1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
3 * new feature: timers based on Zach Beane's excellent timer package
4 * added support for the following external formats: koi8-u,
5 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
6 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
7 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
8 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
9 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
10 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
11 * incompatible change: a threaded SBCL will no longer revert to
12 non-threaded mode on non-NPTL systems, but refuse to start entirely.
13 * bug fix: interrupts are disabled until startup is complete; no
14 more sigsegvs when receiving a signal to soon
15 * optimization: Faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
17 * bug fix: add a workaround for the memory randomization features in
18 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
19 workaround will only be in effect on systems with the proc filesystem
21 * bug fix: printing objects of type HASH-TABLE signals a
22 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
24 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
25 have been running for over 50 days. (reported by Gilbert Baumann)
26 * bug fix: the logic for getting names of functions gets less
27 confused when confronded with alternate-metaclass
28 funcallable-instances. (reported by Cyrus Harmon)
29 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
30 consistent, even on internal alternate-metaclass objects.
32 ** bug fix: parent thread now can be gc'ed even with a live
34 ** bug fix: binding a special with PROGV to no value is not
35 the same as that symbol not having been bound (thanks to
37 ** bug fix: binding specials is thread safe (thanks to
39 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
40 and SB-SPROF do not die with 'no handler for signal XX in
41 interrupt_handle_now(..)' anymore
42 ** bug fix: WITH-TIMEOUT works with multiple running threads
43 * fixed some bugs revealed by Paul Dietz' test suite:
44 ** ENSURE-GENERIC-FUNCTION should take a method class object for
45 the :method-class keyword argument.
47 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
48 * new port: the Solaris operating system on x86 processors is now
49 mostly supported, though some rough edges in the environment
50 remain. (thanks to Daisuke Homma)
51 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
52 space for its heap. (thanks to Thiemo Seufer)
53 * optimization: SBCL on MIPS platforms now supports dynamic-extent
54 closures. (thanks to Thiemo Seufer)
55 * minor incompatible change: eof selects abort in the debugger.
56 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
57 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
59 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
60 deprecated, and will go away in a future revision of SBCL.
61 * minor incompatible change: GC-ON and GC-OFF are no longer
62 implemented with a counter, it does not matter how many times gc
64 * bug fix: discriminating functions for generic function classes
65 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
66 make invalid assumptions about method precedence order. (reported
68 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
69 platforms. (thanks to Thiemo Seufer)
70 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
71 * bug fix: correct pathname printing with printer escaping is on.
72 (thanks to Kevin Reid)
73 * bug fix: complex VOP definitions in "user-space" no longer trigger
74 package locks. (reported by Zach Beane)
75 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
76 for funcallable-instance functions, and is no different from
78 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
79 characters in the current readtable. (reported by Nicholas Neuss)
80 * bug fix: SUBTYPEP on various CONS types returns more a more
81 accurate acknowledgment of its certainty.
82 * optimizations: REMOVE-DUPLICATES now runs in linear time on
83 lists in some cases. This partially fixes bug 384.
84 * flush all standard streams before prompting in the REPL and the
86 * bug fix: signal handling and triggering gc do not conflict
87 directly anymore, in particular a high frequency sb-sprof does
88 not prevent gc from running
89 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
90 approximation for timezone and DST information between the
91 universal time epoch and the smallest negative 32-bit time_t.
92 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
93 year 1899 so long as the encoded time is non-negative.
94 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
95 the correct answer for non-constant rotate amounts when performing
96 an inline 32-bit rotation.
98 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
99 there is only one thread in the session
100 ** bug fix: memory leak for streams created in one thread and
101 written to in another
102 ** bug fix: lockup when compiled with gcc4
103 ** bug fix: race that allows the gc to be triggered when gc is
105 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
106 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
108 ** bug fix: finished threads are gc'ed properly
109 * fixed some bugs revealed by Paul Dietz' test suite:
110 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
111 has arguments with a different set of applicable methods from
112 the orignal arguments.
113 ** The type error thrown by MAP now has a correct expected-type
115 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
116 name a compiled function.
117 ** Three MISC tests where a large bignum was improperly coerced to
118 a float that couldn't represent that bignum during type
119 derivation were fixed.
120 ** SUBTYPEP can now handle types involving the negation of a
121 list-form FUNCTION type.
122 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
123 as (COMPLEX (AND RATIO (NOT FIXNUM))).
124 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
126 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
127 * New feature: Experimental support for bivalent streams: streams
128 opened with :element-type :default now allow character and binary
129 (unsigned-byte 8) I/O
130 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
131 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
133 * bug fix: no more highly sporadic "couldn't check whether ~S is
134 readable" when reading a stream and an interrupt hits in the middle
135 of a select system call
136 * compiler better recognizes complex arrays (reported by Eduardo
138 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
139 for variables on SBCL built with linkage-tables. (reported by Luis
141 * various error reporting improvements.
142 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
143 (Thanks to Thiemo Seufer)
144 * bug fix: nested reader invokations work correctly
145 * bug fix: it is possible to have simultaneous references to foreign
146 code and foreign data with the same name.
148 ** added x86-64 support
149 ** incompatible change: the threading api now works with thread
150 objects instead of thread ids
151 ** bug fix: threads are protected from signals and interruption when
152 starting up or going down
153 ** bug fix: a race where an exiting thread could lose its stack to gc
154 ** bug fix: don't halt on infinite error in threads if possible
155 ** fixed numerous gc deadlocks introduced in the pthread merge
156 ** bug fix: fixed thread safety issues in read and print
157 ** bug fix: debugger doesn't hang on session lock if interrupted at
158 an inappropriate moment
159 ** bug fix: run-program is now thread safe(r)
160 ** bug fix: inner with-recursive-lock no longer releases the mutex
161 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
162 * fixed some bugs revealed by Paul Dietz' test suite:
163 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
164 EXPECTED-TYPE slots filled.
165 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
167 ** MAKE-CONDITION accepts classes as type-designators.
168 ** COMPILE may never return NIL.
169 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
170 range before calling Unix time functions
172 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
173 * numerous signal handling fixes to increase stability
174 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
175 * minor incompatible change: we now correctly canonize default
176 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
177 as per the MOP, rather than the historical (INITARG INITFUNCTION
178 INITFORM). (reported by Bruno Haible)
179 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
180 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
181 for more information.
182 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
183 pathname is a directory pathname.
184 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
185 * dynamic space size on PPC has been increased to 768Mb. (thanks to
187 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
188 :METACLASS argument in addition to a class name. (reported by
189 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
190 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
191 (reported by Bruno Haible; patch from Gerd Moellmann)
192 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
194 * bug fix: more cleanups to the floating point exception handling on
195 x86-64 (thanks to James Knight)
196 * bug fix: the compiler does not try to constant fold calls of
197 COERCE and BIT-* functions when they return freshly constructed
198 objects. (reported by Nikodemus Siivola and Paul Dietz)
199 * optimization: improved the MIPS versions of generic (in the
200 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
201 * optimization: direct conversion of (unsigned-byte 32) to floats on
202 the PowerPC platform.
203 * optimization: structure instances with raw slots now use less
204 memory, and probably show better memory locality. (thanks to
206 * optimization: DYNAMIC-EXTENT declarations for lists and closures
207 are treated as requests for stack allocation on the x86-64,
208 Alpha-32, PPC and SPARC platforms.
209 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
210 the wrong file descriptor; implementation of SOCKET-OPEN-P.
211 (thanks to Tony Martinez)
213 ** gcing a dead thread can no longer lead to lockups
214 ** threads block signals until they are set up properly
215 ** errno is no longer shared by threads
216 ** interrupt-thread restores the eflags register on x86
217 ** fixed some lockups due to gc/thread interaction
218 * fixed some bugs revealed by Paul Dietz' test suite:
219 ** invalid dotted lists no longer raise a read error when
221 ** signal an error if a symbol that names a declaration is used
222 as the name of a type, or vice versa
223 ** allow using the (declare (typespec var*)) abbreviation for
224 (declare (type typespec var*)) with all type specifiers
225 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
226 ** always return NIL from PPRINT-POP when OBJECT is NIL
227 ** don't signal errors when pretty-printing malformed LABELS,
228 FLET or MACROLET forms
229 ** declarations in a DOLIST body are also in scope for the
231 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
233 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
236 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
237 * fixed cross-compiler leakages that prevented building a 32-bit
238 target with a 64-bit host compiler.
239 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
240 opened with :IF-EXISTS :OVERWRITE.
241 * fixed bug 281: error for an invalid qualifier in a short-form method
242 combination method is not signalled until the faulty method is called.
243 * bug fix: iteration variable type inferrer failed to deal with open
244 intervals. (reported by Alan Shields)
245 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
246 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
247 or more runtime options were provided to the sbcl binary.
248 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
249 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
250 * optimization: there's now a fast-path for fixnum arguments in the
251 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
252 * optimization: the code generated on x86-64 is more compact thanks
253 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
254 * bug fix: floating-point exception handling now partly works on
255 x86-64. (Thanks to James Knight)
256 * improvement to the MIPS backend from Thiemo Seufer: C-style
257 64-bit long long arguments and return values to alien functions
259 * fixed some bugs revealed by Paul Dietz' test suite:
260 ** the type-error signalled from WARN has a filled-in DATUM slot.
261 ** the type-error required when a stream is not associated with
262 a file has the stream as its datum.
263 ** type-errors on single-floats on x86-64 no longer have
264 :INVALID-OBJECT as the datum
265 ** the type-errors signalled for invalid function names now have
266 a correct expected type
267 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
268 for structures defined with a :TYPE. Documentation strings for
269 typed structures are no longer immediately discarded
270 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
271 broadcast streams and synonym streams. FILE-LENGTH now also works
272 on broadcast streams.
274 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
275 * incompatible change: the --noprogrammer option, deprecated since
276 version 0.7.5, has been removed. Please use the equivalent
277 --disable-debugger option instead.
278 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
280 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
281 inoperational for a while now) has been completely removed.
282 * null lexical environments are now printed as #<NULL-LEXENV>,
283 significantly reducing the amount of clutter in typical backtraces.
284 * documentation on weak pointers, finalization, and after GC hooks
285 has been added to the manual.
286 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
287 as other specialized array types, is much faster. SUBSEQ and
288 COPY-SEQ on such arrays have also been sped up.
289 * optimization: EQL is now more efficient when at least other argument
290 is known to be of type (OR FIXNUM (NOT NUMBER)).
291 * fixed bug: the runtime is now less vulnerable to changes in the
292 size of the SBCL object on OS X, and virtual memory is reserved for
293 all spaces that need to be at a fixed address.
294 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
295 * fixed bug: finalizers and after GC hooks that cause consing are now
297 * fixed bug: compiler error messages and summaries are now printed to
298 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
299 * fixed inference of the upper bound of an iteration variable.
300 (reported by Rajat Datta).
301 * fixed bug 211e: calling local functions with duplicated constant
302 keyword argument no longer causes a bogus style warning about an
304 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
305 ftype declaration to be disregarded. (reported by Dave Roberts)
306 * fixed bug 373: caused by erronous compilation of references to alien
307 variables in the runtime on ppc/darwin.
308 * fixed bug 376: CONJUGATE type deriver.
309 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
310 (reported by Baughn on #lisp)
311 * a cross-compiler bug on non-x86 platforms has been identified and
312 fixed. (thanks to Bruno Haible)
313 * improvements to the MIPS runtime code for increased stability.
314 (thanks to Thiemo Seufer)
315 * increased the maximimum compact environment size to allow
316 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
317 * partial workaround for bug 135: don't name the function we're
318 calling for hairy cases of EVAL, so as not to accumulate one environment
319 entry per call to EVAL. (reported by Kevin Reid)
320 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
321 * optimization: type testing for non-vector arrays should be faster.
322 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
324 * bug fix: setting 31st element of a bit vector to zero did not work
326 * bug fix: redefining a class definition which failed due to a
327 previous accessor / function clash now works (but see BUGS entry
328 #380 for more problems in this area). (thanks to Zach Beane)
329 * the long form of DEFINE-METHOD-COMBINATION disables method group
330 checking when given a single method group with pattern *. (thanks
332 * on x86 compiler supports stack allocation of results of simple
333 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
334 * support for building and running on Mac OS X 10.4 "Tiger" has been added
335 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
336 * fixed some bugs related to Unicode integration:
337 ** the restarts for recovering from input and output encoding
338 errors only appear when there is in fact such an error to
340 * fixed some bugs revealed by Paul Dietz' test suite:
341 ** MISC.549 and similar: late transformation of unsafe type
342 assertions into derived types caused unexpected code
344 ** SCALE-FLOAT type deriver is less wrong.
345 ** type derivers for EXP, LOG and similar functions compute result
346 types for complex arguments better.
347 ** (MISC.563) CONJUGATE type deriver works for very restricted
349 ** out-of-line type testers for character strings are available.
350 ** EQUAL compiler transform understands specialness of objects
352 ** accessing double-floats stored on the stack now works on x86-64.
353 ** debugger internals could sometimes create invalid lispobjs,
354 resulting in GC crashes.
355 ** MISC.548: type check weakening can convert required type into
357 ** initialization forms for bindings are not in scope of free special
360 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
361 * incompatible change: thread support for non-NPTL systems has
362 been removed - locking is buggy and unreliable. A threaded
363 SBCL build will now warn at startup and refuse to create
364 new threads, unless futex support is detected in the kernel
365 * incompatible change: the top level REPL now has only an ABORT
366 restart associated with it, not TOPLEVEL and ABORT as it used to.
367 TOP and TOPLEVEL are now available as debugger commands for
368 returning to the top level.
369 * incompatible change: forms evaluated in the REPL now use the
370 global optimization policy.
371 * incompatible change: user- and system-initialization files are
372 no longer processed with LOAD, but by READ and EVAL; hence the
373 global optimization policy, startup package, readtable, etc,
375 * The .fasl file format number has been incremented because of
376 various incompatible changes.
377 * internal entry point details and argument counts no longer appear
378 in backtraces unless explicitly requested by setting
379 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
380 * built-in and standard functions no longer have names like "top
381 level local call to FOO".
382 * fixed bug 32: functions defined in non-null lexical environments
383 now have more legible printed representation
384 * fixed bug 33: functions defined in non-null lexical environemnts
385 are now more amenable to inspection by INSPECT.
386 * workaround for bug 354: XEPs no longer appear in backtraces unless
387 explicitly requested.
388 * fixed bug: receiving the signal which results from attempting to
389 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
390 SIGBUS on other *BSDs) on architectures where the C stack is also
391 the Lisp stack (x86 and x86-64) from foreign code no longer leads
392 to debugger confusion or wild instability. (reported by Cheuksan
394 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
395 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
396 (reported by Lutz Euler)
397 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
398 compiled code. (reported by Paul Dietz)
399 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
400 specialized parameter to a method to something that is not TYPEP
401 the specializer is now possible.
402 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
403 face of package deletion.
404 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
405 pathname. (reported by David Tolpin on c.l.l)
406 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
407 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
408 STRING-STREAM and FILE-STREAM, respectively.
409 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
410 than previously; in addition, its test suite is now run on build.
411 (thanks to Luke Gorrie)
412 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
413 * optimization: added a immediate representation for single-floats
415 * fixed some bugs revealed by Paul Dietz' test suite:
416 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
417 ** attempting to create a package with a colliding nickname causes
418 correctable errors to be signalled.
419 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
420 ** several x86-64 backend bugs related to sign-extension of immediate
423 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
424 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
425 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
427 * fixed loading of multiply forward-referenced layouts.
428 (thanks to Cheuksan Wang)
429 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
430 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
431 (see COMPILE-FILE documentation for details of :PRINT option)
432 * fixed bugs 19 and 317: fixed-format floating point printing is
433 more accurate. This also fixes a bug reported by Adam Warner
434 related to the ~@F format directive.
435 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
437 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
438 dispatch table if the from-char is a dispatch macro character.
439 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
440 by the wordsize no longer ignore the last word. (reported by Lutz
442 * fixed bug in type checking of dynamic-extent variables. (reported
444 * optimization: sequence traversal functions use their freedom to
445 coerce function designators to functions.
446 * optimization: code with many calls to CLOS methods specialized on
447 CLOS classes has had redundant type checks removed.
448 * optimization: added declarations to speed up operations that access
449 the internal character database (for example STRING-UPCASE)
450 * optimization: comparison operations between floats and sufficiently small
451 fixnums no longer create extra rationals
452 * fixed some bugs related to Unicode integration:
453 ** portions of multibyte characters at the end of buffers for
454 character-based file input are correctly transferred to the
455 start of the buffer at the next read.
456 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
457 passing it through to OPEN.
458 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
459 argument given to internal calls to OPEN.
