported by Eduardo
49 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
50 for variables on SBCL built with linkage-tables. (reported by Luis
52 * various error reporting improvements.
53 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
54 (Thanks to Thiemo Seufer)
55 * bug fix: nested reader invokations work correctly
56 * bug fix: it is possible to have simultaneous references to foreign
57 code and foreign data with the same name.
59 ** added x86-64 support
60 ** incompatible change: the threading api now works with thread
61 objects instead of thread ids
62 ** bug fix: threads are protected from signals and interruption when
63 starting up or going down
64 ** bug fix: a race where an exiting thread could lose its stack to gc
65 ** bug fix: don't halt on infinite error in threads if possible
66 ** fixed numerous gc deadlocks introduced in the pthread merge
67 ** bug fix: fixed thread safety issues in read and print
68 ** bug fix: debugger doesn't hang on session lock if interrupted at
69 an inappropriate moment
70 ** bug fix: run-program is now thread safe(r)
71 ** bug fix: inner with-recursive-lock no longer releases the mutex
72 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
73 * fixed some bugs revealed by Paul Dietz' test suite:
74 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
75 EXPECTED-TYPE slots filled.
76 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
78 ** MAKE-CONDITION accepts classes as type-designators.
79 ** COMPILE may never return NIL.
80 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
81 range before calling Unix time functions
83 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
84 * numerous signal handling fixes to increase stability
85 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
86 * minor incompatible change: we now correctly canonize default
87 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
88 as per the MOP, rather than the historical (INITARG INITFUNCTION
89 INITFORM). (reported by Bruno Haible)
90 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
91 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
93 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
94 pathname is a directory pathname.
95 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
96 * dynamic space size on PPC has been increased to 768Mb. (thanks to
98 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
99 :METACLASS argument in addition to a class name. (reported by
100 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
101 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
102 (reported by Bruno Haible; patch from Gerd Moellmann)
103 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
105 * bug fix: more cleanups to the floating point exception handling on
106 x86-64 (thanks to James Knight)
107 * bug fix: the compiler does not try to constant fold calls of
108 COERCE and BIT-* functions when they return freshly constructed
109 objects. (reported by Nikodemus Siivola and Paul Dietz)
110 * optimization: improved the MIPS versions of generic (in the
111 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
112 * optimization: direct conversion of (unsigned-byte 32) to floats on
113 the PowerPC platform.
114 * optimization: structure instances with raw slots now use less
115 memory, and probably show better memory locality. (thanks to
117 * optimization: DYNAMIC-EXTENT declarations for lists and closures
118 are treated as requests for stack allocation on the x86-64,
119 Alpha-32, PPC and SPARC platforms.
120 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
121 the wrong file descriptor; implementation of SOCKET-OPEN-P.
122 (thanks to Tony Martinez)
124 ** gcing a dead thread can no longer lead to lockups
125 ** threads block signals until they are set up properly
126 ** errno is no longer shared by threads
127 ** interrupt-thread restores the eflags register on x86
128 ** fixed some lockups due to gc/thread interaction
129 * fixed some bugs revealed by Paul Dietz' test suite:
130 ** invalid dotted lists no longer raise a read error when
132 ** signal an error if a symbol that names a declaration is used
133 as the name of a type, or vice versa
134 ** allow using the (declare (typespec var*)) abbreviation for
135 (declare (type typespec var*)) with all type specifiers
136 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
137 ** always return NIL from PPRINT-POP when OBJECT is NIL
138 ** don't signal errors when pretty-printing malformed LABELS,
139 FLET or MACROLET forms
140 ** declarations in a DOLIST body are also in scope for the
142 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
144 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
147 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
148 * fixed cross-compiler leakages that prevented building a 32-bit
149 target with a 64-bit host compiler.
150 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
151 opened with :IF-EXISTS :OVERWRITE.
152 * fixed bug 281: error for an invalid qualifier in a short-form method
153 combination method is not signalled until the faulty method is called.
154 * bug fix: iteration variable type inferrer failed to deal with open
155 intervals. (reported by Alan Shields)
156 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
157 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
158 or more runtime options were provided to the sbcl binary.
159 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
160 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
161 * optimization: there's now a fast-path for fixnum arguments in the
162 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
163 * optimization: the code generated on x86-64 is more compact thanks
164 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
165 * bug fix: floating-point exception handling now partly works on
166 x86-64. (Thanks to James Knight)
167 * improvement to the MIPS backend from Thiemo Seufer: C-style
168 64-bit long long arguments and return values to alien functions
170 * fixed some bugs revealed by Paul Dietz' test suite:
171 ** the type-error signalled from WARN has a filled-in DATUM slot.
172 ** the type-error required when a stream is not associated with
173 a file has the stream as its datum.
174 ** type-errors on single-floats on x86-64 no longer have
175 :INVALID-OBJECT as the datum
176 ** the type-errors signalled for invalid function names now have
177 a correct expected type
178 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
179 for structures defined with a :TYPE. Documentation strings for
180 typed structures are no longer immediately discarded
181 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
182 broadcast streams and synonym streams. FILE-LENGTH now also works
183 on broadcast streams.
185 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
186 * incompatible change: the --noprogrammer option, deprecated since
187 version 0.7.5, has been removed. Please use the equivalent
188 --disable-debugger option instead.
189 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
191 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
192 inoperational for a while now) has been completely removed.
193 * null lexical environments are now printed as #<NULL-LEXENV>,
194 significantly reducing the amount of clutter in typical backtraces.
195 * documentation on weak pointers, finalization, and after GC hooks
196 has been added to the manual.
197 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
198 as other specialized array types, is much faster. SUBSEQ and
199 COPY-SEQ on such arrays have also been sped up.
200 * optimization: EQL is now more efficient when at least other argument
201 is known to be of type (OR FIXNUM (NOT NUMBER)).
202 * fixed bug: the runtime is now less vulnerable to changes in the
203 size of the SBCL object on OS X, and virtual memory is reserved for
204 all spaces that need to be at a fixed address.
205 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
206 * fixed bug: finalizers and after GC hooks that cause consing are now
208 * fixed bug: compiler error messages and summaries are now printed to
209 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
210 * fixed inference of the upper bound of an iteration variable.
211 (reported by Rajat Datta).
212 * fixed bug 211e: calling local functions with duplicated constant
213 keyword argument no longer causes a bogus style warning about an
215 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
216 ftype declaration to be disregarded. (reported by Dave Roberts)
217 * fixed bug 373: caused by erronous compilation of references to alien
218 variables in the runtime on ppc/darwin.
219 * fixed bug 376: CONJUGATE type deriver.
220 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
221 (reported by Baughn on #lisp)
222 * a cross-compiler bug on non-x86 platforms has been identified and
223 fixed. (thanks to Bruno Haible)
224 * improvements to the MIPS runtime code for increased stability.
225 (thanks to Thiemo Seufer)
226 * increased the maximimum compact environment size to allow
227 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
228 * partial workaround for bug 135: don't name the function we're
229 calling for hairy cases of EVAL, so as not to accumulate one environment
230 entry per call to EVAL. (reported by Kevin Reid)
231 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
232 * optimization: type testing for non-vector arrays should be faster.
233 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
235 * bug fix: setting 31st element of a bit vector to zero did not work
237 * bug fix: redefining a class definition which failed due to a
238 previous accessor / function clash now works (but see BUGS entry
239 #380 for more problems in this area). (thanks to Zach Beane)
240 * the long form of DEFINE-METHOD-COMBINATION disables method group
241 checking when given a single method group with pattern *. (thanks
243 * on x86 compiler supports stack allocation of results of simple
244 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
245 * support for building and running on Mac OS X 10.4 "Tiger" has been added
246 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
247 * fixed some bugs related to Unicode integration:
248 ** the restarts for recovering from input and output encoding
249 errors only appear when there is in fact such an error to
251 * fixed some bugs revealed by Paul Dietz' test suite:
252 ** MISC.549 and similar: late transformation of unsafe type
253 assertions into derived types caused unexpected code
255 ** SCALE-FLOAT type deriver is less wrong.
256 ** type derivers for EXP, LOG and similar functions compute result
257 types for complex arguments better.
258 ** (MISC.563) CONJUGATE type deriver works for very restricted
260 ** out-of-line type testers for character strings are available.
261 ** EQUAL compiler transform understands specialness of objects
263 ** accessing double-floats stored on the stack now works on x86-64.
264 ** debugger internals could sometimes create invalid lispobjs,
265 resulting in GC crashes.
266 ** MISC.548: type check weakening can convert required type into
268 ** initialization forms for bindings are not in scope of free special
271 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
272 * incompatible change: thread support for non-NPTL systems has
273 been removed - locking is buggy and unreliable. A threaded
274 SBCL build will now warn at startup and refuse to create
275 new threads, unless futex support is detected in the kernel
276 * incompatible change: the top level REPL now has only an ABORT
277 restart associated with it, not TOPLEVEL and ABORT as it used to.
278 TOP and TOPLEVEL are now available as debugger commands for
279 returning to the top level.
280 * incompatible change: forms evaluated in the REPL now use the
281 global optimization policy.
282 * incompatible change: user- and system-initialization files are
283 no longer processed with LOAD, but by READ and EVAL; hence the
284 global optimization policy, startup package, readtable, etc,
286 * The .fasl file format number has been incremented because of
287 various incompatible changes.
288 * internal entry point details and argument counts no longer appear
289 in backtraces unless explicitly requested by setting
290 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
291 * built-in and standard functions no longer have names like "top
292 level local call to FOO".
293 * fixed bug 32: functions defined in non-null lexical environments
294 now have more legible printed representation
295 * fixed bug 33: functions defined in non-null lexical environemnts
296 are now more amenable to inspection by INSPECT.
297 * workaround for bug 354: XEPs no longer appear in backtraces unless
298 explicitly requested.
299 * fixed bug: receiving the signal which results from attempting to
300 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
301 SIGBUS on other *BSDs) on architectures where the C stack is also
302 the Lisp stack (x86 and x86-64) from foreign code no longer leads
303 to debugger confusion or wild instability. (reported by Cheuksan
305 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
306 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
307 (reported by Lutz Euler)
308 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
309 compiled code. (reported by Paul Dietz)
310 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
311 specialized parameter to a method to something that is not TYPEP
312 the specializer is now possible.
313 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
314 face of package deletion.
315 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
316 pathname. (reported by David Tolpin on c.l.l)
317 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
318 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
319 STRING-STREAM and FILE-STREAM, respectively.
320 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
321 than previously; in addition, its test suite is now run on build.
322 (thanks to Luke Gorrie)
323 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
324 * optimization: added a immediate representation for single-floats
326 * fixed some bugs revealed by Paul Dietz' test suite:
327 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
328 ** attempting to create a package with a colliding nickname causes
329 correctable errors to be signalled.
330 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
331 ** several x86-64 backend bugs related to sign-extension of immediate
334 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
335 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
336 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
338 * fixed loading of multiply forward-referenced layouts.
339 (thanks to Cheuksan Wang)
340 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
341 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
342 (see COMPILE-FILE documentation for details of :PRINT option)
343 * fixed bugs 19 and 317: fixed-format floating point printing is
344 more accurate. This also fixes a bug reported by Adam Warner
345 related to the ~@F format directive.
346 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
348 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
349 dispatch table if the from-char is a dispatch macro character.
350 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
351 by the wordsize no longer ignore the last word. (reported by Lutz
353 * fixed bug in type checking of dynamic-extent variables. (reported
355 * optimization: sequence traversal functions use their freedom to
356 coerce function designators to functions.
357 * optimization: code with many calls to CLOS methods specialized on
358 CLOS classes has had redundant type checks removed.
359 * optimization: added declarations to speed up operations that access
360 the internal character database (for example STRING-UPCASE)
361 * optimization: comparison operations between floats and sufficiently small
362 fixnums no longer create extra rationals
363 * fixed some bugs related to Unicode integration:
364 ** portions of multibyte characters at the end of buffers for
365 character-based file input are correctly transferred to the
366 start of the buffer at the next read.
367 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
368 passing it through to OPEN.
369 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
370 argument given to internal calls to OPEN.
371 ** the built-in comment readers (introduced by character sequences
372 ";" and "#|") are more forgiving to encoding errors; they will
373 STYLE-WARN and then attempt to resync the stream at a character
374 boundary. (thanks to Teemu Kalvas)
375 * fixed some bugs in the x86-64 port:
376 ** Negative short int return values from c-calls are sign-extended
378 ** The stack is aligned to 16-bytes for c-calls, as required by
379 the ABI. (thanks to Cheuksan Wang)
380 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
381 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
382 for enums. (thanks to Vincent Arkesteijn)
383 ** Multiple small optimizations and bugfixes for floating point
385 * fixed some bugs revealed by Paul Dietz' test suite:
386 ** Space, Tab, Linefeed, Return and Page have the invalid
387 secondary constituent character trait.
388 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
390 ** WITH-INPUT-FROM-STRING should only update the index place on
392 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
394 ** Bit-array operations (BIT-AND and similar) worked incorrectly
395 with one-dimensional arrays with fill pointers.
396 ** TYPE-OF failed on a complex with an integer realpart and a
398 ** compiler failure during type inference for the code of form
399 (IF (EQL X (THE ...)) ...) (MISC.535).
401 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
402 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
403 platform. The port supports 61-bit fixnums, large memory spaces
404 and reloading shared object files.
405 * enhancement: saving cores with foreign code loaded is now
406 supported on ppc/Darwin in addition to the previously supported
408 * enhancement: the statistical profiler now walks deeper into the
409 call stack for more meaningful call-graphs and accrued time
410 reports (x86/x86-64 only). It also now reports time spent in
412 * enhancement: it is now possible to trace most individual methods
413 of a generic function in addition to tracing the generic function
415 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
417 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
418 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
419 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
420 produces an error. (thanks to Vincent Arkesteijn)
421 * bug fix: NAMESTRING on pathnames with :WILD components in their
422 directories works correctly. (thanks to Artem V. Andreev)
423 * fixed bug 125: compiler preserves identity of closures. (reported
425 * bug fixed: functions with &REST arguments sometimes failed with
426 "Undefined function" when compiled with (DEBUG 3). (reported by
428 * bug fix: overflow during compiling of setting element of a bit
429 vector with constant index and value. (reported by Timmy Douglas)
430 * build fix: fixed the dependence on *LOAD-PATHNAME* and
431 *COMPILE-FILE-PATHNAME* being absolute pathnames.
432 * on x86 compiler partially supports stack allocation of dynamic-extent
434 * GO and RETURN-FROM do not check the extent of their exit points
435 when compiled with SAFETY 0.
436 * fixed some bugs related to Unicode integration:
437 ** encoding and decoding errors are now much more robustly
438 handled; it should now be possible to recover even from invalid
439 input or output to the terminal. (thanks to Teemu Kalvas)
440 ** provided a first cut at implementing STRING-TO-OCTETS and
441 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
442 ** altered the SB-MD5 contributed module slightly, changing the
443 interface just enough for it to be supportable for builds where
444 lisp characters are not eight bits.
445 * fixed some bugs revealed by Paul Dietz' test suite:
446 ** the FORMATTER-generated functions for ~V[ conditionals require
447 the correct number of arguments.
448 ** READ-FROM-STRING returns the mandated second value when applied
449 to displaced strings.
450 ** the #\Rubout and #\Backspace characters are treated as invalid
451 constituent characters by the tokenizer.
453 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
454 * new feature: reloading changed shared object files with
455 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
456 * new feature: references to foreign variables and functions
457 can now be compiled and loaded before the corresponding shared
458 object file is loaded, as long as the foreign definitions are
459 available at runtime.
460 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
461 supported. (thanks to Dan Debertin)
462 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
463 just keywords. (thanks to Vincent Arkesteijn)
464 * fixed bug #331: structure-class instances corresponding to
465 DEFSTRUCT forms are now created eagerly.
466 * fixed bug #345: backtraces from calls to undefined functions work
467 on x86 as well. Related bug #61 is now also partially fixed on x86
468 (backtraces from throws to unknown catch tags.)
469 * bug fix: lambda-list parsing is now stricter vrt. order and number
470 of lambda-list keywords.