460 ** the built-in comment readers (introduced by character sequences
461 ";" and "#|") are more forgiving to encoding errors; they will
462 STYLE-WARN and then attempt to resync the stream at a character
463 boundary. (thanks to Teemu Kalvas)
464 * fixed some bugs in the x86-64 port:
465 ** Negative short int return values from c-calls are sign-extended
467 ** The stack is aligned to 16-bytes for c-calls, as required by
468 the ABI. (thanks to Cheuksan Wang)
469 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
470 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
471 for enums. (thanks to Vincent Arkesteijn)
472 ** Multiple small optimizations and bugfixes for floating point
474 * fixed some bugs revealed by Paul Dietz' test suite:
475 ** Space, Tab, Linefeed, Return and Page have the invalid
476 secondary constituent character trait.
477 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
479 ** WITH-INPUT-FROM-STRING should only update the index place on
481 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
483 ** Bit-array operations (BIT-AND and similar) worked incorrectly
484 with one-dimensional arrays with fill pointers.
485 ** TYPE-OF failed on a complex with an integer realpart and a
487 ** compiler failure during type inference for the code of form
488 (IF (EQL X (THE ...)) ...) (MISC.535).
490 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
491 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
492 platform. The port supports 61-bit fixnums, large memory spaces
493 and reloading shared object files.
494 * enhancement: saving cores with foreign code loaded is now
495 supported on ppc/Darwin in addition to the previously supported
497 * enhancement: the statistical profiler now walks deeper into the
498 call stack for more meaningful call-graphs and accrued time
499 reports (x86/x86-64 only). It also now reports time spent in
501 * enhancement: it is now possible to trace most individual methods
502 of a generic function in addition to tracing the generic function
504 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
506 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
507 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
508 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
509 produces an error. (thanks to Vincent Arkesteijn)
510 * bug fix: NAMESTRING on pathnames with :WILD components in their
511 directories works correctly. (thanks to Artem V. Andreev)
512 * fixed bug 125: compiler preserves identity of closures. (reported
514 * bug fixed: functions with &REST arguments sometimes failed with
515 "Undefined function" when compiled with (DEBUG 3). (reported by
517 * bug fix: overflow during compiling of setting element of a bit
518 vector with constant index and value. (reported by Timmy Douglas)
519 * build fix: fixed the dependence on *LOAD-PATHNAME* and
520 *COMPILE-FILE-PATHNAME* being absolute pathnames.
521 * on x86 compiler partially supports stack allocation of dynamic-extent
523 * GO and RETURN-FROM do not check the extent of their exit points
524 when compiled with SAFETY 0.
525 * fixed some bugs related to Unicode integration:
526 ** encoding and decoding errors are now much more robustly
527 handled; it should now be possible to recover even from invalid
528 input or output to the terminal. (thanks to Teemu Kalvas)
529 ** provided a first cut at implementing STRING-TO-OCTETS and
530 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
531 ** altered the SB-MD5 contributed module slightly, changing the
532 interface just enough for it to be supportable for builds where
533 lisp characters are not eight bits.
534 * fixed some bugs revealed by Paul Dietz' test suite:
535 ** the FORMATTER-generated functions for ~V[ conditionals require
536 the correct number of arguments.
537 ** READ-FROM-STRING returns the mandated second value when applied
538 to displaced strings.
539 ** the #\Rubout and #\Backspace characters are treated as invalid
540 constituent characters by the tokenizer.
542 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
543 * new feature: reloading changed shared object files with
544 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
545 * new feature: references to foreign variables and functions
546 can now be compiled and loaded before the corresponding shared
547 object file is loaded, as long as the foreign definitions are
548 available at runtime.
549 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
550 supported. (thanks to Dan Debertin)
551 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
552 just keywords. (thanks to Vincent Arkesteijn)
553 * fixed bug #331: structure-class instances corresponding to
554 DEFSTRUCT forms are now created eagerly.
555 * fixed bug #345: backtraces from calls to undefined functions work
556 on x86 as well. Related bug #61 is now also partially fixed on x86
557 (backtraces from throws to unknown catch tags.)
558 * bug fix: lambda-list parsing is now stricter vrt. order and number
559 of lambda-list keywords.
560 * bug fix: as specified by AMOP, an error is signalled if a
561 class-option appears multiple times in a DEFCLASS form. (reported
563 * bug fix: RANDOM can be compiled when the compiler derives the type
564 of its numeric argument as a disjoint set of small integers.
565 (reported by Paul Dietz)
566 * bug fix: starting a core saved with shared objects loaded when
567 those objects are not available no longer causes threaded SBCL to
568 hang. (reported by Sean Ross)
569 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
571 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
572 argument, which must always be NIL. (reported by Kalle Niemitalo)
573 * bug fix: printing 1.0d+23 no longer results in an error.
574 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
575 * bug fix: structure slot setters preserve evaluation order. (thanks
577 * fixed some bugs related to Unicode integration:
578 ** RUN-PROGRAM can allow its child to take input from a Lisp
579 stream. (reported by Stefan Scholl)
580 ** ASDF-INSTALL successfully downloads .tar.gz files in all
581 locales. (reported by Ken Causey)
582 * fixed some bugs revealed by Paul Dietz' test suite:
583 ** INCF, DECF and REMF evaluate their place form as specified in
585 ** FORMATTER expands ~{ ~} iteration directives with V or #
586 parameters correctly.
587 ** FORMATTER deals with the ~@[ ~] conditional directive where the
588 consequent uses no arguments correctly.
589 ** the system has a partial understanding of the (COMPLEX RATIO)
591 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
594 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
595 * new feature: a build-time option (controlled by the :SB-UNICODE
596 keyword feature, enabled by default) for building the system with
597 support for the entire 21-bit character space defined by the
599 * new feature: the system now has rudimentary external-format
600 support; the primary user-visible change at this time is that
601 characters with the high bit set (such as lower-case-e-acute) will
602 print correctly to a terminal in a UTF-8 environment.
603 * minor incompatible change: BASE-CHAR no longer names a class;
604 however, CHARACTER continues to do so, as required by ANSI.
605 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
606 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
607 has been moved to the SB-EXT package (temporarily re-exported from
609 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
610 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
612 * fixed bug #351: better error-handlind and reporting for malformed
614 * fixed bug #350: bignum-printing is now more memory-efficient,
615 allowing printing of very large bignums, eg. (expt 2 10000000).
616 (reported by Bruno Haible)
617 * fixed bug #302: better primitive-type selection for intersection
619 * fixed bug #308: non-graphic characters now all have names, as
620 required. (reported by Bruno Haible)
621 * bug fix: redefining a class with different superclasses now correctly
622 removes it from the direct-subclasses of its previous superclasses.
623 (reported by David Morse)
624 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
625 new value now works. (reported by Bruno Haible)
626 * bug fix: correct canonicalization of multiple non-standard slot
627 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
628 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
629 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
630 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
631 now exists, an signals an error.
632 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
633 during macroexpansion for non-iteration variables as well. (reported
634 by Bruno Haible for CMUCL)
635 * bug fix: Cyclic structures and unprintable objects in compiler
636 messages no longer cause errors. (reported by Bruno Haible)
637 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
638 and READ-FROM-STRING all now return a primary value of NIL if
639 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
640 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
641 (reported by Bruno Haible for CMUCL)
642 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
643 do not designate a condition. (reported by Bruno Haible for
645 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
646 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
647 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
648 permitting resolution in favour of any of the conflicting symbols.
649 (reported by Bruno Haible for CMUCL)
650 * bug fix: EQUAL compiler optimizations is less aggressive on
651 strings which can potentially compare true despite having distinct
652 specialized array element types.
653 * bug fix: unit enumerations can be defined without dividing by
654 zero. (thanks to Vincent Arkesteijn)
655 * FORMAT compile-time argument count checking has been enhanced.
656 (report from Bruno Haible for CMUCL)
657 * a partial workaround for the bug 262: the compiler does not try to
658 inline-expand a local function doing RETURN-FROM from a deleted
659 BLOCK. (thanks to Peter Denno for the bug report and to David
660 Wragg for the simple test case)
661 * fixed some bugs revealed by Paul Dietz' test suite:
662 ** RENAME-PACKAGE allows all package designators as new package
664 ** constraint propagation and lambda variable substitution are
665 more cautious in dealing with partially deleted code.
666 ** compiler sometimes lost reoptimization passes.
667 ** CERROR, given a condition as condition designator, passes its
668 remaining arguments to the continue format control without
670 ** Case-altering FORMAT directives work correctly on non-ASCII
672 ** The REAL type specifier handles bounds outside the floating
673 point ranges without signalling FLOATING-POINT-OVERFLOW.
674 ** Functions with IR1-transformations can create intercomponent
675 references to global functions.
676 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
678 ** FORMAT ~R treats a nil value for its first parameter correctly.
680 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
681 * enhancement: saving cores with foreign code loaded is now
682 supported on x86/NetBSD and sparc/Linux in addition to the previously
684 * bug fix: on some platforms repeated installations caused multiple
685 copies of HTML documentation to be installed -- should not happen
686 any more. (reported by Stefan Scholl)
687 * bug fix: parsing self-recursive alien record types multiple times
688 no longer causes infinite recursion. (reported by Thomas F. Burdick,
689 original patch by Helmut Eller for CMUCL)
690 * bug fix: stack-exhaustion detection works now on NetBSD as well.
691 (thanks to Richard Kreuter)
692 * bug fix: defining classes whose accessors are methods on existing
693 generic functions in other (locked) packages no longer signals
694 bogus package lock violations. (reported by François-René Rideau)
695 * bug fix: special variables as DEFMETHOD parameters no longer have
696 associated bogus type declarations. (reported by David Wragg and
698 * bug fix: read-write consistency on streams of element-type
699 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
700 * bug fix: redefiniton of the only method of a generic function with
701 no DEFGENERIC no longer emits a full WARNING. In addition,
702 redefinition of generic functions with no DEFGENERIC to an
703 incompatible lambda list now signals an error. (thanks to Zach
705 * bug fix: DEFGENERIC now works even when there's a function of the
706 same name in an enclosing lexical environment. (thanks to Zach
708 * fixed compiler failure, caused by instrumenting code during
709 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
710 * optimization: added loop analysis and improved register allocation
711 * fixed some bugs revealed by Paul Dietz' test suite:
712 ** POSITION on displaced vectors with non-zero displacement
713 returns the right answer.
714 ** (SIMPLE-STRING) is a valid type specifier for sequence
716 ** *PRINT-LEVEL* handling for slotless structures is pedantically
718 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
719 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
721 ** The justification version of the FORMAT ~< directive treats
722 non-zero minpad parameter correctly.
724 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
725 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
726 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
727 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
728 the supported interface.
729 * new feature: Single-stepping of code compiled with DEBUG 2 or
730 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
731 * new feature: saving cores with foreign code loaded is now
732 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
733 Timothy Moore's work for CMUCL)
734 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
735 parameters to * instead of NIL if no initform is supplied.
736 (reported by Johan Bockgaard)
737 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
738 correctly when FUNCALL appears as the car of the form. Note:
739 despite this FUNCALL forms are not currently subject to
740 compiler-macro expansion. (port of Raymond Toy's fix for the
741 same from CMUCL, reported by Johan Bockgaard)
742 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
743 (thanks for Teemu Kalvas)
744 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
745 case instead of terminating the iteration (thanks for Julian
746 Squires, Sean Champ and Raymond Toy)
747 * bug fix: incorrect expansion of defgeneric that caused a style
748 warning. (thanks for Zach Beane)
749 * bug fix: slot accessor effective method computation works properly
750 for classes with multiple non-standard applicable methods on
751 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
752 * on x86 compiler supports stack allocation of results of LIST and
753 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
754 CMUCL implementation by Gerd Moellmann)
755 * fixed some bugs revealed by Paul Dietz' test suite:
756 ** FORMAT strings with both the ~<~:;~> form of the justification
757 directive and pretty-printing directives cause an error.
759 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
760 * incompatible change: the internal functions
761 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
762 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
763 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
764 instead of the old functions.
765 * new feature: on platforms where "dladdr" is available foreign
766 function names now appear in backtraces. (based on Helmut Eller's
768 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
769 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
771 * documentation: ASDF manual is now distributed with SBCL.
772 * bug fix: SBCL can now load its contributed modules with REQUIRE
773 even if the system-provided entries have been removed from
774 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
776 * bug fix: backtraces involving undefined functions or assembly
777 routines are more informative. (thanks to Brian Downing)
778 * bug fix: mutually referent alien structures now work correctly.
779 (reported by Rick Taube)
780 * bug fix: structures defined by WITH-ALIEN can be referred to
781 within other definitions in the same WITH-ALIEN.
782 * bug fix: division operators (MOD, TRUNCATE and the like) with
783 constant zero divisors and integer dividends no longer generate
785 * bug fix: provide default methods for INPUT-STREAM-P and
786 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
787 * bug fix: improve the detection and resolution of MOP metacycles
788 (where effective method computation depends on the generic function
789 having its effective method computed). (reported by Bruno Haible)
790 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
791 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
792 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
793 * bug fix: pathnames with relative directory components can be
794 represented relative to default pathnames.
795 * optimization: in taking the GCD of bignums, reduce the two bignums
796 to approximately the same size (using Euclid's algorithm) before
797 applying the more sophisticated binary GCD. (thanks to Juho
799 * optimization: COUNT on bitvectors now operates word-at-a-time.
800 * optimization: ASH with a positive, but not necessarily constant,
801 (leftwards) shift, when in a modular context, is compiled to a
803 * fixed some bugs revealed by Paul Dietz' test suite:
804 ** FORMAT variable parameters ("~V<char>") are defaulted properly
805 if the corresponding argument is NIL.
806 ** FORMAT directives accepting a minpad parameter treat negative
808 ** literal commas as character parameters to FORMAT directives are
810 ** literal spaces directly after ~<Newline> directives within a
811 format-logical-block (~:< ~@:>) do not induce :FILL-style
812 conditional newlines.
813 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
814 as specified: it no longer includes conditional newlines).
815 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
817 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
818 :PER-LINE-PREFIX argument does not evaluate to a string.
820 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
821 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
822 the manual for details; for now, package locks can be disabled by
823 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
824 if no major problems are found then it is likely that they will be
825 compiled in unconditionally.
826 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
827 unsupported operators on all platforms. To load a shared library
828 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
829 object file, link it into a shared library outside of SBCL and
830 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
832 * bug workaround: ROOM T and the GC were not working together
833 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
834 reporting facility aren't preserved. That reporting has been
835 disabled, so now until and unless someone figures out how to make it
836 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
837 * minor incompatible change: as threatened around sbcl-0.8.0, the
838 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
839 The symbols are also exported from SB-PCL for backwards
840 compatibility, but more so than before SB-PCL should be treated as
841 an implementation-internal package.
842 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
844 * the SB-SPROF contrib now works on (most) non-x86 architectures.
845 It is known as of this release not to work on the Alpha, however.
846 * fixed bug #167: errors signalled due to illegal syntax in method
847 bodies are now more legible.
848 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
849 designators and can hence be used with TYPEP.
850 * fixed bug #333: CHECK-TYPE now ensures that the type error
851 signalled, if any, has the right object to be accessed by
852 TYPE-ERROR-DATUM. (reported by Tony Martinez)
853 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
854 5.1.2.3 for multiple-value place subforms. (reported by Kalle
856 * fixed bug #334: programmatic addition of slots using specialized
857 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
858 and :ALLOCATION :CLASS slots.
859 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
860 float exponents. (rereported by Peter Seibel)
861 * fixed a bug: #\Space (and other whitespace characters) are no
862 longer considered to be macro characters in standard syntax by
864 * fixed bug: initialization of condition class metaobjects no longer
865 causes an instance of the condition to be created. (reported by
867 * fixed bug: it is now possible to have slots such that
868 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
869 description is neither :INSTANCE nor :CLASS.
870 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
871 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
872 SLOT-VALUE-USING-CLASS), and no longer causes errors with
873 non-standard SLOT-DEFINITION-ALLOCATION values.
874 * fixed bugs: various race conditions handling exiting threads.
875 CL-PPCRE's thread torture test now passes.
876 * fixed arguably-a-bug: GC time proportional to number of threads in
877 system even when most of them are idle
878 * optimization: improved performance of BIT and SBIT on bit-vectors.
879 * .fasl file incompatibility: The fasl file version number has
880 been incremented because of changes associated with package locks.
882 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
883 * minor incompatible change: the system no longer provides
884 optimization hints (conditions which are TYPEP
885 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
886 mode; these hints are still emitted when higher SPEED optimization
888 * new contrib module: a sampling profiler (profiling by statistical
889 sampling, rather than by instrumenting functions) is available as
890 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
891 * the behaviour of the standard function ED is now customizeable by
892 third parties through a hook variable: see ED's documentation
893 string for information on the protocol.
894 * the compiler no longer emits efficiency notes for (FUNCALL X)
895 when the type of X is uncertain under default optimization
897 * fixed bug 276: mutating a binding of a specialized parameter to a
898 method to something that is not TYPEP the specializer is now
900 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
901 exhibit the required behaviour on the x86 platform. (thanks to
902 Peter van Eynde, Eric Marsden and Bruno Haible)
903 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
904 even for difficult arguments. (reported by Peter Graves)
905 * fixed bug 141a: the backquote printer now descends quoted
907 * fixed another bug in backquote printing: no more destructive
908 modification of the form's list structure. (reported by Brian
910 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
911 move between its address being taken and the call to
912 interrupt_thread, fixing a crashing race condition.
913 * the SB-POSIX contrib implementation has been adjusted so that it
914 no longer exhibits ridiculously poor performance when constructing
915 instances corresponding to C structs.
917 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
918 * minor incompatible change: the sb-grovel contrib now treats C
919 structures as alien (in the sense of SB-ALIEN) objects rather than
920 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
921 has implications for memory management of client code
922 (sb-grovel-returned objects must now be manually managed) and for
923 type safety (alien objects now have full types).
924 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
925 used to control emission of compiler diagnostics, rather than the
926 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
927 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
928 quality should be considered deprecated.
929 * install.sh now installs the user manual as well
930 * (not quite a new documentable feature, but worth considering in
931 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
932 beginnings of a semantically meaningful condition hierarchy is
933 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
935 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
936 designator as the defaults argument.
937 * fixed bug: Displaced arrays whose displaced-to array has become
938 too small now cause ARRAY-DIMENSION to signal an error, providing
939 for safer bounds-checking. (reported by Bruno Haible)
940 * fixed bug: DEFCLASS slot definitions with identical :READER and
941 :WRITER names now signal a reasonable error. (reported by Thomas
943 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
945 * fixed bug: Invalid slot specification errors now print correctly.
946 (thanks to Zach Beane)
947 * fixed bug 320: Shared to local slot value transfers in class
948 redefinitions now happen corrently from superclasses as
949 well. (reported by Bruno Haible)
950 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
952 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
953 improper lists correctly. (reported by Bruno Haible)
954 * fixed bug 313: source-transform for <fun-name> was erroneously
955 applied to a call of a value of a variable with name <fun-name>.
956 (reported by Antonio Menezes Leitao)
957 * fixed bug 307: The obsolete instance protocol ensures that
958 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
959 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
960 remove dead unknown-values globs from the middle of the stack.
961 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
962 easier to resume long computations after using *BREAK-ON-SIGNALS*
963 to diagnose and fix failures. (thanks to Nikodemus Siivola)
964 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
965 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
966 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
967 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
968 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
970 * fixed a (fairly theoretical) bug in string printing: if
971 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
972 does not have array-element-type equal to the most general string
974 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
975 function name. (thanks to Nikodemus Siivola)
976 * fixed bug: compile-time format string checker failed on
977 non-closed ~{. (reported by Thomas F Burdick)
978 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
979 don't warn on legal loop code involving more than one aggregate
980 boolean. (thanks to Nikodemus Siivola)
981 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
982 in keyword position to MAKE-PACKAGE was not being reported as
984 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
985 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
987 * fixed bugs 280 and 312: the checking for multiple definitions in a
988 file is less likely to become confused by uses of inline
990 * fixed bug: the #S reader macro performs the keyword coercion
991 specified for slot names. (reported by Kalle Niemitalo)
992 * fixed bug: lambda lists may contain symbols whose names start with
993 & but are not lambda-list-keywords; their occurrence triggers a
995 * fixed bug 321: define-method-combination argument lambda lists do
996 not cause invalid code to be generated when &optional or &aux
997 variables have default values. (reported by Bruno Haible)
998 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
999 have CLOS classes; this appears also to have fixed reported
1000 instability in stack exhaustion detection.
1001 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1002 MUFFLE-WARNING when no associated restart is present is now
1004 * optimization: rearranged the expansion of various defining macros
1005 so that each expands into only one top-level form in a
1006 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1008 * optimization: used a previously little-used slot in symbols to
1009 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1011 * fixed some bugs revealed by Paul Dietz' test suite:
1012 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1014 ** Fixed an optimization bug related to inheritance of initforms
1015 from local to shared slots.
1016 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1017 ** CAST optimizer forgot to flush argument derived type.
1018 ** print/read consistency on floats is now orders of magnitude
1019 more likely. (thanks also to Bruno Haible for a similar report
1021 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1022 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1023 for objects of type REAL. Make it so.
1024 ** SXHASH is specified (infelicitously) to respect similarity,
1025 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1026 it so. (thanks to Markus Ziegler)
1027 ** on the Alpha, the compiler succeeds in compiling functions
1028 returning a known number of arguments greater than 63.
1029 ** fixed handling of invalid NIL arguments in keyword position
1031 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1033 ** provided a readably-printable representation for RANDOM-STATE
1035 ** ensured that pathnames, where they have a namestring, always
1036 print using #P"..." syntax.
1038 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1039 * Support for the forthcoming 2.0 version of the NetBSD kernel
1040 running on x86 hardware has been added. (thanks to Perry
1041 E. Metzger most immediately, and others for their past work)
1042 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1043 current release version; SBCL's OpenBSD support had been broken
1044 since about the time of OpenBSD's switch to ELF binary format.)
1045 * [placeholder for DX summary]
1046 ** user code with &REST lists declared dynamic-extent, under high
1047 speed or space and low safety and debug optimization policy.
1048 * The manual has been converted to Texinfo format and the debugger
1049 chapter from the cmucl manual has been added.
1050 * A facility has been added to extract documentation strings from
1051 sbcl and store them as Texinfo-formatted snippets for inclusion in
1052 the manual (via Texinfo's @include directive)
1053 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1054 the test case to Dave Roberts)
1055 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1056 pointers. (reported by Sean Ross)
1057 * bug fix: PROFILE output is printed nicely even for large numerical
1058 values. (thanks to Zach Beane)
1059 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1060 greater than 32 handle EOF correctly.
1061 * bug fix: on X86 an immediate argument of the IMUL instruction is
1062 correctly printed in disassembly. (thanks to Lutz Euler)
1063 * bug fix: class slots in redefined classes preserve their old
1064 values. (thanks to Bruno Haible and Nikodemus Siivola)
1065 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1066 succeeds. (reported by Marco Baringer)
1067 * fixed some bugs revealed by Paul Dietz' test suite:
1068 ** READ-SEQUENCE now works on ECHO-STREAMs.
1069 ** RATIONALIZE works more according to its specification. (thanks
1072 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1073 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1074 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1075 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1076 should matter to you only if you rebind the printer control
1077 variables and then find you want different bindings in the
1078 debugger than in the ordinary execution of your program.)
1079 * The runtime build system has been tweaked to support building
1080 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1081 assembler and linker. (thanks to Nikodemus Siivola)
1082 * Unbound, undefined, undeclared variables now trigger full
1083 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1084 more often programmer error than deliberate exploitation of undefined
1086 * optimization: the hash algorithm for strings has changed to one
1087 that is less vulnerable to spurious collisions. (thanks to Juho
1089 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1090 needless bounds checking. (thanks to Juho Snellman)
1091 * optimization: implemented multiplication as a modular
1092 (UNSIGNED-BYTE 32) operation on the PPC backend.
1093 * fixed some bugs revealed by Paul Dietz' test suite:
1094 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1096 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1097 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1098 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1100 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1102 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1104 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1105 their output stream on EOF from read.
1106 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1107 have been read to end-of-file.
1108 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1110 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1111 description of determination of which consecutive characters
1113 ** Printing the "Space" character with escaping on now yields "#\\ ",
1114 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1115 ** Reading floating-point numbers with *READ-BASE* set to a number
1116 less than 10 works correctly.
1117 ** Reading floating-point numbers with *READ-BASE* set to a number
1118 more than 10 works correctly.
1119 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1120 the readtable currently in effect.
1122 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1123 * minor incompatible change: parsing of namestrings on a physical
1124 (Unix) host has changed; numbers after the final #\. in a
1125 namestring are no longer interpreted as a version field. This is
1126 intented to be largely invisible to the user, except that the
1127 meaning of the namestring "*.*.*" has changed: it now refers to a
1128 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1129 should usually be replaced by
1130 (make-pathname :name :wild :type :wild :version :wild)
1131 with the added benefit that this is more likely to be portable.
1132 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1133 option to OPEN now signals an error if the file being opened
1134 exists; this may have an impact on existing code.
1135 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1136 signals received, on the PowerPC platforms (both Linux and
1137 Darwin). (thanks to Pierre Mai for pointing out the location of
1139 * several fixes on OS X: The system now builds and runs cleanly on
1140 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1141 (thanks to Brian Mastenbrook)
1142 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1143 second-resolution: integer multiples of 1/3600 between -24 and 24.
1144 (thanks to Vincent Arkesteijn)
1145 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1146 type when called with 1 argument; PEEK-CHAR checked type of
1147 PEEK-TYPE only after having read first character from a
1148 stream. (reported by Peter Graves)
1149 * bug fix: the garbage collector now has much better locality
1150 behaviour, and in particular no longer treats all memory as being
1151 exhausted when many small objects point to each other in a deeply
1153 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1154 recognized as being TYPEP their class.
1155 * bug fix: the PUSHNEW documentation string has been corrected.
1156 (thanks to Vincent Arkesteijn)
1157 * bug fix: defaulting of the value for the last of an atypically
1158 large number of multiple values being bound was not being
1159 performed correctly on the Alpha or PPC platforms
1160 * optimization: implemented multiplication as a modular
1161 (UNSIGNED-BYTE 32) operation on the x86 backend.
1162 * optimization: SEARCH on simple-base-strings can now be open-coded.
1163 (see also contrib/compiler-extras.lisp for inspiration for
1164 teaching the compiler about the Boyer-Moore algorithm).
1165 * value, returned by MAX (and MIN) called with several EQUALP, but
1166 not EQL, arguments now does not depend on compiler settings.
1167 * fixed some bugs revealed by Paul Dietz' test suite:
1168 ** in stack analysis liveness information is propagated from
1169 non-local entry points.
1170 ** pathwise CAST removing failed when the CAST node did not start
1172 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1173 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1175 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1176 host is already defined.
1177 ** RENAME-FILE works on streams instead of signalling an internal
1179 ** PEEK-CHAR uses the current readtable when determining whether
1180 or not a character is whitespace.
1181 ** MERGE-PATHNAMES handles the case when the pathname does not
1182 specify a name while the default-pathname specifies a version.
1183 ** Pathnames now stand a better chance of respecting print/read
1185 ** Attempting to use standardized file system operators with a
1186 pathname with invalid :DIRECTORY components signals a
1188 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1190 ** DIRECTORY on logical pathnames is more correct.
1191 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1192 signal a TYPE-ERROR if their argument is not a stream.
1193 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1194 designator argument does not designate a stream.
1195 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1196 examining the synonym.
1197 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1199 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1200 element-type larger than ([UN]SIGNED-BYTE 32).
1202 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1203 * When built with the :SB-FUTEX feature, threaded builds now take
1204 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1205 for faster/more reliable mutex and condition variable support.
1206 * Incompatible change (but one you probably shouldn't have been using
1207 anyway): the interface and code for arbitrating between multiple
1208 threads in the same user session has been redesigned.
1209 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1210 error when called without an explicit environment argument.
1211 (thanks to Nikodemus Siivola)
1212 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1213 confused about their position. (thanks to Adam Warner and Gerd
1215 * bug fix: Pretty printing backquoted forms with unquotations in the
1216 argument list position of various code constructs such as LAMBDA
1217 now works correctly. (reported by Paul Dietz)
1218 * bug fix: Pretty printing unquotations no longer loses all
1219 stream position information.
1220 * optimization: performance of string output streams is now less
1221 poor for multiple small sequence writes.
1222 * optimization: performance of CSUBTYPEP in the presence of complex
1223 expressions involving CONS and NOT many times has been improved.
1224 (reported by Paul Dietz)
1225 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1227 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1228 streams for the REPL, for the debugger, and so on. (thanks to
1230 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1231 * fixed some bugs revealed by Paul Dietz' test suite:
1232 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1233 is ignored if it is too small, rather than propagating through
1234 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1235 ** extremely complex negations of CONS types were not being
1236 sufficiently canonicalized, leading to inconsistencies in
1238 ** VALUES tranformer lost derived type.
1240 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1241 * fixed a bootstrapping bug: the build process no longer assumes
1242 that the various BOOLE-related constants have the same value in
1243 host and target lisps. (noted by Paul Dietz' test suite on an
1244 SBCL binary built from CLISP)
1245 * The system can now be dynamically linked on the MIPS platform,
1246 which enables dynamic loading of foreign code from Lisp. (thanks
1247 to Ralf Baechle for discussions on the MIPS ABI)
1248 * The system now records debugging information for its own source
1249 files in a filesystem-position-independent manner, relative to
1250 the "SYS" logical host.
1251 * fixed a compiler bug: MV-LET convertion did not check references
1252 to the "max args" entry point. (reported by Brian Downing)
1253 * tweaked disassembly notes to be less confident about proclaiming
1254 some instruction as an LRA. (thanks to Brian Downing)
1255 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1256 now each have their own history, command character, and other
1257 characteristics. (thanks to David Lichteblau)
1258 * fixed some bugs revealed by Paul Dietz' test suite:
1259 ** compiler failure in compiling LOGAND expressions including a
1261 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1262 shift greater than 32.
1263 ** FLUSH-DEST did not mark blocks for type check regeneration.
1264 ** HANDLER-CASE failed to accept declarations in handler clauses
1265 in some circumstances.
1267 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1268 * New code in contrib/sb-introspect (still probably not entirely
1269 stable yet) provides some support for smart Lisp development
1270 environments like SLIME.
1271 * The conditions signalled for errors occurring when loading .fasl
1272 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1273 in a way which should help ASDF recover gracefully.
1274 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1275 stuff has been cleaned up. If you code contrib/ stuff, this might
1276 affect you, and you can look at contrib/README, contrib/STANDARDS,
1277 and/or the 0.8.4.27 diff to check.
1278 * In full calls the compiler now does not generate checks for declared
1279 argument types for all arguments.
1280 * various threading fixes
1281 ** and some experimental patches which didn't make it into
1282 the main tree for this release, but which are shipped in
1283 contrib/experimental-thread.patch as a possible fix for some
1284 failures (deadlock, spinning...) in GC-intensive multithreaded
1286 * fixed PPC build problem (source code incompatibility of different
1287 library versions): added offsetof() hackery which attempts to divine
1288 where glibc maintainers put uc_mcontext today
1289 * fixed bug 282: compiler does not trust type assertions while passing
1290 arguments to a full call.
1291 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1292 &OPTIONAL VALUES type parameter.
1293 * fix bug 214: algorithm for noting rejected templates is now more
1294 similar to that of template seletion. (also reported by rydis on
1296 * fixed bug 141b: printing backquoted information readably and prettily
1297 inserts a space where necessary.
1298 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1299 (CHARACTER 10) are now reported as errors, rather than propagated
1300 as unknown types. (reported by piso on #lisp)
1301 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1302 with values NIL and :ERROR. (thanks to Milan Zamazal)
1303 * fixed bug 191c: CLOS now does proper keyword argument checking as
1304 described in CLHS 7.6.5 and 7.6.5.1.
1305 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1306 longer raise an error; further, using a list as a for-as-arithmetic
1307 counter now raises a meaningful error.
1308 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1309 now understood by sequence creation functions such as MAKE-SEQUENCE
1311 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1312 a STRING-INPUT-STREAM.
1313 * compiler enhancement: SIGNUM is now better able to derive the type
1315 * type declarations inside WITH-SLOTS are checked. (reported by
1317 * fixed some bugs revealed by Paul Dietz' test suite:
1318 ** incorrect optimization of TRUNCATE for a positive first
1319 argument and negative second.
1320 ** compiler failure in let-convertion during flushing dead code.
1321 ** compiler failure while deriving type of TRUNCATE on an
1322 interval, containing 0.
1323 ** ASH of a negative bignum by a negative bignum count now returns
1325 ** intersection of CONS types now canonicalizes properly, fixing
1326 inconsistencies in SUBTYPEP.
1328 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1329 * incompatible change: The --disable-debugger command line
1330 option now clobbers the debugger at a more fundamental
1331 level, by redefining #'INVOKE-DEBUGGER instead of by
1332 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1333 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1334 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1335 enter the debugger REPL and then suffer recursive errors
1336 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1337 while under the new system, BREAK in --disable-debugger mode
1338 terminates the system just as an unhandled error would.