471 * bug fix: as specified by AMOP, an error is signalled if a
472 class-option appears multiple times in a DEFCLASS form. (reported
474 * bug fix: RANDOM can be compiled when the compiler derives the type
475 of its numeric argument as a disjoint set of small integers.
476 (reported by Paul Dietz)
477 * bug fix: starting a core saved with shared objects loaded when
478 those objects are not available no longer causes threaded SBCL to
479 hang. (reported by Sean Ross)
480 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
482 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
483 argument, which must always be NIL. (reported by Kalle Niemitalo)
484 * bug fix: printing 1.0d+23 no longer results in an error.
485 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
486 * bug fix: structure slot setters preserve evaluation order. (thanks
488 * fixed some bugs related to Unicode integration:
489 ** RUN-PROGRAM can allow its child to take input from a Lisp
490 stream. (reported by Stefan Scholl)
491 ** ASDF-INSTALL successfully downloads .tar.gz files in all
492 locales. (reported by Ken Causey)
493 * fixed some bugs revealed by Paul Dietz' test suite:
494 ** INCF, DECF and REMF evaluate their place form as specified in
496 ** FORMATTER expands ~{ ~} iteration directives with V or #
497 parameters correctly.
498 ** FORMATTER deals with the ~@[ ~] conditional directive where the
499 consequent uses no arguments correctly.
500 ** the system has a partial understanding of the (COMPLEX RATIO)
502 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
505 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
506 * new feature: a build-time option (controlled by the :SB-UNICODE
507 keyword feature, enabled by default) for building the system with
508 support for the entire 21-bit character space defined by the
510 * new feature: the system now has rudimentary external-format
511 support; the primary user-visible change at this time is that
512 characters with the high bit set (such as lower-case-e-acute) will
513 print correctly to a terminal in a UTF-8 environment.
514 * minor incompatible change: BASE-CHAR no longer names a class;
515 however, CHARACTER continues to do so, as required by ANSI.
516 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
517 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
518 has been moved to the SB-EXT package (temporarily re-exported from
520 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
521 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
523 * fixed bug #351: better error-handlind and reporting for malformed
525 * fixed bug #350: bignum-printing is now more memory-efficient,
526 allowing printing of very large bignums, eg. (expt 2 10000000).
527 (reported by Bruno Haible)
528 * fixed bug #302: better primitive-type selection for intersection
530 * fixed bug #308: non-graphic characters now all have names, as
531 required. (reported by Bruno Haible)
532 * bug fix: redefining a class with different superclasses now correctly
533 removes it from the direct-subclasses of its previous superclasses.
534 (reported by David Morse)
535 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
536 new value now works. (reported by Bruno Haible)
537 * bug fix: correct canonicalization of multiple non-standard slot
538 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
539 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
540 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
541 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
542 now exists, an signals an error.
543 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
544 during macroexpansion for non-iteration variables as well. (reported
545 by Bruno Haible for CMUCL)
546 * bug fix: Cyclic structures and unprintable objects in compiler
547 messages no longer cause errors. (reported by Bruno Haible)
548 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
549 and READ-FROM-STRING all now return a primary value of NIL if
550 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
551 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
552 (reported by Bruno Haible for CMUCL)
553 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
554 do not designate a condition. (reported by Bruno Haible for
556 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
557 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
558 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
559 permitting resolution in favour of any of the conflicting symbols.
560 (reported by Bruno Haible for CMUCL)
561 * bug fix: EQUAL compiler optimizations is less aggressive on
562 strings which can potentially compare true despite having distinct
563 specialized array element types.
564 * bug fix: unit enumerations can be defined without dividing by
565 zero. (thanks to Vincent Arkesteijn)
566 * FORMAT compile-time argument count checking has been enhanced.
567 (report from Bruno Haible for CMUCL)
568 * a partial workaround for the bug 262: the compiler does not try to
569 inline-expand a local function doing RETURN-FROM from a deleted
570 BLOCK. (thanks to Peter Denno for the bug report and to David
571 Wragg for the simple test case)
572 * fixed some bugs revealed by Paul Dietz' test suite:
573 ** RENAME-PACKAGE allows all package designators as new package
575 ** constraint propagation and lambda variable substitution are
576 more cautious in dealing with partially deleted code.
577 ** compiler sometimes lost reoptimization passes.
578 ** CERROR, given a condition as condition designator, passes its
579 remaining arguments to the continue format control without
581 ** Case-altering FORMAT directives work correctly on non-ASCII
583 ** The REAL type specifier handles bounds outside the floating
584 point ranges without signalling FLOATING-POINT-OVERFLOW.
585 ** Functions with IR1-transformations can create intercomponent
586 references to global functions.
587 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
589 ** FORMAT ~R treats a nil value for its first parameter correctly.
591 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
592 * enhancement: saving cores with foreign code loaded is now
593 supported on x86/NetBSD and sparc/Linux in addition to the previously
595 * bug fix: on some platforms repeated installations caused multiple
596 copies of HTML documentation to be installed -- should not happen
597 any more. (reported by Stefan Scholl)
598 * bug fix: parsing self-recursive alien record types multiple times
599 no longer causes infinite recursion. (reported by Thomas F. Burdick,
600 original patch by Helmut Eller for CMUCL)
601 * bug fix: stack-exhaustion detection works now on NetBSD as well.
602 (thanks to Richard Kreuter)
603 * bug fix: defining classes whose accessors are methods on existing
604 generic functions in other (locked) packages no longer signals
605 bogus package lock violations. (reported by François-René Rideau)
606 * bug fix: special variables as DEFMETHOD parameters no longer have
607 associated bogus type declarations. (reported by David Wragg and
609 * bug fix: read-write consistency on streams of element-type
610 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
611 * bug fix: redefiniton of the only method of a generic function with
612 no DEFGENERIC no longer emits a full WARNING. In addition,
613 redefinition of generic functions with no DEFGENERIC to an
614 incompatible lambda list now signals an error. (thanks to Zach
616 * bug fix: DEFGENERIC now works even when there's a function of the
617 same name in an enclosing lexical environment. (thanks to Zach
619 * fixed compiler failure, caused by instrumenting code during
620 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
621 * optimization: added loop analysis and improved register allocation
622 * fixed some bugs revealed by Paul Dietz' test suite:
623 ** POSITION on displaced vectors with non-zero displacement
624 returns the right answer.
625 ** (SIMPLE-STRING) is a valid type specifier for sequence
627 ** *PRINT-LEVEL* handling for slotless structures is pedantically
629 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
630 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
632 ** The justification version of the FORMAT ~< directive treats
633 non-zero minpad parameter correctly.
635 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
636 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
637 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
638 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
639 the supported interface.
640 * new feature: Single-stepping of code compiled with DEBUG 2 or
641 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
642 * new feature: saving cores with foreign code loaded is now
643 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
644 Timothy Moore's work for CMUCL)
645 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
646 parameters to * instead of NIL if no initform is supplied.
647 (reported by Johan Bockgaard)
648 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
649 correctly when FUNCALL appears as the car of the form. Note:
650 despite this FUNCALL forms are not currently subject to
651 compiler-macro expansion. (port of Raymond Toy's fix for the
652 same from CMUCL, reported by Johan Bockgaard)
653 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
654 (thanks for Teemu Kalvas)
655 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
656 case instead of terminating the iteration (thanks for Julian
657 Squires, Sean Champ and Raymond Toy)
658 * bug fix: incorrect expansion of defgeneric that caused a style
659 warning. (thanks for Zach Beane)
660 * bug fix: slot accessor effective method computation works properly
661 for classes with multiple non-standard applicable methods on
662 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
663 * on x86 compiler supports stack allocation of results of LIST and
664 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
665 CMUCL implementation by Gerd Moellmann)
666 * fixed some bugs revealed by Paul Dietz' test suite:
667 ** FORMAT strings with both the ~<~:;~> form of the justification
668 directive and pretty-printing directives cause an error.
670 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
671 * incompatible change: the internal functions
672 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
673 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
674 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
675 instead of the old functions.
676 * new feature: on platforms where "dladdr" is available foreign
677 function names now appear in backtraces. (based on Helmut Eller's
679 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
680 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
682 * documentation: ASDF manual is now distributed with SBCL.
683 * bug fix: SBCL can now load its contributed modules with REQUIRE
684 even if the system-provided entries have been removed from
685 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
687 * bug fix: backtraces involving undefined functions or assembly
688 routines are more informative. (thanks to Brian Downing)
689 * bug fix: mutually referent alien structures now work correctly.
690 (reported by Rick Taube)
691 * bug fix: structures defined by WITH-ALIEN can be referred to
692 within other definitions in the same WITH-ALIEN.
693 * bug fix: division operators (MOD, TRUNCATE and the like) with
694 constant zero divisors and integer dividends no longer generate
696 * bug fix: provide default methods for INPUT-STREAM-P and
697 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
698 * bug fix: improve the detection and resolution of MOP metacycles
699 (where effective method computation depends on the generic function
700 having its effective method computed). (reported by Bruno Haible)
701 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
702 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
703 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
704 * bug fix: pathnames with relative directory components can be
705 represented relative to default pathnames.
706 * optimization: in taking the GCD of bignums, reduce the two bignums
707 to approximately the same size (using Euclid's algorithm) before
708 applying the more sophisticated binary GCD. (thanks to Juho
710 * optimization: COUNT on bitvectors now operates word-at-a-time.
711 * optimization: ASH with a positive, but not necessarily constant,
712 (leftwards) shift, when in a modular context, is compiled to a
714 * fixed some bugs revealed by Paul Dietz' test suite:
715 ** FORMAT variable parameters ("~V<char>") are defaulted properly
716 if the corresponding argument is NIL.
717 ** FORMAT directives accepting a minpad parameter treat negative
719 ** literal commas as character parameters to FORMAT directives are
721 ** literal spaces directly after ~<Newline> directives within a
722 format-logical-block (~:< ~@:>) do not induce :FILL-style
723 conditional newlines.
724 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
725 as specified: it no longer includes conditional newlines).
726 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
728 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
729 :PER-LINE-PREFIX argument does not evaluate to a string.
731 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
732 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
733 the manual for details; for now, package locks can be disabled by
734 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
735 if no major problems are found then it is likely that they will be
736 compiled in unconditionally.
737 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
738 unsupported operators on all platforms. To load a shared library
739 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
740 object file, link it into a shared library outside of SBCL and
741 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
743 * bug workaround: ROOM T and the GC were not working together
744 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
745 reporting facility aren't preserved. That reporting has been
746 disabled, so now until and unless someone figures out how to make it
747 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
748 * minor incompatible change: as threatened around sbcl-0.8.0, the
749 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
750 The symbols are also exported from SB-PCL for backwards
751 compatibility, but more so than before SB-PCL should be treated as
752 an implementation-internal package.
753 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
755 * the SB-SPROF contrib now works on (most) non-x86 architectures.
756 It is known as of this release not to work on the Alpha, however.
757 * fixed bug #167: errors signalled due to illegal syntax in method
758 bodies are now more legible.
759 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
760 designators and can hence be used with TYPEP.
761 * fixed bug #333: CHECK-TYPE now ensures that the type error
762 signalled, if any, has the right object to be accessed by
763 TYPE-ERROR-DATUM. (reported by Tony Martinez)
764 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
765 5.1.2.3 for multiple-value place subforms. (reported by Kalle
767 * fixed bug #334: programmatic addition of slots using specialized
768 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
769 and :ALLOCATION :CLASS slots.
770 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
771 float exponents. (rereported by Peter Seibel)
772 * fixed a bug: #\Space (and other whitespace characters) are no
773 longer considered to be macro characters in standard syntax by
775 * fixed bug: initialization of condition class metaobjects no longer
776 causes an instance of the condition to be created. (reported by
778 * fixed bug: it is now possible to have slots such that
779 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
780 description is neither :INSTANCE nor :CLASS.
781 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
782 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
783 SLOT-VALUE-USING-CLASS), and no longer causes errors with
784 non-standard SLOT-DEFINITION-ALLOCATION values.
785 * fixed bugs: various race conditions handling exiting threads.
786 CL-PPCRE's thread torture test now passes.
787 * fixed arguably-a-bug: GC time proportional to number of threads in
788 system even when most of them are idle
789 * optimization: improved performance of BIT and SBIT on bit-vectors.
790 * .fasl file incompatibility: The fasl file version number has
791 been incremented because of changes associated with package locks.
793 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
794 * minor incompatible change: the system no longer provides
795 optimization hints (conditions which are TYPEP
796 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
797 mode; these hints are still emitted when higher SPEED optimization
799 * new contrib module: a sampling profiler (profiling by statistical
800 sampling, rather than by instrumenting functions) is available as
801 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
802 * the behaviour of the standard function ED is now customizeable by
803 third parties through a hook variable: see ED's documentation
804 string for information on the protocol.
805 * the compiler no longer emits efficiency notes for (FUNCALL X)
806 when the type of X is uncertain under default optimization
808 * fixed bug 276: mutating a binding of a specialized parameter to a
809 method to something that is not TYPEP the specializer is now
811 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
812 exhibit the required behaviour on the x86 platform. (thanks to
813 Peter van Eynde, Eric Marsden and Bruno Haible)
814 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
815 even for difficult arguments. (reported by Peter Graves)
816 * fixed bug 141a: the backquote printer now descends quoted
818 * fixed another bug in backquote printing: no more destructive
819 modification of the form's list structure. (reported by Brian
821 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
822 move between its address being taken and the call to
823 interrupt_thread, fixing a crashing race condition.
824 * the SB-POSIX contrib implementation has been adjusted so that it
825 no longer exhibits ridiculously poor performance when constructing
826 instances corresponding to C structs.
828 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
829 * minor incompatible change: the sb-grovel contrib now treats C
830 structures as alien (in the sense of SB-ALIEN) objects rather than
831 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
832 has implications for memory management of client code
833 (sb-grovel-returned objects must now be manually managed) and for
834 type safety (alien objects now have full types).
835 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
836 used to control emission of compiler diagnostics, rather than the
837 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
838 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
839 quality should be considered deprecated.
840 * install.sh now installs the user manual as well
841 * (not quite a new documentable feature, but worth considering in
842 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
843 beginnings of a semantically meaningful condition hierarchy is
844 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
846 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
847 designator as the defaults argument.
848 * fixed bug: Displaced arrays whose displaced-to array has become
849 too small now cause ARRAY-DIMENSION to signal an error, providing
850 for safer bounds-checking. (reported by Bruno Haible)
851 * fixed bug: DEFCLASS slot definitions with identical :READER and
852 :WRITER names now signal a reasonable error. (reported by Thomas
854 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
856 * fixed bug: Invalid slot specification errors now print correctly.
857 (thanks to Zach Beane)
858 * fixed bug 320: Shared to local slot value transfers in class
859 redefinitions now happen corrently from superclasses as
860 well. (reported by Bruno Haible)
861 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
863 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
864 improper lists correctly. (reported by Bruno Haible)
865 * fixed bug 313: source-transform for <fun-name> was erroneously
866 applied to a call of a value of a variable with name <fun-name>.
867 (reported by Antonio Menezes Leitao)
868 * fixed bug 307: The obsolete instance protocol ensures that
869 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
870 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
871 remove dead unknown-values globs from the middle of the stack.
872 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
873 easier to resume long computations after using *BREAK-ON-SIGNALS*
874 to diagnose and fix failures. (thanks to Nikodemus Siivola)
875 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
876 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
877 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
878 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
879 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
881 * fixed a (fairly theoretical) bug in string printing: if
882 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
883 does not have array-element-type equal to the most general string
885 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
886 function name. (thanks to Nikodemus Siivola)
887 * fixed bug: compile-time format string checker failed on
888 non-closed ~{. (reported by Thomas F Burdick)
889 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
890 don't warn on legal loop code involving more than one aggregate
891 boolean. (thanks to Nikodemus Siivola)
892 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
893 in keyword position to MAKE-PACKAGE was not being reported as
895 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
896 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
898 * fixed bugs 280 and 312: the checking for multiple definitions in a
899 file is less likely to become confused by uses of inline
901 * fixed bug: the #S reader macro performs the keyword coercion
902 specified for slot names. (reported by Kalle Niemitalo)
903 * fixed bug: lambda lists may contain symbols whose names start with
904 & but are not lambda-list-keywords; their occurrence triggers a
906 * fixed bug 321: define-method-combination argument lambda lists do
907 not cause invalid code to be generated when &optional or &aux
908 variables have default values. (reported by Bruno Haible)
909 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
910 have CLOS classes; this appears also to have fixed reported
911 instability in stack exhaustion detection.
912 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
913 MUFFLE-WARNING when no associated restart is present is now
915 * optimization: rearranged the expansion of various defining macros
916 so that each expands into only one top-level form in a
917 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
919 * optimization: used a previously little-used slot in symbols to
920 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
922 * fixed some bugs revealed by Paul Dietz' test suite:
923 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
925 ** Fixed an optimization bug related to inheritance of initforms
926 from local to shared slots.
927 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
928 ** CAST optimizer forgot to flush argument derived type.
929 ** print/read consistency on floats is now orders of magnitude
930 more likely. (thanks also to Bruno Haible for a similar report
932 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
933 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
934 for objects of type REAL. Make it so.
935 ** SXHASH is specified (infelicitously) to respect similarity,
936 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
937 it so. (thanks to Markus Ziegler)
938 ** on the Alpha, the compiler succeeds in compiling functions
939 returning a known number of arguments greater than 63.
940 ** fixed handling of invalid NIL arguments in keyword position
942 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
944 ** provided a readably-printable representation for RANDOM-STATE
946 ** ensured that pathnames, where they have a namestring, always
947 print using #P"..." syntax.
949 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
950 * Support for the forthcoming 2.0 version of the NetBSD kernel
951 running on x86 hardware has been added. (thanks to Perry
952 E. Metzger most immediately, and others for their past work)
953 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
954 current release version; SBCL's OpenBSD support had been broken
955 since about the time of OpenBSD's switch to ELF binary format.)
956 * [placeholder for DX summary]
957 ** user code with &REST lists declared dynamic-extent, under high
958 speed or space and low safety and debug optimization policy.
959 * The manual has been converted to Texinfo format and the debugger
960 chapter from the cmucl manual has been added.
961 * A facility has been added to extract documentation strings from
962 sbcl and store them as Texinfo-formatted snippets for inclusion in
963 the manual (via Texinfo's @include directive)
964 * bug fix: compiler emitted division in optimized DEREF. (thanks for
965 the test case to Dave Roberts)
966 * bug fix: multidimensional simple arrays loaded from FASLs had fill
967 pointers. (reported by Sean Ross)
968 * bug fix: PROFILE output is printed nicely even for large numerical
969 values. (thanks to Zach Beane)
970 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
971 greater than 32 handle EOF correctly.
972 * bug fix: on X86 an immediate argument of the IMUL instruction is
973 correctly printed in disassembly. (thanks to Lutz Euler)
974 * bug fix: class slots in redefined classes preserve their old
975 values. (thanks to Bruno Haible and Nikodemus Siivola)
976 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
977 succeeds. (reported by Marco Baringer)
978 * fixed some bugs revealed by Paul Dietz' test suite:
979 ** READ-SEQUENCE now works on ECHO-STREAMs.
980 ** RATIONALIZE works more according to its specification. (thanks
983 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
984 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
985 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
986 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
987 should matter to you only if you rebind the printer control
988 variables and then find you want different bindings in the
989 debugger than in the ordinary execution of your program.)
990 * The runtime build system has been tweaked to support building
991 (on SPARC/SunOS) using a C compiler which invokes Sun's own
992 assembler and linker. (thanks to Nikodemus Siivola)
993 * Unbound, undefined, undeclared variables now trigger full
994 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
995 more often programmer error than deliberate exploitation of undefined
997 * optimization: the hash algorithm for strings has changed to one
998 that is less vulnerable to spurious collisions. (thanks to Juho
1000 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1001 needless bounds checking. (thanks to Juho Snellman)
1002 * optimization: implemented multiplication as a modular
1003 (UNSIGNED-BYTE 32) operation on the PPC backend.
1004 * fixed some bugs revealed by Paul Dietz' test suite:
1005 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1007 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1008 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1009 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1011 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1013 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1015 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1016 their output stream on EOF from read.
1017 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1018 have been read to end-of-file.
1019 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1021 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1022 description of determination of which consecutive characters
1024 ** Printing the "Space" character with escaping on now yields "#\\ ",
1025 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1026 ** Reading floating-point numbers with *READ-BASE* set to a number
1027 less than 10 works correctly.
1028 ** Reading floating-point numbers with *READ-BASE* set to a number
1029 more than 10 works correctly.
1030 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1031 the readtable currently in effect.
1033 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1034 * minor incompatible change: parsing of namestrings on a physical
1035 (Unix) host has changed; numbers after the final #\. in a
1036 namestring are no longer interpreted as a version field. This is
1037 intented to be largely invisible to the user, except that the
1038 meaning of the namestring "*.*.*" has changed: it now refers to a
1039 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1040 should usually be replaced by
1041 (make-pathname :name :wild :type :wild :version :wild)
1042 with the added benefit that this is more likely to be portable.
1043 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1044 option to OPEN now signals an error if the file being opened
1045 exists; this may have an impact on existing code.
1046 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1047 signals received, on the PowerPC platforms (both Linux and
1048 Darwin). (thanks to Pierre Mai for pointing out the location of
1050 * several fixes on OS X: The system now builds and runs cleanly on
1051 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1052 (thanks to Brian Mastenbrook)
1053 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1054 second-resolution: integer multiples of 1/3600 between -24 and 24.
1055 (thanks to Vincent Arkesteijn)
1056 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1057 type when called with 1 argument; PEEK-CHAR checked type of
1058 PEEK-TYPE only after having read first character from a
1059 stream. (reported by Peter Graves)
1060 * bug fix: the garbage collector now has much better locality
1061 behaviour, and in particular no longer treats all memory as being
1062 exhausted when many small objects point to each other in a deeply
1064 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1065 recognized as being TYPEP their class.
1066 * bug fix: the PUSHNEW documentation string has been corrected.
1067 (thanks to Vincent Arkesteijn)
1068 * bug fix: defaulting of the value for the last of an atypically
1069 large number of multiple values being bound was not being
1070 performed correctly on the Alpha or PPC platforms
1071 * optimization: implemented multiplication as a modular
1072 (UNSIGNED-BYTE 32) operation on the x86 backend.
1073 * optimization: SEARCH on simple-base-strings can now be open-coded.
1074 (see also contrib/compiler-extras.lisp for inspiration for
1075 teaching the compiler about the Boyer-Moore algorithm).
1076 * value, returned by MAX (and MIN) called with several EQUALP, but
1077 not EQL, arguments now does not depend on compiler settings.
1078 * fixed some bugs revealed by Paul Dietz' test suite:
1079 ** in stack analysis liveness information is propagated from
1080 non-local entry points.
1081 ** pathwise CAST removing failed when the CAST node did not start
1083 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1084 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1086 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1087 host is already defined.
1088 ** RENAME-FILE works on streams instead of signalling an internal
1090 ** PEEK-CHAR uses the current readtable when determining whether
1091 or not a character is whitespace.
1092 ** MERGE-PATHNAMES handles the case when the pathname does not
1093 specify a name while the default-pathname specifies a version.
1094 ** Pathnames now stand a better chance of respecting print/read
1096 ** Attempting to use standardized file system operators with a
1097 pathname with invalid :DIRECTORY components signals a
1099 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1101 ** DIRECTORY on logical pathnames is more correct.
1102 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1103 signal a TYPE-ERROR if their argument is not a stream.
1104 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1105 designator argument does not designate a stream.
1106 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1107 examining the synonym.
1108 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1110 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1111 element-type larger than ([UN]SIGNED-BYTE 32).
1113 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1114 * When built with the :SB-FUTEX feature, threaded builds now take
1115 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1116 for faster/more reliable mutex and condition variable support.
1117 * Incompatible change (but one you probably shouldn't have been using
1118 anyway): the interface and code for arbitrating between multiple
1119 threads in the same user session has been redesigned.
1120 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1121 error when called without an explicit environment argument.
1122 (thanks to Nikodemus Siivola)
1123 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1124 confused about their position. (thanks to Adam Warner and Gerd
1126 * bug fix: Pretty printing backquoted forms with unquotations in the
1127 argument list position of various code constructs such as LAMBDA
1128 now works correctly. (reported by Paul Dietz)
1129 * bug fix: Pretty printing unquotations no longer loses all
1130 stream position information.
1131 * optimization: performance of string output streams is now less
1132 poor for multiple small sequence writes.
1133 * optimization: performance of CSUBTYPEP in the presence of complex
1134 expressions involving CONS and NOT many times has been improved.
1135 (reported by Paul Dietz)
1136 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1138 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1139 streams for the REPL, for the debugger, and so on. (thanks to
1141 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1142 * fixed some bugs revealed by Paul Dietz' test suite:
1143 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1144 is ignored if it is too small, rather than propagating through
1145 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1146 ** extremely complex negations of CONS types were not being
1147 sufficiently canonicalized, leading to inconsistencies in
1149 ** VALUES tranformer lost derived type.
1151 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1152 * fixed a bootstrapping bug: the build process no longer assumes
1153 that the various BOOLE-related constants have the same value in
1154 host and target lisps. (noted by Paul Dietz' test suite on an
1155 SBCL binary built from CLISP)
1156 * The system can now be dynamically linked on the MIPS platform,
1157 which enables dynamic loading of foreign code from Lisp. (thanks
1158 to Ralf Baechle for discussions on the MIPS ABI)
1159 * The system now records debugging information for its own source
1160 files in a filesystem-position-independent manner, relative to
1161 the "SYS" logical host.
1162 * fixed a compiler bug: MV-LET convertion did not check references
1163 to the "max args" entry point. (reported by Brian Downing)
1164 * tweaked disassembly notes to be less confident about proclaiming
1165 some instruction as an LRA. (thanks to Brian Downing)
1166 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1167 now each have their own history, command character, and other
1168 characteristics. (thanks to David Lichteblau)
1169 * fixed some bugs revealed by Paul Dietz' test suite:
1170 ** compiler failure in compiling LOGAND expressions including a
1172 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1173 shift greater than 32.
1174 ** FLUSH-DEST did not mark blocks for type check regeneration.
1175 ** HANDLER-CASE failed to accept declarations in handler clauses
1176 in some circumstances.
1178 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1179 * New code in contrib/sb-introspect (still probably not entirely
1180 stable yet) provides some support for smart Lisp development
1181 environments like SLIME.
1182 * The conditions signalled for errors occurring when loading .fasl
1183 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1184 in a way which should help ASDF recover gracefully.
1185 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1186 stuff has been cleaned up. If you code contrib/ stuff, this might
1187 affect you, and you can look at contrib/README, contrib/STANDARDS,
1188 and/or the 0.8.4.27 diff to check.
1189 * In full calls the compiler now does not generate checks for declared
1190 argument types for all arguments.
1191 * various threading fixes
1192 ** and some experimental patches which didn't make it into
1193 the main tree for this release, but which are shipped in
1194 contrib/experimental-thread.patch as a possible fix for some
1195 failures (deadlock, spinning...) in GC-intensive multithreaded
1197 * fixed PPC build problem (source code incompatibility of different
1198 library versions): added offsetof() hackery which attempts to divine
1199 where glibc maintainers put uc_mcontext today
1200 * fixed bug 282: compiler does not trust type assertions while passing
1201 arguments to a full call.
1202 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1203 &OPTIONAL VALUES type parameter.
1204 * fix bug 214: algorithm for noting rejected templates is now more
1205 similar to that of template seletion. (also reported by rydis on
1207 * fixed bug 141b: printing backquoted information readably and prettily
1208 inserts a space where necessary.
1209 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1210 (CHARACTER 10) are now reported as errors, rather than propagated
1211 as unknown types. (reported by piso on #lisp)
1212 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1213 with values NIL and :ERROR. (thanks to Milan Zamazal)
1214 * fixed bug 191c: CLOS now does proper keyword argument checking as
1215 described in CLHS 7.6.5 and 7.6.5.1.
1216 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1217 longer raise an error; further, using a list as a for-as-arithmetic
1218 counter now raises a meaningful error.
1219 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1220 now understood by sequence creation functions such as MAKE-SEQUENCE
1222 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1223 a STRING-INPUT-STREAM.
1224 * compiler enhancement: SIGNUM is now better able to derive the type
1226 * type declarations inside WITH-SLOTS are checked. (reported by
1228 * fixed some bugs revealed by Paul Dietz' test suite:
1229 ** incorrect optimization of TRUNCATE for a positive first
1230 argument and negative second.
1231 ** compiler failure in let-convertion during flushing dead code.
1232 ** compiler failure while deriving type of TRUNCATE on an
1233 interval, containing 0.
1234 ** ASH of a negative bignum by a negative bignum count now returns
1236 ** intersection of CONS types now canonicalizes properly, fixing
1237 inconsistencies in SUBTYPEP.
1239 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1240 * incompatible change: The --disable-debugger command line
1241 option now clobbers the debugger at a more fundamental
1242 level, by redefining #'INVOKE-DEBUGGER instead of by
1243 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1244 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1245 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1246 enter the debugger REPL and then suffer recursive errors
1247 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1248 while under the new system, BREAK in --disable-debugger mode
1249 terminates the system just as an unhandled error would.
1250 * fixed compiler performance when processing loops with a step >1;
1251 * bug fix: DOCUMENTATION now retrieves generic function
1252 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1253 support has been systematized, and now supports the methods
1254 specified by ANSI, along with a default method and a method for
1255 slot documentation. (reported by Nathan Froyd)
1256 * bug fix: effective methods associated with a generic function are
1257 no longer cached over a change of that generic function's method
1258 combination. (reported by Andreas Fuchs)
1259 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1260 names a non-existent directory. (thanks to Andreas Fuchs)
1261 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1262 the PPC platform, lead to stack corruption; this has been fixed.
1263 (reported by Rainer Joswig)
1264 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1265 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1266 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1267 arithmetic in general, is now much more reliable on the PPC
1269 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1270 platform now returns the right answer.
1271 * optimization: restored some effective method precomputation in
1272 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1273 precomputation is now tunable.
1274 * optimization: compiler-internal data structure use has been
1275 reviewed, and changes have been made that should improve the
1276 performance of the compiler by about 20%.
1277 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1278 simple-base-strings and simple-bit-vectors is improved.
1279 * optimization: the optimization of 32-bit logical and arithmetic
1280 functions introduced in version 0.8.3 on the x86 has been
1281 implemented on the mips, ppc and sparc platforms; an
1282 implementation of the same facility, but for 64-bit arithmetic,
1283 has been added for the alpha.
1284 * microoptimization: the compiler is better able to make use of the
1285 x86 LEA instruction for multiplication by constants.
1286 * bug fix: in some situations compiler did not report usage of
1287 generic arithmetic in (SPEED 3) policy.
1288 * bug 145b fix: compiler used wrong type specifier while converting
1289 MEMBER-types to numeric.
1290 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1292 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1293 index argument now works properly on the Alpha platform.
1294 * bug fix: floating point exception treatment on the Alpha platform
1296 * bug fix: FILE-POSITION works much better on string input and
1297 output streams. (thanks to Nikodemus Siivola)
1298 * bug fix: many threading/garbage collection symptoms sorted.
1299 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1300 might be pseudo-atomic.
1301 * internal change: Stopping for GC is now done with signals not
1302 ptrace. GC is now done in whichever thread wanted it, instead of
1304 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1306 * bug fix: result form in DO is not contained in the implicit
1308 * incompatible change: ICR structure is changed; the value part of
1309 CONTINUATION is now called LVAR; corresponding functions are
1310 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1311 * added type deriver for ISQRT (thanks to Robert E. Brown).
1312 * bug fix: better support for loading from the command line when an
1313 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1315 * fixed some bugs revealed by Paul Dietz' test suite:
1316 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1317 ** ROUND and FROUND now give the right answer when given very
1318 small float arguments.
1319 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1321 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1322 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1323 ** LCM should return a non-negative integer.