1339 * fixed compiler performance when processing loops with a step >1;
1340 * bug fix: DOCUMENTATION now retrieves generic function
1341 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1342 support has been systematized, and now supports the methods
1343 specified by ANSI, along with a default method and a method for
1344 slot documentation. (reported by Nathan Froyd)
1345 * bug fix: effective methods associated with a generic function are
1346 no longer cached over a change of that generic function's method
1347 combination. (reported by Andreas Fuchs)
1348 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1349 names a non-existent directory. (thanks to Andreas Fuchs)
1350 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1351 the PPC platform, lead to stack corruption; this has been fixed.
1352 (reported by Rainer Joswig)
1353 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1354 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1355 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1356 arithmetic in general, is now much more reliable on the PPC
1358 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1359 platform now returns the right answer.
1360 * optimization: restored some effective method precomputation in
1361 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1362 precomputation is now tunable.
1363 * optimization: compiler-internal data structure use has been
1364 reviewed, and changes have been made that should improve the
1365 performance of the compiler by about 20%.
1366 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1367 simple-base-strings and simple-bit-vectors is improved.
1368 * optimization: the optimization of 32-bit logical and arithmetic
1369 functions introduced in version 0.8.3 on the x86 has been
1370 implemented on the mips, ppc and sparc platforms; an
1371 implementation of the same facility, but for 64-bit arithmetic,
1372 has been added for the alpha.
1373 * microoptimization: the compiler is better able to make use of the
1374 x86 LEA instruction for multiplication by constants.
1375 * bug fix: in some situations compiler did not report usage of
1376 generic arithmetic in (SPEED 3) policy.
1377 * bug 145b fix: compiler used wrong type specifier while converting
1378 MEMBER-types to numeric.
1379 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1381 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1382 index argument now works properly on the Alpha platform.
1383 * bug fix: floating point exception treatment on the Alpha platform
1385 * bug fix: FILE-POSITION works much better on string input and
1386 output streams. (thanks to Nikodemus Siivola)
1387 * bug fix: many threading/garbage collection symptoms sorted.
1388 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1389 might be pseudo-atomic.
1390 * internal change: Stopping for GC is now done with signals not
1391 ptrace. GC is now done in whichever thread wanted it, instead of
1393 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1395 * bug fix: result form in DO is not contained in the implicit
1397 * incompatible change: ICR structure is changed; the value part of
1398 CONTINUATION is now called LVAR; corresponding functions are
1399 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1400 * added type deriver for ISQRT (thanks to Robert E. Brown).
1401 * bug fix: better support for loading from the command line when an
1402 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1404 * fixed some bugs revealed by Paul Dietz' test suite:
1405 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1406 ** ROUND and FROUND now give the right answer when given very
1407 small float arguments.
1408 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1410 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1411 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1412 ** LCM should return a non-negative integer.
1413 ** PARSE-INTEGER returned the index of a terminator instead of the
1414 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1415 ** PARSE-INTEGER returned an incorrect index being applied to a
1417 ** LCM with two arguments of 0 returns 0 rather than signalling
1419 ** unsigned addition of a 32-bit constant with the high bit set no
1420 longer causes an internal compiler error on the x86.
1421 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1422 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1423 with negative last argument.
1424 ** byte specifiers with zero size and position no longer cause
1425 an error during type derivation.
1426 ** bignum multiplication on the Alpha platform now returns the
1428 * porting: The system now builds on SuSE AMD64, although it still
1429 generates a 32-bit binary.
1430 * .fasl file incompatibility: The fasl file version number has
1431 been incremented (because of the changes to internal compiler
1432 data structures referred to above).
1434 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1435 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1436 more accurately, on the Darwin kernel running on PowerPC hardware.
1437 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1438 * Compiler code deletion notes now signal a condition of type
1439 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1440 an associated MUFFLE-WARNING restart.
1441 * The compiler now performs limited argument count validation of
1442 constant format strings in FORMAT, and where appropriate in ERROR,
1443 CERROR and WARN. (thanks to Gerd Moellmann)
1444 * New ASDF-INSTALL contrib can be used for automatic download and
1445 installation of third-party Lisp code from CCLAN or other sites
1447 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1448 which forces another thread to execute a function supplied by the
1450 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1451 now accept and act upon their :ELEMENT-TYPE keyword argument.
1452 (reported by Martin Atzmueller, Edi Weitz)
1453 * bug fix: FILE-POSITION now accepts position designators up to
1454 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1455 is the greater. (thanks to Patrik Nordebo)
1456 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1457 Kalle Olavi Niemitalo)
1458 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1459 printable. (reported by Eric Marsden)
1460 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1461 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1462 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1463 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1464 (thanks to Stig Sandoe)
1465 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1466 UNREAD-CHAR now correctly includes the unread character in the
1467 target sequence. (thanks to Gerd Moellmann)
1468 * bug fix in threaded builds: the system can now be suspended and
1469 resumed by shell job control with minimal disruption.
1470 * bug fixes in times and timezones >2038AD
1471 * better handling of "where is GNU make?" problem in build scripts
1472 (thanks to Nikodemus Siivola)
1473 * new optimization: inside a named function any reference to a
1474 function with the same name is considered to be a self-reference;
1475 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1476 optimization quality.
1477 * new optimization on x86: logical functions and + now have
1478 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1479 used when the result is truncated to 32 bits.
1480 * VALUES declaration is partially enabled.
1481 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1482 * fixed some bugs revealed by Paul Dietz' test suite:
1483 ** The system now obeys the constraint imposed by
1484 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1485 types form a lattice under type intersection.
1486 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1487 ** ASSOC now ignores NIL elements in an alist.
1488 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1489 and (1+ MOST-POSITIVE-FIXNUM) answers.
1490 ** The addition of a method with invalid qualifiers to a generic
1491 function does not cause an error to be signalled immediately;
1492 a warning is signalled, and the error is generated only on
1493 calling the generic function.
1494 * changed .fasl file version number, in part to add type codes for
1495 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1496 obscure ANSI requirements
1498 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1499 * fixed bug 148: failure to inline-expand a local function left
1500 garbage, confusing the compiler.
1501 * fixed bugs 3cd: structure slot readers perform type check if the
1502 slot can have an invalid value (i.e. it is either not initialized
1503 or can be written with a less specific slot writer).
1504 * bug fix: the compiler now traps array references to elements off
1505 the end of an array; previously, the bounds checking in some
1506 circumstances could go off-by-one.
1507 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1509 * type declarations for array element types now obey the description
1510 on the CLHS page "Declaration TYPE", as per discussions on
1511 sbcl-help around 2003-05-08. This means that a declaration
1512 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1513 declaration, all references to BAR will be asserted or assumed
1514 (with THE, so dependent on compiler policy) to involve objects of
1515 type FOO. Note that no such declaration is implied in
1516 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1517 * declared types of functions from the "Conditions"
1518 chapter. (reported by Paul Dietz)
1519 * bug fix: CERROR accepts a function as its first argument.
1520 * bug fix: NTH an NTHCDR accept a bignum as index
1521 arguments. (reported by Adam Warner)
1522 * optimization: character compare routines now optimize comparing
1523 against a constant character. (reported by Gilbert Baumann)
1524 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1525 argument now works properly on the MIPS platform.
1526 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1527 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1528 declared to have a functional type, does not cause an error
1530 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1531 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1532 * fixed bug: if last continuation of a deleted block has a
1533 destination, this destination should be deleted too. (reported by
1535 * fixed a bug in the bootstrap process: the host compiler's values
1536 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1537 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1538 test case from Patrik Nordebo)
1539 * improved the ability of the disassembler on the PPC platform to
1540 provide helpful disassembly notes.
1541 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1542 the class in more cases than previously.
1543 * bug fix: FILE-POSITION now understands :START and :END for
1544 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1545 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1546 8) data. (thanks to David Lichteblau)
1547 * bug fix: it is possible to add a method to a generic function
1548 without lambda list.
1549 * bug fix: reader failed to signal END-OF-FILE inside an
1550 object representation. (reported by Nikodemus Siivola)
1551 * fixed some bugs revealed by Paul Dietz' test suite:
1552 ** LAST and [N]BUTLAST should accept a bignum.
1553 ** condition slot accessors are methods.
1554 ** (VECTOR NIL) is a subtype of STRING.
1556 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1557 * minor incompatible change: some nonsensical specialized lambda
1558 lists (used in DEFMETHOD) which were previously ignored now signal
1560 * minor incompatible change: the system is now aware of the types of
1561 variables in the COMMON-LISP package, and will signal errors for
1562 most violations of these type constraints (where previously they
1563 were silently accepted).
1564 * minor incompatible change: COMPILE-FILE now uses the freedom
1565 afforded (ANSI 3.2.2.3) to use derived function types for
1566 functions defined in the same file. This also permits the system
1567 to warn on static type mismatches and function
1568 redefinition. (Currently it does not work with high DEBUG level.)
1569 * minor incompatible change: VALUES declaration is disabled.
1570 * When issuing notes, the compiler now signals a condition of type
1571 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1572 restart for use in user handlers. It is expected that the
1573 COMPILER-NOTE condition will eventually become a condition
1574 supertype to a hierarchy of note types, which will then be
1575 handleable in a similar fashion. However, at the moment, no such
1576 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1577 but it's an implementation detail, not a classification for the
1579 * Changes in type checking closed the following bugs:
1580 ** type checking of unused values (192b, 194d, 203);
1581 ** template selection based on unsafe type assertions (192c, 236);
1582 ** type checking in branches (194bc).
1583 * A short form of VALUES type specifier has ANSI meaning (it has
1584 increased the number of situations when SBCL cannot perform type
1586 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1587 keywords or constants is permissible.
1588 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1589 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1590 classes. (thanks to Antonio Martinez)
1591 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1592 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1593 argument to a pretty-print stream. Instead, it leaves any such
1594 operations to DESCRIBE-OBJECT methods.
1595 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1597 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1598 lambda lists. (thanks to David Lichteblau)
1599 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1600 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1601 (thanks to Gerd Moellmann)
1602 * fixed bug 246: increased compilation speed of long
1603 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1605 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1607 * DEFCONSTANT now throws a condition of type
1608 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1609 constant to a non-EQL value; CONTINUE and ABORT restarts
1610 respectively change and preserve the value.
1611 * fixed bug 63: The code walker, part of the implementation of CLOS,
1612 is now better at handling symbol macros.
1613 * bug fix: There is no longer an internal implementation type named
1614 CL:LENGTH. (reported by Raymond Toy)
1615 * bug fix: In macro-like defining macros/special operators the
1616 implicit block does not enclose the lambda list.
1617 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1618 enclosed in parentheses) are not suitable as type specifiers, and
1619 their use properly signals an error now.
1620 * bug fix: An argument count mismatch for a type specifier in code
1621 being compiled no longer causes an unhandled error at compile
1622 time, but signals a compile-time warning.
1623 * fixed simple vector readable printing
1624 * bug fix: DESCRIBE takes more care over whether the class
1625 precedence list slot of a class is bound before accessing it.
1626 (reported by Markus Krummenacker)
1627 * bug fix: FORMATTER can successfully compile pretty-printer format
1628 strings which use variants of the ~* directive inside.
1629 * bug fix: SEARCH now applies its TEST predicate to the elements of
1630 the arguments in the correct order. (thanks to Wolfhard Buss)
1631 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1632 in type inference. (thanks to Robert E. Brown)
1633 * bug fix: Reading in symbols with an explicit package name of ""
1634 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1635 * fixed some bugs revealed by Paul Dietz' test suite:
1636 ** NIL is now allowed as a structure slot name.
1637 ** Arbitrary numbers, not just REALs, are allowed in certain
1638 circumstances in LOOP for-as-arithmetic clauses.
1639 ** Multiple class redefinitions before slot access no longer
1640 causes a type error.
1641 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1642 association between the name and a class.
1643 ** Generic functions with non-standard method-combination and over
1644 five methods all of which return constants no longer return NIL
1645 after the first few invocations. (thanks to Gerd Moellmann)
1646 ** CALL-NEXT-METHOD with no arguments now passes the original
1647 values of the arguments, even in the presence of assignment.
1648 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1650 ** DISASSEMBLE works with closures and funcallable instances.
1651 ** ADD-METHOD now returns the generic function, not the new method.
1652 ** FIND-METHOD signals an error if the lengths of the specializers
1653 is incompatible with the generic function, even if the ERRORP
1655 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1656 which its argument is a member.
1657 ** DEFCLASS only redefines the class named by its class-name
1658 argument if that name is the proper name of the class;
1659 otherwise, it creates a new class.
1660 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1661 of the UNBOUND-SLOT condition to the name of the slot.
1662 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1663 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1664 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1665 SLOT-MAKUNBOUND in the specified fashion.
1667 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1668 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1669 cross-compilation host. As a consequence, we can now bootstrap our
1670 way up to SBCL starting with a bare gcc toolchain and human-readable
1671 source code (first the source to CLISP, then the source to SBCL).
1672 * A contributed module containing a partial implementation of the
1673 simple-streams interface has been included. (thanks to Rudi
1675 * A contributed module implementing the RFC1321 Message Digest
1676 Algorithm, known as MD5, has been included.
1677 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1678 no longer has any effect, as the code controlled by this feature
1679 has been deleted. (As far as we know, no-one has ever built using
1680 this feature, and its semantics were confused in any case).
1681 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1682 work on conditions (as required by the ANSI specification),
1683 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1684 expected behaviour on conditions. Users should note, however,
1685 that such behaviour is not required by the ANSI specification,
1686 and so use of this behaviour may render their code unportable.
1687 * fixed some bugs revealed by Paul Dietz' test suite:
1688 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1690 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1691 ** SXHASH on condition objects no longer returns NIL.
1692 ** :ALLOCATION :CLASS slots are better treated; their values are
1693 updated on class redefinition, and initforms inherited from
1694 superclasses are applied.
1695 ** REMOVE-METHOD returns its generic function argument even when
1696 no method was removed.
1697 ** SHARED-INITIALIZE now initializes the values of the requested
1698 slots, including those with :ALLOCATION :CLASS.
1699 ** ALLOCATE-INSTANCE now works on structure classes defined via
1700 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1702 ** SLOT-EXISTS-P now works on conditions, as well as structures
1704 ** MAKE-LOAD-FORM now has the required methods on
1705 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1706 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1707 keyword, and now implements the SLOT-NAMES argument.
1708 ** methods with &OPTIONAL arguments no longer allow too many
1709 arguments to be passed in the call without error.
1710 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1711 option is consistent with the required arguments of the generic
1712 function lambda list.
1713 * bug fix: REQUIRE accepts a string designator. (Thanks to
1715 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1716 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1717 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1718 * bug fix: The compiler now checks for duplicated variables in macro
1720 * bug fix: SETQ on globals returns the correct value.
1721 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1722 not a COMPILER-ERROR (followed by some other strange error on
1723 choosing the CONTINUE restart).
1724 * bug fix: make.sh and friends are now more consistent in the way that
1725 they look for GNU "make".
1727 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1728 * experimental native threads support (on x86 Linux >=2.4 only).
1729 This is not compiled in by default: you need to add :SB-THREAD to
1730 the target features. See the "Beyond ANSI" chapter of the manual
1732 * fix for longstanding nonANSIism: The old distinction between
1733 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1734 The return value from CL:FIND-CLASS is now a CLOS class, and
1735 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1736 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1737 * An interface to the MetaObject Protocol, as described in Kiczales,
1738 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1739 MIT Press, 1991, is available from the SB-MOP package.
1740 * incompatible change: the SB-PCL package should now be considered
1741 a private implementation detail, and no longer a semi-private MOP
1743 * minor incompatible change: due to rearrangement for threads, the
1744 control stack and binding stack are now allocated at arbitrary
1745 addresses instead of being hardcoded per-port. Users affected by
1746 this probably have to be doing advanced things with shared
1747 libraries, and will know who they are.
1748 * minor incompatible change: Previously, all --eval forms used were
1749 processed with READ before any of them were processed with EVAL.
1750 Now each --eval form is processed with both READ and EVAL before
1751 the next --eval form is processed. (Thus package operations like
1752 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1753 work as the user might reasonably expect.)
1754 * minor incompatible change: *STANDARD-INPUT* is now only an
1755 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1757 * minor incompatible change: Y-OR-N-P is now character-oriented,
1758 not line oriented. Also, YES-OR-NO-P now works without errors.
1759 (thanks to Antonio Martinez)
1760 * sb-aclrepl module improvements: an integrated inspector, added
1761 repl features, and a bug fix to :trace command.