1324 ** PARSE-INTEGER returned the index of a terminator instead of the
1325 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1326 ** PARSE-INTEGER returned an incorrect index being applied to a
1328 ** LCM with two arguments of 0 returns 0 rather than signalling
1330 ** unsigned addition of a 32-bit constant with the high bit set no
1331 longer causes an internal compiler error on the x86.
1332 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1333 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1334 with negative last argument.
1335 ** byte specifiers with zero size and position no longer cause
1336 an error during type derivation.
1337 ** bignum multiplication on the Alpha platform now returns the
1339 * porting: The system now builds on SuSE AMD64, although it still
1340 generates a 32-bit binary.
1341 * .fasl file incompatibility: The fasl file version number has
1342 been incremented (because of the changes to internal compiler
1343 data structures referred to above).
1345 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1346 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1347 more accurately, on the Darwin kernel running on PowerPC hardware.
1348 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1349 * Compiler code deletion notes now signal a condition of type
1350 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1351 an associated MUFFLE-WARNING restart.
1352 * The compiler now performs limited argument count validation of
1353 constant format strings in FORMAT, and where appropriate in ERROR,
1354 CERROR and WARN. (thanks to Gerd Moellmann)
1355 * New ASDF-INSTALL contrib can be used for automatic download and
1356 installation of third-party Lisp code from CCLAN or other sites
1358 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1359 which forces another thread to execute a function supplied by the
1361 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1362 now accept and act upon their :ELEMENT-TYPE keyword argument.
1363 (reported by Martin Atzmueller, Edi Weitz)
1364 * bug fix: FILE-POSITION now accepts position designators up to
1365 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1366 is the greater. (thanks to Patrik Nordebo)
1367 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1368 Kalle Olavi Niemitalo)
1369 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1370 printable. (reported by Eric Marsden)
1371 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1372 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1373 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1374 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1375 (thanks to Stig Sandoe)
1376 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1377 UNREAD-CHAR now correctly includes the unread character in the
1378 target sequence. (thanks to Gerd Moellmann)
1379 * bug fix in threaded builds: the system can now be suspended and
1380 resumed by shell job control with minimal disruption.
1381 * bug fixes in times and timezones >2038AD
1382 * better handling of "where is GNU make?" problem in build scripts
1383 (thanks to Nikodemus Siivola)
1384 * new optimization: inside a named function any reference to a
1385 function with the same name is considered to be a self-reference;
1386 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1387 optimization quality.
1388 * new optimization on x86: logical functions and + now have
1389 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1390 used when the result is truncated to 32 bits.
1391 * VALUES declaration is partially enabled.
1392 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1393 * fixed some bugs revealed by Paul Dietz' test suite:
1394 ** The system now obeys the constraint imposed by
1395 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1396 types form a lattice under type intersection.
1397 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1398 ** ASSOC now ignores NIL elements in an alist.
1399 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1400 and (1+ MOST-POSITIVE-FIXNUM) answers.
1401 ** The addition of a method with invalid qualifiers to a generic
1402 function does not cause an error to be signalled immediately;
1403 a warning is signalled, and the error is generated only on
1404 calling the generic function.
1405 * changed .fasl file version number, in part to add type codes for
1406 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1407 obscure ANSI requirements
1409 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1410 * fixed bug 148: failure to inline-expand a local function left
1411 garbage, confusing the compiler.
1412 * fixed bugs 3cd: structure slot readers perform type check if the
1413 slot can have an invalid value (i.e. it is either not initialized
1414 or can be written with a less specific slot writer).
1415 * bug fix: the compiler now traps array references to elements off
1416 the end of an array; previously, the bounds checking in some
1417 circumstances could go off-by-one.
1418 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1420 * type declarations for array element types now obey the description
1421 on the CLHS page "Declaration TYPE", as per discussions on
1422 sbcl-help around 2003-05-08. This means that a declaration
1423 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1424 declaration, all references to BAR will be asserted or assumed
1425 (with THE, so dependent on compiler policy) to involve objects of
1426 type FOO. Note that no such declaration is implied in
1427 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1428 * declared types of functions from the "Conditions"
1429 chapter. (reported by Paul Dietz)
1430 * bug fix: CERROR accepts a function as its first argument.
1431 * bug fix: NTH an NTHCDR accept a bignum as index
1432 arguments. (reported by Adam Warner)
1433 * optimization: character compare routines now optimize comparing
1434 against a constant character. (reported by Gilbert Baumann)
1435 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1436 argument now works properly on the MIPS platform.
1437 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1438 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1439 declared to have a functional type, does not cause an error
1441 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1442 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1443 * fixed bug: if last continuation of a deleted block has a
1444 destination, this destination should be deleted too. (reported by
1446 * fixed a bug in the bootstrap process: the host compiler's values
1447 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1448 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1449 test case from Patrik Nordebo)
1450 * improved the ability of the disassembler on the PPC platform to
1451 provide helpful disassembly notes.
1452 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1453 the class in more cases than previously.
1454 * bug fix: FILE-POSITION now understands :START and :END for
1455 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1456 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1457 8) data. (thanks to David Lichteblau)
1458 * bug fix: it is possible to add a method to a generic function
1459 without lambda list.
1460 * bug fix: reader failed to signal END-OF-FILE inside an
1461 object representation. (reported by Nikodemus Siivola)
1462 * fixed some bugs revealed by Paul Dietz' test suite:
1463 ** LAST and [N]BUTLAST should accept a bignum.
1464 ** condition slot accessors are methods.
1465 ** (VECTOR NIL) is a subtype of STRING.
1467 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1468 * minor incompatible change: some nonsensical specialized lambda
1469 lists (used in DEFMETHOD) which were previously ignored now signal
1471 * minor incompatible change: the system is now aware of the types of
1472 variables in the COMMON-LISP package, and will signal errors for
1473 most violations of these type constraints (where previously they
1474 were silently accepted).
1475 * minor incompatible change: COMPILE-FILE now uses the freedom
1476 afforded (ANSI 3.2.2.3) to use derived function types for
1477 functions defined in the same file. This also permits the system
1478 to warn on static type mismatches and function
1479 redefinition. (Currently it does not work with high DEBUG level.)
1480 * minor incompatible change: VALUES declaration is disabled.
1481 * When issuing notes, the compiler now signals a condition of type
1482 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1483 restart for use in user handlers. It is expected that the
1484 COMPILER-NOTE condition will eventually become a condition
1485 supertype to a hierarchy of note types, which will then be
1486 handleable in a similar fashion. However, at the moment, no such
1487 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1488 but it's an implementation detail, not a classification for the
1490 * Changes in type checking closed the following bugs:
1491 ** type checking of unused values (192b, 194d, 203);
1492 ** template selection based on unsafe type assertions (192c, 236);
1493 ** type checking in branches (194bc).
1494 * A short form of VALUES type specifier has ANSI meaning (it has
1495 increased the number of situations when SBCL cannot perform type
1497 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1498 keywords or constants is permissible.
1499 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1500 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1501 classes. (thanks to Antonio Martinez)
1502 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1503 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1504 argument to a pretty-print stream. Instead, it leaves any such
1505 operations to DESCRIBE-OBJECT methods.
1506 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1508 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1509 lambda lists. (thanks to David Lichteblau)
1510 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1511 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1512 (thanks to Gerd Moellmann)
1513 * fixed bug 246: increased compilation speed of long
1514 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1516 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1518 * DEFCONSTANT now throws a condition of type
1519 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1520 constant to a non-EQL value; CONTINUE and ABORT restarts
1521 respectively change and preserve the value.
1522 * fixed bug 63: The code walker, part of the implementation of CLOS,
1523 is now better at handling symbol macros.
1524 * bug fix: There is no longer an internal implementation type named
1525 CL:LENGTH. (reported by Raymond Toy)
1526 * bug fix: In macro-like defining macros/special operators the
1527 implicit block does not enclose the lambda list.
1528 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1529 enclosed in parentheses) are not suitable as type specifiers, and
1530 their use properly signals an error now.
1531 * bug fix: An argument count mismatch for a type specifier in code
1532 being compiled no longer causes an unhandled error at compile
1533 time, but signals a compile-time warning.
1534 * fixed simple vector readable printing
1535 * bug fix: DESCRIBE takes more care over whether the class
1536 precedence list slot of a class is bound before accessing it.
1537 (reported by Markus Krummenacker)
1538 * bug fix: FORMATTER can successfully compile pretty-printer format
1539 strings which use variants of the ~* directive inside.
1540 * bug fix: SEARCH now applies its TEST predicate to the elements of
1541 the arguments in the correct order. (thanks to Wolfhard Buss)
1542 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1543 in type inference. (thanks to Robert E. Brown)
1544 * bug fix: Reading in symbols with an explicit package name of ""
1545 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1546 * fixed some bugs revealed by Paul Dietz' test suite:
1547 ** NIL is now allowed as a structure slot name.
1548 ** Arbitrary numbers, not just REALs, are allowed in certain
1549 circumstances in LOOP for-as-arithmetic clauses.
1550 ** Multiple class redefinitions before slot access no longer
1551 causes a type error.
1552 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1553 association between the name and a class.
1554 ** Generic functions with non-standard method-combination and over
1555 five methods all of which return constants no longer return NIL
1556 after the first few invocations. (thanks to Gerd Moellmann)
1557 ** CALL-NEXT-METHOD with no arguments now passes the original
1558 values of the arguments, even in the presence of assignment.
1559 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1561 ** DISASSEMBLE works with closures and funcallable instances.
1562 ** ADD-METHOD now returns the generic function, not the new method.
1563 ** FIND-METHOD signals an error if the lengths of the specializers
1564 is incompatible with the generic function, even if the ERRORP
1566 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1567 which its argument is a member.
1568 ** DEFCLASS only redefines the class named by its class-name
1569 argument if that name is the proper name of the class;
1570 otherwise, it creates a new class.
1571 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1572 of the UNBOUND-SLOT condition to the name of the slot.
1573 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1574 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1575 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1576 SLOT-MAKUNBOUND in the specified fashion.
1578 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1579 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1580 cross-compilation host. As a consequence, we can now bootstrap our
1581 way up to SBCL starting with a bare gcc toolchain and human-readable
1582 source code (first the source to CLISP, then the source to SBCL).
1583 * A contributed module containing a partial implementation of the
1584 simple-streams interface has been included. (thanks to Rudi
1586 * A contributed module implementing the RFC1321 Message Digest
1587 Algorithm, known as MD5, has been included.
1588 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1589 no longer has any effect, as the code controlled by this feature
1590 has been deleted. (As far as we know, no-one has ever built using
1591 this feature, and its semantics were confused in any case).
1592 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1593 work on conditions (as required by the ANSI specification),
1594 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1595 expected behaviour on conditions. Users should note, however,
1596 that such behaviour is not required by the ANSI specification,
1597 and so use of this behaviour may render their code unportable.
1598 * fixed some bugs revealed by Paul Dietz' test suite:
1599 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1601 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1602 ** SXHASH on condition objects no longer returns NIL.
1603 ** :ALLOCATION :CLASS slots are better treated; their values are
1604 updated on class redefinition, and initforms inherited from
1605 superclasses are applied.
1606 ** REMOVE-METHOD returns its generic function argument even when
1607 no method was removed.
1608 ** SHARED-INITIALIZE now initializes the values of the requested
1609 slots, including those with :ALLOCATION :CLASS.
1610 ** ALLOCATE-INSTANCE now works on structure classes defined via
1611 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1613 ** SLOT-EXISTS-P now works on conditions, as well as structures
1615 ** MAKE-LOAD-FORM now has the required methods on
1616 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1617 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1618 keyword, and now implements the SLOT-NAMES argument.
1619 ** methods with &OPTIONAL arguments no longer allow too many
1620 arguments to be passed in the call without error.
1621 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1622 option is consistent with the required arguments of the generic
1623 function lambda list.
1624 * bug fix: REQUIRE accepts a string designator. (Thanks to
1626 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1627 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1628 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1629 * bug fix: The compiler now checks for duplicated variables in macro
1631 * bug fix: SETQ on globals returns the correct value.
1632 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1633 not a COMPILER-ERROR (followed by some other strange error on
1634 choosing the CONTINUE restart).
1635 * bug fix: make.sh and friends are now more consistent in the way that
1636 they look for GNU "make".
1638 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1639 * experimental native threads support (on x86 Linux >=2.4 only).
1640 This is not compiled in by default: you need to add :SB-THREAD to
1641 the target features. See the "Beyond ANSI" chapter of the manual
1643 * fix for longstanding nonANSIism: The old distinction between
1644 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1645 The return value from CL:FIND-CLASS is now a CLOS class, and
1646 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1647 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1648 * An interface to the MetaObject Protocol, as described in Kiczales,
1649 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1650 MIT Press, 1991, is available from the SB-MOP package.
1651 * incompatible change: the SB-PCL package should now be considered
1652 a private implementation detail, and no longer a semi-private MOP
1654 * minor incompatible change: due to rearrangement for threads, the
1655 control stack and binding stack are now allocated at arbitrary
1656 addresses instead of being hardcoded per-port. Users affected by
1657 this probably have to be doing advanced things with shared
1658 libraries, and will know who they are.
1659 * minor incompatible change: Previously, all --eval forms used were
1660 processed with READ before any of them were processed with EVAL.
1661 Now each --eval form is processed with both READ and EVAL before
1662 the next --eval form is processed. (Thus package operations like
1663 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1664 work as the user might reasonably expect.)
1665 * minor incompatible change: *STANDARD-INPUT* is now only an
1666 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1668 * minor incompatible change: Y-OR-N-P is now character-oriented,
1669 not line oriented. Also, YES-OR-NO-P now works without errors.
1670 (thanks to Antonio Martinez)
1671 * sb-aclrepl module improvements: an integrated inspector, added
1672 repl features, and a bug fix to :trace command.
1673 * Known functions, which cannot be open coded by the backend, are
1674 considered to be able to check types of their arguments. (fixing
1675 a bug report by Nathan J. Froyd)
1676 * fixed a bug in computing method discriminating functions: It is
1677 now possible to define methods specialized on classes which have
1678 forward-referenced superclasses. (thanks to Gerd Moellmann)
1679 * fixed evaluation order in optional entries (reported by Gilbert
1681 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1682 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1683 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1684 required-by-AMOP NAME argument, as well as CLASS and
1685 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1686 * fixed bug 20: DEFMETHOD can define methods using names that are
1687 not the proper names of classes to designate class specializers.
1688 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1689 Unix file descriptors, instead of blowing up. (thanks to Antonio
1691 * Garbage collection refactoring: user-visible change is that a
1692 call to the GC function during WITHOUT-GCING will not do garbage
1693 collection until the end of the WITHOUT-GCING. If you were doing
1694 this you were probably losing anyway.
1695 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1696 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1697 TYPEP the latter but not the former.
1698 * The compiler issues a full WARNING for calls to undefined functions
1699 with names from the CL package.
1700 * MAP-INTO for a vector destination is open coded. (reported by
1701 Brian Downing on c.l.l)
1702 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1703 documentation string.
1704 * fixed some bugs revealed by Paul Dietz' test suite:
1705 ** COPY-ALIST now signals an error if its argument is a dotted
1707 ** Condition slots are now accessed more correctly in the presence
1708 of multiple initargs for a given slot.
1709 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1710 exclude restarts of the same name associated with a different
1712 ** DEFCLASS of forward-referenced classes with another
1713 forward-referenced class in the superclasses list no longer
1715 ** Condition slots are now initialized once each, not multiple
1716 times. (thanks to Gerd Moellmann)
1717 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1719 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1720 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1721 arguments contain duplicated elements.
1722 ** RESTART-CASE understands local macros.
1723 ** RESTART-CASE associates exactly its own restarts with a condition.
1724 ** ENDP in safe mode checks its argument to be of type LIST.
1725 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1726 in question is unbound.
1727 ** Optimized MAKE-INSTANCE functions no longer cause internal
1728 assertion failures in the presence of duplicate initargs.
1729 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1730 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1731 propagated even in the ctor optimized implementation of
1733 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1735 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1736 * a better implementation of SXHASH on (simple) bit vectors,
1737 measured both in execution speed and in distribution of results
1738 over the positive fixnums, has been installed. Likewise, a better
1739 implementation of EQUAL for simple bit vectors is now available.
1740 * fixed CEILING optimization for a divisor of form 2^k.
1741 * fixed bug 240 (emitting extra style warnings "using the lexical
1742 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1743 by Antonio Martinez)
1744 * fixed SXHASH, giving different results for NIL depending on type
1745 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1746 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1747 variable. (found by Rolf Wester)
1748 * SBCL does not ignore type declarations for special
1749 variables. (reported by rif on c.l.l 2003-03-05)
1750 * some bug fixes in contrib/sb-aclrepl/
1751 * fixed some bugs revealed by Paul Dietz' test suite:
1752 ** a bug in the CONS type specifier, whereby the CAR and CDR
1753 types got intertwined, has been fixed;
1754 ** the type system is now able to reason about the interaction
1755 between INTEGER and RATIO types more completely;
1756 ** APPEND, [N]REVERSE and NRECONC check that those their
1757 arguments, which must be proper lists, are really so;
1758 ** An array specialized to be unable to hold elements has been
1759 implemented, as required -- yes, really -- by ANSI;
1760 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1761 on malformed property lists;
1763 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1764 * incompatible packaging change: in line with Unix convention,
1765 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1766 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1767 * REQUIRE and PROVIDE are now optionally capable of doing something
1768 useful. See the documentation string for REQUIRE.
1769 * infrastructure for a managed SBCL contrib system: contributed
1770 modules in this release include:
1771 ** the ASDF system definition facility;
1772 ** an interface to the BSD Sockets API;
1773 ** an ACL-like convenience interface to the repl;
1774 (thanks to Kevin Rosenberg)
1775 ** an implementation of ROTATE-BYTE, with efficient implementation
1777 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1778 gives a closer approximation to the right answer than previously.
1779 (thanks to Raymond Toy)
1780 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1781 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1782 optional environment arguments, as required by ANSI.
1783 * fixed bugs in other functions taking environment objects, allowing
1784 calls with an explicit NIL environment argument to be compiled
1786 * fixed bug 228: primary return values from
1787 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1788 COMPILE or FUNCTION.
1789 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1790 :TYPE will no longer signal errors on innocuous objects.
1791 * fixed bug 231b: SETQ is better at respecting type declarations in
1792 the lexical environment.
1793 * fixed a bug in DEFCLASS: classes named by symbols with no or
1794 unprintable packages can now be defined.
1795 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1796 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1797 * fixed compiler failure related to checking types of functions.
1798 (reported by Robert E. Brown)
1799 * the compiler is now much more consistent in its error-checking
1800 treatment of bounding index arguments to sequence functions: in
1801 (SAFETY 3) code, errors will be signalled in almost all cases if
1802 invalid sequence bounding indices are passed to functions defined
1803 by ANSI to operate on sequences.
1804 * fixed a bug in the build procedure: documentation of SBCL-specific
1805 packages is now preserved and available in the final Lisp image.
1806 * lifted FDEFINITION lookup out of loops in the implementation of
1807 many list operations. (thanks to Robert E. Brown)
1808 * fixed a bug in the reader: the #n# reader macro now works for
1809 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1810 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1811 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1812 E. Brown and Tony Martinez respectively)
1813 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1814 SLOT-BOUNDP: in particular, it is now able to optimize them much
1815 better, and is now not vulnerable to having packages renamed.
1816 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1817 always called when a slot is not present in an instance. (thanks
1819 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1820 longer signal an error on unreading a character following EOF on
1821 the previous constituent stream. (thanks to Tony Martinez)
1822 * fixed some bugs revealed by Paul Dietz' test suite:
1823 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1824 not just nonnegative fixnums;
1825 ** the logical bit-array operators such as BIT-AND now accept an
1826 explicit NIL for their "opt-arg" argument (to indicate a
1827 freshly-consed result bit-array);
1828 ** ELT now signals an error on an invalid sequence index in safe
1830 ** the type system is now cleverer about negations of numeric
1831 types, and consequently understands the BIGNUM and RATIO types
1833 ** the type system is now cleverer about the interaction between
1834 INTEGER and RATIO types: while bugs still remain, many more
1835 cases are accurately computed;
1836 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1837 if it is in the last clause;
1838 ** CONSTANTLY now correctly returns a side-effect-free function in
1840 ** DECLARE is no longer treated as a special-operator; in
1841 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1842 * incremented fasl file version number due to the change in the
1843 DEFSTRUCT-SLOT-DESCRIPTION structure.
1845 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1846 * minor incompatible change: code processed by the "interpreter" or
1847 EVAL now has a compilation optimization policy of (DEBUG 2)
1848 (changed from (DEBUG 1)) to improve debuggability of interactive
1849 development, and to allow the use of the debug RETURN command in
1851 * an experimental implementation of the RETURN command for the
1852 debugger has been included. (thanks to Frederik Kuivinen)
1853 * fixed bug 62: constraints were not propagated into a loop.
1854 * fixed bug in embedded calls of SORT (reported and investigated by
1856 * fixed some bugs revealed by Paul Dietz' test suite:
1857 ** printing and reading of arrays with some dimensions having
1858 length 0 (thanks to Gerd Moellmann);
1859 ** BOA constructor with &AUX argument without a default value does
1860 not cause a type error;
1861 ** CONSTANTP now returns true for all self-evaluating objects.
1863 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1864 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1865 accessors that are related by inheritance, as specified in the
1866 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1867 Valtteri Vuorikoski)
1868 * The compiler is now able to inline functions that were defined in
1869 a complex lexical environment (e.g. inside a MACROLET).
1870 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1872 * Support for the upcoming FreeBSD-5.0 release has been included.
1873 (thanks to Dag-Erling Smorgrav)
1874 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1875 effect when it is not in a toplevel context.
1876 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1877 stand a better chance of being correct. (thanks to Gerd
1879 * fixed bug in COERCE, which now signals an error on coercing a
1880 rational to a bounded real type which excludes the expected
1882 * The compiler is now able to derive types more accurately from the
1883 COERCE and COMPILE functions.
1884 * fixed bug 223: functional binding is considered to be constant
1885 only for symbols in the CL package.
1886 * fixed bug 231: SETQ did not check the type of a variable being set
1887 (reported by Robert E. Brown)
1888 * A new optimization for MAKE-INSTANCE has been included, fixing
1889 various bugs (including relating to :ALLOCATION :CLASS slots and
1890 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1892 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1893 ** As required by ANSI, LOOP now disallows anonymous collection
1894 clauses such as COLLECT I in conjunction with aggregate boolean
1895 clauses such as THEREIS (= I 1);
1896 ** LOOP now signals an error when any variable is reused in the
1897 same loop (including the potentially useful construct analogous
1898 to WITH A = 1 WITH A = (1+ A);
1899 ** IT is only a special loop symbol within the first clause of a
1900 conditional loop clause;
1901 ** LOOP with a typed iteration variable over a hashtable now
1902 signals a type error iff it should.
1903 * fixed some other bugs revealed by Paul Dietz' test suite:
1904 ** FILE-STREAM now names the class previously known as FD-STREAM;
1905 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1906 argument) no longer signals an error;
1907 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1908 of the slot symbol, rather than using the current package
1909 ((:CONC-NAME "") continues to intern the slot's name in the
1911 * incremented fasl file version number, because of the incompatible
1912 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1913 of the new implementation of DEFINE-COMPILER-MACRO.
1915 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1916 * Support for building SBCL for MIPS platforms running in
1917 little-endian mode has now been checked in, and basic
1918 functionality on said platforms verified.
1919 * minor incompatible change: PCL now records the pathname of a file
1920 in which methods and the like are defined, rather than its
1922 * minor incompatible change: TRUENAME now considers the truename of
1923 a file naming a directory to be the pathname with :DIRECTORY
1924 component indicating that directory.
1925 * minor incompatible change: a NAMED clause in the extended form of
1926 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1927 reason for the previous behaviour is unclear.
1928 * more systematization and improvement of CLOS and MOP conformance
1929 in PCL (thanks to Gerd Moellman and Pierre Mai):
1930 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1932 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1933 primary methods with no specializers;
1934 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1936 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1937 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1938 FORWARD-REFERENCED-CLASSes; error reporting on
1939 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1941 ** SXHASH on CLOS instances now uses a slot internal to the
1942 instance to return different numbers on distinct instances,
1943 while preserving the same return value through invocations of
1945 ** DEFMETHOD signals errors when methods with longer incongruent
1946 lambda lists are added to generic functions;
1947 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1948 CLASS, as specified in AMOP;
1949 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1950 on the order returned by the primary method for classes of
1951 class STANDARD-CLASS;
1952 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1953 * fixed some bugs shown by Paul Dietz' test suite:
1954 ** DOLIST puts its body in TAGBODY;
1955 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1957 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1958 value producing form;
1959 ** if more variables are given to PROGV than values, extra
1960 variables are bound and made to have no value;
1961 ** NSUBSTITUTE on list arguments gets the right answer with
1963 ** ELT signals an error of type TYPE-ERROR when the index argument
1964 is not a valid sequence index;
1965 ** LOOP signals (at macroexpansion time) an error of type
1966 PROGRAM-ERROR when duplicate variable names are found;
1967 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1968 ** FUNCALL of special-operators now cause an error of type
1970 ** PSETQ now works as required in the presence of side-effecting
1971 symbol-macro places;
1972 ** NCONC accepts any object as its last argument;
1973 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1975 ** loop-for-as-package does not require a package to be explicitely
1977 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1979 * fixed bug 166: compiler preserves "there is a way to go"
1980 invariant when deleting code.
1981 * fixed bug 172: macro lambda lists with required arguments after
1982 &REST arguments now cause an error to be signalled. (thanks to
1984 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1985 bug. (thanks to Matthew Danish)
1986 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1988 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1989 arguments are lexically rebound. (thanks to Gerd Moellmann and
1991 * fixed bug 194: error messages are now more informative when there
1992 is no primary method applicable in a call to a generic
1993 function. (thanks to Gerd Moellmann)
1994 * fixed bug in command line argument checking (thanks to Julian
1996 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1997 (thanks to Matthew Danish)
1998 * incremented fasl file version number, because of the
1999 SXHASH-related changes in the layout of CLOS data structures
2001 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2002 * minor incompatible change: The runtime (the Unix executable named
2003 "sbcl") is now much pickier about the .core files it will load.
2004 Essentially it now requires .core files to descend from the same
2005 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2006 as the runtime does. (The intent is to prevent the crashes which
2007 can occur, and which can even be reported as mysterious failures,
2008 when people patch the sources or change the build parameters
2009 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2010 sbcl and .core files.)
2011 * fixed bug: VALUES-LIST is no longer optimized away.
2012 * fixed bug 142: The FFI conversion of C string values to Lisp
2013 string values no longer conses excessively. (thanks to Nathan
2014 Froyd porting Raymond Toy's fix to CMU CL)
2015 * began to systematize and improve MOP conformance in PCL (thanks to
2016 Nathan Froyd, Gerd Moellman and Pierre Mai):
2017 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2019 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2020 ** FINALIZE-INHERITANCE is now called on class finalization;
2021 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2022 argument precedence order.
2023 * fixed bug 202: The compiler no longer fails on functions whose
2024 derived types contradict their declared type.
2025 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2026 so it can be non-toplevel.
2027 * The fasl file version number has changed (because of the new
2028 implementation of DEFMACRO).
2029 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2030 safe code, that any length requirement by their type-specifier
2031 argument is valid. The exceptions to this are described in bug
2033 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2034 functions now check the types of their inputs as required by ANSI.
2035 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2036 introduced symbol is DECLAREd to be SPECIAL.
2037 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2038 * fixed bug: :COUNT argument to sequence functions may be negative
2039 * fixed bug: body of DO-SYMBOLS may contain declarations
2040 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2041 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2042 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2043 symbol macro only once
2044 * fixed printing of call frame when argument list is unavailable
2045 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2046 * compiler no longer signals WARNING on unknown keyword
2049 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2050 * A beta-quality port to the mips architecture running Linux,
2051 based on the old CMUCL backend, has been made. It has been tested
2052 on a big-endian kernel, and works sufficiently well to be able to
2053 rebuild itself; it has not been tested in little-endian mode.
2054 * fixed an inconsistency between gencgc.c and purify.c which made
2055 dumping/loading .core files unreliable
2056 * fixed bug 120a: The compiler now deals correctly with IFs where
2057 the consequent is the same as the alternative, instead of
2058 misderiving the return type. (thanks to Alexey Dejneka)
2059 * fixed bug 113: Logical pathnames are now dumpable (the logical
2060 host is resolved at load-time, throwing an error if it is not
2062 * fixed bug 174: FORMAT's error message is slightly clearer when a
2063 non-printing character is used in a format directive.
2064 * fixed several bugs in compiler checking of type declarations, i.e.
2065 violations of the Python "declarations are assertions" principle
2066 (thanks to Alexey Dejneka)
2067 * fixed several bugs in PCL's error checking (thanks to Gerd
2069 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2071 * fixed bug in compilation of functions as first class values
2072 (thanks to Antonio Martinez-Shotton)
2073 * The compiler's handling TYPE-ERRORs which it can prove will
2074 inevitably happen at runtime has been cleaned up and corrected
2075 in several ways. (thanks to Alexey Dejneka)
2076 * improved argument type checking for various basic arithmetic
2077 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2078 lavished on them in the past that they can be compiled in many
2079 ways in different special cases
2080 * fixed bug 181: compiler checks validity of user supplied type
2082 * cleaned up code flushing in optimization: Function calls which
2083 should signal errors for safety purposes (e.g. which ANSI says
2084 should signal errors when their arguments are of incorrect type)
2085 are no longer optimized away.
2086 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2087 * incremented fasl file version number, because changes in the
2088 implementation of sequence functions like COERCE caused
2089 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2090 in old inline expansions) to become undefined. (Actually these
2091 changes were later undone, so we might very well be binary
2092 compatible with 0.7.7 after all, but leaving the version number
2093 incremented seemed like the simplest and most conservative
2096 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2097 * An alpha-quality port to the parisc architecture running Linux,
2098 based on the old CMUCL backend, has been made. This, even more so
2099 than the other backends, should be considered still a work in
2100 progress; known problems include that the Linux kernel in 64-bit
2101 mode does not propagate the correct sigcontext structure to
2102 userspace, and consequently SBCL on a parisc64 kernel will not
2104 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2105 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2106 suppressed.) Also now that the compiler is looking at declarations
2107 in the environment, it checks optimization declarations as well,
2108 and suppresses inlining when (> DEBUG SPEED).
2109 * More fixes have been made to treatment of floating point exception
2110 treatment and other Unix signals. In particular, floating point
2111 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2112 * The detection and handling of control stack exhaustion (infinite
2113 or very deeply nested recursion) has changed. Stack exhaustion
2114 detection is now done by write-protecting pages at the OS level
2115 and applies at all optimization settings; when found, a
2116 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2117 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2118 be caught using IGNORE-ERRORS.
2119 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2120 correctly when transforming recursive LABELS functions to LETs.
2121 (thanks to Alexey Dejneka)
2122 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2123 that are names of constants or global variables.
2124 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2125 alien routines with docstrings.
2126 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2127 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2129 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2130 (thanks to Eric Marsden)
2131 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2132 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2133 * Bug fix: The SPARC backend can now compile functions involving
2134 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2135 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2136 to a functions expecting a PATHNAME-DESIGNATOR.
2137 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2138 lambda lists. (thanks to Alexey Dejneka)
2139 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2140 functions so that EXPORTing the name of the class doesn't cause
2141 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2142 to Antonio Martinez for reporting this)
2143 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2144 constant-folded without failing an assertion. (thanks to Einar
2145 Floystad Dorum for reporting this)
2146 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2147 types are now dealt with more correctly by the compiler.
2148 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2149 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2150 behaviour analogous to RENAME-FILE. This puts its behaviour more
2151 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2153 * The fasl file version number has changed again. (because of the
2154 bug fix involving the names of PCL MAKE-INSTANCE functions)
2156 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2157 * bug fix: Floating point exceptions are treated much more
2158 consistently on the x86/Linux and PPC/Linux platforms.
2159 * Array initialization with :INITIAL-ELEMENT is now much faster for
2160 cases when the compiler cannot open code the array creation, but
2161 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2162 array accesses have also seen a speed increase.