1762 * Known functions, which cannot be open coded by the backend, are
1763 considered to be able to check types of their arguments. (fixing
1764 a bug report by Nathan J. Froyd)
1765 * fixed a bug in computing method discriminating functions: It is
1766 now possible to define methods specialized on classes which have
1767 forward-referenced superclasses. (thanks to Gerd Moellmann)
1768 * fixed evaluation order in optional entries (reported by Gilbert
1770 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1771 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1772 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1773 required-by-AMOP NAME argument, as well as CLASS and
1774 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1775 * fixed bug 20: DEFMETHOD can define methods using names that are
1776 not the proper names of classes to designate class specializers.
1777 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1778 Unix file descriptors, instead of blowing up. (thanks to Antonio
1780 * Garbage collection refactoring: user-visible change is that a
1781 call to the GC function during WITHOUT-GCING will not do garbage
1782 collection until the end of the WITHOUT-GCING. If you were doing
1783 this you were probably losing anyway.
1784 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1785 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1786 TYPEP the latter but not the former.
1787 * The compiler issues a full WARNING for calls to undefined functions
1788 with names from the CL package.
1789 * MAP-INTO for a vector destination is open coded. (reported by
1790 Brian Downing on c.l.l)
1791 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1792 documentation string.
1793 * fixed some bugs revealed by Paul Dietz' test suite:
1794 ** COPY-ALIST now signals an error if its argument is a dotted
1796 ** Condition slots are now accessed more correctly in the presence
1797 of multiple initargs for a given slot.
1798 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1799 exclude restarts of the same name associated with a different
1801 ** DEFCLASS of forward-referenced classes with another
1802 forward-referenced class in the superclasses list no longer
1804 ** Condition slots are now initialized once each, not multiple
1805 times. (thanks to Gerd Moellmann)
1806 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1808 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1809 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1810 arguments contain duplicated elements.
1811 ** RESTART-CASE understands local macros.
1812 ** RESTART-CASE associates exactly its own restarts with a condition.
1813 ** ENDP in safe mode checks its argument to be of type LIST.
1814 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1815 in question is unbound.
1816 ** Optimized MAKE-INSTANCE functions no longer cause internal
1817 assertion failures in the presence of duplicate initargs.
1818 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1819 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1820 propagated even in the ctor optimized implementation of
1822 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1824 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1825 * a better implementation of SXHASH on (simple) bit vectors,
1826 measured both in execution speed and in distribution of results
1827 over the positive fixnums, has been installed. Likewise, a better
1828 implementation of EQUAL for simple bit vectors is now available.
1829 * fixed CEILING optimization for a divisor of form 2^k.
1830 * fixed bug 240 (emitting extra style warnings "using the lexical
1831 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1832 by Antonio Martinez)
1833 * fixed SXHASH, giving different results for NIL depending on type
1834 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1835 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1836 variable. (found by Rolf Wester)
1837 * SBCL does not ignore type declarations for special
1838 variables. (reported by rif on c.l.l 2003-03-05)
1839 * some bug fixes in contrib/sb-aclrepl/
1840 * fixed some bugs revealed by Paul Dietz' test suite:
1841 ** a bug in the CONS type specifier, whereby the CAR and CDR
1842 types got intertwined, has been fixed;
1843 ** the type system is now able to reason about the interaction
1844 between INTEGER and RATIO types more completely;
1845 ** APPEND, [N]REVERSE and NRECONC check that those their
1846 arguments, which must be proper lists, are really so;
1847 ** An array specialized to be unable to hold elements has been
1848 implemented, as required -- yes, really -- by ANSI;
1849 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1850 on malformed property lists;
1852 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1853 * incompatible packaging change: in line with Unix convention,
1854 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1855 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1856 * REQUIRE and PROVIDE are now optionally capable of doing something
1857 useful. See the documentation string for REQUIRE.
1858 * infrastructure for a managed SBCL contrib system: contributed
1859 modules in this release include:
1860 ** the ASDF system definition facility;
1861 ** an interface to the BSD Sockets API;
1862 ** an ACL-like convenience interface to the repl;
1863 (thanks to Kevin Rosenberg)
1864 ** an implementation of ROTATE-BYTE, with efficient implementation
1866 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1867 gives a closer approximation to the right answer than previously.
1868 (thanks to Raymond Toy)
1869 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1870 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1871 optional environment arguments, as required by ANSI.
1872 * fixed bugs in other functions taking environment objects, allowing
1873 calls with an explicit NIL environment argument to be compiled
1875 * fixed bug 228: primary return values from
1876 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1877 COMPILE or FUNCTION.
1878 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1879 :TYPE will no longer signal errors on innocuous objects.
1880 * fixed bug 231b: SETQ is better at respecting type declarations in
1881 the lexical environment.
1882 * fixed a bug in DEFCLASS: classes named by symbols with no or
1883 unprintable packages can now be defined.
1884 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1885 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1886 * fixed compiler failure related to checking types of functions.
1887 (reported by Robert E. Brown)
1888 * the compiler is now much more consistent in its error-checking
1889 treatment of bounding index arguments to sequence functions: in
1890 (SAFETY 3) code, errors will be signalled in almost all cases if
1891 invalid sequence bounding indices are passed to functions defined
1892 by ANSI to operate on sequences.
1893 * fixed a bug in the build procedure: documentation of SBCL-specific
1894 packages is now preserved and available in the final Lisp image.
1895 * lifted FDEFINITION lookup out of loops in the implementation of
1896 many list operations. (thanks to Robert E. Brown)
1897 * fixed a bug in the reader: the #n# reader macro now works for
1898 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1899 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1900 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1901 E. Brown and Tony Martinez respectively)
1902 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1903 SLOT-BOUNDP: in particular, it is now able to optimize them much
1904 better, and is now not vulnerable to having packages renamed.
1905 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1906 always called when a slot is not present in an instance. (thanks
1908 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1909 longer signal an error on unreading a character following EOF on
1910 the previous constituent stream. (thanks to Tony Martinez)
1911 * fixed some bugs revealed by Paul Dietz' test suite:
1912 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1913 not just nonnegative fixnums;
1914 ** the logical bit-array operators such as BIT-AND now accept an
1915 explicit NIL for their "opt-arg" argument (to indicate a
1916 freshly-consed result bit-array);
1917 ** ELT now signals an error on an invalid sequence index in safe
1919 ** the type system is now cleverer about negations of numeric
1920 types, and consequently understands the BIGNUM and RATIO types
1922 ** the type system is now cleverer about the interaction between
1923 INTEGER and RATIO types: while bugs still remain, many more
1924 cases are accurately computed;
1925 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1926 if it is in the last clause;
1927 ** CONSTANTLY now correctly returns a side-effect-free function in
1929 ** DECLARE is no longer treated as a special-operator; in
1930 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1931 * incremented fasl file version number due to the change in the
1932 DEFSTRUCT-SLOT-DESCRIPTION structure.
1934 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1935 * minor incompatible change: code processed by the "interpreter" or
1936 EVAL now has a compilation optimization policy of (DEBUG 2)
1937 (changed from (DEBUG 1)) to improve debuggability of interactive
1938 development, and to allow the use of the debug RETURN command in
1940 * an experimental implementation of the RETURN command for the
1941 debugger has been included. (thanks to Frederik Kuivinen)
1942 * fixed bug 62: constraints were not propagated into a loop.
1943 * fixed bug in embedded calls of SORT (reported and investigated by
1945 * fixed some bugs revealed by Paul Dietz' test suite:
1946 ** printing and reading of arrays with some dimensions having
1947 length 0 (thanks to Gerd Moellmann);
1948 ** BOA constructor with &AUX argument without a default value does
1949 not cause a type error;
1950 ** CONSTANTP now returns true for all self-evaluating objects.
1952 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1953 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1954 accessors that are related by inheritance, as specified in the
1955 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1956 Valtteri Vuorikoski)
1957 * The compiler is now able to inline functions that were defined in
1958 a complex lexical environment (e.g. inside a MACROLET).
1959 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1961 * Support for the upcoming FreeBSD-5.0 release has been included.
1962 (thanks to Dag-Erling Smorgrav)
1963 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1964 effect when it is not in a toplevel context.
1965 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1966 stand a better chance of being correct. (thanks to Gerd
1968 * fixed bug in COERCE, which now signals an error on coercing a
1969 rational to a bounded real type which excludes the expected
1971 * The compiler is now able to derive types more accurately from the
1972 COERCE and COMPILE functions.
1973 * fixed bug 223: functional binding is considered to be constant
1974 only for symbols in the CL package.
1975 * fixed bug 231: SETQ did not check the type of a variable being set
1976 (reported by Robert E. Brown)
1977 * A new optimization for MAKE-INSTANCE has been included, fixing
1978 various bugs (including relating to :ALLOCATION :CLASS slots and
1979 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1981 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1982 ** As required by ANSI, LOOP now disallows anonymous collection
1983 clauses such as COLLECT I in conjunction with aggregate boolean
1984 clauses such as THEREIS (= I 1);
1985 ** LOOP now signals an error when any variable is reused in the
1986 same loop (including the potentially useful construct analogous
1987 to WITH A = 1 WITH A = (1+ A);
1988 ** IT is only a special loop symbol within the first clause of a
1989 conditional loop clause;
1990 ** LOOP with a typed iteration variable over a hashtable now
1991 signals a type error iff it should.
1992 * fixed some other bugs revealed by Paul Dietz' test suite:
1993 ** FILE-STREAM now names the class previously known as FD-STREAM;
1994 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1995 argument) no longer signals an error;
1996 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1997 of the slot symbol, rather than using the current package
1998 ((:CONC-NAME "") continues to intern the slot's name in the
2000 * incremented fasl file version number, because of the incompatible
2001 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2002 of the new implementation of DEFINE-COMPILER-MACRO.
2004 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2005 * Support for building SBCL for MIPS platforms running in
2006 little-endian mode has now been checked in, and basic
2007 functionality on said platforms verified.
2008 * minor incompatible change: PCL now records the pathname of a file
2009 in which methods and the like are defined, rather than its
2011 * minor incompatible change: TRUENAME now considers the truename of
2012 a file naming a directory to be the pathname with :DIRECTORY
2013 component indicating that directory.
2014 * minor incompatible change: a NAMED clause in the extended form of
2015 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2016 reason for the previous behaviour is unclear.
2017 * more systematization and improvement of CLOS and MOP conformance
2018 in PCL (thanks to Gerd Moellman and Pierre Mai):
2019 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2021 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2022 primary methods with no specializers;
2023 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2025 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2026 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2027 FORWARD-REFERENCED-CLASSes; error reporting on
2028 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2030 ** SXHASH on CLOS instances now uses a slot internal to the
2031 instance to return different numbers on distinct instances,
2032 while preserving the same return value through invocations of
2034 ** DEFMETHOD signals errors when methods with longer incongruent
2035 lambda lists are added to generic functions;
2036 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2037 CLASS, as specified in AMOP;
2038 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2039 on the order returned by the primary method for classes of
2040 class STANDARD-CLASS;
2041 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2042 * fixed some bugs shown by Paul Dietz' test suite:
2043 ** DOLIST puts its body in TAGBODY;
2044 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2046 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2047 value producing form;
2048 ** if more variables are given to PROGV than values, extra
2049 variables are bound and made to have no value;
2050 ** NSUBSTITUTE on list arguments gets the right answer with
2052 ** ELT signals an error of type TYPE-ERROR when the index argument
2053 is not a valid sequence index;
2054 ** LOOP signals (at macroexpansion time) an error of type
2055 PROGRAM-ERROR when duplicate variable names are found;
2056 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2057 ** FUNCALL of special-operators now cause an error of type
2059 ** PSETQ now works as required in the presence of side-effecting
2060 symbol-macro places;
2061 ** NCONC accepts any object as its last argument;
2062 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2064 ** loop-for-as-package does not require a package to be explicitely
2066 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2068 * fixed bug 166: compiler preserves "there is a way to go"
2069 invariant when deleting code.
2070 * fixed bug 172: macro lambda lists with required arguments after
2071 &REST arguments now cause an error to be signalled. (thanks to
2073 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2074 bug. (thanks to Matthew Danish)
2075 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2077 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2078 arguments are lexically rebound. (thanks to Gerd Moellmann and
2080 * fixed bug 194: error messages are now more informative when there
2081 is no primary method applicable in a call to a generic
2082 function. (thanks to Gerd Moellmann)
2083 * fixed bug in command line argument checking (thanks to Julian
2085 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2086 (thanks to Matthew Danish)
2087 * incremented fasl file version number, because of the
2088 SXHASH-related changes in the layout of CLOS data structures
2090 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2091 * minor incompatible change: The runtime (the Unix executable named
2092 "sbcl") is now much pickier about the .core files it will load.
2093 Essentially it now requires .core files to descend from the same
2094 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2095 as the runtime does. (The intent is to prevent the crashes which
2096 can occur, and which can even be reported as mysterious failures,
2097 when people patch the sources or change the build parameters
2098 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2099 sbcl and .core files.)
2100 * fixed bug: VALUES-LIST is no longer optimized away.
2101 * fixed bug 142: The FFI conversion of C string values to Lisp
2102 string values no longer conses excessively. (thanks to Nathan
2103 Froyd porting Raymond Toy's fix to CMU CL)
2104 * began to systematize and improve MOP conformance in PCL (thanks to
2105 Nathan Froyd, Gerd Moellman and Pierre Mai):
2106 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2108 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2109 ** FINALIZE-INHERITANCE is now called on class finalization;
2110 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2111 argument precedence order.
2112 * fixed bug 202: The compiler no longer fails on functions whose
2113 derived types contradict their declared type.
2114 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2115 so it can be non-toplevel.
2116 * The fasl file version number has changed (because of the new
2117 implementation of DEFMACRO).
2118 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2119 safe code, that any length requirement by their type-specifier
2120 argument is valid. The exceptions to this are described in bug
2122 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2123 functions now check the types of their inputs as required by ANSI.
2124 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2125 introduced symbol is DECLAREd to be SPECIAL.
2126 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2127 * fixed bug: :COUNT argument to sequence functions may be negative
2128 * fixed bug: body of DO-SYMBOLS may contain declarations
2129 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2130 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2131 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2132 symbol macro only once
2133 * fixed printing of call frame when argument list is unavailable
2134 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2135 * compiler no longer signals WARNING on unknown keyword
2138 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2139 * A beta-quality port to the mips architecture running Linux,
2140 based on the old CMUCL backend, has been made. It has been tested
2141 on a big-endian kernel, and works sufficiently well to be able to
2142 rebuild itself; it has not been tested in little-endian mode.
2143 * fixed an inconsistency between gencgc.c and purify.c which made
2144 dumping/loading .core files unreliable
2145 * fixed bug 120a: The compiler now deals correctly with IFs where
2146 the consequent is the same as the alternative, instead of
2147 misderiving the return type. (thanks to Alexey Dejneka)
2148 * fixed bug 113: Logical pathnames are now dumpable (the logical
2149 host is resolved at load-time, throwing an error if it is not
2151 * fixed bug 174: FORMAT's error message is slightly clearer when a
2152 non-printing character is used in a format directive.
2153 * fixed several bugs in compiler checking of type declarations, i.e.
2154 violations of the Python "declarations are assertions" principle
2155 (thanks to Alexey Dejneka)
2156 * fixed several bugs in PCL's error checking (thanks to Gerd
2158 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2160 * fixed bug in compilation of functions as first class values
2161 (thanks to Antonio Martinez-Shotton)
2162 * The compiler's handling TYPE-ERRORs which it can prove will
2163 inevitably happen at runtime has been cleaned up and corrected
2164 in several ways. (thanks to Alexey Dejneka)
2165 * improved argument type checking for various basic arithmetic
2166 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2167 lavished on them in the past that they can be compiled in many
2168 ways in different special cases
2169 * fixed bug 181: compiler checks validity of user supplied type
2171 * cleaned up code flushing in optimization: Function calls which
2172 should signal errors for safety purposes (e.g. which ANSI says
2173 should signal errors when their arguments are of incorrect type)
2174 are no longer optimized away.
2175 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2176 * incremented fasl file version number, because changes in the
2177 implementation of sequence functions like COERCE caused
2178 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2179 in old inline expansions) to become undefined. (Actually these
2180 changes were later undone, so we might very well be binary
2181 compatible with 0.7.7 after all, but leaving the version number
2182 incremented seemed like the simplest and most conservative
2185 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2186 * An alpha-quality port to the parisc architecture running Linux,
2187 based on the old CMUCL backend, has been made. This, even more so
2188 than the other backends, should be considered still a work in
2189 progress; known problems include that the Linux kernel in 64-bit
2190 mode does not propagate the correct sigcontext structure to
2191 userspace, and consequently SBCL on a parisc64 kernel will not
2193 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2194 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2195 suppressed.) Also now that the compiler is looking at declarations
2196 in the environment, it checks optimization declarations as well,
2197 and suppresses inlining when (> DEBUG SPEED).