2163 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2164 specified. (This was at the root of some bad interactions between
2165 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2167 * bug fix: Internal error arguments for undefined functions are now
2168 computed correctly on the PPC/Linux platform.
2169 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2170 Raymond Toy's patch for CMU CL)
2171 * Support for the Solaris 9 operating environment has been included
2172 (thanks to Daniel Merritt)
2173 * A very ugly but hopefully complete draft of the missing FFI chapter
2174 of the manual has been created by reformatting the corresponding
2175 CMU CL manual chapter into (currently very ugly and incoherent)
2176 DocBook and bringing it up to date for SBCL behavior. Thus, the
2177 manual is now essentially complete, at least by my extreme
2178 once-and-only-once standards, whereby it's acceptable to refer to
2179 the doc strings of SB-EXT functions as the primary documentation.
2180 * The fasl file version number has changed again, due to cleanup of
2181 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2182 is no longer a static symbol.)
2184 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2185 * SBCL now builds with OpenMCL (version 0.12) as the
2186 cross-compilation host; also, more progress has been made toward
2187 bootstrapping under CLISP.
2188 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2190 * bug 158 fixed: The compiler can now deal with integer loop
2191 increments different from 1; fixing this turned out also to fix
2193 * bug 169 fixed: no more bogus warnings about using lexical bindings
2194 despite the presence of perfectly good SPECIAL declarations (thanks
2195 to David Lichteblau)
2196 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2197 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2198 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2200 * bug 180 fixed: Method combination specifications no longer ignore
2201 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2202 * bug fix: Structure type predicate functions now check their argument
2203 count as they should.
2204 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2205 correctly. (thanks to Pierre Mai)
2206 * minor incompatible change: The --noprogrammer option is deprecated
2207 in favor of the new --disable-debugger option, which is very similar.
2208 (The major difference is that it takes effect at a slightly different
2209 time at startup, causing handling of errors in --sysinit and
2210 --userinit files will be affected differently.) The
2211 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2212 been added to allow this functionality to be controlled from ordinary
2213 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2214 maintainers, who might want to run non-interactive scripts to
2215 build SBCL cores which will later be used interactively.)
2216 * minor incompatible change: The LOAD function no longer, when given
2217 a wild pathname to load, loads all files matching that pathname.
2218 Instead, an error of type FILE-ERROR is signalled.
2220 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2221 * bug 147 fixed: The compiler preserves its block link/count
2222 invariants more correctly now so that it doesn't crash. (thanks
2224 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2226 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2227 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2228 no-value-for-this-character as NIL (as specified by ANSI).
2229 * HOST-NAMESTRING on physical pathnames now returns a string that is
2230 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2231 (thanks to Christophe Rhodes)
2232 * The Alpha port handles icache flushing more correctly. (thanks to
2234 * More progress has been made toward bootstrapping under CLISP. (thanks
2235 to Christophe Rhodes)
2236 * The fasl file format has changed again, because dynamic loading
2237 on OpenBSD (which has non-ELF object files) motivated some cleanups
2238 in the way that foreign symbols are transformed and passed around.
2239 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2240 is no longer treated as whitespace by the reader, but instead as
2241 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2242 |A
\7fB|, instead of A as it used to.
2244 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2245 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2246 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2247 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2248 this as of 0.7.1.45, but was left out of the previous news section
2249 (thanks to Dan Barlow)
2250 * SBCL now runs on the Solaris operating system on SPARC architectures
2251 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2252 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2253 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2254 * SPARC backend cleanups, allowing builds of cores optimized for V8
2255 and V9 SPARCS, and also emission of code targeted to a particular
2256 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2258 * SBCL is closer to bootstrapping under CLISP, thanks to various
2259 fixes by Christophe Rhodes.
2260 * The fasl file format has changed again, to allow the compiler's
2261 INFO database to support symbol macros.
2262 * The user manual (in doc/) is formatted into HTML more nicely.
2263 (thanks to coreythomas)
2264 * The system is smarter about SUBTYPEP relationships, especially
2265 those involving NOT types (including types such as ATOM which are
2266 represented internally using NOT types). Thus SUBTYPEP is less
2267 likely to return (VALUES NIL NIL) in general, and in particular
2268 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2270 * The fasl file format has changed again, because the internal
2271 representation of types now includes a new slot to support the new
2272 SUBTYPEP-of-NOT-types logic.
2273 * (not a change in the main branch of SBCL, but a related prototype
2274 which can hopefully be merged into the main branch of SBCL in the
2275 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2276 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2277 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2278 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2279 dispatch macro character combinations. (thanks to Alexey Dejneka)
2280 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2281 that superficially look like logical namestrings correctly.
2282 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2284 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2285 default supplied-p) for &optional and &key arguments. (thanks to
2288 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2289 * incompatible change: The compiler is now less aggressive about
2290 tail call optimization, doing it only when (> SPACE DEBUG) or
2291 (> SPEED DEBUG). (This is an incompatible change because there are
2292 programs which relied on the old CMU-CL-style behavior to optimize
2293 away their unbounded recursion which will now die of stack overflow.)
2294 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2295 for non-GENCGC systems has been increased to 20M (since that
2296 seems much closer to the likely performance optimum for modern
2297 systems than the old 4M value was)
2298 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2299 and *DEBUG-PRINT-LEVEL*
2300 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2301 of CMU CL's support for SPARC, and various endianness and other
2302 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2303 * new syntactic sugar for the Unix command line: --load foo.bar is now
2304 an alternate notation for --eval '(load "foo.bar")'.
2306 ** The system now detects stack overflow and handles it gracefully,
2307 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2308 optimization settings. (This is a good thing in general, and
2309 its introduction in this version should be particularly timely
2310 for anyone whose code fails because of suppression of tail
2312 ** The system now hunts for the C variable "environ" in a more
2313 devious way, to avoid segfaults when the C library version differs
2314 between compile time and run time. (thanks to Christophe Rhodes)
2315 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2316 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2317 ** The compiler no longer issues bogus style warnings for undefined
2318 classes in the same source file as the DEFCLASSes which defined
2319 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2321 ** fixes in CONDITION class precedence list for undefined function
2322 errors (thanks to Alexei Dejneka)
2323 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2324 correctly. (thanks to Dan Barlow)
2325 ** portability fixes aiming at bootstrapping under CLISP (thanks
2326 to Dave McDonald and Christophe Rhodes)
2327 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2328 ** fixes in type translation and and type inference (thanks to
2330 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2331 ** various fixes in the new ports (thanks to Dan Barlow)
2332 * several changes related to debugging:
2333 ** suppression of tail recursion, as noted above
2334 ** stack overflow detection, as noted above
2335 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2336 is now the default. (For some time encapsulation has been more
2337 reliable than the breakpoint-based :ENCAPSULATE NIL
2338 implementation, at least on X86 systems; and I just noticed that
2339 encapsulation also seems closer to the spirit of the ANSI
2342 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2344 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2345 up properly again. (There was a packaging bug in 0.7.0 which
2346 left their definitions in SB-SYS::LOAD-FOREIGN and
2347 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2348 vital for most things which interface to C-level interfaces,
2349 like extensions working with sockets or databases or
2350 Perl-compatible regexes or whatever, and the need to fix
2351 this bug is the main reason that 0.7.1 was released so
2353 ** DEFGENERIC is now choosier about the methods it redefines, so that
2354 reLOADing a previously-LOADed file containing DEFGENERICs does
2355 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2356 cycle now works as it should. (thanks to Alexey Dejneka)
2357 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2358 fixed by Christophe Rhodes. (He actually submitted this patch
2359 months ago, and I delayed until after 0.7.0.)
2360 ** Bug 111 (internal compiler confusion about runtime checks on
2361 FUNCTION types) was fixed by Alexey Dejneka.
2362 * Some internal cleanups (getting rid of variables which aren't
2363 needed now that the byte interpreter is gone) caused the fasl
2364 file format number to change again.
2366 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2367 * major incompatible change: The default fasl file extension, i.e. the
2368 default extension for files produced by COMPILE-FILE, has changed
2369 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2371 ** There are many changes in the implementation of the compiler.
2372 SBCL is now essentially a compiler-only implementation of ANSI
2373 Common Lisp. EVAL still "interprets" a few special cases, but
2374 almost all the interesting cases are handled by creating
2375 a LAMBDA expression, calling COMPILE on it, then calling
2376 FUNCALL on the result.
2377 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2378 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2379 Since the code is newer, there might still be some new bugs
2380 (though not as many as before Martin Atzmueller's fixes:-). But
2381 the new code is substantially simpler and clearer, and hopefully
2382 any remaining bugs will be simpler, less fundamental, and more
2383 fixable then the bugs in the old code.
2384 ** The revised compiler is still a little unsteady on its feet.
2386 *** The debugging information it produces (particularly the names
2387 of FUNCTION objects) is sometimes much less useful than what
2388 the old compiler produced.
2389 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2390 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2391 within a MACROLET) has been temporarily weakened.
2392 ** There are new compiler optimizations for various functions:
2393 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2394 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2395 *** the math functions TRUNCATE, FLOOR, and CEILING
2396 *** the function-of-all-trades COERCE
2397 Mostly these should be transparent, but there's one
2398 potentially-annoying problem (bug 117): when the compiler
2399 inline-expands a function and does type analysis on the result,
2400 it can create control paths which have type mismatches, and
2401 when it can't prove that those control paths aren't taken,
2402 it will issue WARNINGs about the type mismatches. This is
2403 a particular problem in practice for the new sequence functions.
2404 It's not clear how this should be fixed, and for now, a
2405 workaround is given in the entry for 117 in the BUGS file.
2406 ** (Because of the interaction between the two previous items --
2407 occasional inlining problems and new inline expansions -- some
2408 of the new sequence function optimizations won't really kick in
2409 completely until debugging information, and then inlining, are
2410 straightened out in some future version.)
2411 * minor incompatible changes:
2412 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2413 DIRECTORY no longer implicitly promotes NIL slots of its
2414 pathname argument to :WILD. In particular, when you ask for the
2415 contents of a directory (which you used to be able to do without
2416 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2417 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2418 ** changes in behavior that ANSI explicitly defines to be
2419 implementation dependent:
2420 *** The new compiler-only implementation still conforms with ANSI,
2421 but acts a little different than before. Besides the obvious
2422 changes in performance tradeoffs (that the cost per form passed
2423 to EVAL has gone up, and the cost per form executed by EVAL
2424 has gone down), the behavior of the system changes a little
2425 because there are no longer any interpreted function objects.
2426 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2427 e.g. doing COMPILE on the output of interactive DEFUN is
2429 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2431 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2432 has changed from (:CL) to NIL.
2433 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2434 e.g. #\Newline and #\Tab, don't have names specified in the
2435 ANSI Common Lisp standard, is now based on their ASCII symbolic
2436 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2437 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2438 are no longer used for output.
2439 ** changes in internal implementation constants:
2440 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2441 4 million. (If your application spends a lot of time GCing and
2442 you have a lot of RAM, you might want to experiment with
2443 increasing it even more.)
2444 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2445 However, almost all old code should still continue to work without
2446 immediate update, as SB-C-CALL is now a (deprecated) nickname
2448 ** Old operator names in the style DEF-FOO are now deprecated in
2449 favor of new corresponding names DEFINE-FOO, for consistency with
2450 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2451 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2452 internal symbols, but a few supported extensions like
2453 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2454 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2455 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2456 etc. as you get deeper into recursive calls to the debugger
2457 command loop, instead of the old "5]", "5]]", "5]]]"
2458 sequence. (I was motivated to do this when squabbles between
2459 ILISP and SBCL left me very deeply nested in the debugger. In the
2460 short term, this change will probably provoke more ILISP/SBCL
2461 squabbles, but hopefully it will be an improvement in the long run.)
2462 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2463 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2464 function, since declaiming return types involving aliens is
2465 (1) annoyingly messy to do by hand and (2) vital to efficient
2466 compilation of code which calls such functions.
2467 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2468 longer reexported by the SB-EXT package. They're solely useful
2469 for alien code, so it seems more logical that you should get
2470 them from the SB-ALIEN package, not in SB-EXT.
2471 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2472 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2473 features. Instead, the code that they used to control is always
2474 built into the system.
2475 * many other bug fixes
2476 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2477 advantage of the new EVAL-WHEN stuff and to clean them up in
2478 general, and they are now more ANSI-compliant in a number of
2479 ways. Martin Atzmueller is responsible for a lot of this.
2480 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2482 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2483 *** correct ERROR type for various file operations
2484 *** some fixes for Lisp streams
2485 *** DEFMETHOD syntax checking
2486 *** changing old weird representation of debug information as
2487 strings (which, among their other deficiencies, don't transform
2488 correctly when you rename packages, and don't change their
2489 print representation when you change things like *PACKAGE*
2490 and *PRINT-LENGTH*) to symbols and lists of symbols
2491 He also made several improvements and fixed several bugs in DESCRIBE.
2492 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2494 *** misbehavior of WRITE-STRING/WRITE-LINE
2495 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2496 and several other LOOP problems as well
2497 *** DIRECTORY when similar filenames are present
2498 *** DEFGENERIC with :METHOD options
2499 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2500 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2501 *** argument ordering in FIND with :TEST option
2502 *** mishandled package designator argument in APROPOS-LIST
2503 *** various problems in the backquote readmacro
2504 *** a bug in APROPOS
2505 *** probably some others that I'm not describing very well here,
2506 since the CVS log documents them by reference to sbcl-devel
2507 messages, and the SourceForge archives aren't working well.:-(
2508 ** Dan Barlow improved the Alpha port (and is making progress on the
2509 PPC port, for those of you who think different).
2510 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2511 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2512 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2513 continues to work on the SPARC port (for those of you in a position
2514 to look down upon our little PC-compatible boxes from a great height).
2515 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2516 and uses that copy, rather than the current dynamic value, when
2517 it's trying to decide whether to truncate output. Thus e.g.
2518 (let ((*print-lines* 50))
2519 (pprint-logical-block (stream nil)
2521 (let ((*print-lines* 8))
2522 (print (aref possiblybigthings i) stream)))))
2523 should now truncate the logical block only at 50 lines, instead of
2524 often truncating it at 8 lines, as it did before.
2525 * The doc/cmucl/ directory, containing old CMU CL documentation
2526 from the time of the fork, is no longer part of the base system.
2527 SourceForge has shut down its anonymous FTP service, and with it
2528 my original plan for distributing the old CMU CL documentation
2529 there. For now, if you need these files you can download an old
2530 SBCL source release and extract them from it.
2531 * The fasl file version number changed again, for dozens of reasons,
2532 some of which are apparent above.
2534 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2535 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2536 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2537 patch, so that DISASSEMBLE output is much nicer.
2538 * The code in the SB-PROFILE package now seems reasonably stable.
2539 I still haven't decided what the final interface should look like
2540 (I'd like PROFILE to interact cleanly with TRACE, since both
2541 facilities use function encapsulation) but if you have a need
2542 for profiling now, you can probably use it successfully with
2543 the current CMU-CL-style interface.
2544 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2545 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2546 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2547 link now returns the dangling link itself, and for similar
2548 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2549 link itself. (In these cases the old code signalled an error and
2550 looped endlessly, respectively.) Thus, DIRECTORY now works even
2551 in the presence of dangling and cyclic symbolic links.
2552 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2553 is now a supported extension again, since the consensus on
2554 sbcl-devel was that it can be useful for ordinary development
2555 work, not just for debugging SBCL itself.
2556 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2557 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2558 that currently-defined functions might be redefined later with
2559 different return types.
2560 * Hash tables can be printed readably, as inspired by CMU CL code
2561 of Eric Marsden and SBCL code of Martin Atzmueller.
2562 * better error handling in CLOS method combination, thanks to
2563 Martin Atzmueller porting Pierre Mai's CMU CL patches
2564 * more overflow fixes for >16Mbyte I/O buffers
2565 * A bug in READ has been fixed, so that now a single Ctrl-D
2566 character suffices to cause end-of-file on character streams.
2567 In particular, now you only need one Ctrl-D at the command
2568 line (not two) to exit SBCL.
2569 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2571 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2572 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2573 does the right thing.