2198 * More fixes have been made to treatment of floating point exception
2199 treatment and other Unix signals. In particular, floating point
2200 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2201 * The detection and handling of control stack exhaustion (infinite
2202 or very deeply nested recursion) has changed. Stack exhaustion
2203 detection is now done by write-protecting pages at the OS level
2204 and applies at all optimization settings; when found, a
2205 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2206 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2207 be caught using IGNORE-ERRORS.
2208 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2209 correctly when transforming recursive LABELS functions to LETs.
2210 (thanks to Alexey Dejneka)
2211 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2212 that are names of constants or global variables.
2213 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2214 alien routines with docstrings.
2215 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2216 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2218 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2219 (thanks to Eric Marsden)
2220 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2221 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2222 * Bug fix: The SPARC backend can now compile functions involving
2223 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2224 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2225 to a functions expecting a PATHNAME-DESIGNATOR.
2226 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2227 lambda lists. (thanks to Alexey Dejneka)
2228 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2229 functions so that EXPORTing the name of the class doesn't cause
2230 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2231 to Antonio Martinez for reporting this)
2232 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2233 constant-folded without failing an assertion. (thanks to Einar
2234 Floystad Dorum for reporting this)
2235 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2236 types are now dealt with more correctly by the compiler.
2237 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2238 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2239 behaviour analogous to RENAME-FILE. This puts its behaviour more
2240 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2242 * The fasl file version number has changed again. (because of the
2243 bug fix involving the names of PCL MAKE-INSTANCE functions)
2245 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2246 * bug fix: Floating point exceptions are treated much more
2247 consistently on the x86/Linux and PPC/Linux platforms.
2248 * Array initialization with :INITIAL-ELEMENT is now much faster for
2249 cases when the compiler cannot open code the array creation, but
2250 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2251 array accesses have also seen a speed increase.
2252 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2253 specified. (This was at the root of some bad interactions between
2254 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2256 * bug fix: Internal error arguments for undefined functions are now
2257 computed correctly on the PPC/Linux platform.
2258 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2259 Raymond Toy's patch for CMU CL)
2260 * Support for the Solaris 9 operating environment has been included
2261 (thanks to Daniel Merritt)
2262 * A very ugly but hopefully complete draft of the missing FFI chapter
2263 of the manual has been created by reformatting the corresponding
2264 CMU CL manual chapter into (currently very ugly and incoherent)
2265 DocBook and bringing it up to date for SBCL behavior. Thus, the
2266 manual is now essentially complete, at least by my extreme
2267 once-and-only-once standards, whereby it's acceptable to refer to
2268 the doc strings of SB-EXT functions as the primary documentation.
2269 * The fasl file version number has changed again, due to cleanup of
2270 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2271 is no longer a static symbol.)
2273 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2274 * SBCL now builds with OpenMCL (version 0.12) as the
2275 cross-compilation host; also, more progress has been made toward
2276 bootstrapping under CLISP.
2277 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2279 * bug 158 fixed: The compiler can now deal with integer loop
2280 increments different from 1; fixing this turned out also to fix
2282 * bug 169 fixed: no more bogus warnings about using lexical bindings
2283 despite the presence of perfectly good SPECIAL declarations (thanks
2284 to David Lichteblau)
2285 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2286 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2287 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2289 * bug 180 fixed: Method combination specifications no longer ignore
2290 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2291 * bug fix: Structure type predicate functions now check their argument
2292 count as they should.
2293 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2294 correctly. (thanks to Pierre Mai)
2295 * minor incompatible change: The --noprogrammer option is deprecated
2296 in favor of the new --disable-debugger option, which is very similar.
2297 (The major difference is that it takes effect at a slightly different
2298 time at startup, causing handling of errors in --sysinit and
2299 --userinit files will be affected differently.) The
2300 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2301 been added to allow this functionality to be controlled from ordinary
2302 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2303 maintainers, who might want to run non-interactive scripts to
2304 build SBCL cores which will later be used interactively.)
2305 * minor incompatible change: The LOAD function no longer, when given
2306 a wild pathname to load, loads all files matching that pathname.
2307 Instead, an error of type FILE-ERROR is signalled.
2309 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2310 * bug 147 fixed: The compiler preserves its block link/count
2311 invariants more correctly now so that it doesn't crash. (thanks
2313 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2315 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2316 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2317 no-value-for-this-character as NIL (as specified by ANSI).
2318 * HOST-NAMESTRING on physical pathnames now returns a string that is
2319 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2320 (thanks to Christophe Rhodes)
2321 * The Alpha port handles icache flushing more correctly. (thanks to
2323 * More progress has been made toward bootstrapping under CLISP. (thanks
2324 to Christophe Rhodes)
2325 * The fasl file format has changed again, because dynamic loading
2326 on OpenBSD (which has non-ELF object files) motivated some cleanups
2327 in the way that foreign symbols are transformed and passed around.
2328 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2329 is no longer treated as whitespace by the reader, but instead as
2330 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2331 |A
\7fB|, instead of A as it used to.
2333 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2334 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2335 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2336 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2337 this as of 0.7.1.45, but was left out of the previous news section
2338 (thanks to Dan Barlow)
2339 * SBCL now runs on the Solaris operating system on SPARC architectures
2340 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2341 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2342 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2343 * SPARC backend cleanups, allowing builds of cores optimized for V8
2344 and V9 SPARCS, and also emission of code targeted to a particular
2345 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2347 * SBCL is closer to bootstrapping under CLISP, thanks to various
2348 fixes by Christophe Rhodes.
2349 * The fasl file format has changed again, to allow the compiler's
2350 INFO database to support symbol macros.
2351 * The user manual (in doc/) is formatted into HTML more nicely.
2352 (thanks to coreythomas)
2353 * The system is smarter about SUBTYPEP relationships, especially
2354 those involving NOT types (including types such as ATOM which are
2355 represented internally using NOT types). Thus SUBTYPEP is less
2356 likely to return (VALUES NIL NIL) in general, and in particular
2357 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2359 * The fasl file format has changed again, because the internal
2360 representation of types now includes a new slot to support the new
2361 SUBTYPEP-of-NOT-types logic.
2362 * (not a change in the main branch of SBCL, but a related prototype
2363 which can hopefully be merged into the main branch of SBCL in the
2364 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2365 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2366 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2367 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2368 dispatch macro character combinations. (thanks to Alexey Dejneka)
2369 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2370 that superficially look like logical namestrings correctly.
2371 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2373 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2374 default supplied-p) for &optional and &key arguments. (thanks to
2377 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2378 * incompatible change: The compiler is now less aggressive about
2379 tail call optimization, doing it only when (> SPACE DEBUG) or
2380 (> SPEED DEBUG). (This is an incompatible change because there are
2381 programs which relied on the old CMU-CL-style behavior to optimize
2382 away their unbounded recursion which will now die of stack overflow.)
2383 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2384 for non-GENCGC systems has been increased to 20M (since that
2385 seems much closer to the likely performance optimum for modern
2386 systems than the old 4M value was)
2387 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2388 and *DEBUG-PRINT-LEVEL*
2389 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2390 of CMU CL's support for SPARC, and various endianness and other
2391 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2392 * new syntactic sugar for the Unix command line: --load foo.bar is now
2393 an alternate notation for --eval '(load "foo.bar")'.
2395 ** The system now detects stack overflow and handles it gracefully,
2396 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2397 optimization settings. (This is a good thing in general, and
2398 its introduction in this version should be particularly timely
2399 for anyone whose code fails because of suppression of tail
2401 ** The system now hunts for the C variable "environ" in a more
2402 devious way, to avoid segfaults when the C library version differs
2403 between compile time and run time. (thanks to Christophe Rhodes)
2404 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2405 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2406 ** The compiler no longer issues bogus style warnings for undefined
2407 classes in the same source file as the DEFCLASSes which defined
2408 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2410 ** fixes in CONDITION class precedence list for undefined function
2411 errors (thanks to Alexei Dejneka)
2412 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2413 correctly. (thanks to Dan Barlow)
2414 ** portability fixes aiming at bootstrapping under CLISP (thanks
2415 to Dave McDonald and Christophe Rhodes)
2416 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2417 ** fixes in type translation and and type inference (thanks to
2419 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2420 ** various fixes in the new ports (thanks to Dan Barlow)
2421 * several changes related to debugging:
2422 ** suppression of tail recursion, as noted above
2423 ** stack overflow detection, as noted above
2424 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2425 is now the default. (For some time encapsulation has been more
2426 reliable than the breakpoint-based :ENCAPSULATE NIL
2427 implementation, at least on X86 systems; and I just noticed that
2428 encapsulation also seems closer to the spirit of the ANSI
2431 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2433 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2434 up properly again. (There was a packaging bug in 0.7.0 which
2435 left their definitions in SB-SYS::LOAD-FOREIGN and
2436 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2437 vital for most things which interface to C-level interfaces,
2438 like extensions working with sockets or databases or
2439 Perl-compatible regexes or whatever, and the need to fix
2440 this bug is the main reason that 0.7.1 was released so
2442 ** DEFGENERIC is now choosier about the methods it redefines, so that
2443 reLOADing a previously-LOADed file containing DEFGENERICs does
2444 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2445 cycle now works as it should. (thanks to Alexey Dejneka)
2446 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2447 fixed by Christophe Rhodes. (He actually submitted this patch
2448 months ago, and I delayed until after 0.7.0.)
2449 ** Bug 111 (internal compiler confusion about runtime checks on
2450 FUNCTION types) was fixed by Alexey Dejneka.
2451 * Some internal cleanups (getting rid of variables which aren't
2452 needed now that the byte interpreter is gone) caused the fasl
2453 file format number to change again.
2455 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2456 * major incompatible change: The default fasl file extension, i.e. the
2457 default extension for files produced by COMPILE-FILE, has changed
2458 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2460 ** There are many changes in the implementation of the compiler.
2461 SBCL is now essentially a compiler-only implementation of ANSI
2462 Common Lisp. EVAL still "interprets" a few special cases, but
2463 almost all the interesting cases are handled by creating
2464 a LAMBDA expression, calling COMPILE on it, then calling
2465 FUNCALL on the result.
2466 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2467 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2468 Since the code is newer, there might still be some new bugs
2469 (though not as many as before Martin Atzmueller's fixes:-). But
2470 the new code is substantially simpler and clearer, and hopefully
2471 any remaining bugs will be simpler, less fundamental, and more
2472 fixable then the bugs in the old code.
2473 ** The revised compiler is still a little unsteady on its feet.
2475 *** The debugging information it produces (particularly the names
2476 of FUNCTION objects) is sometimes much less useful than what
2477 the old compiler produced.
2478 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2479 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2480 within a MACROLET) has been temporarily weakened.
2481 ** There are new compiler optimizations for various functions:
2482 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2483 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2484 *** the math functions TRUNCATE, FLOOR, and CEILING
2485 *** the function-of-all-trades COERCE
2486 Mostly these should be transparent, but there's one
2487 potentially-annoying problem (bug 117): when the compiler
2488 inline-expands a function and does type analysis on the result,
2489 it can create control paths which have type mismatches, and
2490 when it can't prove that those control paths aren't taken,
2491 it will issue WARNINGs about the type mismatches. This is
2492 a particular problem in practice for the new sequence functions.
2493 It's not clear how this should be fixed, and for now, a
2494 workaround is given in the entry for 117 in the BUGS file.
2495 ** (Because of the interaction between the two previous items --
2496 occasional inlining problems and new inline expansions -- some
2497 of the new sequence function optimizations won't really kick in
2498 completely until debugging information, and then inlining, are
2499 straightened out in some future version.)
2500 * minor incompatible changes:
2501 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2502 DIRECTORY no longer implicitly promotes NIL slots of its
2503 pathname argument to :WILD. In particular, when you ask for the
2504 contents of a directory (which you used to be able to do without
2505 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2506 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2507 ** changes in behavior that ANSI explicitly defines to be
2508 implementation dependent:
2509 *** The new compiler-only implementation still conforms with ANSI,
2510 but acts a little different than before. Besides the obvious
2511 changes in performance tradeoffs (that the cost per form passed
2512 to EVAL has gone up, and the cost per form executed by EVAL
2513 has gone down), the behavior of the system changes a little
2514 because there are no longer any interpreted function objects.
2515 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2516 e.g. doing COMPILE on the output of interactive DEFUN is
2518 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2520 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2521 has changed from (:CL) to NIL.
2522 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2523 e.g. #\Newline and #\Tab, don't have names specified in the
2524 ANSI Common Lisp standard, is now based on their ASCII symbolic
2525 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2526 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2527 are no longer used for output.
2528 ** changes in internal implementation constants:
2529 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2530 4 million. (If your application spends a lot of time GCing and
2531 you have a lot of RAM, you might want to experiment with
2532 increasing it even more.)
2533 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2534 However, almost all old code should still continue to work without
2535 immediate update, as SB-C-CALL is now a (deprecated) nickname
2537 ** Old operator names in the style DEF-FOO are now deprecated in
2538 favor of new corresponding names DEFINE-FOO, for consistency with
2539 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2540 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2541 internal symbols, but a few supported extensions like
2542 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2543 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2544 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2545 etc. as you get deeper into recursive calls to the debugger
2546 command loop, instead of the old "5]", "5]]", "5]]]"
2547 sequence. (I was motivated to do this when squabbles between
2548 ILISP and SBCL left me very deeply nested in the debugger. In the
2549 short term, this change will probably provoke more ILISP/SBCL
2550 squabbles, but hopefully it will be an improvement in the long run.)
2551 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2552 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2553 function, since declaiming return types involving aliens is
2554 (1) annoyingly messy to do by hand and (2) vital to efficient
2555 compilation of code which calls such functions.
2556 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2557 longer reexported by the SB-EXT package. They're solely useful
2558 for alien code, so it seems more logical that you should get
2559 them from the SB-ALIEN package, not in SB-EXT.
2560 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2561 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2562 features. Instead, the code that they used to control is always
2563 built into the system.
2564 * many other bug fixes
2565 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2566 advantage of the new EVAL-WHEN stuff and to clean them up in
2567 general, and they are now more ANSI-compliant in a number of
2568 ways. Martin Atzmueller is responsible for a lot of this.
2569 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2571 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2572 *** correct ERROR type for various file operations
2573 *** some fixes for Lisp streams
2574 *** DEFMETHOD syntax checking
2575 *** changing old weird representation of debug information as
2576 strings (which, among their other deficiencies, don't transform
2577 correctly when you rename packages, and don't change their
2578 print representation when you change things like *PACKAGE*
2579 and *PRINT-LENGTH*) to symbols and lists of symbols
2580 He also made several improvements and fixed several bugs in DESCRIBE.
2581 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2583 *** misbehavior of WRITE-STRING/WRITE-LINE
2584 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2585 and several other LOOP problems as well
2586 *** DIRECTORY when similar filenames are present
2587 *** DEFGENERIC with :METHOD options
2588 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2589 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2590 *** argument ordering in FIND with :TEST option
2591 *** mishandled package designator argument in APROPOS-LIST
2592 *** various problems in the backquote readmacro
2593 *** a bug in APROPOS
2594 *** probably some others that I'm not describing very well here,
2595 since the CVS log documents them by reference to sbcl-devel
2596 messages, and the SourceForge archives aren't working well.:-(
2597 ** Dan Barlow improved the Alpha port (and is making progress on the
2598 PPC port, for those of you who think different).
2599 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2600 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2601 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2602 continues to work on the SPARC port (for those of you in a position
2603 to look down upon our little PC-compatible boxes from a great height).
2604 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2605 and uses that copy, rather than the current dynamic value, when
2606 it's trying to decide whether to truncate output. Thus e.g.
2607 (let ((*print-lines* 50))
2608 (pprint-logical-block (stream nil)
2610 (let ((*print-lines* 8))
2611 (print (aref possiblybigthings i) stream)))))
2612 should now truncate the logical block only at 50 lines, instead of
2613 often truncating it at 8 lines, as it did before.
2614 * The doc/cmucl/ directory, containing old CMU CL documentation
2615 from the time of the fork, is no longer part of the base system.
2616 SourceForge has shut down its anonymous FTP service, and with it
2617 my original plan for distributing the old CMU CL documentation
2618 there. For now, if you need these files you can download an old
2619 SBCL source release and extract them from it.
2620 * The fasl file version number changed again, for dozens of reasons,
2621 some of which are apparent above.
2623 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2624 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2625 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2626 patch, so that DISASSEMBLE output is much nicer.
2627 * The code in the SB-PROFILE package now seems reasonably stable.
2628 I still haven't decided what the final interface should look like
2629 (I'd like PROFILE to interact cleanly with TRACE, since both
2630 facilities use function encapsulation) but if you have a need
2631 for profiling now, you can probably use it successfully with
2632 the current CMU-CL-style interface.