2574 * The implementation of some type tests, especially for CONDITION
2575 types, is now tidier and maybe faster, due to CMU CL code
2576 originally by Douglas Crosher, ported by Martin Atzmueller.
2577 * Some math functions have been fixed, and there are new
2578 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2579 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2580 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2581 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2582 likely to be built into sbcl-0.7.0.)
2583 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2584 CL, ported to SBCL by Martin Atzmueller)
2585 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2586 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2587 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2588 * There's a new slam.sh hack to shorten the edit/compile/debug
2589 cycle for low-level changes to SBCL itself, and a new
2590 :SB-AFTER-XC-CORE target feature to control the generation of
2591 the after-xc.core file needed by slam.sh.
2592 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2593 COMPILE-FILE is no longer supported, so that now every function
2594 gets an entry point, so that block compilation looks a little
2595 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2596 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2597 returns the number of bytes consed since the system started,
2598 rather than the number consed since the first time the function
2599 was called. (The new definition parallels ANSI functions like
2600 CL:GET-INTERNAL-RUN-TIME.)
2601 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2602 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2603 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2604 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2605 * Fasl file version numbers are now independent of the target CPU,
2606 since historically most system changes which required version
2607 number changes have affected all CPUs equally. Similarly,
2608 the byte fasl file version is now equal to the ordinary
2611 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2612 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2613 is no longer recognized. I apologize for this, because it was
2614 listed in SB-EXT as a supported extension, but I found that
2615 its existing behavior was poorly specified, as well as incorrectly
2616 specified, and it looked like too much of a mess to straighten it
2617 out. I have enough on my hands trying to get ANSI stuff to work..
2618 * many patches ported from CMU CL by Martin Atzmueller, with
2619 half a dozen bug fixes in pretty-printing and the debugger, and
2620 half a dozen others elsewhere
2621 * fixed bug 13: Floating point infinities are now supported again.
2622 They might still be a little bit flaky, but thanks to bug reports
2623 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2624 as flaky as they were.
2625 * The --noprogrammer command line option is now supported. (Its
2626 behavior is slightly different in detail from what the old man
2627 page claimed it would do, but it's still appropriate under the
2628 same circumstances that the man page talks about.)
2629 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2630 are now supported, and enabled by default. Thus, the compiler can
2631 handle many floating point and complex operations much less
2632 inefficiently. (Thus e.g. you can implement a complex FFT
2634 * The compiler now detects type mismatches between DECLAIM FTYPE
2635 and DEFUN better, and implements CHECK-TYPE more correctly, and
2636 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2637 thanks to patches from Martin Atzmueller.
2638 * various fixes to make the cross-compiler more portable to
2639 ANSI-conforming-but-different cross-compilation hosts (notably
2640 Lispworks for Windows, following bug reports from Arthur Lemmens)
2641 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2642 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2643 fixed thanks to Pierre Mai's CMU CL patches.
2644 * fixes to keep the system from overflowing internal counters when
2645 it tries to use i/o buffers larger than 16M bytes
2646 * fixed bug 45a: Various internal functions required to support
2647 complex special functions have been merged from CMU CL sources.
2648 (When I was first setting up SBCL, I misunderstood a compile-time
2649 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2650 * improved support for type intersection and union, fixing bug 12
2651 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2652 more obscure bugs as well
2653 * some steps toward byte-compiling non-performance-critical
2654 parts of the system, courtesy of patches from Martin Atzmueller
2655 * Christophe Rhodes has made some debian packages of sbcl at
2656 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2657 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2658 stable, but are nonetheless usable. When he's ready, I'd be happy
2659 to add them to the SourceForge "File Releases" section. (And if
2660 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2661 * new fasl file format version number (because of changes in
2662 internal representation of (OR ..) types to accommodate the new
2663 support for (AND ..) types, among other things)
2665 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2666 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2668 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2670 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2671 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2672 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2673 * DESCRIBE now works on structure objects again.
2674 * Most function call argument type mismatches are now handled as
2675 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2676 know whether the function will be redefined before the call is
2677 executed. (The compiler could flag local calls with full WARNINGs,
2678 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2679 it doesn't keep track of enough information to know whether calls
2680 are local in this sense.)
2681 * Compiler output is now more verbose, with messages truncated
2682 later than before. (There should be some supported way for users
2683 to override the default verbosity, but I haven't decided how to
2684 provide it yet, so this behavior is still controlled by the internal
2685 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2686 src/compiler/ir1util.lisp.)
2687 * Fasl file format version numbers have increased again, because
2688 support for the Gray streams extension changes the layout of the
2689 system's STREAM objects.
2690 * The Gray subclassable streams extension now works, thanks to a
2691 patch from Martin Atzmueller.
2692 * The full LOAD-FOREIGN extension (not just the primitive
2693 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2694 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2695 but like most other programs, it defaults to copying the Unix
2696 environment from the original process instead of starting the
2697 new process in an empty environment.
2698 * Extensions which manipulate the Unix environment now support
2699 an :ENVIRONMENT keyword option which doesn't smash case or
2700 do other bad things. The CMU-CL-style :ENV option is retained
2701 for porting convenience.
2702 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2703 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2705 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2707 * A patch from Martin Atzmueller seems to have solved the SIGINT
2708 problem, and as far as we know, signal-handling now works cleanly.
2709 (If you find any new bugs, please report them!)
2710 * The system no longer defaults Lisp source file names to types
2711 ".l", ".cl", or ".lsp", but only to ".lisp".
2712 * The compiler no longer uses special default file extensions for
2713 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2714 seems to expect a single default extension for all compiled code,
2715 and there's no compelling reason to try to stretch the standard
2716 to allow two different extensions.) Instead, byte-compiled files
2717 default to the same extension as native-compiled files.
2718 * Fasl file format version numbers have increased again, because
2719 a rearrangement of internal implementation packages made some
2720 dumped symbols in old fasl files unreadable in new cores.
2721 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2722 many fewer weird special cases.
2723 * Bug #17 (differing COMPILE-FILE behavior between logical and
2724 physical pathnames) has been fixed, and some related misbehavior too,
2725 thanks to a patch from Martin Atzmueller.
2726 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2727 by Tim Moore, ported to SBCL by Martin Atzmueller.
2728 * Martin Atzmueller fixed several filesystem-related problems,
2729 including bug #36, in part by porting CMU CL patches, which were
2730 written in part by Paul Werkowski.
2731 * More compiler warnings in src/runtime/ are gone, thanks to
2732 more patches from Martin Atzmueller.
2733 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2736 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2738 * DESCRIBE now works on CONDITION objects.
2739 * The debugger now handles errors which arise when trying to print
2740 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2742 * The build system now uses an additional file, customize-target-features.lisp,
2743 to allow local modifications to the target *FEATURES* list. (The point of
2744 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2745 debugging features enabled, without having to worry about propagating it
2746 into everyone's system when I do a "cvs update".) When no
2747 customize-target-features.lisp file exists, the target *FEATURES* list
2748 should be constructed the same way as before.
2749 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2750 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2751 warn about behavior which is undefined under the ANSI spec. Note: This
2752 is specified by ANSI, but it's not very popular with programmers.
2753 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2754 macro in the SBCL sources for an example of a workaround which you
2755 might use to make portable ANSI-standard code which does what you want.
2756 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2757 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2758 when it's used as a non-toplevel form. (This is required in order
2759 to implement the DEFCONSTANT-EQX macro.)
2760 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2761 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2762 funny things, due to bugs in the implementation of EVAL-WHEN
2763 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2764 non-toplevel DEFCONSTANTs are uncommon.)
2765 * The core file version number and fasl file version number have been
2766 incremented, because the old noncompliant DEFCONSTANT behavior involved
2767 calling functions which no longer exist, and because I also took the
2768 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2769 * fixed bug 1 (error handling before read-eval-print loop starts), and
2770 redid debugger restarts and related debugger commands somewhat while
2772 ** The QUIT debugger command is gone, since it did something
2773 rather different than the SB-EXT:QUIT command, and since it never
2774 worked properly outside the main toplevel read/eval/print loop.
2775 Invoking the new TOPLEVEL restart provides the same functionality.
2776 ** The GO debugger command is also gone, since you can just invoke
2777 the CONTINUE restart directly instead.
2778 ** The TOP debugger command is also gone, since it's redundant with the
2779 FRAME 0 command, and since it interfered with abbreviations for the
2781 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2783 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2784 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2785 some porting work by Martin Atzmueller)
2786 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2787 with a boatload of patches which he ported to SBCL. Now that those
2789 ** The system tries to make sure that its low-priority messages
2790 are prefixed by semicolons, to help people who like to use
2791 syntax highlighting in their ILISP buffer. (This patch
2792 was originally due to Raymond Toy.)
2793 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2794 patches originally written by Raymond Toy.
2795 ** The compiler understands coercion between single-value and
2796 multiple-VALUES type expressions better, getting rid of some very
2797 weird behavior, thanks to patches originally by Robert MacLachlan
2798 and Douglas Crosher.
2799 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2800 lambda lists, thanks to a patch originally by Pierre Mai.
2801 ** The system no longer bogusly warns about "abbreviated type
2803 ** The compiler gets less confused by inlining and RETURN-FROM,
2804 thanks to some patches originally by Tim Moore.
2805 ** The system no longer hangs when dumping circular lists to fasl
2806 files, thanks to a patch originally from Douglas Crosher.
2807 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2808 undefined function error.
2809 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2810 in structure slot initforms) for now, documented workaround instead:-|
2811 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2812 * fixed bug 5: added stubs for various Gray stream functions called
2813 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2814 installed, at least appropriate type errors are generated
2815 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2816 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2817 consistently in DEFMETHOD forms.
2818 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2819 it doesn't seem to affect SBCL after all
2820 * The C runtime system now builds with better optimization and many
2821 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2823 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2825 * The system is now under CVS at SourceForge (instead of the
2826 CVS repository on my home machine).
2827 * The new signal handling code has been tweaked to treat register
2828 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2829 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2830 downstream, e.g. in debug-int.lisp, has implicit dependencies
2831 on the unsignedness of integer representation of machine words,
2832 and that caused the system to bomb out with infinite regress
2833 when trying to recover from type errors involving signed values,
2834 e.g. (BUTLAST '(1 2 3) -1).)
2835 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2836 (This was one of the bugs Peter Van Eynde reported back in July.)
2837 * The system now uses code inspired by Colin Walters' O(N)
2838 implementation of MAP (from the cmucl-imp@cons.org mailing
2839 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2840 inline the MAP operation, and there is more than one
2841 sequence argument to the MAP call (so that it can't just
2842 do ETYPECASE once and for all based on the type of the
2843 single sequence argument). (The old non-inline implementation
2844 of the general M-argument sequence-of-length-N case required
2845 O(M*N*N) time when any of the sequence arguments were LISTs.)
2846 * The QUIT :UNIX-CODE keyword argument has been renamed to
2847 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2848 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2850 * Raymond Wiker's patches to port dynamic loading from Linux to
2851 FreeBSD have been added.
2852 * The BUGS file is now more nearly up to date, thanks in large part
2853 to Martin Atzmueller's review of it.
2854 * The debugger now flushes standard output streams before it begins
2855 its output ("debugger invoked" and so forth).
2856 * The core version number and fasl file version number have both
2857 been incremented, because of incompatible changes in the layout
2859 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2860 used to not be called for a saved Lisp image.)
2861 * Martin Atzmueller's version of a patch to fix a compiler crash,
2862 as posted on sbcl-devel 13 September 2000, has been installed.
2863 * Instead of installing Martin Atzmueller's patch for the
2864 compiler transform for SUBSEQ, I deleted the compiler transform,
2865 and transforms for some similar consing operations.
2866 * A bug in signal handling which kept TRACE from working on OpenBSD
2868 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2869 away by constant folding
2870 * The system now defines its address space constants in one place
2871 (in the Lisp sources), and propagates them automatically elsewhere
2872 (through GENESIS and the sbcl.h file). Therefore, patching the
2873 address map is less unnecessarily tedious and error-prone. The
2874 Lisp names of address space constants have also been systematized.
2875 * CVS tags like dollar-Header-dollar have been removed from
2876 the sources, because they have never saved me trouble and
2877 they've been source of trouble working with patches and other
2878 diff-related operations.
2879 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2880 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2882 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2884 * The system has been ported to OpenBSD.
2885 * The system now compiles with a simple "sh make.sh" on the systems
2886 that it's supported on. I.e., now you no longer need to tweak
2887 text in the target-features.lisp-expr and symlinks in src/runtime/
2888 by hand, the make.sh takes care of it for you.
2889 * The system is no longer so grossly inefficient when compiling code
2890 involving vectors implemented as general (not simple) vectors (VECTOR T),
2891 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2892 which dares to use the various sequence functions on non-simple
2893 vectors, takes less of a performance hit.
2894 * There is now a primitive type predicate VECTOR-T-P
2895 to test for the (VECTOR T) type, so that e.g.
2896 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2897 can now be compiled with some semblance of efficiency. (The old code
2898 turned the type declaration into a full call to %TYPEP at runtime!)
2899 * AREF on (VECTOR T) is still not fast, since it's still compiled
2900 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2901 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2902 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2903 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2904 without making full calls to SUBTYPEP at runtime.
2905 (Some analogous efficiency issues for non-simple vectors specialized to
2906 element types other than T, or for non-simple multidimensional arrays,
2907 have not been addressed. They could almost certainly be handled the
2908 same way if anyone is motivated to do so.)
2909 * The changes in array handling break binary compatibility, so
2910 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2911 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2912 instead of (VALUES T T).
2913 * By following the instructions that Dan Barlow posted to sbcl-devel
2914 on 2 July 2000, I was able to enable primitive dynamic object
2915 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2916 functionality is not implemented (since it calls ld to resolve
2917 library references automatically, requiring RUN-PROGRAM for its
2918 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2919 try to resolve library references) is now supported.
2920 * The system now flushes the standard output streams when it terminates,
2921 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2922 them at several other probably-convenient times, e.g. in each pass of
2923 the toplevel read-eval-print loop, and after evaluating a form given
2924 as an "--eval" command-line option. (These changes were motivated by a
2925 discussion of stream flushing issues on cmucl-imp in August 2000.)
2926 * The source transform for TYPEP of array types no longer assumes
2927 that an array whose element type is a not-yet-defined type
2928 is implemented as an array of T, but instead punts, so that the
2929 type will be interpreted at runtime.
2930 * There is now some support for cross-compiling in make.sh: each of
2931 the phases of make.sh has its own script. (This should be transparent
2932 to people doing ordinary, non-cross-compile builds.)
2933 * Since my laptop doesn't have hundreds of megabytes of memory like
2934 my desktop machine, I became more motivated to do some items on
2935 my to-do list in order to reduce the size of the system a little:
2936 ** Arrange for various needed-only-at-cold-init things to be
2937 uninterned after cold init. To support this, those things have
2938 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2939 symbols with such names are now uninterned after cold init).
2940 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2941 fasl files for cold load.
2942 ** Remove the old compiler structure pooling code (which used to
2943 be conditional on the target feature :SB-ALLOC) completely.
2944 ** Redo the representation of some data in cold init to be more compact.
2945 (I also looked into supporting byte compiled code at bootstrap time,
2946 which would probably reduce the size of the system a lot, but that
2947 looked too complicated, so I punted for now.)
2948 * The maximum signal nesting depth in the src/runtime/ support code has
2949 been reduced from 4096 to 256. (I don't know any reason for the very
2950 large old value. If the new smaller value turns out to break something,
2951 I'll probably just bump it back up.)
2952 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2954 * Many, many bugs reported by Peter Van Eynde have been added to
2955 the BUGS list; some have even been fixed.
2956 * While enabling dynamic object file loading, I tried to make the
2957 code easier to understand, renaming various functions and variables
2958 with less ambiguous names, and changing some function calling
2959 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2960 * While trying to figure out how to do the OpenBSD port, I tried to
2961 clean up some of the code in src/runtime/. In particular, I dropped
2962 support for non-POSIX signal handling, added various comments,
2963 tweaked the code to reduce the number of compilation warnings, and
2964 renamed some files to increase consistency.
2965 * To support the new automatic configuration functionality in make.sh,
2966 the source file target-features.lisp-expr has been replaced with the
2967 source file base-target-features.lisp-expr and the machine-generated
2968 file local-target-features.lisp-expr.
2969 * fixed a stupid quoting error in make.sh so that using CMU CL
2970 "lisp -batch" as cross-compilation host works again
2972 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2974 * DESCRIBE no longer tries to call itself recursively to describe
2975 bound/fbound values, so that it no longer fails on symbols which are
2976 bound to themselves (like keywords, T, and NIL).
2977 * DESCRIBE now works on generic functions.
2978 * The printer now prints less-screwed-up representations of closures
2979 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2980 * A private symbol is used instead of the :EMPTY keyword previously
2981 used to mark empty slots in hash tables. Thus
2982 (DEFVAR *HT* (MAKE-HASH-TABLE))
2983 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2984 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2985 now does what ANSI says that it should. (You can still get
2986 similar noncompliant behavior if bang on the hash table
2987 implementation with all the symbols you get back from
2988 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2989 This breaks binary compatibility, since tests for equality to
2990 :EMPTY are wired into things like the macroexpansion of
2991 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2993 * There's now a minimal placeholder implementation for CL:STEP,
2994 as required by ANSI.
2995 * An obscure bug in the interaction of the normal compiler, the byte
2996 compiler, inlining, and structure predicates has been patched
2997 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2998 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3000 * Missing ordinary arguments in a macro call are now detected even
3001 when the macro lambda list contains &KEY or &REST.
3002 * The debugger no longer complains about encountering the top of the
3003 stack when you type "FRAME 0" to explicitly instruct it to go to
3004 the top of the stack. And it now prints the frame you request even
3005 if it's the current frame (instead of saying "You are here.").
3006 * As specified by ANSI, the system now always prints keywords
3007 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3009 * The default initial SIZE of HASH-TABLEs is now smaller.
3010 * Type information from CLOS class dispatch is now propagated
3011 into DEFMETHOD bodies, so that e.g.
3012 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3014 is now basically equivalent to
3015 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3016 (DECLARE (TYPE SINGLE-FLOAT X))
3018 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3019 operation, without having to do run-time type dispatch.
3020 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3021 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3022 * Since I don't seem to be making big file reorganizations very often
3023 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3024 is overflowing my ability to conveniently back them up), I've finally
3025 checked the system into CVS. (The CVS repository is on my home system,
3026 not at SourceForge -- putting it on SourceForge might come later.)
3027 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3028 high-level GC-NOTIFY-FOO functions send their output. (There's
3029 still very little control of where low-level verbose GC functions
3030 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3031 less than it used to -- the GC-NOTIFY-FOO functions are now under
3032 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3033 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3034 in only one place in the source code, and propagates it automatically
3035 everywhere that it's needed. Thus e.g. when I bump the version from
3036 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3037 * The C source files now include boilerplate legalese and documentation
3038 at the head of each file (just as the Lisp source files already did).
3039 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3040 to his page will be replaced with a link to his new CLiki service.
3042 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3044 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3045 * The build process now looks for GNU make under the default name "gmake",
3046 instead of "make" as it used to. If GNU make is not available as "gmake"
3047 on your system, you can change this default behavior by setting the
3048 GNUMAKE environment variable.
3049 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3050 can build without error under CMU CL.
3052 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3054 * There is now a partial SBCL user manual (with some new text and some
3055 text cribbed from the CMU CL manual).
3056 * The beginnings of a profiler have been added (starting with the
3057 CMU CL profiler and simplifying and cleaning up). Eventually the
3058 main interface should be through the TRACE macro, but for now,
3059 it's still accessed through vaguely CMU-CL-style functions and macros
3060 exported from the package SB-PROFILE.
3061 * Some problems left over from porting CMU CL to the new
3062 cross-compilation bootstrap process have been cleaned up:
3063 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3064 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3065 part when getting rid of the compiler *BACKEND* stuff.)
3066 ** The value of *NULL-TYPE* was screwed up, because it was
3067 being initialized before the type system knew the final
3068 definition of the 'NULL type. This screwed up several key
3069 optimizations in the compiler, causing inefficiency in all sorts
3070 of places. (I found it because I wanted to understand why
3071 GET-INTERNAL-RUN-TIME was consing.)
3072 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3073 PROCLAIM FTYPE information. Unfortunately this broke binary
3074 compatibility again, since now the forms output by DEFGENERIC
3075 to refer to functions which didn't exist in 0.6.3.
3076 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3077 can use the new (as of 0.6.3) transform for SOME into MAP into
3079 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3080 values to (INTEGER 0 1000000), so that the time code will no longer
3081 occasionally get blown up by Linux returning 1000000 microseconds
3082 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3083 its output conform to the ANSI spec. (Alas, this makes its output
3084 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3085 * A full call to MAP NIL with a single sequence argument no longer conses.
3086 * fixes to problems pointed out by Martin Atzmueller:
3087 * The manual page no longer talks about multiprocessing as though
3088 it were currently supported.
3089 * The ILISP support patches have been removed from the distribution,
3090 because as of version 5.10.1, ILISP now supports SBCL without us
3091 having to maintain patches.
3092 * added a modified version of Raymond Toy's recent CMU CL patch for
3093 EQUALP comparison of HASH-TABLE
3095 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3097 * The system still can't cross-compile itself with
3098 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3099 entails), but at least it can compile more of itself that way
3100 than it used to be able to, and various buglets which were uncovered
3101 by trying to cross-compile itself that way have now been fixed.
3102 * This release breaks binary compatibility again. This time
3103 at least I've incremented the FASL file format version to 2, so that the
3104 problem can be detected reliably instead of just causing weird errors.
3105 * various new style warnings:
3106 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3107 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3108 ** using the lexical binding of a variable named in the *FOO* style
3109 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3110 as specified by ANSI.
3111 * *RANDOM-STATE* is no longer automatically initialized from
3112 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3113 default behavior of the system is to repeat its behavior every time
3114 it's run. If you'd like to change this behavior, you can always
3115 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3116 old convention there was no comparably easy way to get the system to
3117 repeat its behavior every time it was run.
3118 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3119 FTYPE declarations has been removed, in favor of their ANSI
3120 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3121 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3122 the types of their sequence arguments can be determined at compile time.
3123 This is done through a new open code expansion for MAP which eliminates
3124 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3125 argument types can be determined at compile time.
3126 * The optimizer now transforms COERCE into an identity operation when it
3127 can prove that the coerced object is already of the correct type. (This
3128 can be a win for machine generated code, including the output of other
3129 optimization transforms, such as the MAP transform above.)
3130 * Credit information has been moved from source file headers into CREDITS.
3131 * Source file headers have been made more standard.
3132 * The CASE macro now compiles without complaining even when it has
3135 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3137 * (Note that the way that the PCL macroexpansions were rewritten
3138 to accommodate the change in DEFGENERIC below breaks binary
3139 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3140 not run under sbcl-0.6.2. Once we get out of alpha releases,
3141 i.e. hit release 1.0.0, we'll probably try to maintain binary
3142 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3143 and sbcl-1.4.4. Until then, however, it might be fairly common
3144 for maintenance releases to break binary compatibility.)
3145 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3147 * The reporting of unhandled signals has been changed to print some
3148 explanatory text as well as the report form. (Previously only
3149 the report form was printed.)
3150 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3151 it defines, so that the compiler no longer issues undefined function
3152 warnings for compiled-but-not-yet-loaded generic functions.
3153 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3154 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3155 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3156 section of the ANSI spec are supported.
3157 * The "" nickname for the "KEYWORD" package has been removed.
3158 The reader still handles symbol tokens which begin with a package marker
3159 as keywords, but it doesn't expose its mechanism for doing so in the
3160 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3161 * The system now issues STYLE-WARNINGs for contradictory TYPE
3162 proclamations. (Warnings for contradictory FTYPE proclamations would
3163 be nice too, but those can't be done usefully unless the type system
3164 is made smarter about FUNCTION types.)
3165 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3166 systematized, so that "*target-*.lisp is supposed to exist only on the
3167 target and imply that there's a related file which exists on the
3168 host, and *host-*.lisp is supposed to exist only on the host and imply
3169 that there's a related file which exists on the target. This involves a
3170 lot of renaming. Hopefully the acute confusion caused by the renaming
3171 will be justified by the reduction in chronic confusion..
3172 ** runtime-type.lisp -> early-target-type.lisp
3173 ** target-type.lisp -> late-target-type.lisp
3174 ** early-host-format.lisp -> early-format.lisp
3175 ** late-host-format.lisp -> late-format.lisp
3176 ** host-error.lisp -> misc-error.lisp
3177 ** early-error.lisp -> early-target-error.lisp
3178 ** late-error.lisp -> late-target-error.lisp
3179 ** host-defboot.lisp -> early-defboot.lisp
3180 ** code/misc.lisp -> code/target-misc.lisp
3181 ** code/host-misc.lisp -> code/misc.lisp
3182 ** code/numbers.lisp -> code/target-numbers.lisp
3183 ** code/early-numbers.lisp -> numbers.lisp
3184 ** early-host-type.lisp -> early-type.lisp
3185 ** late-host-type.lisp -> late-type.lisp
3186 ** host-typep.lisp -> typep.lisp
3187 ** load.lisp -> target-load.lisp
3188 ** host-load.lisp -> load.lisp
3189 ** host-disassem.lisp -> disassem.lisp
3190 ** host-insts.lisp -> insts.lisp
3191 ** byte-comp.lisp -> target-byte-comp.lisp
3192 ** host-byte-comp.lisp -> byte-comp.lisp
3193 ** host-signal.lisp -> signal.lisp
3194 ** host-defstruct.lisp -> defstruct.lisp
3195 ** late-target-type.lisp -> deftypes-for-target.lisp
3196 Furthermore, several other previously target-only files foo.lisp (e.g.
3197 hash-table.lisp and random.lisp) have been split into a target-and-host
3198 foo.lisp file and a target-only target-foo.lisp file, with their key type
3199 definitions in the target-and-host part, so that the cross-compiler will
3200 know more about target types.
3201 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3202 gone away. In their place are various *BACKEND-FOO* variables
3203 corresponding to the slots of the old structure.
3204 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3205 into the target SBCL has been fixed.
3206 * The chill.lisp system for loading cold code into a running SBCL
3208 * Support for the CMU CL "scavenger hook" extension has been removed.
3209 (It was undocumented and unused in the CMU CL sources that SBCL was
3210 derived from, and stale in sbcl-0.6.1.)
3211 * Various errors in the cross-compiler type system were detected
3212 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3213 (enabling various consistency checks). Many of them were fixed,
3214 but some hard problems remain, so the compiler is back to
3215 running without *TYPE-SYSTEM-INITIALIZED* for now.
3216 * As part of the cross-compiler type system cleanup, I implemented
3217 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3218 * I have started adding UNCROSS calls throughout the type system
3219 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3220 on cross-compiler input files will be able to go away, and various
3222 * CONSTANTP now returns true for quoted forms (as explicitly required
3225 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3227 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3228 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3229 (MAKE-LIST -1). In the longer term, it ought to become true
3230 that declarations are assertions even at SAFETY 1. For now, it's not
3231 quite true even at SAFETY 3, but it's at least more nearly true..
3232 (Note that this change seems to increases the size of the system by
3233 O(5%) and to decrease the speed of the compiler by 20% or more.)
3234 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3235 for the problem of printing dozens of lines of distracting information
3236 about low-level system machinery as part of the top stack frame
3237 on entry to the debugger when an undefined function was called.
3238 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3239 is not reset to COMMON-LISP-USER.
3240 * Compilation of stuff related to dyncount.lisp has been made conditional
3241 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3242 smaller. The various dyncount-related symbols have been moved into
3243 a new "SB-DYNCOUNT" package.
3244 * tty-inspect.lisp has been renamed to inspect.lisp.
3245 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3246 feature has gone away. (When we eventually port to other flavors of
3247 libc and/or Unix, we'll try to make the differences between flavors
3248 invisible at the user level.)
3249 * Various other *FEATURES* tags, and/or their associated conditionals,
3250 have been removed if obsolescent, or given better documentation, or
3251 sometimes given more-mnemonic names.
3253 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3255 * tidied up "make.sh" script
3256 * tidied up system directory structure
3257 * better "clean.sh" behavior
3258 * added doc/FOR-CMUCL-DEVELOPERS
3259 * many many small tweaks to output format, e.g. removing possibly-confusing
3260 trailing #\. character in DESCRIBE-INSTANCE
3261 * (EQUALP #\A 'A) no longer signals an error.
3262 * new hashing code, including EQUALP hashing
3263 * tidied up Lisp initialization and toplevel
3264 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3265 * command line argument processing
3266 * added POSIX-GETENV function to deal with Unix-ish environment variables
3267 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3268 terminating SBCL on EOF
3269 * non-verbose GC by default
3270 * There is no more "sbcl" shell script; the sbcl file is now the C
3271 runtime executable (just like CMU CL).
3272 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3274 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3275 symbol and package information as Lisp native symbol and package objects
3276 instead of strings naming symbols and strings naming packages. This way,
3277 whenever packages are renamed (as in warm init), debug information is
3278 transformed along with everything else.
3279 * tweaked the optimization policy declarations which control the building
3280 of SBCL itself. Now, among other things, the system no longer saves
3281 source location debugging information. (This helps two problems at once
3282 by reducing SBCL size and by keeping SBCL from trying to look for its
3283 sources -- which may not exist -- when reporting errors.)
3284 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3285 debugging and testing purposes
3286 * cleaned up printing, making the printer call PRINT-OBJECT for
3287 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3288 methods, giving nearly-ANSI behavior
3289 * converted almost all special variables to use *FOO* naming convention
3290 * deleted PARSE-TIME functionality, since it can be done portably
3291 * moved some files out of cold init into warm init
3292 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3294 * regularized formatting of source files
3295 * added an install.sh script
3296 * fixed ridiculous memory usage of cross-compiler by making
3297 compiler/alloc.lisp not try to do pooling unless it can hook
3298 itself into the GC of the cross-compilation host. Now the system
3299 builds nicely on my old laptop.
3300 * added :SB-ALLOC in target-features.lisp-expr
3301 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3302 was not implemented)
3303 * re-did condition handling and note reporting in the compiler. Notes
3304 are no longer handled by signalling conditions. Style warnings
3305 and warnings are handled more correctly and reported in such a way
3306 that it's easy to find one or the other in your output (so that you
3307 can e.g. figure out which of many problems caused COMPILE-FILE to
3309 * changed the severity of several compiler warnings from full WARNING
3310 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3311 compiler note reporting so that it doesn't use the condition system
3312 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3313 COMPILE-FILE command)
3314 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3315 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3316 no longer supported, since their implementation was deeply intertwingled
3317 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3318 * removed bogus "support" for compiler macros named (SETF FOO), and
3319 removed the compiler macro for SETF INFO (but only after making a fool
3320 of myself on the cmucl-imp mailing list by posting a bogus patch for
3321 DEFINE-COMPILER-MACRO..)
3322 * Compiled files containing forms which have side effects on the Lisp
3323 reader (such as DEFPACKAGE forms) are now handled more correctly.
3324 (Compiler queuing of top level lambdas has been suppressed by setting
3325 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3326 * deleted various currently-unused source files, e.g. gengc.lisp. They
3327 may be added back at some point e.g. when porting to other architectures,
3328 but until they are it's distracting to distribute them and to try to
3330 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3331 there were so many of them they're just distractions, and UNCROSS is
3332 known to be able to handle the current sources
3333 * moved PROFILE functionality into TRACE, so that it will be clear
3334 how the wrapping and unwrapping of functions when you profile them
3335 interacts with the wrapping and unwrapping of functions when you
3336 trace them. (Actually, the functionality isn't there yet, but at least
3337 the interface specification is there. Hopefully, the functionality will
3338 arrive with some future maintenance release.)
3339 * removed host-oops.lisp
3340 * changed signature of QUIT function to allow UNIX-CODE argument
3341 * fixed READ-SEQUENCE bug
3342 * tweaked verbose GC output so that it looks more like the progress
3343 output that ANSI specifies for functions like LOAD
3344 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3345 * added <http://sbcl.sourceforge.com> to the banner information printed by