2633 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2634 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2635 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2636 link now returns the dangling link itself, and for similar
2637 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2638 link itself. (In these cases the old code signalled an error and
2639 looped endlessly, respectively.) Thus, DIRECTORY now works even
2640 in the presence of dangling and cyclic symbolic links.
2641 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2642 is now a supported extension again, since the consensus on
2643 sbcl-devel was that it can be useful for ordinary development
2644 work, not just for debugging SBCL itself.
2645 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2646 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2647 that currently-defined functions might be redefined later with
2648 different return types.
2649 * Hash tables can be printed readably, as inspired by CMU CL code
2650 of Eric Marsden and SBCL code of Martin Atzmueller.
2651 * better error handling in CLOS method combination, thanks to
2652 Martin Atzmueller porting Pierre Mai's CMU CL patches
2653 * more overflow fixes for >16Mbyte I/O buffers
2654 * A bug in READ has been fixed, so that now a single Ctrl-D
2655 character suffices to cause end-of-file on character streams.
2656 In particular, now you only need one Ctrl-D at the command
2657 line (not two) to exit SBCL.
2658 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2660 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2661 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2662 does the right thing.
2663 * The implementation of some type tests, especially for CONDITION
2664 types, is now tidier and maybe faster, due to CMU CL code
2665 originally by Douglas Crosher, ported by Martin Atzmueller.
2666 * Some math functions have been fixed, and there are new
2667 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2668 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2669 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2670 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2671 likely to be built into sbcl-0.7.0.)
2672 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2673 CL, ported to SBCL by Martin Atzmueller)
2674 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2675 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2676 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2677 * There's a new slam.sh hack to shorten the edit/compile/debug
2678 cycle for low-level changes to SBCL itself, and a new
2679 :SB-AFTER-XC-CORE target feature to control the generation of
2680 the after-xc.core file needed by slam.sh.
2681 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2682 COMPILE-FILE is no longer supported, so that now every function
2683 gets an entry point, so that block compilation looks a little
2684 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2685 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2686 returns the number of bytes consed since the system started,
2687 rather than the number consed since the first time the function
2688 was called. (The new definition parallels ANSI functions like
2689 CL:GET-INTERNAL-RUN-TIME.)
2690 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2691 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2692 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2693 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2694 * Fasl file version numbers are now independent of the target CPU,
2695 since historically most system changes which required version
2696 number changes have affected all CPUs equally. Similarly,
2697 the byte fasl file version is now equal to the ordinary
2700 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2701 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2702 is no longer recognized. I apologize for this, because it was
2703 listed in SB-EXT as a supported extension, but I found that
2704 its existing behavior was poorly specified, as well as incorrectly
2705 specified, and it looked like too much of a mess to straighten it
2706 out. I have enough on my hands trying to get ANSI stuff to work..
2707 * many patches ported from CMU CL by Martin Atzmueller, with
2708 half a dozen bug fixes in pretty-printing and the debugger, and
2709 half a dozen others elsewhere
2710 * fixed bug 13: Floating point infinities are now supported again.
2711 They might still be a little bit flaky, but thanks to bug reports
2712 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2713 as flaky as they were.
2714 * The --noprogrammer command line option is now supported. (Its
2715 behavior is slightly different in detail from what the old man
2716 page claimed it would do, but it's still appropriate under the
2717 same circumstances that the man page talks about.)
2718 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2719 are now supported, and enabled by default. Thus, the compiler can
2720 handle many floating point and complex operations much less
2721 inefficiently. (Thus e.g. you can implement a complex FFT
2723 * The compiler now detects type mismatches between DECLAIM FTYPE
2724 and DEFUN better, and implements CHECK-TYPE more correctly, and
2725 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2726 thanks to patches from Martin Atzmueller.
2727 * various fixes to make the cross-compiler more portable to
2728 ANSI-conforming-but-different cross-compilation hosts (notably
2729 Lispworks for Windows, following bug reports from Arthur Lemmens)
2730 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2731 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2732 fixed thanks to Pierre Mai's CMU CL patches.
2733 * fixes to keep the system from overflowing internal counters when
2734 it tries to use i/o buffers larger than 16M bytes
2735 * fixed bug 45a: Various internal functions required to support
2736 complex special functions have been merged from CMU CL sources.
2737 (When I was first setting up SBCL, I misunderstood a compile-time
2738 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2739 * improved support for type intersection and union, fixing bug 12
2740 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2741 more obscure bugs as well
2742 * some steps toward byte-compiling non-performance-critical
2743 parts of the system, courtesy of patches from Martin Atzmueller
2744 * Christophe Rhodes has made some debian packages of sbcl at
2745 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2746 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2747 stable, but are nonetheless usable. When he's ready, I'd be happy
2748 to add them to the SourceForge "File Releases" section. (And if
2749 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2750 * new fasl file format version number (because of changes in
2751 internal representation of (OR ..) types to accommodate the new
2752 support for (AND ..) types, among other things)
2754 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2755 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2757 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2759 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2760 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2761 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2762 * DESCRIBE now works on structure objects again.
2763 * Most function call argument type mismatches are now handled as
2764 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2765 know whether the function will be redefined before the call is
2766 executed. (The compiler could flag local calls with full WARNINGs,
2767 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2768 it doesn't keep track of enough information to know whether calls
2769 are local in this sense.)
2770 * Compiler output is now more verbose, with messages truncated
2771 later than before. (There should be some supported way for users
2772 to override the default verbosity, but I haven't decided how to
2773 provide it yet, so this behavior is still controlled by the internal
2774 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2775 src/compiler/ir1util.lisp.)
2776 * Fasl file format version numbers have increased again, because
2777 support for the Gray streams extension changes the layout of the
2778 system's STREAM objects.
2779 * The Gray subclassable streams extension now works, thanks to a
2780 patch from Martin Atzmueller.
2781 * The full LOAD-FOREIGN extension (not just the primitive
2782 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2783 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2784 but like most other programs, it defaults to copying the Unix
2785 environment from the original process instead of starting the
2786 new process in an empty environment.
2787 * Extensions which manipulate the Unix environment now support
2788 an :ENVIRONMENT keyword option which doesn't smash case or
2789 do other bad things. The CMU-CL-style :ENV option is retained
2790 for porting convenience.
2791 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2792 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2794 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2796 * A patch from Martin Atzmueller seems to have solved the SIGINT
2797 problem, and as far as we know, signal-handling now works cleanly.
2798 (If you find any new bugs, please report them!)
2799 * The system no longer defaults Lisp source file names to types
2800 ".l", ".cl", or ".lsp", but only to ".lisp".
2801 * The compiler no longer uses special default file extensions for
2802 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2803 seems to expect a single default extension for all compiled code,
2804 and there's no compelling reason to try to stretch the standard
2805 to allow two different extensions.) Instead, byte-compiled files
2806 default to the same extension as native-compiled files.
2807 * Fasl file format version numbers have increased again, because
2808 a rearrangement of internal implementation packages made some
2809 dumped symbols in old fasl files unreadable in new cores.
2810 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2811 many fewer weird special cases.
2812 * Bug #17 (differing COMPILE-FILE behavior between logical and
2813 physical pathnames) has been fixed, and some related misbehavior too,
2814 thanks to a patch from Martin Atzmueller.
2815 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2816 by Tim Moore, ported to SBCL by Martin Atzmueller.
2817 * Martin Atzmueller fixed several filesystem-related problems,
2818 including bug #36, in part by porting CMU CL patches, which were
2819 written in part by Paul Werkowski.
2820 * More compiler warnings in src/runtime/ are gone, thanks to
2821 more patches from Martin Atzmueller.
2822 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2825 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2827 * DESCRIBE now works on CONDITION objects.
2828 * The debugger now handles errors which arise when trying to print
2829 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2831 * The build system now uses an additional file, customize-target-features.lisp,
2832 to allow local modifications to the target *FEATURES* list. (The point of
2833 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2834 debugging features enabled, without having to worry about propagating it
2835 into everyone's system when I do a "cvs update".) When no
2836 customize-target-features.lisp file exists, the target *FEATURES* list
2837 should be constructed the same way as before.
2838 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2839 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2840 warn about behavior which is undefined under the ANSI spec. Note: This
2841 is specified by ANSI, but it's not very popular with programmers.
2842 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2843 macro in the SBCL sources for an example of a workaround which you
2844 might use to make portable ANSI-standard code which does what you want.
2845 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2846 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2847 when it's used as a non-toplevel form. (This is required in order
2848 to implement the DEFCONSTANT-EQX macro.)
2849 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2850 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2851 funny things, due to bugs in the implementation of EVAL-WHEN
2852 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2853 non-toplevel DEFCONSTANTs are uncommon.)
2854 * The core file version number and fasl file version number have been
2855 incremented, because the old noncompliant DEFCONSTANT behavior involved
2856 calling functions which no longer exist, and because I also took the
2857 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2858 * fixed bug 1 (error handling before read-eval-print loop starts), and
2859 redid debugger restarts and related debugger commands somewhat while
2861 ** The QUIT debugger command is gone, since it did something
2862 rather different than the SB-EXT:QUIT command, and since it never
2863 worked properly outside the main toplevel read/eval/print loop.
2864 Invoking the new TOPLEVEL restart provides the same functionality.
2865 ** The GO debugger command is also gone, since you can just invoke
2866 the CONTINUE restart directly instead.
2867 ** The TOP debugger command is also gone, since it's redundant with the
2868 FRAME 0 command, and since it interfered with abbreviations for the
2870 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2872 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2873 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2874 some porting work by Martin Atzmueller)
2875 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2876 with a boatload of patches which he ported to SBCL. Now that those
2878 ** The system tries to make sure that its low-priority messages
2879 are prefixed by semicolons, to help people who like to use
2880 syntax highlighting in their ILISP buffer. (This patch
2881 was originally due to Raymond Toy.)
2882 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2883 patches originally written by Raymond Toy.
2884 ** The compiler understands coercion between single-value and
2885 multiple-VALUES type expressions better, getting rid of some very
2886 weird behavior, thanks to patches originally by Robert MacLachlan
2887 and Douglas Crosher.
2888 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2889 lambda lists, thanks to a patch originally by Pierre Mai.
2890 ** The system no longer bogusly warns about "abbreviated type
2892 ** The compiler gets less confused by inlining and RETURN-FROM,
2893 thanks to some patches originally by Tim Moore.
2894 ** The system no longer hangs when dumping circular lists to fasl
2895 files, thanks to a patch originally from Douglas Crosher.
2896 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2897 undefined function error.
2898 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2899 in structure slot initforms) for now, documented workaround instead:-|
2900 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2901 * fixed bug 5: added stubs for various Gray stream functions called
2902 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2903 installed, at least appropriate type errors are generated
2904 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2905 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2906 consistently in DEFMETHOD forms.
2907 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2908 it doesn't seem to affect SBCL after all
2909 * The C runtime system now builds with better optimization and many
2910 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2912 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2914 * The system is now under CVS at SourceForge (instead of the
2915 CVS repository on my home machine).
2916 * The new signal handling code has been tweaked to treat register
2917 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2918 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2919 downstream, e.g. in debug-int.lisp, has implicit dependencies
2920 on the unsignedness of integer representation of machine words,
2921 and that caused the system to bomb out with infinite regress
2922 when trying to recover from type errors involving signed values,
2923 e.g. (BUTLAST '(1 2 3) -1).)
2924 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2925 (This was one of the bugs Peter Van Eynde reported back in July.)
2926 * The system now uses code inspired by Colin Walters' O(N)
2927 implementation of MAP (from the cmucl-imp@cons.org mailing
2928 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2929 inline the MAP operation, and there is more than one
2930 sequence argument to the MAP call (so that it can't just
2931 do ETYPECASE once and for all based on the type of the
2932 single sequence argument). (The old non-inline implementation
2933 of the general M-argument sequence-of-length-N case required
2934 O(M*N*N) time when any of the sequence arguments were LISTs.)
2935 * The QUIT :UNIX-CODE keyword argument has been renamed to
2936 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2937 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2939 * Raymond Wiker's patches to port dynamic loading from Linux to
2940 FreeBSD have been added.
2941 * The BUGS file is now more nearly up to date, thanks in large part
2942 to Martin Atzmueller's review of it.
2943 * The debugger now flushes standard output streams before it begins
2944 its output ("debugger invoked" and so forth).
2945 * The core version number and fasl file version number have both
2946 been incremented, because of incompatible changes in the layout
2948 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2949 used to not be called for a saved Lisp image.)
2950 * Martin Atzmueller's version of a patch to fix a compiler crash,
2951 as posted on sbcl-devel 13 September 2000, has been installed.
2952 * Instead of installing Martin Atzmueller's patch for the
2953 compiler transform for SUBSEQ, I deleted the compiler transform,
2954 and transforms for some similar consing operations.
2955 * A bug in signal handling which kept TRACE from working on OpenBSD
2957 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2958 away by constant folding
2959 * The system now defines its address space constants in one place
2960 (in the Lisp sources), and propagates them automatically elsewhere
2961 (through GENESIS and the sbcl.h file). Therefore, patching the
2962 address map is less unnecessarily tedious and error-prone. The
2963 Lisp names of address space constants have also been systematized.
2964 * CVS tags like dollar-Header-dollar have been removed from
2965 the sources, because they have never saved me trouble and
2966 they've been source of trouble working with patches and other
2967 diff-related operations.
2968 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2969 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2971 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2973 * The system has been ported to OpenBSD.
2974 * The system now compiles with a simple "sh make.sh" on the systems
2975 that it's supported on. I.e., now you no longer need to tweak
2976 text in the target-features.lisp-expr and symlinks in src/runtime/
2977 by hand, the make.sh takes care of it for you.
2978 * The system is no longer so grossly inefficient when compiling code
2979 involving vectors implemented as general (not simple) vectors (VECTOR T),
2980 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2981 which dares to use the various sequence functions on non-simple
2982 vectors, takes less of a performance hit.
2983 * There is now a primitive type predicate VECTOR-T-P
2984 to test for the (VECTOR T) type, so that e.g.
2985 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2986 can now be compiled with some semblance of efficiency. (The old code
2987 turned the type declaration into a full call to %TYPEP at runtime!)
2988 * AREF on (VECTOR T) is still not fast, since it's still compiled
2989 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2990 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2991 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2992 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2993 without making full calls to SUBTYPEP at runtime.
2994 (Some analogous efficiency issues for non-simple vectors specialized to
2995 element types other than T, or for non-simple multidimensional arrays,
2996 have not been addressed. They could almost certainly be handled the
2997 same way if anyone is motivated to do so.)
2998 * The changes in array handling break binary compatibility, so
2999 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3000 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3001 instead of (VALUES T T).
3002 * By following the instructions that Dan Barlow posted to sbcl-devel
3003 on 2 July 2000, I was able to enable primitive dynamic object
3004 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3005 functionality is not implemented (since it calls ld to resolve
3006 library references automatically, requiring RUN-PROGRAM for its
3007 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3008 try to resolve library references) is now supported.
3009 * The system now flushes the standard output streams when it terminates,
3010 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3011 them at several other probably-convenient times, e.g. in each pass of
3012 the toplevel read-eval-print loop, and after evaluating a form given
3013 as an "--eval" command-line option. (These changes were motivated by a
3014 discussion of stream flushing issues on cmucl-imp in August 2000.)
3015 * The source transform for TYPEP of array types no longer assumes
3016 that an array whose element type is a not-yet-defined type
3017 is implemented as an array of T, but instead punts, so that the
3018 type will be interpreted at runtime.
3019 * There is now some support for cross-compiling in make.sh: each of
3020 the phases of make.sh has its own script. (This should be transparent
3021 to people doing ordinary, non-cross-compile builds.)
3022 * Since my laptop doesn't have hundreds of megabytes of memory like
3023 my desktop machine, I became more motivated to do some items on
3024 my to-do list in order to reduce the size of the system a little:
3025 ** Arrange for various needed-only-at-cold-init things to be
3026 uninterned after cold init. To support this, those things have
3027 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3028 symbols with such names are now uninterned after cold init).
3029 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3030 fasl files for cold load.
3031 ** Remove the old compiler structure pooling code (which used to
3032 be conditional on the target feature :SB-ALLOC) completely.
3033 ** Redo the representation of some data in cold init to be more compact.
3034 (I also looked into supporting byte compiled code at bootstrap time,
3035 which would probably reduce the size of the system a lot, but that
3036 looked too complicated, so I punted for now.)
3037 * The maximum signal nesting depth in the src/runtime/ support code has
3038 been reduced from 4096 to 256. (I don't know any reason for the very
3039 large old value. If the new smaller value turns out to break something,
3040 I'll probably just bump it back up.)
3041 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3043 * Many, many bugs reported by Peter Van Eynde have been added to
3044 the BUGS list; some have even been fixed.
3045 * While enabling dynamic object file loading, I tried to make the
3046 code easier to understand, renaming various functions and variables
3047 with less ambiguous names, and changing some function calling
3048 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3049 * While trying to figure out how to do the OpenBSD port, I tried to
3050 clean up some of the code in src/runtime/. In particular, I dropped
3051 support for non-POSIX signal handling, added various comments,
3052 tweaked the code to reduce the number of compilation warnings, and
3053 renamed some files to increase consistency.
3054 * To support the new automatic configuration functionality in make.sh,
3055 the source file target-features.lisp-expr has been replaced with the
3056 source file base-target-features.lisp-expr and the machine-generated
3057 file local-target-features.lisp-expr.
3058 * fixed a stupid quoting error in make.sh so that using CMU CL
3059 "lisp -batch" as cross-compilation host works again
3061 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3063 * DESCRIBE no longer tries to call itself recursively to describe
3064 bound/fbound values, so that it no longer fails on symbols which are
3065 bound to themselves (like keywords, T, and NIL).
3066 * DESCRIBE now works on generic functions.
3067 * The printer now prints less-screwed-up representations of closures
3068 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3069 * A private symbol is used instead of the :EMPTY keyword previously
3070 used to mark empty slots in hash tables. Thus
3071 (DEFVAR *HT* (MAKE-HASH-TABLE))
3072 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3073 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3074 now does what ANSI says that it should. (You can still get
3075 similar noncompliant behavior if bang on the hash table
3076 implementation with all the symbols you get back from
3077 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3078 This breaks binary compatibility, since tests for equality to
3079 :EMPTY are wired into things like the macroexpansion of
3080 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3082 * There's now a minimal placeholder implementation for CL:STEP,
3083 as required by ANSI.
3084 * An obscure bug in the interaction of the normal compiler, the byte
3085 compiler, inlining, and structure predicates has been patched
3086 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3087 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3089 * Missing ordinary arguments in a macro call are now detected even
3090 when the macro lambda list contains &KEY or &REST.
3091 * The debugger no longer complains about encountering the top of the
3092 stack when you type "FRAME 0" to explicitly instruct it to go to
3093 the top of the stack. And it now prints the frame you request even
3094 if it's the current frame (instead of saying "You are here.").
3095 * As specified by ANSI, the system now always prints keywords
3096 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3098 * The default initial SIZE of HASH-TABLEs is now smaller.
3099 * Type information from CLOS class dispatch is now propagated
3100 into DEFMETHOD bodies, so that e.g.
3101 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3103 is now basically equivalent to
3104 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3105 (DECLARE (TYPE SINGLE-FLOAT X))
3107 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3108 operation, without having to do run-time type dispatch.
3109 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3110 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3111 * Since I don't seem to be making big file reorganizations very often
3112 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3113 is overflowing my ability to conveniently back them up), I've finally
3114 checked the system into CVS. (The CVS repository is on my home system,
3115 not at SourceForge -- putting it on SourceForge might come later.)
3116 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3117 high-level GC-NOTIFY-FOO functions send their output. (There's
3118 still very little control of where low-level verbose GC functions
3119 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3120 less than it used to -- the GC-NOTIFY-FOO functions are now under
3121 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3122 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3123 in only one place in the source code, and propagates it automatically
3124 everywhere that it's needed. Thus e.g. when I bump the version from
3125 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3126 * The C source files now include boilerplate legalese and documentation
3127 at the head of each file (just as the Lisp source files already did).
3128 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3129 to his page will be replaced with a link to his new CLiki service.
3131 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3133 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3134 * The build process now looks for GNU make under the default name "gmake",
3135 instead of "make" as it used to. If GNU make is not available as "gmake"
3136 on your system, you can change this default behavior by setting the
3137 GNUMAKE environment variable.
3138 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3139 can build without error under CMU CL.
3141 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3143 * There is now a partial SBCL user manual (with some new text and some
3144 text cribbed from the CMU CL manual).
3145 * The beginnings of a profiler have been added (starting with the
3146 CMU CL profiler and simplifying and cleaning up). Eventually the
3147 main interface should be through the TRACE macro, but for now,
3148 it's still accessed through vaguely CMU-CL-style functions and macros
3149 exported from the package SB-PROFILE.
3150 * Some problems left over from porting CMU CL to the new
3151 cross-compilation bootstrap process have been cleaned up:
3152 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3153 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3154 part when getting rid of the compiler *BACKEND* stuff.)
3155 ** The value of *NULL-TYPE* was screwed up, because it was
3156 being initialized before the type system knew the final
3157 definition of the 'NULL type. This screwed up several key
3158 optimizations in the compiler, causing inefficiency in all sorts
3159 of places. (I found it because I wanted to understand why
3160 GET-INTERNAL-RUN-TIME was consing.)
3161 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3162 PROCLAIM FTYPE information. Unfortunately this broke binary
3163 compatibility again, since now the forms output by DEFGENERIC
3164 to refer to functions which didn't exist in 0.6.3.
3165 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3166 can use the new (as of 0.6.3) transform for SOME into MAP into
3168 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3169 values to (INTEGER 0 1000000), so that the time code will no longer
3170 occasionally get blown up by Linux returning 1000000 microseconds
3171 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3172 its output conform to the ANSI spec. (Alas, this makes its output
3173 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3174 * A full call to MAP NIL with a single sequence argument no longer conses.
3175 * fixes to problems pointed out by Martin Atzmueller:
3176 * The manual page no longer talks about multiprocessing as though
3177 it were currently supported.
3178 * The ILISP support patches have been removed from the distribution,
3179 because as of version 5.10.1, ILISP now supports SBCL without us
3180 having to maintain patches.
3181 * added a modified version of Raymond Toy's recent CMU CL patch for
3182 EQUALP comparison of HASH-TABLE
3184 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3186 * The system still can't cross-compile itself with
3187 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3188 entails), but at least it can compile more of itself that way
3189 than it used to be able to, and various buglets which were uncovered
3190 by trying to cross-compile itself that way have now been fixed.
3191 * This release breaks binary compatibility again. This time
3192 at least I've incremented the FASL file format version to 2, so that the
3193 problem can be detected reliably instead of just causing weird errors.
3194 * various new style warnings:
3195 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3196 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3197 ** using the lexical binding of a variable named in the *FOO* style
3198 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3199 as specified by ANSI.
3200 * *RANDOM-STATE* is no longer automatically initialized from
3201 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3202 default behavior of the system is to repeat its behavior every time
3203 it's run. If you'd like to change this behavior, you can always
3204 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3205 old convention there was no comparably easy way to get the system to
3206 repeat its behavior every time it was run.
3207 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3208 FTYPE declarations has been removed, in favor of their ANSI
3209 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3210 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3211 the types of their sequence arguments can be determined at compile time.
3212 This is done through a new open code expansion for MAP which eliminates
3213 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3214 argument types can be determined at compile time.
3215 * The optimizer now transforms COERCE into an identity operation when it
3216 can prove that the coerced object is already of the correct type. (This
3217 can be a win for machine generated code, including the output of other
3218 optimization transforms, such as the MAP transform above.)
3219 * Credit information has been moved from source file headers into CREDITS.
3220 * Source file headers have been made more standard.
3221 * The CASE macro now compiles without complaining even when it has
3224 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3226 * (Note that the way that the PCL macroexpansions were rewritten
3227 to accommodate the change in DEFGENERIC below breaks binary
3228 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3229 not run under sbcl-0.6.2. Once we get out of alpha releases,
3230 i.e. hit release 1.0.0, we'll probably try to maintain binary
3231 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3232 and sbcl-1.4.4. Until then, however, it might be fairly common
3233 for maintenance releases to break binary compatibility.)
3234 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3236 * The reporting of unhandled signals has been changed to print some
3237 explanatory text as well as the report form. (Previously only
3238 the report form was printed.)
3239 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3240 it defines, so that the compiler no longer issues undefined function
3241 warnings for compiled-but-not-yet-loaded generic functions.
3242 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3243 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3244 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3245 section of the ANSI spec are supported.
3246 * The "" nickname for the "KEYWORD" package has been removed.
3247 The reader still handles symbol tokens which begin with a package marker
3248 as keywords, but it doesn't expose its mechanism for doing so in the
3249 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3250 * The system now issues STYLE-WARNINGs for contradictory TYPE
3251 proclamations. (Warnings for contradictory FTYPE proclamations would
3252 be nice too, but those can't be done usefully unless the type system
3253 is made smarter about FUNCTION types.)
3254 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3255 systematized, so that "*target-*.lisp is supposed to exist only on the
3256 target and imply that there's a related file which exists on the
3257 host, and *host-*.lisp is supposed to exist only on the host and imply
3258 that there's a related file which exists on the target. This involves a
3259 lot of renaming. Hopefully the acute confusion caused by the renaming
3260 will be justified by the reduction in chronic confusion..
3261 ** runtime-type.lisp -> early-target-type.lisp
3262 ** target-type.lisp -> late-target-type.lisp
3263 ** early-host-format.lisp -> early-format.lisp
3264 ** late-host-format.lisp -> late-format.lisp
3265 ** host-error.lisp -> misc-error.lisp
3266 ** early-error.lisp -> early-target-error.lisp
3267 ** late-error.lisp -> late-target-error.lisp
3268 ** host-defboot.lisp -> early-defboot.lisp
3269 ** code/misc.lisp -> code/target-misc.lisp
3270 ** code/host-misc.lisp -> code/misc.lisp
3271 ** code/numbers.lisp -> code/target-numbers.lisp
3272 ** code/early-numbers.lisp -> numbers.lisp
3273 ** early-host-type.lisp -> early-type.lisp
3274 ** late-host-type.lisp -> late-type.lisp
3275 ** host-typep.lisp -> typep.lisp
3276 ** load.lisp -> target-load.lisp
3277 ** host-load.lisp -> load.lisp
3278 ** host-disassem.lisp -> disassem.lisp
3279 ** host-insts.lisp -> insts.lisp
3280 ** byte-comp.lisp -> target-byte-comp.lisp
3281 ** host-byte-comp.lisp -> byte-comp.lisp
3282 ** host-signal.lisp -> signal.lisp
3283 ** host-defstruct.lisp -> defstruct.lisp
3284 ** late-target-type.lisp -> deftypes-for-target.lisp
3285 Furthermore, several other previously target-only files foo.lisp (e.g.
3286 hash-table.lisp and random.lisp) have been split into a target-and-host
3287 foo.lisp file and a target-only target-foo.lisp file, with their key type
3288 definitions in the target-and-host part, so that the cross-compiler will
3289 know more about target types.
3290 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3291 gone away. In their place are various *BACKEND-FOO* variables
3292 corresponding to the slots of the old structure.
3293 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3294 into the target SBCL has been fixed.
3295 * The chill.lisp system for loading cold code into a running SBCL
3297 * Support for the CMU CL "scavenger hook" extension has been removed.
3298 (It was undocumented and unused in the CMU CL sources that SBCL was
3299 derived from, and stale in sbcl-0.6.1.)
3300 * Various errors in the cross-compiler type system were detected
3301 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3302 (enabling various consistency checks). Many of them were fixed,
3303 but some hard problems remain, so the compiler is back to
3304 running without *TYPE-SYSTEM-INITIALIZED* for now.
3305 * As part of the cross-compiler type system cleanup, I implemented
3306 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3307 * I have started adding UNCROSS calls throughout the type system
3308 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3309 on cross-compiler input files will be able to go away, and various
3311 * CONSTANTP now returns true for quoted forms (as explicitly required
3314 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3316 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3317 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3318 (MAKE-LIST -1). In the longer term, it ought to become true
3319 that declarations are assertions even at SAFETY 1. For now, it's not
3320 quite true even at SAFETY 3, but it's at least more nearly true..
3321 (Note that this change seems to increases the size of the system by
3322 O(5%) and to decrease the speed of the compiler by 20% or more.)
3323 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3324 for the problem of printing dozens of lines of distracting information
3325 about low-level system machinery as part of the top stack frame
3326 on entry to the debugger when an undefined function was called.
3327 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3328 is not reset to COMMON-LISP-USER.
3329 * Compilation of stuff related to dyncount.lisp has been made conditional
3330 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3331 smaller. The various dyncount-related symbols have been moved into
3332 a new "SB-DYNCOUNT" package.
3333 * tty-inspect.lisp has been renamed to inspect.lisp.
3334 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3335 feature has gone away. (When we eventually port to other flavors of
3336 libc and/or Unix, we'll try to make the differences between flavors
3337 invisible at the user level.)
3338 * Various other *FEATURES* tags, and/or their associated conditionals,
3339 have been removed if obsolescent, or given better documentation, or
3340 sometimes given more-mnemonic names.
3342 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3344 * tidied up "make.sh" script
3345 * tidied up system directory structure
3346 * better "clean.sh" behavior
3347 * added doc/FOR-CMUCL-DEVELOPERS
3348 * many many small tweaks to output format, e.g. removing possibly-confusing
3349 trailing #\. character in DESCRIBE-INSTANCE
3350 * (EQUALP #\A 'A) no longer signals an error.
3351 * new hashing code, including EQUALP hashing
3352 * tidied up Lisp initialization and toplevel
3353 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3354 * command line argument processing
3355 * added POSIX-GETENV function to deal with Unix-ish environment variables
3356 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3357 terminating SBCL on EOF
3358 * non-verbose GC by default
3359 * There is no more "sbcl" shell script; the sbcl file is now the C
3360 runtime executable (just like CMU CL).
3361 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3363 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3364 symbol and package information as Lisp native symbol and package objects
3365 instead of strings naming symbols and strings naming packages. This way,
3366 whenever packages are renamed (as in warm init), debug information is
3367 transformed along with everything else.
3368 * tweaked the optimization policy declarations which control the building
3369 of SBCL itself. Now, among other things, the system no longer saves
3370 source location debugging information. (This helps two problems at once
3371 by reducing SBCL size and by keeping SBCL from trying to look for its
3372 sources -- which may not exist -- when reporting errors.)
3373 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3374 debugging and testing purposes
3375 * cleaned up printing, making the printer call PRINT-OBJECT for
3376 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3377 methods, giving nearly-ANSI behavior
3378 * converted almost all special variables to use *FOO* naming convention
3379 * deleted PARSE-TIME functionality, since it can be done portably
3380 * moved some files out of cold init into warm init
3381 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3383 * regularized formatting of source files
3384 * added an install.sh script
3385 * fixed ridiculous memory usage of cross-compiler by making
3386 compiler/alloc.lisp not try to do pooling unless it can hook
3387 itself into the GC of the cross-compilation host. Now the system
3388 builds nicely on my old laptop.
3389 * added :SB-ALLOC in target-features.lisp-expr
3390 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3391 was not implemented)
3392 * re-did condition handling and note reporting in the compiler. Notes
3393 are no longer handled by signalling conditions. Style warnings
3394 and warnings are handled more correctly and reported in such a way
3395 that it's easy to find one or the other in your output (so that you
3396 can e.g. figure out which of many problems caused COMPILE-FILE to
3398 * changed the severity of several compiler warnings from full WARNING
3399 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3400 compiler note reporting so that it doesn't use the condition system
3401 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3402 COMPILE-FILE command)
3403 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3404 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3405 no longer supported, since their implementation was deeply intertwingled
3406 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3407 * removed bogus "support" for compiler macros named (SETF FOO), and
3408 removed the compiler macro for SETF INFO (but only after making a fool
3409 of myself on the cmucl-imp mailing list by posting a bogus patch for
3410 DEFINE-COMPILER-MACRO..)
3411 * Compiled files containing forms which have side effects on the Lisp
3412 reader (such as DEFPACKAGE forms) are now handled more correctly.
3413 (Compiler queuing of top level lambdas has been suppressed by setting
3414 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3415 * deleted various currently-unused source files, e.g. gengc.lisp. They
3416 may be added back at some point e.g. when porting to other architectures,
3417 but until they are it's distracting to distribute them and to try to
3419 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3420 there were so many of them they're just distractions, and UNCROSS is
3421 known to be able to handle the current sources
3422 * moved PROFILE functionality into TRACE, so that it will be clear
3423 how the wrapping and unwrapping of functions when you profile them
3424 interacts with the wrapping and unwrapping of functions when you
3425 trace them. (Actually, the functionality isn't there yet, but at least
3426 the interface specification is there. Hopefully, the functionality will
3427 arrive with some future maintenance release.)
3428 * removed host-oops.lisp
3429 * changed signature of QUIT function to allow UNIX-CODE argument
3430 * fixed READ-SEQUENCE bug
3431 * tweaked verbose GC output so that it looks more like the progress
3432 output that ANSI specifies for functions like LOAD
3433 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3434 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob