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)
30 ** bug fix: parent thread now can be gc'ed even with a live
32 ** bug fix: binding a special with PROGV to no value is not
33 the same as that symbol not having been bound (thanks to
35 ** bug fix: binding specials is thread safe (thanks to
37 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
38 and SB-SPROF do not die with 'no handler for signal XX in
39 interrupt_handle_now(..)' anymore
40 ** bug fix: WITH-TIMEOUT works with multiple running threads
41 * fixed some bugs revealed by Paul Dietz' test suite:
42 ** ENSURE-GENERIC-FUNCTION should take a method class object for
43 the :method-class keyword argument.
45 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
46 * new port: the Solaris operating system on x86 processors is now
47 mostly supported, though some rough edges in the environment
48 remain. (thanks to Daisuke Homma)
49 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
50 space for its heap. (thanks to Thiemo Seufer)
51 * optimization: SBCL on MIPS platforms now supports dynamic-extent
52 closures. (thanks to Thiemo Seufer)
53 * minor incompatible change: eof selects abort in the debugger.
54 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
55 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
57 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
58 deprecated, and will go away in a future revision of SBCL.
59 * minor incompatible change: GC-ON and GC-OFF are no longer
60 implemented with a counter, it does not matter how many times gc
62 * bug fix: discriminating functions for generic function classes
63 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
64 make invalid assumptions about method precedence order. (reported
66 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
67 platforms. (thanks to Thiemo Seufer)
68 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
69 * bug fix: correct pathname printing with printer escaping is on.
70 (thanks to Kevin Reid)
71 * bug fix: complex VOP definitions in "user-space" no longer trigger
72 package locks. (reported by Zach Beane)
73 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
74 for funcallable-instance functions, and is no different from
76 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
77 characters in the current readtable. (reported by Nicholas Neuss)
78 * bug fix: SUBTYPEP on various CONS types returns more a more
79 accurate acknowledgment of its certainty.
80 * optimizations: REMOVE-DUPLICATES now runs in linear time on
81 lists in some cases. This partially fixes bug 384.
82 * flush all standard streams before prompting in the REPL and the
84 * bug fix: signal handling and triggering gc do not conflict
85 directly anymore, in particular a high frequency sb-sprof does
86 not prevent gc from running
87 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
88 approximation for timezone and DST information between the
89 universal time epoch and the smallest negative 32-bit time_t.
90 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
91 year 1899 so long as the encoded time is non-negative.
92 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
93 the correct answer for non-constant rotate amounts when performing
94 an inline 32-bit rotation.
96 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
97 there is only one thread in the session
98 ** bug fix: memory leak for streams created in one thread and
100 ** bug fix: lockup when compiled with gcc4
101 ** bug fix: race that allows the gc to be triggered when gc is
103 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
104 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
106 ** bug fix: finished threads are gc'ed properly
107 * fixed some bugs revealed by Paul Dietz' test suite:
108 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
109 has arguments with a different set of applicable methods from
110 the orignal arguments.
111 ** The type error thrown by MAP now has a correct expected-type
113 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
114 name a compiled function.
115 ** Three MISC tests where a large bignum was improperly coerced to
116 a float that couldn't represent that bignum during type
117 derivation were fixed.
118 ** SUBTYPEP can now handle types involving the negation of a
119 list-form FUNCTION type.
120 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
121 as (COMPLEX (AND RATIO (NOT FIXNUM))).
122 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
124 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
125 * New feature: Experimental support for bivalent streams: streams
126 opened with :element-type :default now allow character and binary
127 (unsigned-byte 8) I/O
128 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
129 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
131 * bug fix: no more highly sporadic "couldn't check whether ~S is
132 readable" when reading a stream and an interrupt hits in the middle
133 of a select system call
134 * compiler better recognizes complex arrays (reported by Eduardo
136 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
137 for variables on SBCL built with linkage-tables. (reported by Luis
139 * various error reporting improvements.
140 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
141 (Thanks to Thiemo Seufer)
142 * bug fix: nested reader invokations work correctly
143 * bug fix: it is possible to have simultaneous references to foreign
144 code and foreign data with the same name.
146 ** added x86-64 support
147 ** incompatible change: the threading api now works with thread
148 objects instead of thread ids
149 ** bug fix: threads are protected from signals and interruption when
150 starting up or going down
151 ** bug fix: a race where an exiting thread could lose its stack to gc
152 ** bug fix: don't halt on infinite error in threads if possible
153 ** fixed numerous gc deadlocks introduced in the pthread merge
154 ** bug fix: fixed thread safety issues in read and print
155 ** bug fix: debugger doesn't hang on session lock if interrupted at
156 an inappropriate moment
157 ** bug fix: run-program is now thread safe(r)
158 ** bug fix: inner with-recursive-lock no longer releases the mutex
159 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
160 * fixed some bugs revealed by Paul Dietz' test suite:
161 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
162 EXPECTED-TYPE slots filled.
163 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
165 ** MAKE-CONDITION accepts classes as type-designators.
166 ** COMPILE may never return NIL.
167 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
168 range before calling Unix time functions
170 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
171 * numerous signal handling fixes to increase stability
172 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
173 * minor incompatible change: we now correctly canonize default
174 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
175 as per the MOP, rather than the historical (INITARG INITFUNCTION
176 INITFORM). (reported by Bruno Haible)
177 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
178 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
179 for more information.
180 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
181 pathname is a directory pathname.
182 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
183 * dynamic space size on PPC has been increased to 768Mb. (thanks to
185 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
186 :METACLASS argument in addition to a class name. (reported by
187 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
188 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
189 (reported by Bruno Haible; patch from Gerd Moellmann)
190 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
192 * bug fix: more cleanups to the floating point exception handling on
193 x86-64 (thanks to James Knight)
194 * bug fix: the compiler does not try to constant fold calls of
195 COERCE and BIT-* functions when they return freshly constructed
196 objects. (reported by Nikodemus Siivola and Paul Dietz)
197 * optimization: improved the MIPS versions of generic (in the
198 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
199 * optimization: direct conversion of (unsigned-byte 32) to floats on
200 the PowerPC platform.
201 * optimization: structure instances with raw slots now use less
202 memory, and probably show better memory locality. (thanks to
204 * optimization: DYNAMIC-EXTENT declarations for lists and closures
205 are treated as requests for stack allocation on the x86-64,
206 Alpha-32, PPC and SPARC platforms.
207 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
208 the wrong file descriptor; implementation of SOCKET-OPEN-P.
209 (thanks to Tony Martinez)
211 ** gcing a dead thread can no longer lead to lockups
212 ** threads block signals until they are set up properly
213 ** errno is no longer shared by threads
214 ** interrupt-thread restores the eflags register on x86
215 ** fixed some lockups due to gc/thread interaction
216 * fixed some bugs revealed by Paul Dietz' test suite:
217 ** invalid dotted lists no longer raise a read error when
219 ** signal an error if a symbol that names a declaration is used
220 as the name of a type, or vice versa
221 ** allow using the (declare (typespec var*)) abbreviation for
222 (declare (type typespec var*)) with all type specifiers
223 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
224 ** always return NIL from PPRINT-POP when OBJECT is NIL
225 ** don't signal errors when pretty-printing malformed LABELS,
226 FLET or MACROLET forms
227 ** declarations in a DOLIST body are also in scope for the
229 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
231 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
234 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
235 * fixed cross-compiler leakages that prevented building a 32-bit
236 target with a 64-bit host compiler.
237 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
238 opened with :IF-EXISTS :OVERWRITE.
239 * fixed bug 281: error for an invalid qualifier in a short-form method
240 combination method is not signalled until the faulty method is called.
241 * bug fix: iteration variable type inferrer failed to deal with open
242 intervals. (reported by Alan Shields)
243 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
244 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
245 or more runtime options were provided to the sbcl binary.
246 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
247 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
248 * optimization: there's now a fast-path for fixnum arguments in the
249 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
250 * optimization: the code generated on x86-64 is more compact thanks
251 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
252 * bug fix: floating-point exception handling now partly works on
253 x86-64. (Thanks to James Knight)
254 * improvement to the MIPS backend from Thiemo Seufer: C-style
255 64-bit long long arguments and return values to alien functions
257 * fixed some bugs revealed by Paul Dietz' test suite:
258 ** the type-error signalled from WARN has a filled-in DATUM slot.
259 ** the type-error required when a stream is not associated with
260 a file has the stream as its datum.
261 ** type-errors on single-floats on x86-64 no longer have
262 :INVALID-OBJECT as the datum
263 ** the type-errors signalled for invalid function names now have
264 a correct expected type
265 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
266 for structures defined with a :TYPE. Documentation strings for
267 typed structures are no longer immediately discarded
268 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
269 broadcast streams and synonym streams. FILE-LENGTH now also works
270 on broadcast streams.
272 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
273 * incompatible change: the --noprogrammer option, deprecated since
274 version 0.7.5, has been removed. Please use the equivalent
275 --disable-debugger option instead.
276 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
278 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
279 inoperational for a while now) has been completely removed.
280 * null lexical environments are now printed as #<NULL-LEXENV>,
281 significantly reducing the amount of clutter in typical backtraces.
282 * documentation on weak pointers, finalization, and after GC hooks
283 has been added to the manual.
284 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
285 as other specialized array types, is much faster. SUBSEQ and
286 COPY-SEQ on such arrays have also been sped up.
287 * optimization: EQL is now more efficient when at least other argument
288 is known to be of type (OR FIXNUM (NOT NUMBER)).
289 * fixed bug: the runtime is now less vulnerable to changes in the
290 size of the SBCL object on OS X, and virtual memory is reserved for
291 all spaces that need to be at a fixed address.
292 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
293 * fixed bug: finalizers and after GC hooks that cause consing are now
295 * fixed bug: compiler error messages and summaries are now printed to
296 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
297 * fixed inference of the upper bound of an iteration variable.
298 (reported by Rajat Datta).
299 * fixed bug 211e: calling local functions with duplicated constant
300 keyword argument no longer causes a bogus style warning about an
302 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
303 ftype declaration to be disregarded. (reported by Dave Roberts)
304 * fixed bug 373: caused by erronous compilation of references to alien
305 variables in the runtime on ppc/darwin.
306 * fixed bug 376: CONJUGATE type deriver.
307 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
308 (reported by Baughn on #lisp)
309 * a cross-compiler bug on non-x86 platforms has been identified and
310 fixed. (thanks to Bruno Haible)
311 * improvements to the MIPS runtime code for increased stability.
312 (thanks to Thiemo Seufer)
313 * increased the maximimum compact environment size to allow
314 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
315 * partial workaround for bug 135: don't name the function we're
316 calling for hairy cases of EVAL, so as not to accumulate one environment
317 entry per call to EVAL. (reported by Kevin Reid)
318 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
319 * optimization: type testing for non-vector arrays should be faster.
320 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
322 * bug fix: setting 31st element of a bit vector to zero did not work
324 * bug fix: redefining a class definition which failed due to a
325 previous accessor / function clash now works (but see BUGS entry
326 #380 for more problems in this area). (thanks to Zach Beane)
327 * the long form of DEFINE-METHOD-COMBINATION disables method group
328 checking when given a single method group with pattern *. (thanks
330 * on x86 compiler supports stack allocation of results of simple
331 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
332 * support for building and running on Mac OS X 10.4 "Tiger" has been added
333 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
334 * fixed some bugs related to Unicode integration:
335 ** the restarts for recovering from input and output encoding
336 errors only appear when there is in fact such an error to
338 * fixed some bugs revealed by Paul Dietz' test suite:
339 ** MISC.549 and similar: late transformation of unsafe type
340 assertions into derived types caused unexpected code
342 ** SCALE-FLOAT type deriver is less wrong.
343 ** type derivers for EXP, LOG and similar functions compute result
344 types for complex arguments better.
345 ** (MISC.563) CONJUGATE type deriver works for very restricted
347 ** out-of-line type testers for character strings are available.
348 ** EQUAL compiler transform understands specialness of objects
350 ** accessing double-floats stored on the stack now works on x86-64.
351 ** debugger internals could sometimes create invalid lispobjs,
352 resulting in GC crashes.
353 ** MISC.548: type check weakening can convert required type into
355 ** initialization forms for bindings are not in scope of free special
358 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
359 * incompatible change: thread support for non-NPTL systems has
360 been removed - locking is buggy and unreliable. A threaded
361 SBCL build will now warn at startup and refuse to create
362 new threads, unless futex support is detected in the kernel
363 * incompatible change: the top level REPL now has only an ABORT
364 restart associated with it, not TOPLEVEL and ABORT as it used to.
365 TOP and TOPLEVEL are now available as debugger commands for
366 returning to the top level.
367 * incompatible change: forms evaluated in the REPL now use the
368 global optimization policy.
369 * incompatible change: user- and system-initialization files are
370 no longer processed with LOAD, but by READ and EVAL; hence the
371 global optimization policy, startup package, readtable, etc,
373 * The .fasl file format number has been incremented because of
374 various incompatible changes.
375 * internal entry point details and argument counts no longer appear
376 in backtraces unless explicitly requested by setting
377 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
378 * built-in and standard functions no longer have names like "top
379 level local call to FOO".
380 * fixed bug 32: functions defined in non-null lexical environments
381 now have more legible printed representation
382 * fixed bug 33: functions defined in non-null lexical environemnts
383 are now more amenable to inspection by INSPECT.
384 * workaround for bug 354: XEPs no longer appear in backtraces unless
385 explicitly requested.
386 * fixed bug: receiving the signal which results from attempting to
387 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
388 SIGBUS on other *BSDs) on architectures where the C stack is also
389 the Lisp stack (x86 and x86-64) from foreign code no longer leads
390 to debugger confusion or wild instability. (reported by Cheuksan
392 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
393 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
394 (reported by Lutz Euler)
395 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
396 compiled code. (reported by Paul Dietz)
397 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
398 specialized parameter to a method to something that is not TYPEP
399 the specializer is now possible.
400 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
401 face of package deletion.
402 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
403 pathname. (reported by David Tolpin on c.l.l)
404 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
405 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
406 STRING-STREAM and FILE-STREAM, respectively.
407 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
408 than previously; in addition, its test suite is now run on build.
409 (thanks to Luke Gorrie)
410 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
411 * optimization: added a immediate representation for single-floats
413 * fixed some bugs revealed by Paul Dietz' test suite:
414 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
415 ** attempting to create a package with a colliding nickname causes
416 correctable errors to be signalled.
417 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
418 ** several x86-64 backend bugs related to sign-extension of immediate
421 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
422 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
423 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
425 * fixed loading of multiply forward-referenced layouts.
426 (thanks to Cheuksan Wang)
427 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
428 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
429 (see COMPILE-FILE documentation for details of :PRINT option)
430 * fixed bugs 19 and 317: fixed-format floating point printing is
431 more accurate. This also fixes a bug reported by Adam Warner
432 related to the ~@F format directive.
433 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
435 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
436 dispatch table if the from-char is a dispatch macro character.
437 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
438 by the wordsize no longer ignore the last word. (reported by Lutz
440 * fixed bug in type checking of dynamic-extent variables. (reported
442 * optimization: sequence traversal functions use their freedom to
443 coerce function designators to functions.
444 * optimization: code with many calls to CLOS methods specialized on
445 CLOS classes has had redundant type checks removed.
446 * optimization: added declarations to speed up operations that access
447 the internal character database (for example STRING-UPCASE)
448 * optimization: comparison operations between floats and sufficiently small
449 fixnums no longer create extra rationals
450 * fixed some bugs related to Unicode integration:
451 ** portions of multibyte characters at the end of buffers for
452 character-based file input are correctly transferred to the
453 start of the buffer at the next read.
454 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
455 passing it through to OPEN.
456 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
457 argument given to internal calls to OPEN.
458 ** the built-in comment readers (introduced by character sequences
459 ";" and "#|") are more forgiving to encoding errors; they will
460 STYLE-WARN and then attempt to resync the stream at a character
461 boundary. (thanks to Teemu Kalvas)
462 * fixed some bugs in the x86-64 port:
463 ** Negative short int return values from c-calls are sign-extended
465 ** The stack is aligned to 16-bytes for c-calls, as required by
466 the ABI. (thanks to Cheuksan Wang)
467 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
468 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
469 for enums. (thanks to Vincent Arkesteijn)
470 ** Multiple small optimizations and bugfixes for floating point
472 * fixed some bugs revealed by Paul Dietz' test suite:
473 ** Space, Tab, Linefeed, Return and Page have the invalid
474 secondary constituent character trait.
475 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
477 ** WITH-INPUT-FROM-STRING should only update the index place on
479 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
481 ** Bit-array operations (BIT-AND and similar) worked incorrectly
482 with one-dimensional arrays with fill pointers.
483 ** TYPE-OF failed on a complex with an integer realpart and a
485 ** compiler failure during type inference for the code of form
486 (IF (EQL X (THE ...)) ...) (MISC.535).
488 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
489 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
490 platform. The port supports 61-bit fixnums, large memory spaces
491 and reloading shared object files.
492 * enhancement: saving cores with foreign code loaded is now
493 supported on ppc/Darwin in addition to the previously supported
495 * enhancement: the statistical profiler now walks deeper into the
496 call stack for more meaningful call-graphs and accrued time
497 reports (x86/x86-64 only). It also now reports time spent in
499 * enhancement: it is now possible to trace most individual methods
500 of a generic function in addition to tracing the generic function
502 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
504 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
505 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
506 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
507 produces an error. (thanks to Vincent Arkesteijn)
508 * bug fix: NAMESTRING on pathnames with :WILD components in their
509 directories works correctly. (thanks to Artem V. Andreev)
510 * fixed bug 125: compiler preserves identity of closures. (reported
512 * bug fixed: functions with &REST arguments sometimes failed with
513 "Undefined function" when compiled with (DEBUG 3). (reported by
515 * bug fix: overflow during compiling of setting element of a bit
516 vector with constant index and value. (reported by Timmy Douglas)
517 * build fix: fixed the dependence on *LOAD-PATHNAME* and
518 *COMPILE-FILE-PATHNAME* being absolute pathnames.
519 * on x86 compiler partially supports stack allocation of dynamic-extent
521 * GO and RETURN-FROM do not check the extent of their exit points
522 when compiled with SAFETY 0.
523 * fixed some bugs related to Unicode integration:
524 ** encoding and decoding errors are now much more robustly
525 handled; it should now be possible to recover even from invalid
526 input or output to the terminal. (thanks to Teemu Kalvas)
527 ** provided a first cut at implementing STRING-TO-OCTETS and
528 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
529 ** altered the SB-MD5 contributed module slightly, changing the
530 interface just enough for it to be supportable for builds where
531 lisp characters are not eight bits.
532 * fixed some bugs revealed by Paul Dietz' test suite:
533 ** the FORMATTER-generated functions for ~V[ conditionals require
534 the correct number of arguments.
535 ** READ-FROM-STRING returns the mandated second value when applied
536 to displaced strings.
537 ** the #\Rubout and #\Backspace characters are treated as invalid
538 constituent characters by the tokenizer.
540 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
541 * new feature: reloading changed shared object files with
542 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
543 * new feature: references to foreign variables and functions
544 can now be compiled and loaded before the corresponding shared
545 object file is loaded, as long as the foreign definitions are
546 available at runtime.
547 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
548 supported. (thanks to Dan Debertin)
549 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
550 just keywords. (thanks to Vincent Arkesteijn)
551 * fixed bug #331: structure-class instances corresponding to
552 DEFSTRUCT forms are now created eagerly.
553 * fixed bug #345: backtraces from calls to undefined functions work
554 on x86 as well. Related bug #61 is now also partially fixed on x86
555 (backtraces from throws to unknown catch tags.)
556 * bug fix: lambda-list parsing is now stricter vrt. order and number
557 of lambda-list keywords.
558 * bug fix: as specified by AMOP, an error is signalled if a
559 class-option appears multiple times in a DEFCLASS form. (reported
561 * bug fix: RANDOM can be compiled when the compiler derives the type
562 of its numeric argument as a disjoint set of small integers.
563 (reported by Paul Dietz)
564 * bug fix: starting a core saved with shared objects loaded when
565 those objects are not available no longer causes threaded SBCL to
566 hang. (reported by Sean Ross)
567 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
569 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
570 argument, which must always be NIL. (reported by Kalle Niemitalo)
571 * bug fix: printing 1.0d+23 no longer results in an error.
572 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
573 * bug fix: structure slot setters preserve evaluation order. (thanks
575 * fixed some bugs related to Unicode integration:
576 ** RUN-PROGRAM can allow its child to take input from a Lisp
577 stream. (reported by Stefan Scholl)
578 ** ASDF-INSTALL successfully downloads .tar.gz files in all
579 locales. (reported by Ken Causey)
580 * fixed some bugs revealed by Paul Dietz' test suite:
581 ** INCF, DECF and REMF evaluate their place form as specified in
583 ** FORMATTER expands ~{ ~} iteration directives with V or #
584 parameters correctly.
585 ** FORMATTER deals with the ~@[ ~] conditional directive where the
586 consequent uses no arguments correctly.
587 ** the system has a partial understanding of the (COMPLEX RATIO)
589 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
592 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
593 * new feature: a build-time option (controlled by the :SB-UNICODE
594 keyword feature, enabled by default) for building the system with
595 support for the entire 21-bit character space defined by the
597 * new feature: the system now has rudimentary external-format
598 support; the primary user-visible change at this time is that
599 characters with the high bit set (such as lower-case-e-acute) will
600 print correctly to a terminal in a UTF-8 environment.
601 * minor incompatible change: BASE-CHAR no longer names a class;
602 however, CHARACTER continues to do so, as required by ANSI.
603 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
604 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
605 has been moved to the SB-EXT package (temporarily re-exported from
607 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
608 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
610 * fixed bug #351: better error-handlind and reporting for malformed
612 * fixed bug #350: bignum-printing is now more memory-efficient,
613 allowing printing of very large bignums, eg. (expt 2 10000000).
614 (reported by Bruno Haible)
615 * fixed bug #302: better primitive-type selection for intersection
617 * fixed bug #308: non-graphic characters now all have names, as
618 required. (reported by Bruno Haible)
619 * bug fix: redefining a class with different superclasses now correctly
620 removes it from the direct-subclasses of its previous superclasses.
621 (reported by David Morse)
622 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
623 new value now works. (reported by Bruno Haible)
624 * bug fix: correct canonicalization of multiple non-standard slot
625 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
626 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
627 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
628 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
629 now exists, an signals an error.
630 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
631 during macroexpansion for non-iteration variables as well. (reported
632 by Bruno Haible for CMUCL)
633 * bug fix: Cyclic structures and unprintable objects in compiler
634 messages no longer cause errors. (reported by Bruno Haible)
635 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
636 and READ-FROM-STRING all now return a primary value of NIL if
637 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
638 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
639 (reported by Bruno Haible for CMUCL)
640 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
641 do not designate a condition. (reported by Bruno Haible for
643 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
644 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
645 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
646 permitting resolution in favour of any of the conflicting symbols.
647 (reported by Bruno Haible for CMUCL)
648 * bug fix: EQUAL compiler optimizations is less aggressive on
649 strings which can potentially compare true despite having distinct
650 specialized array element types.
651 * bug fix: unit enumerations can be defined without dividing by
652 zero. (thanks to Vincent Arkesteijn)
653 * FORMAT compile-time argument count checking has been enhanced.
654 (report from Bruno Haible for CMUCL)
655 * a partial workaround for the bug 262: the compiler does not try to
656 inline-expand a local function doing RETURN-FROM from a deleted
657 BLOCK. (thanks to Peter Denno for the bug report and to David
658 Wragg for the simple test case)
659 * fixed some bugs revealed by Paul Dietz' test suite:
660 ** RENAME-PACKAGE allows all package designators as new package
662 ** constraint propagation and lambda variable substitution are
663 more cautious in dealing with partially deleted code.
664 ** compiler sometimes lost reoptimization passes.
665 ** CERROR, given a condition as condition designator, passes its
666 remaining arguments to the continue format control without
668 ** Case-altering FORMAT directives work correctly on non-ASCII
670 ** The REAL type specifier handles bounds outside the floating
671 point ranges without signalling FLOATING-POINT-OVERFLOW.
672 ** Functions with IR1-transformations can create intercomponent
673 references to global functions.
674 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
676 ** FORMAT ~R treats a nil value for its first parameter correctly.
678 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
679 * enhancement: saving cores with foreign code loaded is now
680 supported on x86/NetBSD and sparc/Linux in addition to the previously
682 * bug fix: on some platforms repeated installations caused multiple
683 copies of HTML documentation to be installed -- should not happen
684 any more. (reported by Stefan Scholl)
685 * bug fix: parsing self-recursive alien record types multiple times
686 no longer causes infinite recursion. (reported by Thomas F. Burdick,
687 original patch by Helmut Eller for CMUCL)
688 * bug fix: stack-exhaustion detection works now on NetBSD as well.
689 (thanks to Richard Kreuter)
690 * bug fix: defining classes whose accessors are methods on existing
691 generic functions in other (locked) packages no longer signals
692 bogus package lock violations. (reported by François-René Rideau)
693 * bug fix: special variables as DEFMETHOD parameters no longer have
694 associated bogus type declarations. (reported by David Wragg and
696 * bug fix: read-write consistency on streams of element-type
697 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
698 * bug fix: redefiniton of the only method of a generic function with
699 no DEFGENERIC no longer emits a full WARNING. In addition,
700 redefinition of generic functions with no DEFGENERIC to an
701 incompatible lambda list now signals an error. (thanks to Zach
703 * bug fix: DEFGENERIC now works even when there's a function of the
704 same name in an enclosing lexical environment. (thanks to Zach
706 * fixed compiler failure, caused by instrumenting code during
707 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
708 * optimization: added loop analysis and improved register allocation
709 * fixed some bugs revealed by Paul Dietz' test suite:
710 ** POSITION on displaced vectors with non-zero displacement
711 returns the right answer.
712 ** (SIMPLE-STRING) is a valid type specifier for sequence
714 ** *PRINT-LEVEL* handling for slotless structures is pedantically
716 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
717 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
719 ** The justification version of the FORMAT ~< directive treats
720 non-zero minpad parameter correctly.
722 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
723 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
724 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
725 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
726 the supported interface.
727 * new feature: Single-stepping of code compiled with DEBUG 2 or
728 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
729 * new feature: saving cores with foreign code loaded is now
730 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
731 Timothy Moore's work for CMUCL)
732 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
733 parameters to * instead of NIL if no initform is supplied.
734 (reported by Johan Bockgaard)
735 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
736 correctly when FUNCALL appears as the car of the form. Note:
737 despite this FUNCALL forms are not currently subject to
738 compiler-macro expansion. (port of Raymond Toy's fix for the
739 same from CMUCL, reported by Johan Bockgaard)
740 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
741 (thanks for Teemu Kalvas)
742 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
743 case instead of terminating the iteration (thanks for Julian
744 Squires, Sean Champ and Raymond Toy)
745 * bug fix: incorrect expansion of defgeneric that caused a style
746 warning. (thanks for Zach Beane)
747 * bug fix: slot accessor effective method computation works properly
748 for classes with multiple non-standard applicable methods on
749 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
750 * on x86 compiler supports stack allocation of results of LIST and
751 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
752 CMUCL implementation by Gerd Moellmann)
753 * fixed some bugs revealed by Paul Dietz' test suite:
754 ** FORMAT strings with both the ~<~:;~> form of the justification
755 directive and pretty-printing directives cause an error.
757 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
758 * incompatible change: the internal functions
759 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
760 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
761 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
762 instead of the old functions.
763 * new feature: on platforms where "dladdr" is available foreign
764 function names now appear in backtraces. (based on Helmut Eller's
766 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
767 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
769 * documentation: ASDF manual is now distributed with SBCL.
770 * bug fix: SBCL can now load its contributed modules with REQUIRE
771 even if the system-provided entries have been removed from
772 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
774 * bug fix: backtraces involving undefined functions or assembly
775 routines are more informative. (thanks to Brian Downing)
776 * bug fix: mutually referent alien structures now work correctly.
777 (reported by Rick Taube)
778 * bug fix: structures defined by WITH-ALIEN can be referred to
779 within other definitions in the same WITH-ALIEN.
780 * bug fix: division operators (MOD, TRUNCATE and the like) with
781 constant zero divisors and integer dividends no longer generate
783 * bug fix: provide default methods for INPUT-STREAM-P and
784 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
785 * bug fix: improve the detection and resolution of MOP metacycles
786 (where effective method computation depends on the generic function
787 having its effective method computed). (reported by Bruno Haible)
788 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
789 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
790 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
791 * bug fix: pathnames with relative directory components can be
792 represented relative to default pathnames.
793 * optimization: in taking the GCD of bignums, reduce the two bignums
794 to approximately the same size (using Euclid's algorithm) before
795 applying the more sophisticated binary GCD. (thanks to Juho
797 * optimization: COUNT on bitvectors now operates word-at-a-time.
798 * optimization: ASH with a positive, but not necessarily constant,
799 (leftwards) shift, when in a modular context, is compiled to a
801 * fixed some bugs revealed by Paul Dietz' test suite:
802 ** FORMAT variable parameters ("~V<char>") are defaulted properly
803 if the corresponding argument is NIL.
804 ** FORMAT directives accepting a minpad parameter treat negative
806 ** literal commas as character parameters to FORMAT directives are
808 ** literal spaces directly after ~<Newline> directives within a
809 format-logical-block (~:< ~@:>) do not induce :FILL-style
810 conditional newlines.
811 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
812 as specified: it no longer includes conditional newlines).
813 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
815 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
816 :PER-LINE-PREFIX argument does not evaluate to a string.
818 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
819 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
820 the manual for details; for now, package locks can be disabled by
821 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
822 if no major problems are found then it is likely that they will be
823 compiled in unconditionally.
824 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
825 unsupported operators on all platforms. To load a shared library
826 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
827 object file, link it into a shared library outside of SBCL and
828 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
830 * bug workaround: ROOM T and the GC were not working together
831 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
832 reporting facility aren't preserved. That reporting has been
833 disabled, so now until and unless someone figures out how to make it
834 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
835 * minor incompatible change: as threatened around sbcl-0.8.0, the
836 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
837 The symbols are also exported from SB-PCL for backwards
838 compatibility, but more so than before SB-PCL should be treated as
839 an implementation-internal package.
840 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
842 * the SB-SPROF contrib now works on (most) non-x86 architectures.
843 It is known as of this release not to work on the Alpha, however.
844 * fixed bug #167: errors signalled due to illegal syntax in method
845 bodies are now more legible.
846 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
847 designators and can hence be used with TYPEP.
848 * fixed bug #333: CHECK-TYPE now ensures that the type error
849 signalled, if any, has the right object to be accessed by
850 TYPE-ERROR-DATUM. (reported by Tony Martinez)
851 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
852 5.1.2.3 for multiple-value place subforms. (reported by Kalle
854 * fixed bug #334: programmatic addition of slots using specialized
855 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
856 and :ALLOCATION :CLASS slots.
857 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
858 float exponents. (rereported by Peter Seibel)
859 * fixed a bug: #\Space (and other whitespace characters) are no
860 longer considered to be macro characters in standard syntax by
862 * fixed bug: initialization of condition class metaobjects no longer
863 causes an instance of the condition to be created. (reported by
865 * fixed bug: it is now possible to have slots such that
866 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
867 description is neither :INSTANCE nor :CLASS.
868 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
869 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
870 SLOT-VALUE-USING-CLASS), and no longer causes errors with
871 non-standard SLOT-DEFINITION-ALLOCATION values.
872 * fixed bugs: various race conditions handling exiting threads.
873 CL-PPCRE's thread torture test now passes.
874 * fixed arguably-a-bug: GC time proportional to number of threads in
875 system even when most of them are idle
876 * optimization: improved performance of BIT and SBIT on bit-vectors.
877 * .fasl file incompatibility: The fasl file version number has
878 been incremented because of changes associated with package locks.
880 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
881 * minor incompatible change: the system no longer provides
882 optimization hints (conditions which are TYPEP
883 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
884 mode; these hints are still emitted when higher SPEED optimization
886 * new contrib module: a sampling profiler (profiling by statistical
887 sampling, rather than by instrumenting functions) is available as
888 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
889 * the behaviour of the standard function ED is now customizeable by
890 third parties through a hook variable: see ED's documentation
891 string for information on the protocol.
892 * the compiler no longer emits efficiency notes for (FUNCALL X)
893 when the type of X is uncertain under default optimization
895 * fixed bug 276: mutating a binding of a specialized parameter to a
896 method to something that is not TYPEP the specializer is now
898 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
899 exhibit the required behaviour on the x86 platform. (thanks to
900 Peter van Eynde, Eric Marsden and Bruno Haible)
901 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
902 even for difficult arguments. (reported by Peter Graves)
903 * fixed bug 141a: the backquote printer now descends quoted
905 * fixed another bug in backquote printing: no more destructive
906 modification of the form's list structure. (reported by Brian
908 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
909 move between its address being taken and the call to
910 interrupt_thread, fixing a crashing race condition.
911 * the SB-POSIX contrib implementation has been adjusted so that it
912 no longer exhibits ridiculously poor performance when constructing
913 instances corresponding to C structs.
915 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
916 * minor incompatible change: the sb-grovel contrib now treats C
917 structures as alien (in the sense of SB-ALIEN) objects rather than
918 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
919 has implications for memory management of client code
920 (sb-grovel-returned objects must now be manually managed) and for
921 type safety (alien objects now have full types).
922 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
923 used to control emission of compiler diagnostics, rather than the
924 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
925 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
926 quality should be considered deprecated.
927 * install.sh now installs the user manual as well
928 * (not quite a new documentable feature, but worth considering in
929 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
930 beginnings of a semantically meaningful condition hierarchy is
931 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
933 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
934 designator as the defaults argument.
935 * fixed bug: Displaced arrays whose displaced-to array has become
936 too small now cause ARRAY-DIMENSION to signal an error, providing
937 for safer bounds-checking. (reported by Bruno Haible)
938 * fixed bug: DEFCLASS slot definitions with identical :READER and
939 :WRITER names now signal a reasonable error. (reported by Thomas
941 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
943 * fixed bug: Invalid slot specification errors now print correctly.
944 (thanks to Zach Beane)
945 * fixed bug 320: Shared to local slot value transfers in class
946 redefinitions now happen corrently from superclasses as
947 well. (reported by Bruno Haible)
948 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
950 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
951 improper lists correctly. (reported by Bruno Haible)
952 * fixed bug 313: source-transform for <fun-name> was erroneously
953 applied to a call of a value of a variable with name <fun-name>.
954 (reported by Antonio Menezes Leitao)
955 * fixed bug 307: The obsolete instance protocol ensures that
956 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
957 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
958 remove dead unknown-values globs from the middle of the stack.
959 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
960 easier to resume long computations after using *BREAK-ON-SIGNALS*
961 to diagnose and fix failures. (thanks to Nikodemus Siivola)
962 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
963 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
964 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
965 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
966 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
968 * fixed a (fairly theoretical) bug in string printing: if
969 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
970 does not have array-element-type equal to the most general string
972 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
973 function name. (thanks to Nikodemus Siivola)
974 * fixed bug: compile-time format string checker failed on
975 non-closed ~{. (reported by Thomas F Burdick)
976 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
977 don't warn on legal loop code involving more than one aggregate
978 boolean. (thanks to Nikodemus Siivola)
979 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
980 in keyword position to MAKE-PACKAGE was not being reported as
982 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
983 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
985 * fixed bugs 280 and 312: the checking for multiple definitions in a
986 file is less likely to become confused by uses of inline
988 * fixed bug: the #S reader macro performs the keyword coercion
989 specified for slot names. (reported by Kalle Niemitalo)
990 * fixed bug: lambda lists may contain symbols whose names start with
991 & but are not lambda-list-keywords; their occurrence triggers a
993 * fixed bug 321: define-method-combination argument lambda lists do
994 not cause invalid code to be generated when &optional or &aux
995 variables have default values. (reported by Bruno Haible)
996 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
997 have CLOS classes; this appears also to have fixed reported
998 instability in stack exhaustion detection.
999 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1000 MUFFLE-WARNING when no associated restart is present is now
1002 * optimization: rearranged the expansion of various defining macros
1003 so that each expands into only one top-level form in a
1004 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1006 * optimization: used a previously little-used slot in symbols to
1007 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1009 * fixed some bugs revealed by Paul Dietz' test suite:
1010 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1012 ** Fixed an optimization bug related to inheritance of initforms
1013 from local to shared slots.
1014 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1015 ** CAST optimizer forgot to flush argument derived type.
1016 ** print/read consistency on floats is now orders of magnitude
1017 more likely. (thanks also to Bruno Haible for a similar report
1019 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1020 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1021 for objects of type REAL. Make it so.
1022 ** SXHASH is specified (infelicitously) to respect similarity,
1023 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1024 it so. (thanks to Markus Ziegler)
1025 ** on the Alpha, the compiler succeeds in compiling functions
1026 returning a known number of arguments greater than 63.
1027 ** fixed handling of invalid NIL arguments in keyword position
1029 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1031 ** provided a readably-printable representation for RANDOM-STATE
1033 ** ensured that pathnames, where they have a namestring, always
1034 print using #P"..." syntax.
1036 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1037 * Support for the forthcoming 2.0 version of the NetBSD kernel
1038 running on x86 hardware has been added. (thanks to Perry
1039 E. Metzger most immediately, and others for their past work)
1040 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1041 current release version; SBCL's OpenBSD support had been broken
1042 since about the time of OpenBSD's switch to ELF binary format.)
1043 * [placeholder for DX summary]
1044 ** user code with &REST lists declared dynamic-extent, under high
1045 speed or space and low safety and debug optimization policy.
1046 * The manual has been converted to Texinfo format and the debugger
1047 chapter from the cmucl manual has been added.
1048 * A facility has been added to extract documentation strings from
1049 sbcl and store them as Texinfo-formatted snippets for inclusion in
1050 the manual (via Texinfo's @include directive)
1051 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1052 the test case to Dave Roberts)
1053 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1054 pointers. (reported by Sean Ross)
1055 * bug fix: PROFILE output is printed nicely even for large numerical
1056 values. (thanks to Zach Beane)
1057 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1058 greater than 32 handle EOF correctly.
1059 * bug fix: on X86 an immediate argument of the IMUL instruction is
1060 correctly printed in disassembly. (thanks to Lutz Euler)
1061 * bug fix: class slots in redefined classes preserve their old
1062 values. (thanks to Bruno Haible and Nikodemus Siivola)
1063 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1064 succeeds. (reported by Marco Baringer)
1065 * fixed some bugs revealed by Paul Dietz' test suite:
1066 ** READ-SEQUENCE now works on ECHO-STREAMs.
1067 ** RATIONALIZE works more according to its specification. (thanks
1070 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1071 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1072 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1073 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1074 should matter to you only if you rebind the printer control
1075 variables and then find you want different bindings in the
1076 debugger than in the ordinary execution of your program.)
1077 * The runtime build system has been tweaked to support building
1078 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1079 assembler and linker. (thanks to Nikodemus Siivola)
1080 * Unbound, undefined, undeclared variables now trigger full
1081 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1082 more often programmer error than deliberate exploitation of undefined
1084 * optimization: the hash algorithm for strings has changed to one
1085 that is less vulnerable to spurious collisions. (thanks to Juho
1087 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1088 needless bounds checking. (thanks to Juho Snellman)
1089 * optimization: implemented multiplication as a modular
1090 (UNSIGNED-BYTE 32) operation on the PPC backend.
1091 * fixed some bugs revealed by Paul Dietz' test suite:
1092 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1094 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1095 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1096 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1098 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1100 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1102 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1103 their output stream on EOF from read.
1104 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1105 have been read to end-of-file.
1106 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1108 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1109 description of determination of which consecutive characters
1111 ** Printing the "Space" character with escaping on now yields "#\\ ",
1112 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1113 ** Reading floating-point numbers with *READ-BASE* set to a number
1114 less than 10 works correctly.
1115 ** Reading floating-point numbers with *READ-BASE* set to a number
1116 more than 10 works correctly.
1117 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1118 the readtable currently in effect.
1120 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1121 * minor incompatible change: parsing of namestrings on a physical
1122 (Unix) host has changed; numbers after the final #\. in a
1123 namestring are no longer interpreted as a version field. This is
1124 intented to be largely invisible to the user, except that the
1125 meaning of the namestring "*.*.*" has changed: it now refers to a
1126 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1127 should usually be replaced by
1128 (make-pathname :name :wild :type :wild :version :wild)
1129 with the added benefit that this is more likely to be portable.
1130 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1131 option to OPEN now signals an error if the file being opened
1132 exists; this may have an impact on existing code.
1133 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1134 signals received, on the PowerPC platforms (both Linux and
1135 Darwin). (thanks to Pierre Mai for pointing out the location of
1137 * several fixes on OS X: The system now builds and runs cleanly on
1138 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1139 (thanks to Brian Mastenbrook)
1140 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1141 second-resolution: integer multiples of 1/3600 between -24 and 24.
1142 (thanks to Vincent Arkesteijn)
1143 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1144 type when called with 1 argument; PEEK-CHAR checked type of
1145 PEEK-TYPE only after having read first character from a
1146 stream. (reported by Peter Graves)
1147 * bug fix: the garbage collector now has much better locality
1148 behaviour, and in particular no longer treats all memory as being
1149 exhausted when many small objects point to each other in a deeply
1151 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1152 recognized as being TYPEP their class.
1153 * bug fix: the PUSHNEW documentation string has been corrected.
1154 (thanks to Vincent Arkesteijn)
1155 * bug fix: defaulting of the value for the last of an atypically
1156 large number of multiple values being bound was not being
1157 performed correctly on the Alpha or PPC platforms
1158 * optimization: implemented multiplication as a modular
1159 (UNSIGNED-BYTE 32) operation on the x86 backend.
1160 * optimization: SEARCH on simple-base-strings can now be open-coded.
1161 (see also contrib/compiler-extras.lisp for inspiration for
1162 teaching the compiler about the Boyer-Moore algorithm).
1163 * value, returned by MAX (and MIN) called with several EQUALP, but
1164 not EQL, arguments now does not depend on compiler settings.
1165 * fixed some bugs revealed by Paul Dietz' test suite:
1166 ** in stack analysis liveness information is propagated from
1167 non-local entry points.
1168 ** pathwise CAST removing failed when the CAST node did not start
1170 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1171 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1173 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1174 host is already defined.
1175 ** RENAME-FILE works on streams instead of signalling an internal
1177 ** PEEK-CHAR uses the current readtable when determining whether
1178 or not a character is whitespace.
1179 ** MERGE-PATHNAMES handles the case when the pathname does not
1180 specify a name while the default-pathname specifies a version.
1181 ** Pathnames now stand a better chance of respecting print/read
1183 ** Attempting to use standardized file system operators with a
1184 pathname with invalid :DIRECTORY components signals a
1186 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1188 ** DIRECTORY on logical pathnames is more correct.
1189 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1190 signal a TYPE-ERROR if their argument is not a stream.
1191 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1192 designator argument does not designate a stream.
1193 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1194 examining the synonym.
1195 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1197 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1198 element-type larger than ([UN]SIGNED-BYTE 32).
1200 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1201 * When built with the :SB-FUTEX feature, threaded builds now take
1202 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1203 for faster/more reliable mutex and condition variable support.
1204 * Incompatible change (but one you probably shouldn't have been using
1205 anyway): the interface and code for arbitrating between multiple
1206 threads in the same user session has been redesigned.
1207 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1208 error when called without an explicit environment argument.
1209 (thanks to Nikodemus Siivola)
1210 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1211 confused about their position. (thanks to Adam Warner and Gerd
1213 * bug fix: Pretty printing backquoted forms with unquotations in the
1214 argument list position of various code constructs such as LAMBDA
1215 now works correctly. (reported by Paul Dietz)
1216 * bug fix: Pretty printing unquotations no longer loses all
1217 stream position information.
1218 * optimization: performance of string output streams is now less
1219 poor for multiple small sequence writes.
1220 * optimization: performance of CSUBTYPEP in the presence of complex
1221 expressions involving CONS and NOT many times has been improved.
1222 (reported by Paul Dietz)
1223 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1225 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1226 streams for the REPL, for the debugger, and so on. (thanks to
1228 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1229 * fixed some bugs revealed by Paul Dietz' test suite:
1230 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1231 is ignored if it is too small, rather than propagating through
1232 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1233 ** extremely complex negations of CONS types were not being
1234 sufficiently canonicalized, leading to inconsistencies in
1236 ** VALUES tranformer lost derived type.
1238 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1239 * fixed a bootstrapping bug: the build process no longer assumes
1240 that the various BOOLE-related constants have the same value in
1241 host and target lisps. (noted by Paul Dietz' test suite on an
1242 SBCL binary built from CLISP)
1243 * The system can now be dynamically linked on the MIPS platform,
1244 which enables dynamic loading of foreign code from Lisp. (thanks
1245 to Ralf Baechle for discussions on the MIPS ABI)
1246 * The system now records debugging information for its own source
1247 files in a filesystem-position-independent manner, relative to
1248 the "SYS" logical host.
1249 * fixed a compiler bug: MV-LET convertion did not check references
1250 to the "max args" entry point. (reported by Brian Downing)
1251 * tweaked disassembly notes to be less confident about proclaiming
1252 some instruction as an LRA. (thanks to Brian Downing)
1253 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1254 now each have their own history, command character, and other
1255 characteristics. (thanks to David Lichteblau)
1256 * fixed some bugs revealed by Paul Dietz' test suite:
1257 ** compiler failure in compiling LOGAND expressions including a
1259 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1260 shift greater than 32.
1261 ** FLUSH-DEST did not mark blocks for type check regeneration.
1262 ** HANDLER-CASE failed to accept declarations in handler clauses
1263 in some circumstances.
1265 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1266 * New code in contrib/sb-introspect (still probably not entirely
1267 stable yet) provides some support for smart Lisp development
1268 environments like SLIME.
1269 * The conditions signalled for errors occurring when loading .fasl
1270 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1271 in a way which should help ASDF recover gracefully.
1272 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1273 stuff has been cleaned up. If you code contrib/ stuff, this might
1274 affect you, and you can look at contrib/README, contrib/STANDARDS,
1275 and/or the 0.8.4.27 diff to check.
1276 * In full calls the compiler now does not generate checks for declared
1277 argument types for all arguments.
1278 * various threading fixes
1279 ** and some experimental patches which didn't make it into
1280 the main tree for this release, but which are shipped in
1281 contrib/experimental-thread.patch as a possible fix for some
1282 failures (deadlock, spinning...) in GC-intensive multithreaded
1284 * fixed PPC build problem (source code incompatibility of different
1285 library versions): added offsetof() hackery which attempts to divine
1286 where glibc maintainers put uc_mcontext today
1287 * fixed bug 282: compiler does not trust type assertions while passing
1288 arguments to a full call.
1289 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1290 &OPTIONAL VALUES type parameter.
1291 * fix bug 214: algorithm for noting rejected templates is now more
1292 similar to that of template seletion. (also reported by rydis on
1294 * fixed bug 141b: printing backquoted information readably and prettily
1295 inserts a space where necessary.
1296 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1297 (CHARACTER 10) are now reported as errors, rather than propagated
1298 as unknown types. (reported by piso on #lisp)
1299 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1300 with values NIL and :ERROR. (thanks to Milan Zamazal)
1301 * fixed bug 191c: CLOS now does proper keyword argument checking as
1302 described in CLHS 7.6.5 and 7.6.5.1.
1303 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1304 longer raise an error; further, using a list as a for-as-arithmetic
1305 counter now raises a meaningful error.
1306 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1307 now understood by sequence creation functions such as MAKE-SEQUENCE
1309 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1310 a STRING-INPUT-STREAM.
1311 * compiler enhancement: SIGNUM is now better able to derive the type
1313 * type declarations inside WITH-SLOTS are checked. (reported by
1315 * fixed some bugs revealed by Paul Dietz' test suite:
1316 ** incorrect optimization of TRUNCATE for a positive first
1317 argument and negative second.
1318 ** compiler failure in let-convertion during flushing dead code.
1319 ** compiler failure while deriving type of TRUNCATE on an
1320 interval, containing 0.
1321 ** ASH of a negative bignum by a negative bignum count now returns
1323 ** intersection of CONS types now canonicalizes properly, fixing
1324 inconsistencies in SUBTYPEP.
1326 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1327 * incompatible change: The --disable-debugger command line
1328 option now clobbers the debugger at a more fundamental
1329 level, by redefining #'INVOKE-DEBUGGER instead of by
1330 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1331 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1332 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1333 enter the debugger REPL and then suffer recursive errors
1334 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1335 while under the new system, BREAK in --disable-debugger mode
1336 terminates the system just as an unhandled error would.
1337 * fixed compiler performance when processing loops with a step >1;
1338 * bug fix: DOCUMENTATION now retrieves generic function
1339 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1340 support has been systematized, and now supports the methods
1341 specified by ANSI, along with a default method and a method for
1342 slot documentation. (reported by Nathan Froyd)
1343 * bug fix: effective methods associated with a generic function are
1344 no longer cached over a change of that generic function's method
1345 combination. (reported by Andreas Fuchs)
1346 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1347 names a non-existent directory. (thanks to Andreas Fuchs)
1348 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1349 the PPC platform, lead to stack corruption; this has been fixed.
1350 (reported by Rainer Joswig)
1351 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1352 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1353 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1354 arithmetic in general, is now much more reliable on the PPC
1356 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1357 platform now returns the right answer.
1358 * optimization: restored some effective method precomputation in
1359 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1360 precomputation is now tunable.
1361 * optimization: compiler-internal data structure use has been
1362 reviewed, and changes have been made that should improve the
1363 performance of the compiler by about 20%.
1364 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1365 simple-base-strings and simple-bit-vectors is improved.
1366 * optimization: the optimization of 32-bit logical and arithmetic
1367 functions introduced in version 0.8.3 on the x86 has been
1368 implemented on the mips, ppc and sparc platforms; an
1369 implementation of the same facility, but for 64-bit arithmetic,
1370 has been added for the alpha.
1371 * microoptimization: the compiler is better able to make use of the
1372 x86 LEA instruction for multiplication by constants.
1373 * bug fix: in some situations compiler did not report usage of
1374 generic arithmetic in (SPEED 3) policy.
1375 * bug 145b fix: compiler used wrong type specifier while converting
1376 MEMBER-types to numeric.
1377 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1379 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1380 index argument now works properly on the Alpha platform.
1381 * bug fix: floating point exception treatment on the Alpha platform
1383 * bug fix: FILE-POSITION works much better on string input and
1384 output streams. (thanks to Nikodemus Siivola)
1385 * bug fix: many threading/garbage collection symptoms sorted.
1386 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1387 might be pseudo-atomic.
1388 * internal change: Stopping for GC is now done with signals not
1389 ptrace. GC is now done in whichever thread wanted it, instead of
1391 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1393 * bug fix: result form in DO is not contained in the implicit
1395 * incompatible change: ICR structure is changed; the value part of
1396 CONTINUATION is now called LVAR; corresponding functions are
1397 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1398 * added type deriver for ISQRT (thanks to Robert E. Brown).
1399 * bug fix: better support for loading from the command line when an
1400 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1402 * fixed some bugs revealed by Paul Dietz' test suite:
1403 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1404 ** ROUND and FROUND now give the right answer when given very
1405 small float arguments.
1406 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1408 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1409 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1410 ** LCM should return a non-negative integer.
1411 ** PARSE-INTEGER returned the index of a terminator instead of the
1412 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1413 ** PARSE-INTEGER returned an incorrect index being applied to a
1415 ** LCM with two arguments of 0 returns 0 rather than signalling
1417 ** unsigned addition of a 32-bit constant with the high bit set no
1418 longer causes an internal compiler error on the x86.
1419 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1420 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1421 with negative last argument.
1422 ** byte specifiers with zero size and position no longer cause
1423 an error during type derivation.
1424 ** bignum multiplication on the Alpha platform now returns the
1426 * porting: The system now builds on SuSE AMD64, although it still
1427 generates a 32-bit binary.
1428 * .fasl file incompatibility: The fasl file version number has
1429 been incremented (because of the changes to internal compiler
1430 data structures referred to above).
1432 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1433 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1434 more accurately, on the Darwin kernel running on PowerPC hardware.
1435 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1436 * Compiler code deletion notes now signal a condition of type
1437 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1438 an associated MUFFLE-WARNING restart.
1439 * The compiler now performs limited argument count validation of
1440 constant format strings in FORMAT, and where appropriate in ERROR,
1441 CERROR and WARN. (thanks to Gerd Moellmann)
1442 * New ASDF-INSTALL contrib can be used for automatic download and
1443 installation of third-party Lisp code from CCLAN or other sites
1445 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1446 which forces another thread to execute a function supplied by the
1448 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1449 now accept and act upon their :ELEMENT-TYPE keyword argument.
1450 (reported by Martin Atzmueller, Edi Weitz)
1451 * bug fix: FILE-POSITION now accepts position designators up to
1452 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1453 is the greater. (thanks to Patrik Nordebo)
1454 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1455 Kalle Olavi Niemitalo)
1456 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1457 printable. (reported by Eric Marsden)
1458 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1459 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1460 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1461 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1462 (thanks to Stig Sandoe)
1463 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1464 UNREAD-CHAR now correctly includes the unread character in the
1465 target sequence. (thanks to Gerd Moellmann)
1466 * bug fix in threaded builds: the system can now be suspended and
1467 resumed by shell job control with minimal disruption.
1468 * bug fixes in times and timezones >2038AD
1469 * better handling of "where is GNU make?" problem in build scripts
1470 (thanks to Nikodemus Siivola)
1471 * new optimization: inside a named function any reference to a
1472 function with the same name is considered to be a self-reference;
1473 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1474 optimization quality.
1475 * new optimization on x86: logical functions and + now have
1476 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1477 used when the result is truncated to 32 bits.
1478 * VALUES declaration is partially enabled.
1479 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1480 * fixed some bugs revealed by Paul Dietz' test suite:
1481 ** The system now obeys the constraint imposed by
1482 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1483 types form a lattice under type intersection.
1484 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1485 ** ASSOC now ignores NIL elements in an alist.
1486 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1487 and (1+ MOST-POSITIVE-FIXNUM) answers.
1488 ** The addition of a method with invalid qualifiers to a generic
1489 function does not cause an error to be signalled immediately;
1490 a warning is signalled, and the error is generated only on
1491 calling the generic function.
1492 * changed .fasl file version number, in part to add type codes for
1493 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1494 obscure ANSI requirements
1496 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1497 * fixed bug 148: failure to inline-expand a local function left
1498 garbage, confusing the compiler.
1499 * fixed bugs 3cd: structure slot readers perform type check if the
1500 slot can have an invalid value (i.e. it is either not initialized
1501 or can be written with a less specific slot writer).
1502 * bug fix: the compiler now traps array references to elements off
1503 the end of an array; previously, the bounds checking in some
1504 circumstances could go off-by-one.
1505 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1507 * type declarations for array element types now obey the description
1508 on the CLHS page "Declaration TYPE", as per discussions on
1509 sbcl-help around 2003-05-08. This means that a declaration
1510 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1511 declaration, all references to BAR will be asserted or assumed
1512 (with THE, so dependent on compiler policy) to involve objects of
1513 type FOO. Note that no such declaration is implied in
1514 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1515 * declared types of functions from the "Conditions"
1516 chapter. (reported by Paul Dietz)
1517 * bug fix: CERROR accepts a function as its first argument.
1518 * bug fix: NTH an NTHCDR accept a bignum as index
1519 arguments. (reported by Adam Warner)
1520 * optimization: character compare routines now optimize comparing
1521 against a constant character. (reported by Gilbert Baumann)
1522 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1523 argument now works properly on the MIPS platform.
1524 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1525 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1526 declared to have a functional type, does not cause an error
1528 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1529 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1530 * fixed bug: if last continuation of a deleted block has a
1531 destination, this destination should be deleted too. (reported by
1533 * fixed a bug in the bootstrap process: the host compiler's values
1534 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1535 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1536 test case from Patrik Nordebo)
1537 * improved the ability of the disassembler on the PPC platform to
1538 provide helpful disassembly notes.
1539 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1540 the class in more cases than previously.
1541 * bug fix: FILE-POSITION now understands :START and :END for
1542 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1543 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1544 8) data. (thanks to David Lichteblau)
1545 * bug fix: it is possible to add a method to a generic function
1546 without lambda list.
1547 * bug fix: reader failed to signal END-OF-FILE inside an
1548 object representation. (reported by Nikodemus Siivola)
1549 * fixed some bugs revealed by Paul Dietz' test suite:
1550 ** LAST and [N]BUTLAST should accept a bignum.
1551 ** condition slot accessors are methods.
1552 ** (VECTOR NIL) is a subtype of STRING.
1554 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1555 * minor incompatible change: some nonsensical specialized lambda
1556 lists (used in DEFMETHOD) which were previously ignored now signal
1558 * minor incompatible change: the system is now aware of the types of
1559 variables in the COMMON-LISP package, and will signal errors for
1560 most violations of these type constraints (where previously they
1561 were silently accepted).
1562 * minor incompatible change: COMPILE-FILE now uses the freedom
1563 afforded (ANSI 3.2.2.3) to use derived function types for
1564 functions defined in the same file. This also permits the system
1565 to warn on static type mismatches and function
1566 redefinition. (Currently it does not work with high DEBUG level.)
1567 * minor incompatible change: VALUES declaration is disabled.
1568 * When issuing notes, the compiler now signals a condition of type
1569 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1570 restart for use in user handlers. It is expected that the
1571 COMPILER-NOTE condition will eventually become a condition
1572 supertype to a hierarchy of note types, which will then be
1573 handleable in a similar fashion. However, at the moment, no such
1574 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1575 but it's an implementation detail, not a classification for the
1577 * Changes in type checking closed the following bugs:
1578 ** type checking of unused values (192b, 194d, 203);
1579 ** template selection based on unsafe type assertions (192c, 236);
1580 ** type checking in branches (194bc).
1581 * A short form of VALUES type specifier has ANSI meaning (it has
1582 increased the number of situations when SBCL cannot perform type
1584 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1585 keywords or constants is permissible.
1586 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1587 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1588 classes. (thanks to Antonio Martinez)
1589 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1590 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1591 argument to a pretty-print stream. Instead, it leaves any such
1592 operations to DESCRIBE-OBJECT methods.
1593 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1595 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1596 lambda lists. (thanks to David Lichteblau)
1597 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1598 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1599 (thanks to Gerd Moellmann)
1600 * fixed bug 246: increased compilation speed of long
1601 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1603 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1605 * DEFCONSTANT now throws a condition of type
1606 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1607 constant to a non-EQL value; CONTINUE and ABORT restarts
1608 respectively change and preserve the value.
1609 * fixed bug 63: The code walker, part of the implementation of CLOS,
1610 is now better at handling symbol macros.
1611 * bug fix: There is no longer an internal implementation type named
1612 CL:LENGTH. (reported by Raymond Toy)
1613 * bug fix: In macro-like defining macros/special operators the
1614 implicit block does not enclose the lambda list.
1615 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1616 enclosed in parentheses) are not suitable as type specifiers, and
1617 their use properly signals an error now.
1618 * bug fix: An argument count mismatch for a type specifier in code
1619 being compiled no longer causes an unhandled error at compile
1620 time, but signals a compile-time warning.
1621 * fixed simple vector readable printing
1622 * bug fix: DESCRIBE takes more care over whether the class
1623 precedence list slot of a class is bound before accessing it.
1624 (reported by Markus Krummenacker)
1625 * bug fix: FORMATTER can successfully compile pretty-printer format
1626 strings which use variants of the ~* directive inside.
1627 * bug fix: SEARCH now applies its TEST predicate to the elements of
1628 the arguments in the correct order. (thanks to Wolfhard Buss)
1629 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1630 in type inference. (thanks to Robert E. Brown)
1631 * bug fix: Reading in symbols with an explicit package name of ""
1632 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1633 * fixed some bugs revealed by Paul Dietz' test suite:
1634 ** NIL is now allowed as a structure slot name.
1635 ** Arbitrary numbers, not just REALs, are allowed in certain
1636 circumstances in LOOP for-as-arithmetic clauses.
1637 ** Multiple class redefinitions before slot access no longer
1638 causes a type error.
1639 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1640 association between the name and a class.
1641 ** Generic functions with non-standard method-combination and over
1642 five methods all of which return constants no longer return NIL
1643 after the first few invocations. (thanks to Gerd Moellmann)
1644 ** CALL-NEXT-METHOD with no arguments now passes the original
1645 values of the arguments, even in the presence of assignment.
1646 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1648 ** DISASSEMBLE works with closures and funcallable instances.
1649 ** ADD-METHOD now returns the generic function, not the new method.
1650 ** FIND-METHOD signals an error if the lengths of the specializers
1651 is incompatible with the generic function, even if the ERRORP
1653 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1654 which its argument is a member.
1655 ** DEFCLASS only redefines the class named by its class-name
1656 argument if that name is the proper name of the class;
1657 otherwise, it creates a new class.
1658 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1659 of the UNBOUND-SLOT condition to the name of the slot.
1660 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1661 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1662 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1663 SLOT-MAKUNBOUND in the specified fashion.
1665 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1666 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1667 cross-compilation host. As a consequence, we can now bootstrap our
1668 way up to SBCL starting with a bare gcc toolchain and human-readable
1669 source code (first the source to CLISP, then the source to SBCL).
1670 * A contributed module containing a partial implementation of the
1671 simple-streams interface has been included. (thanks to Rudi
1673 * A contributed module implementing the RFC1321 Message Digest
1674 Algorithm, known as MD5, has been included.
1675 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1676 no longer has any effect, as the code controlled by this feature
1677 has been deleted. (As far as we know, no-one has ever built using
1678 this feature, and its semantics were confused in any case).
1679 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1680 work on conditions (as required by the ANSI specification),
1681 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1682 expected behaviour on conditions. Users should note, however,
1683 that such behaviour is not required by the ANSI specification,
1684 and so use of this behaviour may render their code unportable.
1685 * fixed some bugs revealed by Paul Dietz' test suite:
1686 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1688 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1689 ** SXHASH on condition objects no longer returns NIL.
1690 ** :ALLOCATION :CLASS slots are better treated; their values are
1691 updated on class redefinition, and initforms inherited from
1692 superclasses are applied.
1693 ** REMOVE-METHOD returns its generic function argument even when
1694 no method was removed.
1695 ** SHARED-INITIALIZE now initializes the values of the requested
1696 slots, including those with :ALLOCATION :CLASS.
1697 ** ALLOCATE-INSTANCE now works on structure classes defined via
1698 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1700 ** SLOT-EXISTS-P now works on conditions, as well as structures
1702 ** MAKE-LOAD-FORM now has the required methods on
1703 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1704 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1705 keyword, and now implements the SLOT-NAMES argument.
1706 ** methods with &OPTIONAL arguments no longer allow too many
1707 arguments to be passed in the call without error.
1708 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1709 option is consistent with the required arguments of the generic
1710 function lambda list.
1711 * bug fix: REQUIRE accepts a string designator. (Thanks to
1713 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1714 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1715 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1716 * bug fix: The compiler now checks for duplicated variables in macro
1718 * bug fix: SETQ on globals returns the correct value.
1719 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1720 not a COMPILER-ERROR (followed by some other strange error on
1721 choosing the CONTINUE restart).
1722 * bug fix: make.sh and friends are now more consistent in the way that
1723 they look for GNU "make".
1725 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1726 * experimental native threads support (on x86 Linux >=2.4 only).
1727 This is not compiled in by default: you need to add :SB-THREAD to
1728 the target features. See the "Beyond ANSI" chapter of the manual
1730 * fix for longstanding nonANSIism: The old distinction between
1731 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1732 The return value from CL:FIND-CLASS is now a CLOS class, and
1733 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1734 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1735 * An interface to the MetaObject Protocol, as described in Kiczales,
1736 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1737 MIT Press, 1991, is available from the SB-MOP package.
1738 * incompatible change: the SB-PCL package should now be considered
1739 a private implementation detail, and no longer a semi-private MOP
1741 * minor incompatible change: due to rearrangement for threads, the
1742 control stack and binding stack are now allocated at arbitrary
1743 addresses instead of being hardcoded per-port. Users affected by
1744 this probably have to be doing advanced things with shared
1745 libraries, and will know who they are.
1746 * minor incompatible change: Previously, all --eval forms used were
1747 processed with READ before any of them were processed with EVAL.
1748 Now each --eval form is processed with both READ and EVAL before
1749 the next --eval form is processed. (Thus package operations like
1750 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1751 work as the user might reasonably expect.)
1752 * minor incompatible change: *STANDARD-INPUT* is now only an
1753 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1755 * minor incompatible change: Y-OR-N-P is now character-oriented,
1756 not line oriented. Also, YES-OR-NO-P now works without errors.
1757 (thanks to Antonio Martinez)
1758 * sb-aclrepl module improvements: an integrated inspector, added
1759 repl features, and a bug fix to :trace command.
1760 * Known functions, which cannot be open coded by the backend, are
1761 considered to be able to check types of their arguments. (fixing
1762 a bug report by Nathan J. Froyd)
1763 * fixed a bug in computing method discriminating functions: It is
1764 now possible to define methods specialized on classes which have
1765 forward-referenced superclasses. (thanks to Gerd Moellmann)
1766 * fixed evaluation order in optional entries (reported by Gilbert
1768 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1769 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1770 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1771 required-by-AMOP NAME argument, as well as CLASS and
1772 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1773 * fixed bug 20: DEFMETHOD can define methods using names that are
1774 not the proper names of classes to designate class specializers.
1775 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1776 Unix file descriptors, instead of blowing up. (thanks to Antonio
1778 * Garbage collection refactoring: user-visible change is that a
1779 call to the GC function during WITHOUT-GCING will not do garbage
1780 collection until the end of the WITHOUT-GCING. If you were doing
1781 this you were probably losing anyway.
1782 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1783 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1784 TYPEP the latter but not the former.
1785 * The compiler issues a full WARNING for calls to undefined functions
1786 with names from the CL package.
1787 * MAP-INTO for a vector destination is open coded. (reported by
1788 Brian Downing on c.l.l)
1789 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1790 documentation string.
1791 * fixed some bugs revealed by Paul Dietz' test suite:
1792 ** COPY-ALIST now signals an error if its argument is a dotted
1794 ** Condition slots are now accessed more correctly in the presence
1795 of multiple initargs for a given slot.
1796 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1797 exclude restarts of the same name associated with a different
1799 ** DEFCLASS of forward-referenced classes with another
1800 forward-referenced class in the superclasses list no longer
1802 ** Condition slots are now initialized once each, not multiple
1803 times. (thanks to Gerd Moellmann)
1804 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1806 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1807 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1808 arguments contain duplicated elements.
1809 ** RESTART-CASE understands local macros.
1810 ** RESTART-CASE associates exactly its own restarts with a condition.
1811 ** ENDP in safe mode checks its argument to be of type LIST.
1812 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1813 in question is unbound.
1814 ** Optimized MAKE-INSTANCE functions no longer cause internal
1815 assertion failures in the presence of duplicate initargs.
1816 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1817 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1818 propagated even in the ctor optimized implementation of
1820 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1822 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1823 * a better implementation of SXHASH on (simple) bit vectors,
1824 measured both in execution speed and in distribution of results
1825 over the positive fixnums, has been installed. Likewise, a better
1826 implementation of EQUAL for simple bit vectors is now available.
1827 * fixed CEILING optimization for a divisor of form 2^k.
1828 * fixed bug 240 (emitting extra style warnings "using the lexical
1829 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1830 by Antonio Martinez)
1831 * fixed SXHASH, giving different results for NIL depending on type
1832 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1833 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1834 variable. (found by Rolf Wester)
1835 * SBCL does not ignore type declarations for special
1836 variables. (reported by rif on c.l.l 2003-03-05)
1837 * some bug fixes in contrib/sb-aclrepl/
1838 * fixed some bugs revealed by Paul Dietz' test suite:
1839 ** a bug in the CONS type specifier, whereby the CAR and CDR
1840 types got intertwined, has been fixed;
1841 ** the type system is now able to reason about the interaction
1842 between INTEGER and RATIO types more completely;
1843 ** APPEND, [N]REVERSE and NRECONC check that those their
1844 arguments, which must be proper lists, are really so;
1845 ** An array specialized to be unable to hold elements has been
1846 implemented, as required -- yes, really -- by ANSI;
1847 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1848 on malformed property lists;
1850 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1851 * incompatible packaging change: in line with Unix convention,
1852 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1853 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1854 * REQUIRE and PROVIDE are now optionally capable of doing something
1855 useful. See the documentation string for REQUIRE.
1856 * infrastructure for a managed SBCL contrib system: contributed
1857 modules in this release include:
1858 ** the ASDF system definition facility;
1859 ** an interface to the BSD Sockets API;
1860 ** an ACL-like convenience interface to the repl;
1861 (thanks to Kevin Rosenberg)
1862 ** an implementation of ROTATE-BYTE, with efficient implementation
1864 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1865 gives a closer approximation to the right answer than previously.
1866 (thanks to Raymond Toy)
1867 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1868 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1869 optional environment arguments, as required by ANSI.
1870 * fixed bugs in other functions taking environment objects, allowing
1871 calls with an explicit NIL environment argument to be compiled
1873 * fixed bug 228: primary return values from
1874 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1875 COMPILE or FUNCTION.
1876 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1877 :TYPE will no longer signal errors on innocuous objects.
1878 * fixed bug 231b: SETQ is better at respecting type declarations in
1879 the lexical environment.
1880 * fixed a bug in DEFCLASS: classes named by symbols with no or
1881 unprintable packages can now be defined.
1882 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1883 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1884 * fixed compiler failure related to checking types of functions.
1885 (reported by Robert E. Brown)
1886 * the compiler is now much more consistent in its error-checking
1887 treatment of bounding index arguments to sequence functions: in
1888 (SAFETY 3) code, errors will be signalled in almost all cases if
1889 invalid sequence bounding indices are passed to functions defined
1890 by ANSI to operate on sequences.
1891 * fixed a bug in the build procedure: documentation of SBCL-specific
1892 packages is now preserved and available in the final Lisp image.
1893 * lifted FDEFINITION lookup out of loops in the implementation of
1894 many list operations. (thanks to Robert E. Brown)
1895 * fixed a bug in the reader: the #n# reader macro now works for
1896 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1897 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1898 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1899 E. Brown and Tony Martinez respectively)
1900 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1901 SLOT-BOUNDP: in particular, it is now able to optimize them much
1902 better, and is now not vulnerable to having packages renamed.
1903 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1904 always called when a slot is not present in an instance. (thanks
1906 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1907 longer signal an error on unreading a character following EOF on
1908 the previous constituent stream. (thanks to Tony Martinez)
1909 * fixed some bugs revealed by Paul Dietz' test suite:
1910 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1911 not just nonnegative fixnums;
1912 ** the logical bit-array operators such as BIT-AND now accept an
1913 explicit NIL for their "opt-arg" argument (to indicate a
1914 freshly-consed result bit-array);
1915 ** ELT now signals an error on an invalid sequence index in safe
1917 ** the type system is now cleverer about negations of numeric
1918 types, and consequently understands the BIGNUM and RATIO types
1920 ** the type system is now cleverer about the interaction between
1921 INTEGER and RATIO types: while bugs still remain, many more
1922 cases are accurately computed;
1923 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1924 if it is in the last clause;
1925 ** CONSTANTLY now correctly returns a side-effect-free function in
1927 ** DECLARE is no longer treated as a special-operator; in
1928 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1929 * incremented fasl file version number due to the change in the
1930 DEFSTRUCT-SLOT-DESCRIPTION structure.
1932 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1933 * minor incompatible change: code processed by the "interpreter" or
1934 EVAL now has a compilation optimization policy of (DEBUG 2)
1935 (changed from (DEBUG 1)) to improve debuggability of interactive
1936 development, and to allow the use of the debug RETURN command in
1938 * an experimental implementation of the RETURN command for the
1939 debugger has been included. (thanks to Frederik Kuivinen)
1940 * fixed bug 62: constraints were not propagated into a loop.
1941 * fixed bug in embedded calls of SORT (reported and investigated by
1943 * fixed some bugs revealed by Paul Dietz' test suite:
1944 ** printing and reading of arrays with some dimensions having
1945 length 0 (thanks to Gerd Moellmann);
1946 ** BOA constructor with &AUX argument without a default value does
1947 not cause a type error;
1948 ** CONSTANTP now returns true for all self-evaluating objects.
1950 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1951 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1952 accessors that are related by inheritance, as specified in the
1953 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1954 Valtteri Vuorikoski)
1955 * The compiler is now able to inline functions that were defined in
1956 a complex lexical environment (e.g. inside a MACROLET).
1957 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1959 * Support for the upcoming FreeBSD-5.0 release has been included.
1960 (thanks to Dag-Erling Smorgrav)
1961 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1962 effect when it is not in a toplevel context.
1963 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1964 stand a better chance of being correct. (thanks to Gerd
1966 * fixed bug in COERCE, which now signals an error on coercing a
1967 rational to a bounded real type which excludes the expected
1969 * The compiler is now able to derive types more accurately from the
1970 COERCE and COMPILE functions.
1971 * fixed bug 223: functional binding is considered to be constant
1972 only for symbols in the CL package.
1973 * fixed bug 231: SETQ did not check the type of a variable being set
1974 (reported by Robert E. Brown)
1975 * A new optimization for MAKE-INSTANCE has been included, fixing
1976 various bugs (including relating to :ALLOCATION :CLASS slots and
1977 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1979 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1980 ** As required by ANSI, LOOP now disallows anonymous collection
1981 clauses such as COLLECT I in conjunction with aggregate boolean
1982 clauses such as THEREIS (= I 1);
1983 ** LOOP now signals an error when any variable is reused in the
1984 same loop (including the potentially useful construct analogous
1985 to WITH A = 1 WITH A = (1+ A);
1986 ** IT is only a special loop symbol within the first clause of a
1987 conditional loop clause;
1988 ** LOOP with a typed iteration variable over a hashtable now
1989 signals a type error iff it should.
1990 * fixed some other bugs revealed by Paul Dietz' test suite:
1991 ** FILE-STREAM now names the class previously known as FD-STREAM;
1992 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1993 argument) no longer signals an error;
1994 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1995 of the slot symbol, rather than using the current package
1996 ((:CONC-NAME "") continues to intern the slot's name in the
1998 * incremented fasl file version number, because of the incompatible
1999 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2000 of the new implementation of DEFINE-COMPILER-MACRO.
2002 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2003 * Support for building SBCL for MIPS platforms running in
2004 little-endian mode has now been checked in, and basic
2005 functionality on said platforms verified.
2006 * minor incompatible change: PCL now records the pathname of a file
2007 in which methods and the like are defined, rather than its
2009 * minor incompatible change: TRUENAME now considers the truename of
2010 a file naming a directory to be the pathname with :DIRECTORY
2011 component indicating that directory.
2012 * minor incompatible change: a NAMED clause in the extended form of
2013 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2014 reason for the previous behaviour is unclear.
2015 * more systematization and improvement of CLOS and MOP conformance
2016 in PCL (thanks to Gerd Moellman and Pierre Mai):
2017 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2019 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2020 primary methods with no specializers;
2021 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2023 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2024 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2025 FORWARD-REFERENCED-CLASSes; error reporting on
2026 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2028 ** SXHASH on CLOS instances now uses a slot internal to the
2029 instance to return different numbers on distinct instances,
2030 while preserving the same return value through invocations of
2032 ** DEFMETHOD signals errors when methods with longer incongruent
2033 lambda lists are added to generic functions;
2034 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2035 CLASS, as specified in AMOP;
2036 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2037 on the order returned by the primary method for classes of
2038 class STANDARD-CLASS;
2039 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2040 * fixed some bugs shown by Paul Dietz' test suite:
2041 ** DOLIST puts its body in TAGBODY;
2042 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2044 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2045 value producing form;
2046 ** if more variables are given to PROGV than values, extra
2047 variables are bound and made to have no value;
2048 ** NSUBSTITUTE on list arguments gets the right answer with
2050 ** ELT signals an error of type TYPE-ERROR when the index argument
2051 is not a valid sequence index;
2052 ** LOOP signals (at macroexpansion time) an error of type
2053 PROGRAM-ERROR when duplicate variable names are found;
2054 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2055 ** FUNCALL of special-operators now cause an error of type
2057 ** PSETQ now works as required in the presence of side-effecting
2058 symbol-macro places;
2059 ** NCONC accepts any object as its last argument;
2060 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2062 ** loop-for-as-package does not require a package to be explicitely
2064 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2066 * fixed bug 166: compiler preserves "there is a way to go"
2067 invariant when deleting code.
2068 * fixed bug 172: macro lambda lists with required arguments after
2069 &REST arguments now cause an error to be signalled. (thanks to
2071 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2072 bug. (thanks to Matthew Danish)
2073 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2075 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2076 arguments are lexically rebound. (thanks to Gerd Moellmann and
2078 * fixed bug 194: error messages are now more informative when there
2079 is no primary method applicable in a call to a generic
2080 function. (thanks to Gerd Moellmann)
2081 * fixed bug in command line argument checking (thanks to Julian
2083 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2084 (thanks to Matthew Danish)
2085 * incremented fasl file version number, because of the
2086 SXHASH-related changes in the layout of CLOS data structures
2088 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2089 * minor incompatible change: The runtime (the Unix executable named
2090 "sbcl") is now much pickier about the .core files it will load.
2091 Essentially it now requires .core files to descend from the same
2092 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2093 as the runtime does. (The intent is to prevent the crashes which
2094 can occur, and which can even be reported as mysterious failures,
2095 when people patch the sources or change the build parameters
2096 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2097 sbcl and .core files.)
2098 * fixed bug: VALUES-LIST is no longer optimized away.
2099 * fixed bug 142: The FFI conversion of C string values to Lisp
2100 string values no longer conses excessively. (thanks to Nathan
2101 Froyd porting Raymond Toy's fix to CMU CL)
2102 * began to systematize and improve MOP conformance in PCL (thanks to
2103 Nathan Froyd, Gerd Moellman and Pierre Mai):
2104 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2106 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2107 ** FINALIZE-INHERITANCE is now called on class finalization;
2108 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2109 argument precedence order.
2110 * fixed bug 202: The compiler no longer fails on functions whose
2111 derived types contradict their declared type.
2112 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2113 so it can be non-toplevel.
2114 * The fasl file version number has changed (because of the new
2115 implementation of DEFMACRO).
2116 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2117 safe code, that any length requirement by their type-specifier
2118 argument is valid. The exceptions to this are described in bug
2120 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2121 functions now check the types of their inputs as required by ANSI.
2122 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2123 introduced symbol is DECLAREd to be SPECIAL.
2124 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2125 * fixed bug: :COUNT argument to sequence functions may be negative
2126 * fixed bug: body of DO-SYMBOLS may contain declarations
2127 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2128 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2129 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2130 symbol macro only once
2131 * fixed printing of call frame when argument list is unavailable
2132 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2133 * compiler no longer signals WARNING on unknown keyword
2136 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2137 * A beta-quality port to the mips architecture running Linux,
2138 based on the old CMUCL backend, has been made. It has been tested
2139 on a big-endian kernel, and works sufficiently well to be able to
2140 rebuild itself; it has not been tested in little-endian mode.
2141 * fixed an inconsistency between gencgc.c and purify.c which made
2142 dumping/loading .core files unreliable
2143 * fixed bug 120a: The compiler now deals correctly with IFs where
2144 the consequent is the same as the alternative, instead of
2145 misderiving the return type. (thanks to Alexey Dejneka)
2146 * fixed bug 113: Logical pathnames are now dumpable (the logical
2147 host is resolved at load-time, throwing an error if it is not
2149 * fixed bug 174: FORMAT's error message is slightly clearer when a
2150 non-printing character is used in a format directive.
2151 * fixed several bugs in compiler checking of type declarations, i.e.
2152 violations of the Python "declarations are assertions" principle
2153 (thanks to Alexey Dejneka)
2154 * fixed several bugs in PCL's error checking (thanks to Gerd
2156 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2158 * fixed bug in compilation of functions as first class values
2159 (thanks to Antonio Martinez-Shotton)
2160 * The compiler's handling TYPE-ERRORs which it can prove will
2161 inevitably happen at runtime has been cleaned up and corrected
2162 in several ways. (thanks to Alexey Dejneka)
2163 * improved argument type checking for various basic arithmetic
2164 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2165 lavished on them in the past that they can be compiled in many
2166 ways in different special cases
2167 * fixed bug 181: compiler checks validity of user supplied type
2169 * cleaned up code flushing in optimization: Function calls which
2170 should signal errors for safety purposes (e.g. which ANSI says
2171 should signal errors when their arguments are of incorrect type)
2172 are no longer optimized away.
2173 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2174 * incremented fasl file version number, because changes in the
2175 implementation of sequence functions like COERCE caused
2176 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2177 in old inline expansions) to become undefined. (Actually these
2178 changes were later undone, so we might very well be binary
2179 compatible with 0.7.7 after all, but leaving the version number
2180 incremented seemed like the simplest and most conservative
2183 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2184 * An alpha-quality port to the parisc architecture running Linux,
2185 based on the old CMUCL backend, has been made. This, even more so
2186 than the other backends, should be considered still a work in
2187 progress; known problems include that the Linux kernel in 64-bit
2188 mode does not propagate the correct sigcontext structure to
2189 userspace, and consequently SBCL on a parisc64 kernel will not
2191 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2192 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2193 suppressed.) Also now that the compiler is looking at declarations
2194 in the environment, it checks optimization declarations as well,
2195 and suppresses inlining when (> DEBUG SPEED).
2196 * More fixes have been made to treatment of floating point exception
2197 treatment and other Unix signals. In particular, floating point
2198 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2199 * The detection and handling of control stack exhaustion (infinite
2200 or very deeply nested recursion) has changed. Stack exhaustion
2201 detection is now done by write-protecting pages at the OS level
2202 and applies at all optimization settings; when found, a
2203 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2204 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2205 be caught using IGNORE-ERRORS.
2206 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2207 correctly when transforming recursive LABELS functions to LETs.
2208 (thanks to Alexey Dejneka)
2209 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2210 that are names of constants or global variables.
2211 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2212 alien routines with docstrings.
2213 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2214 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2216 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2217 (thanks to Eric Marsden)
2218 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2219 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2220 * Bug fix: The SPARC backend can now compile functions involving
2221 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2222 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2223 to a functions expecting a PATHNAME-DESIGNATOR.
2224 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2225 lambda lists. (thanks to Alexey Dejneka)
2226 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2227 functions so that EXPORTing the name of the class doesn't cause
2228 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2229 to Antonio Martinez for reporting this)
2230 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2231 constant-folded without failing an assertion. (thanks to Einar
2232 Floystad Dorum for reporting this)
2233 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2234 types are now dealt with more correctly by the compiler.
2235 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2236 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2237 behaviour analogous to RENAME-FILE. This puts its behaviour more
2238 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2240 * The fasl file version number has changed again. (because of the
2241 bug fix involving the names of PCL MAKE-INSTANCE functions)
2243 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2244 * bug fix: Floating point exceptions are treated much more
2245 consistently on the x86/Linux and PPC/Linux platforms.
2246 * Array initialization with :INITIAL-ELEMENT is now much faster for
2247 cases when the compiler cannot open code the array creation, but
2248 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2249 array accesses have also seen a speed increase.
2250 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2251 specified. (This was at the root of some bad interactions between
2252 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2254 * bug fix: Internal error arguments for undefined functions are now
2255 computed correctly on the PPC/Linux platform.
2256 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2257 Raymond Toy's patch for CMU CL)
2258 * Support for the Solaris 9 operating environment has been included
2259 (thanks to Daniel Merritt)
2260 * A very ugly but hopefully complete draft of the missing FFI chapter
2261 of the manual has been created by reformatting the corresponding
2262 CMU CL manual chapter into (currently very ugly and incoherent)
2263 DocBook and bringing it up to date for SBCL behavior. Thus, the
2264 manual is now essentially complete, at least by my extreme
2265 once-and-only-once standards, whereby it's acceptable to refer to
2266 the doc strings of SB-EXT functions as the primary documentation.
2267 * The fasl file version number has changed again, due to cleanup of
2268 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2269 is no longer a static symbol.)
2271 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2272 * SBCL now builds with OpenMCL (version 0.12) as the
2273 cross-compilation host; also, more progress has been made toward
2274 bootstrapping under CLISP.
2275 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2277 * bug 158 fixed: The compiler can now deal with integer loop
2278 increments different from 1; fixing this turned out also to fix
2280 * bug 169 fixed: no more bogus warnings about using lexical bindings
2281 despite the presence of perfectly good SPECIAL declarations (thanks
2282 to David Lichteblau)
2283 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2284 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2285 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2287 * bug 180 fixed: Method combination specifications no longer ignore
2288 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2289 * bug fix: Structure type predicate functions now check their argument
2290 count as they should.
2291 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2292 correctly. (thanks to Pierre Mai)
2293 * minor incompatible change: The --noprogrammer option is deprecated
2294 in favor of the new --disable-debugger option, which is very similar.
2295 (The major difference is that it takes effect at a slightly different
2296 time at startup, causing handling of errors in --sysinit and
2297 --userinit files will be affected differently.) The
2298 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2299 been added to allow this functionality to be controlled from ordinary
2300 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2301 maintainers, who might want to run non-interactive scripts to
2302 build SBCL cores which will later be used interactively.)
2303 * minor incompatible change: The LOAD function no longer, when given
2304 a wild pathname to load, loads all files matching that pathname.
2305 Instead, an error of type FILE-ERROR is signalled.
2307 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2308 * bug 147 fixed: The compiler preserves its block link/count
2309 invariants more correctly now so that it doesn't crash. (thanks
2311 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2313 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2314 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2315 no-value-for-this-character as NIL (as specified by ANSI).
2316 * HOST-NAMESTRING on physical pathnames now returns a string that is
2317 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2318 (thanks to Christophe Rhodes)
2319 * The Alpha port handles icache flushing more correctly. (thanks to
2321 * More progress has been made toward bootstrapping under CLISP. (thanks
2322 to Christophe Rhodes)
2323 * The fasl file format has changed again, because dynamic loading
2324 on OpenBSD (which has non-ELF object files) motivated some cleanups
2325 in the way that foreign symbols are transformed and passed around.
2326 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2327 is no longer treated as whitespace by the reader, but instead as
2328 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2329 |A
\7fB|, instead of A as it used to.
2331 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2332 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2333 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2334 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2335 this as of 0.7.1.45, but was left out of the previous news section
2336 (thanks to Dan Barlow)
2337 * SBCL now runs on the Solaris operating system on SPARC architectures
2338 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2339 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2340 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2341 * SPARC backend cleanups, allowing builds of cores optimized for V8
2342 and V9 SPARCS, and also emission of code targeted to a particular
2343 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2345 * SBCL is closer to bootstrapping under CLISP, thanks to various
2346 fixes by Christophe Rhodes.
2347 * The fasl file format has changed again, to allow the compiler's
2348 INFO database to support symbol macros.
2349 * The user manual (in doc/) is formatted into HTML more nicely.
2350 (thanks to coreythomas)
2351 * The system is smarter about SUBTYPEP relationships, especially
2352 those involving NOT types (including types such as ATOM which are
2353 represented internally using NOT types). Thus SUBTYPEP is less
2354 likely to return (VALUES NIL NIL) in general, and in particular
2355 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2357 * The fasl file format has changed again, because the internal
2358 representation of types now includes a new slot to support the new
2359 SUBTYPEP-of-NOT-types logic.
2360 * (not a change in the main branch of SBCL, but a related prototype
2361 which can hopefully be merged into the main branch of SBCL in the
2362 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2363 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2364 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2365 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2366 dispatch macro character combinations. (thanks to Alexey Dejneka)
2367 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2368 that superficially look like logical namestrings correctly.
2369 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2371 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2372 default supplied-p) for &optional and &key arguments. (thanks to
2375 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2376 * incompatible change: The compiler is now less aggressive about
2377 tail call optimization, doing it only when (> SPACE DEBUG) or
2378 (> SPEED DEBUG). (This is an incompatible change because there are
2379 programs which relied on the old CMU-CL-style behavior to optimize
2380 away their unbounded recursion which will now die of stack overflow.)
2381 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2382 for non-GENCGC systems has been increased to 20M (since that
2383 seems much closer to the likely performance optimum for modern
2384 systems than the old 4M value was)
2385 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2386 and *DEBUG-PRINT-LEVEL*
2387 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2388 of CMU CL's support for SPARC, and various endianness and other
2389 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2390 * new syntactic sugar for the Unix command line: --load foo.bar is now
2391 an alternate notation for --eval '(load "foo.bar")'.
2393 ** The system now detects stack overflow and handles it gracefully,
2394 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2395 optimization settings. (This is a good thing in general, and
2396 its introduction in this version should be particularly timely
2397 for anyone whose code fails because of suppression of tail
2399 ** The system now hunts for the C variable "environ" in a more
2400 devious way, to avoid segfaults when the C library version differs
2401 between compile time and run time. (thanks to Christophe Rhodes)
2402 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2403 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2404 ** The compiler no longer issues bogus style warnings for undefined
2405 classes in the same source file as the DEFCLASSes which defined
2406 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2408 ** fixes in CONDITION class precedence list for undefined function
2409 errors (thanks to Alexei Dejneka)
2410 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2411 correctly. (thanks to Dan Barlow)
2412 ** portability fixes aiming at bootstrapping under CLISP (thanks
2413 to Dave McDonald and Christophe Rhodes)
2414 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2415 ** fixes in type translation and and type inference (thanks to
2417 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2418 ** various fixes in the new ports (thanks to Dan Barlow)
2419 * several changes related to debugging:
2420 ** suppression of tail recursion, as noted above
2421 ** stack overflow detection, as noted above
2422 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2423 is now the default. (For some time encapsulation has been more
2424 reliable than the breakpoint-based :ENCAPSULATE NIL
2425 implementation, at least on X86 systems; and I just noticed that
2426 encapsulation also seems closer to the spirit of the ANSI
2429 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2431 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2432 up properly again. (There was a packaging bug in 0.7.0 which
2433 left their definitions in SB-SYS::LOAD-FOREIGN and
2434 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2435 vital for most things which interface to C-level interfaces,
2436 like extensions working with sockets or databases or
2437 Perl-compatible regexes or whatever, and the need to fix
2438 this bug is the main reason that 0.7.1 was released so
2440 ** DEFGENERIC is now choosier about the methods it redefines, so that
2441 reLOADing a previously-LOADed file containing DEFGENERICs does
2442 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2443 cycle now works as it should. (thanks to Alexey Dejneka)
2444 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2445 fixed by Christophe Rhodes. (He actually submitted this patch
2446 months ago, and I delayed until after 0.7.0.)
2447 ** Bug 111 (internal compiler confusion about runtime checks on
2448 FUNCTION types) was fixed by Alexey Dejneka.
2449 * Some internal cleanups (getting rid of variables which aren't
2450 needed now that the byte interpreter is gone) caused the fasl
2451 file format number to change again.
2453 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2454 * major incompatible change: The default fasl file extension, i.e. the
2455 default extension for files produced by COMPILE-FILE, has changed
2456 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2458 ** There are many changes in the implementation of the compiler.
2459 SBCL is now essentially a compiler-only implementation of ANSI
2460 Common Lisp. EVAL still "interprets" a few special cases, but
2461 almost all the interesting cases are handled by creating
2462 a LAMBDA expression, calling COMPILE on it, then calling
2463 FUNCALL on the result.
2464 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2465 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2466 Since the code is newer, there might still be some new bugs
2467 (though not as many as before Martin Atzmueller's fixes:-). But
2468 the new code is substantially simpler and clearer, and hopefully
2469 any remaining bugs will be simpler, less fundamental, and more
2470 fixable then the bugs in the old code.
2471 ** The revised compiler is still a little unsteady on its feet.
2473 *** The debugging information it produces (particularly the names
2474 of FUNCTION objects) is sometimes much less useful than what
2475 the old compiler produced.
2476 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2477 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2478 within a MACROLET) has been temporarily weakened.
2479 ** There are new compiler optimizations for various functions:
2480 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2481 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2482 *** the math functions TRUNCATE, FLOOR, and CEILING
2483 *** the function-of-all-trades COERCE
2484 Mostly these should be transparent, but there's one
2485 potentially-annoying problem (bug 117): when the compiler
2486 inline-expands a function and does type analysis on the result,
2487 it can create control paths which have type mismatches, and
2488 when it can't prove that those control paths aren't taken,
2489 it will issue WARNINGs about the type mismatches. This is
2490 a particular problem in practice for the new sequence functions.
2491 It's not clear how this should be fixed, and for now, a
2492 workaround is given in the entry for 117 in the BUGS file.
2493 ** (Because of the interaction between the two previous items --
2494 occasional inlining problems and new inline expansions -- some
2495 of the new sequence function optimizations won't really kick in
2496 completely until debugging information, and then inlining, are
2497 straightened out in some future version.)
2498 * minor incompatible changes:
2499 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2500 DIRECTORY no longer implicitly promotes NIL slots of its
2501 pathname argument to :WILD. In particular, when you ask for the
2502 contents of a directory (which you used to be able to do without
2503 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2504 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2505 ** changes in behavior that ANSI explicitly defines to be
2506 implementation dependent:
2507 *** The new compiler-only implementation still conforms with ANSI,
2508 but acts a little different than before. Besides the obvious
2509 changes in performance tradeoffs (that the cost per form passed
2510 to EVAL has gone up, and the cost per form executed by EVAL
2511 has gone down), the behavior of the system changes a little
2512 because there are no longer any interpreted function objects.
2513 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2514 e.g. doing COMPILE on the output of interactive DEFUN is
2516 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2518 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2519 has changed from (:CL) to NIL.
2520 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2521 e.g. #\Newline and #\Tab, don't have names specified in the
2522 ANSI Common Lisp standard, is now based on their ASCII symbolic
2523 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2524 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2525 are no longer used for output.
2526 ** changes in internal implementation constants:
2527 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2528 4 million. (If your application spends a lot of time GCing and
2529 you have a lot of RAM, you might want to experiment with
2530 increasing it even more.)
2531 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2532 However, almost all old code should still continue to work without
2533 immediate update, as SB-C-CALL is now a (deprecated) nickname
2535 ** Old operator names in the style DEF-FOO are now deprecated in
2536 favor of new corresponding names DEFINE-FOO, for consistency with
2537 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2538 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2539 internal symbols, but a few supported extensions like
2540 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2541 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2542 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2543 etc. as you get deeper into recursive calls to the debugger
2544 command loop, instead of the old "5]", "5]]", "5]]]"
2545 sequence. (I was motivated to do this when squabbles between
2546 ILISP and SBCL left me very deeply nested in the debugger. In the
2547 short term, this change will probably provoke more ILISP/SBCL
2548 squabbles, but hopefully it will be an improvement in the long run.)
2549 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2550 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2551 function, since declaiming return types involving aliens is
2552 (1) annoyingly messy to do by hand and (2) vital to efficient
2553 compilation of code which calls such functions.
2554 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2555 longer reexported by the SB-EXT package. They're solely useful
2556 for alien code, so it seems more logical that you should get
2557 them from the SB-ALIEN package, not in SB-EXT.
2558 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2559 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2560 features. Instead, the code that they used to control is always
2561 built into the system.
2562 * many other bug fixes
2563 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2564 advantage of the new EVAL-WHEN stuff and to clean them up in
2565 general, and they are now more ANSI-compliant in a number of
2566 ways. Martin Atzmueller is responsible for a lot of this.
2567 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2569 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2570 *** correct ERROR type for various file operations
2571 *** some fixes for Lisp streams
2572 *** DEFMETHOD syntax checking
2573 *** changing old weird representation of debug information as
2574 strings (which, among their other deficiencies, don't transform
2575 correctly when you rename packages, and don't change their
2576 print representation when you change things like *PACKAGE*
2577 and *PRINT-LENGTH*) to symbols and lists of symbols
2578 He also made several improvements and fixed several bugs in DESCRIBE.
2579 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2581 *** misbehavior of WRITE-STRING/WRITE-LINE
2582 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2583 and several other LOOP problems as well
2584 *** DIRECTORY when similar filenames are present
2585 *** DEFGENERIC with :METHOD options
2586 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2587 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2588 *** argument ordering in FIND with :TEST option
2589 *** mishandled package designator argument in APROPOS-LIST
2590 *** various problems in the backquote readmacro
2591 *** a bug in APROPOS
2592 *** probably some others that I'm not describing very well here,
2593 since the CVS log documents them by reference to sbcl-devel
2594 messages, and the SourceForge archives aren't working well.:-(
2595 ** Dan Barlow improved the Alpha port (and is making progress on the
2596 PPC port, for those of you who think different).
2597 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2598 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2599 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2600 continues to work on the SPARC port (for those of you in a position
2601 to look down upon our little PC-compatible boxes from a great height).
2602 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2603 and uses that copy, rather than the current dynamic value, when
2604 it's trying to decide whether to truncate output. Thus e.g.
2605 (let ((*print-lines* 50))
2606 (pprint-logical-block (stream nil)
2608 (let ((*print-lines* 8))
2609 (print (aref possiblybigthings i) stream)))))
2610 should now truncate the logical block only at 50 lines, instead of
2611 often truncating it at 8 lines, as it did before.
2612 * The doc/cmucl/ directory, containing old CMU CL documentation
2613 from the time of the fork, is no longer part of the base system.
2614 SourceForge has shut down its anonymous FTP service, and with it
2615 my original plan for distributing the old CMU CL documentation
2616 there. For now, if you need these files you can download an old
2617 SBCL source release and extract them from it.
2618 * The fasl file version number changed again, for dozens of reasons,
2619 some of which are apparent above.
2621 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2622 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2623 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2624 patch, so that DISASSEMBLE output is much nicer.
2625 * The code in the SB-PROFILE package now seems reasonably stable.
2626 I still haven't decided what the final interface should look like
2627 (I'd like PROFILE to interact cleanly with TRACE, since both
2628 facilities use function encapsulation) but if you have a need
2629 for profiling now, you can probably use it successfully with
2630 the current CMU-CL-style interface.
2631 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2632 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2633 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2634 link now returns the dangling link itself, and for similar
2635 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2636 link itself. (In these cases the old code signalled an error and
2637 looped endlessly, respectively.) Thus, DIRECTORY now works even
2638 in the presence of dangling and cyclic symbolic links.
2639 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2640 is now a supported extension again, since the consensus on
2641 sbcl-devel was that it can be useful for ordinary development
2642 work, not just for debugging SBCL itself.
2643 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2644 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2645 that currently-defined functions might be redefined later with
2646 different return types.
2647 * Hash tables can be printed readably, as inspired by CMU CL code
2648 of Eric Marsden and SBCL code of Martin Atzmueller.
2649 * better error handling in CLOS method combination, thanks to
2650 Martin Atzmueller porting Pierre Mai's CMU CL patches
2651 * more overflow fixes for >16Mbyte I/O buffers
2652 * A bug in READ has been fixed, so that now a single Ctrl-D
2653 character suffices to cause end-of-file on character streams.
2654 In particular, now you only need one Ctrl-D at the command
2655 line (not two) to exit SBCL.
2656 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2658 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2659 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2660 does the right thing.
2661 * The implementation of some type tests, especially for CONDITION
2662 types, is now tidier and maybe faster, due to CMU CL code
2663 originally by Douglas Crosher, ported by Martin Atzmueller.
2664 * Some math functions have been fixed, and there are new
2665 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2666 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2667 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2668 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2669 likely to be built into sbcl-0.7.0.)
2670 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2671 CL, ported to SBCL by Martin Atzmueller)
2672 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2673 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2674 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2675 * There's a new slam.sh hack to shorten the edit/compile/debug
2676 cycle for low-level changes to SBCL itself, and a new
2677 :SB-AFTER-XC-CORE target feature to control the generation of
2678 the after-xc.core file needed by slam.sh.
2679 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2680 COMPILE-FILE is no longer supported, so that now every function
2681 gets an entry point, so that block compilation looks a little
2682 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2683 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2684 returns the number of bytes consed since the system started,
2685 rather than the number consed since the first time the function
2686 was called. (The new definition parallels ANSI functions like
2687 CL:GET-INTERNAL-RUN-TIME.)
2688 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2689 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2690 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2691 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2692 * Fasl file version numbers are now independent of the target CPU,
2693 since historically most system changes which required version
2694 number changes have affected all CPUs equally. Similarly,
2695 the byte fasl file version is now equal to the ordinary
2698 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2699 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2700 is no longer recognized. I apologize for this, because it was
2701 listed in SB-EXT as a supported extension, but I found that
2702 its existing behavior was poorly specified, as well as incorrectly
2703 specified, and it looked like too much of a mess to straighten it
2704 out. I have enough on my hands trying to get ANSI stuff to work..
2705 * many patches ported from CMU CL by Martin Atzmueller, with
2706 half a dozen bug fixes in pretty-printing and the debugger, and
2707 half a dozen others elsewhere
2708 * fixed bug 13: Floating point infinities are now supported again.
2709 They might still be a little bit flaky, but thanks to bug reports
2710 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2711 as flaky as they were.
2712 * The --noprogrammer command line option is now supported. (Its
2713 behavior is slightly different in detail from what the old man
2714 page claimed it would do, but it's still appropriate under the
2715 same circumstances that the man page talks about.)
2716 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2717 are now supported, and enabled by default. Thus, the compiler can
2718 handle many floating point and complex operations much less
2719 inefficiently. (Thus e.g. you can implement a complex FFT
2721 * The compiler now detects type mismatches between DECLAIM FTYPE
2722 and DEFUN better, and implements CHECK-TYPE more correctly, and
2723 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2724 thanks to patches from Martin Atzmueller.
2725 * various fixes to make the cross-compiler more portable to
2726 ANSI-conforming-but-different cross-compilation hosts (notably
2727 Lispworks for Windows, following bug reports from Arthur Lemmens)
2728 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2729 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2730 fixed thanks to Pierre Mai's CMU CL patches.
2731 * fixes to keep the system from overflowing internal counters when
2732 it tries to use i/o buffers larger than 16M bytes
2733 * fixed bug 45a: Various internal functions required to support
2734 complex special functions have been merged from CMU CL sources.
2735 (When I was first setting up SBCL, I misunderstood a compile-time
2736 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2737 * improved support for type intersection and union, fixing bug 12
2738 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2739 more obscure bugs as well
2740 * some steps toward byte-compiling non-performance-critical
2741 parts of the system, courtesy of patches from Martin Atzmueller
2742 * Christophe Rhodes has made some debian packages of sbcl at
2743 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2744 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2745 stable, but are nonetheless usable. When he's ready, I'd be happy
2746 to add them to the SourceForge "File Releases" section. (And if
2747 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2748 * new fasl file format version number (because of changes in
2749 internal representation of (OR ..) types to accommodate the new
2750 support for (AND ..) types, among other things)
2752 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2753 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2755 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2757 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2758 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2759 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2760 * DESCRIBE now works on structure objects again.
2761 * Most function call argument type mismatches are now handled as
2762 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2763 know whether the function will be redefined before the call is
2764 executed. (The compiler could flag local calls with full WARNINGs,
2765 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2766 it doesn't keep track of enough information to know whether calls
2767 are local in this sense.)
2768 * Compiler output is now more verbose, with messages truncated
2769 later than before. (There should be some supported way for users
2770 to override the default verbosity, but I haven't decided how to
2771 provide it yet, so this behavior is still controlled by the internal
2772 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2773 src/compiler/ir1util.lisp.)
2774 * Fasl file format version numbers have increased again, because
2775 support for the Gray streams extension changes the layout of the
2776 system's STREAM objects.
2777 * The Gray subclassable streams extension now works, thanks to a
2778 patch from Martin Atzmueller.
2779 * The full LOAD-FOREIGN extension (not just the primitive
2780 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2781 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2782 but like most other programs, it defaults to copying the Unix
2783 environment from the original process instead of starting the
2784 new process in an empty environment.
2785 * Extensions which manipulate the Unix environment now support
2786 an :ENVIRONMENT keyword option which doesn't smash case or
2787 do other bad things. The CMU-CL-style :ENV option is retained
2788 for porting convenience.
2789 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2790 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2792 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2794 * A patch from Martin Atzmueller seems to have solved the SIGINT
2795 problem, and as far as we know, signal-handling now works cleanly.
2796 (If you find any new bugs, please report them!)
2797 * The system no longer defaults Lisp source file names to types
2798 ".l", ".cl", or ".lsp", but only to ".lisp".
2799 * The compiler no longer uses special default file extensions for
2800 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2801 seems to expect a single default extension for all compiled code,
2802 and there's no compelling reason to try to stretch the standard
2803 to allow two different extensions.) Instead, byte-compiled files
2804 default to the same extension as native-compiled files.
2805 * Fasl file format version numbers have increased again, because
2806 a rearrangement of internal implementation packages made some
2807 dumped symbols in old fasl files unreadable in new cores.
2808 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2809 many fewer weird special cases.
2810 * Bug #17 (differing COMPILE-FILE behavior between logical and
2811 physical pathnames) has been fixed, and some related misbehavior too,
2812 thanks to a patch from Martin Atzmueller.
2813 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2814 by Tim Moore, ported to SBCL by Martin Atzmueller.
2815 * Martin Atzmueller fixed several filesystem-related problems,
2816 including bug #36, in part by porting CMU CL patches, which were
2817 written in part by Paul Werkowski.
2818 * More compiler warnings in src/runtime/ are gone, thanks to
2819 more patches from Martin Atzmueller.
2820 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2823 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2825 * DESCRIBE now works on CONDITION objects.
2826 * The debugger now handles errors which arise when trying to print
2827 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2829 * The build system now uses an additional file, customize-target-features.lisp,
2830 to allow local modifications to the target *FEATURES* list. (The point of
2831 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2832 debugging features enabled, without having to worry about propagating it
2833 into everyone's system when I do a "cvs update".) When no
2834 customize-target-features.lisp file exists, the target *FEATURES* list
2835 should be constructed the same way as before.
2836 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2837 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2838 warn about behavior which is undefined under the ANSI spec. Note: This
2839 is specified by ANSI, but it's not very popular with programmers.
2840 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2841 macro in the SBCL sources for an example of a workaround which you
2842 might use to make portable ANSI-standard code which does what you want.
2843 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2844 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2845 when it's used as a non-toplevel form. (This is required in order
2846 to implement the DEFCONSTANT-EQX macro.)
2847 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2848 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2849 funny things, due to bugs in the implementation of EVAL-WHEN
2850 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2851 non-toplevel DEFCONSTANTs are uncommon.)
2852 * The core file version number and fasl file version number have been
2853 incremented, because the old noncompliant DEFCONSTANT behavior involved
2854 calling functions which no longer exist, and because I also took the
2855 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2856 * fixed bug 1 (error handling before read-eval-print loop starts), and
2857 redid debugger restarts and related debugger commands somewhat while
2859 ** The QUIT debugger command is gone, since it did something
2860 rather different than the SB-EXT:QUIT command, and since it never
2861 worked properly outside the main toplevel read/eval/print loop.
2862 Invoking the new TOPLEVEL restart provides the same functionality.
2863 ** The GO debugger command is also gone, since you can just invoke
2864 the CONTINUE restart directly instead.
2865 ** The TOP debugger command is also gone, since it's redundant with the
2866 FRAME 0 command, and since it interfered with abbreviations for the
2868 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2870 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2871 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2872 some porting work by Martin Atzmueller)
2873 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2874 with a boatload of patches which he ported to SBCL. Now that those
2876 ** The system tries to make sure that its low-priority messages
2877 are prefixed by semicolons, to help people who like to use
2878 syntax highlighting in their ILISP buffer. (This patch
2879 was originally due to Raymond Toy.)
2880 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2881 patches originally written by Raymond Toy.
2882 ** The compiler understands coercion between single-value and
2883 multiple-VALUES type expressions better, getting rid of some very
2884 weird behavior, thanks to patches originally by Robert MacLachlan
2885 and Douglas Crosher.
2886 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2887 lambda lists, thanks to a patch originally by Pierre Mai.
2888 ** The system no longer bogusly warns about "abbreviated type
2890 ** The compiler gets less confused by inlining and RETURN-FROM,
2891 thanks to some patches originally by Tim Moore.
2892 ** The system no longer hangs when dumping circular lists to fasl
2893 files, thanks to a patch originally from Douglas Crosher.
2894 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2895 undefined function error.
2896 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2897 in structure slot initforms) for now, documented workaround instead:-|
2898 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2899 * fixed bug 5: added stubs for various Gray stream functions called
2900 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2901 installed, at least appropriate type errors are generated
2902 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2903 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2904 consistently in DEFMETHOD forms.
2905 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2906 it doesn't seem to affect SBCL after all
2907 * The C runtime system now builds with better optimization and many
2908 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2910 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2912 * The system is now under CVS at SourceForge (instead of the
2913 CVS repository on my home machine).
2914 * The new signal handling code has been tweaked to treat register
2915 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2916 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2917 downstream, e.g. in debug-int.lisp, has implicit dependencies
2918 on the unsignedness of integer representation of machine words,
2919 and that caused the system to bomb out with infinite regress
2920 when trying to recover from type errors involving signed values,
2921 e.g. (BUTLAST '(1 2 3) -1).)
2922 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2923 (This was one of the bugs Peter Van Eynde reported back in July.)
2924 * The system now uses code inspired by Colin Walters' O(N)
2925 implementation of MAP (from the cmucl-imp@cons.org mailing
2926 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2927 inline the MAP operation, and there is more than one
2928 sequence argument to the MAP call (so that it can't just
2929 do ETYPECASE once and for all based on the type of the
2930 single sequence argument). (The old non-inline implementation
2931 of the general M-argument sequence-of-length-N case required
2932 O(M*N*N) time when any of the sequence arguments were LISTs.)
2933 * The QUIT :UNIX-CODE keyword argument has been renamed to
2934 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2935 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2937 * Raymond Wiker's patches to port dynamic loading from Linux to
2938 FreeBSD have been added.
2939 * The BUGS file is now more nearly up to date, thanks in large part
2940 to Martin Atzmueller's review of it.
2941 * The debugger now flushes standard output streams before it begins
2942 its output ("debugger invoked" and so forth).
2943 * The core version number and fasl file version number have both
2944 been incremented, because of incompatible changes in the layout
2946 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2947 used to not be called for a saved Lisp image.)
2948 * Martin Atzmueller's version of a patch to fix a compiler crash,
2949 as posted on sbcl-devel 13 September 2000, has been installed.
2950 * Instead of installing Martin Atzmueller's patch for the
2951 compiler transform for SUBSEQ, I deleted the compiler transform,
2952 and transforms for some similar consing operations.
2953 * A bug in signal handling which kept TRACE from working on OpenBSD
2955 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2956 away by constant folding
2957 * The system now defines its address space constants in one place
2958 (in the Lisp sources), and propagates them automatically elsewhere
2959 (through GENESIS and the sbcl.h file). Therefore, patching the
2960 address map is less unnecessarily tedious and error-prone. The
2961 Lisp names of address space constants have also been systematized.
2962 * CVS tags like dollar-Header-dollar have been removed from
2963 the sources, because they have never saved me trouble and
2964 they've been source of trouble working with patches and other
2965 diff-related operations.
2966 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2967 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2969 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2971 * The system has been ported to OpenBSD.
2972 * The system now compiles with a simple "sh make.sh" on the systems
2973 that it's supported on. I.e., now you no longer need to tweak
2974 text in the target-features.lisp-expr and symlinks in src/runtime/
2975 by hand, the make.sh takes care of it for you.
2976 * The system is no longer so grossly inefficient when compiling code
2977 involving vectors implemented as general (not simple) vectors (VECTOR T),
2978 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2979 which dares to use the various sequence functions on non-simple
2980 vectors, takes less of a performance hit.
2981 * There is now a primitive type predicate VECTOR-T-P
2982 to test for the (VECTOR T) type, so that e.g.
2983 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2984 can now be compiled with some semblance of efficiency. (The old code
2985 turned the type declaration into a full call to %TYPEP at runtime!)
2986 * AREF on (VECTOR T) is still not fast, since it's still compiled
2987 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2988 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2989 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2990 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2991 without making full calls to SUBTYPEP at runtime.
2992 (Some analogous efficiency issues for non-simple vectors specialized to
2993 element types other than T, or for non-simple multidimensional arrays,
2994 have not been addressed. They could almost certainly be handled the
2995 same way if anyone is motivated to do so.)
2996 * The changes in array handling break binary compatibility, so
2997 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2998 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2999 instead of (VALUES T T).
3000 * By following the instructions that Dan Barlow posted to sbcl-devel
3001 on 2 July 2000, I was able to enable primitive dynamic object
3002 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3003 functionality is not implemented (since it calls ld to resolve
3004 library references automatically, requiring RUN-PROGRAM for its
3005 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3006 try to resolve library references) is now supported.
3007 * The system now flushes the standard output streams when it terminates,
3008 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3009 them at several other probably-convenient times, e.g. in each pass of
3010 the toplevel read-eval-print loop, and after evaluating a form given
3011 as an "--eval" command-line option. (These changes were motivated by a
3012 discussion of stream flushing issues on cmucl-imp in August 2000.)
3013 * The source transform for TYPEP of array types no longer assumes
3014 that an array whose element type is a not-yet-defined type
3015 is implemented as an array of T, but instead punts, so that the
3016 type will be interpreted at runtime.
3017 * There is now some support for cross-compiling in make.sh: each of
3018 the phases of make.sh has its own script. (This should be transparent
3019 to people doing ordinary, non-cross-compile builds.)
3020 * Since my laptop doesn't have hundreds of megabytes of memory like
3021 my desktop machine, I became more motivated to do some items on
3022 my to-do list in order to reduce the size of the system a little:
3023 ** Arrange for various needed-only-at-cold-init things to be
3024 uninterned after cold init. To support this, those things have
3025 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3026 symbols with such names are now uninterned after cold init).
3027 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3028 fasl files for cold load.
3029 ** Remove the old compiler structure pooling code (which used to
3030 be conditional on the target feature :SB-ALLOC) completely.
3031 ** Redo the representation of some data in cold init to be more compact.
3032 (I also looked into supporting byte compiled code at bootstrap time,
3033 which would probably reduce the size of the system a lot, but that
3034 looked too complicated, so I punted for now.)
3035 * The maximum signal nesting depth in the src/runtime/ support code has
3036 been reduced from 4096 to 256. (I don't know any reason for the very
3037 large old value. If the new smaller value turns out to break something,
3038 I'll probably just bump it back up.)
3039 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3041 * Many, many bugs reported by Peter Van Eynde have been added to
3042 the BUGS list; some have even been fixed.
3043 * While enabling dynamic object file loading, I tried to make the
3044 code easier to understand, renaming various functions and variables
3045 with less ambiguous names, and changing some function calling
3046 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3047 * While trying to figure out how to do the OpenBSD port, I tried to
3048 clean up some of the code in src/runtime/. In particular, I dropped
3049 support for non-POSIX signal handling, added various comments,
3050 tweaked the code to reduce the number of compilation warnings, and
3051 renamed some files to increase consistency.
3052 * To support the new automatic configuration functionality in make.sh,
3053 the source file target-features.lisp-expr has been replaced with the
3054 source file base-target-features.lisp-expr and the machine-generated
3055 file local-target-features.lisp-expr.
3056 * fixed a stupid quoting error in make.sh so that using CMU CL
3057 "lisp -batch" as cross-compilation host works again
3059 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3061 * DESCRIBE no longer tries to call itself recursively to describe
3062 bound/fbound values, so that it no longer fails on symbols which are
3063 bound to themselves (like keywords, T, and NIL).
3064 * DESCRIBE now works on generic functions.
3065 * The printer now prints less-screwed-up representations of closures
3066 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3067 * A private symbol is used instead of the :EMPTY keyword previously
3068 used to mark empty slots in hash tables. Thus
3069 (DEFVAR *HT* (MAKE-HASH-TABLE))
3070 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3071 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3072 now does what ANSI says that it should. (You can still get
3073 similar noncompliant behavior if bang on the hash table
3074 implementation with all the symbols you get back from
3075 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3076 This breaks binary compatibility, since tests for equality to
3077 :EMPTY are wired into things like the macroexpansion of
3078 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3080 * There's now a minimal placeholder implementation for CL:STEP,
3081 as required by ANSI.
3082 * An obscure bug in the interaction of the normal compiler, the byte
3083 compiler, inlining, and structure predicates has been patched
3084 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3085 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3087 * Missing ordinary arguments in a macro call are now detected even
3088 when the macro lambda list contains &KEY or &REST.
3089 * The debugger no longer complains about encountering the top of the
3090 stack when you type "FRAME 0" to explicitly instruct it to go to
3091 the top of the stack. And it now prints the frame you request even
3092 if it's the current frame (instead of saying "You are here.").
3093 * As specified by ANSI, the system now always prints keywords
3094 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3096 * The default initial SIZE of HASH-TABLEs is now smaller.
3097 * Type information from CLOS class dispatch is now propagated
3098 into DEFMETHOD bodies, so that e.g.
3099 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3101 is now basically equivalent to
3102 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3103 (DECLARE (TYPE SINGLE-FLOAT X))
3105 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3106 operation, without having to do run-time type dispatch.
3107 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3108 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3109 * Since I don't seem to be making big file reorganizations very often
3110 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3111 is overflowing my ability to conveniently back them up), I've finally
3112 checked the system into CVS. (The CVS repository is on my home system,
3113 not at SourceForge -- putting it on SourceForge might come later.)
3114 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3115 high-level GC-NOTIFY-FOO functions send their output. (There's
3116 still very little control of where low-level verbose GC functions
3117 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3118 less than it used to -- the GC-NOTIFY-FOO functions are now under
3119 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3120 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3121 in only one place in the source code, and propagates it automatically
3122 everywhere that it's needed. Thus e.g. when I bump the version from
3123 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3124 * The C source files now include boilerplate legalese and documentation
3125 at the head of each file (just as the Lisp source files already did).
3126 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3127 to his page will be replaced with a link to his new CLiki service.
3129 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3131 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3132 * The build process now looks for GNU make under the default name "gmake",
3133 instead of "make" as it used to. If GNU make is not available as "gmake"
3134 on your system, you can change this default behavior by setting the
3135 GNUMAKE environment variable.
3136 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3137 can build without error under CMU CL.
3139 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3141 * There is now a partial SBCL user manual (with some new text and some
3142 text cribbed from the CMU CL manual).
3143 * The beginnings of a profiler have been added (starting with the
3144 CMU CL profiler and simplifying and cleaning up). Eventually the
3145 main interface should be through the TRACE macro, but for now,
3146 it's still accessed through vaguely CMU-CL-style functions and macros
3147 exported from the package SB-PROFILE.
3148 * Some problems left over from porting CMU CL to the new
3149 cross-compilation bootstrap process have been cleaned up:
3150 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3151 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3152 part when getting rid of the compiler *BACKEND* stuff.)
3153 ** The value of *NULL-TYPE* was screwed up, because it was
3154 being initialized before the type system knew the final
3155 definition of the 'NULL type. This screwed up several key
3156 optimizations in the compiler, causing inefficiency in all sorts
3157 of places. (I found it because I wanted to understand why
3158 GET-INTERNAL-RUN-TIME was consing.)
3159 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3160 PROCLAIM FTYPE information. Unfortunately this broke binary
3161 compatibility again, since now the forms output by DEFGENERIC
3162 to refer to functions which didn't exist in 0.6.3.
3163 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3164 can use the new (as of 0.6.3) transform for SOME into MAP into
3166 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3167 values to (INTEGER 0 1000000), so that the time code will no longer
3168 occasionally get blown up by Linux returning 1000000 microseconds
3169 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3170 its output conform to the ANSI spec. (Alas, this makes its output
3171 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3172 * A full call to MAP NIL with a single sequence argument no longer conses.
3173 * fixes to problems pointed out by Martin Atzmueller:
3174 * The manual page no longer talks about multiprocessing as though
3175 it were currently supported.
3176 * The ILISP support patches have been removed from the distribution,
3177 because as of version 5.10.1, ILISP now supports SBCL without us
3178 having to maintain patches.
3179 * added a modified version of Raymond Toy's recent CMU CL patch for
3180 EQUALP comparison of HASH-TABLE
3182 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3184 * The system still can't cross-compile itself with
3185 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3186 entails), but at least it can compile more of itself that way
3187 than it used to be able to, and various buglets which were uncovered
3188 by trying to cross-compile itself that way have now been fixed.
3189 * This release breaks binary compatibility again. This time
3190 at least I've incremented the FASL file format version to 2, so that the
3191 problem can be detected reliably instead of just causing weird errors.
3192 * various new style warnings:
3193 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3194 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3195 ** using the lexical binding of a variable named in the *FOO* style
3196 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3197 as specified by ANSI.
3198 * *RANDOM-STATE* is no longer automatically initialized from
3199 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3200 default behavior of the system is to repeat its behavior every time
3201 it's run. If you'd like to change this behavior, you can always
3202 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3203 old convention there was no comparably easy way to get the system to
3204 repeat its behavior every time it was run.
3205 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3206 FTYPE declarations has been removed, in favor of their ANSI
3207 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3208 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3209 the types of their sequence arguments can be determined at compile time.
3210 This is done through a new open code expansion for MAP which eliminates
3211 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3212 argument types can be determined at compile time.
3213 * The optimizer now transforms COERCE into an identity operation when it
3214 can prove that the coerced object is already of the correct type. (This
3215 can be a win for machine generated code, including the output of other
3216 optimization transforms, such as the MAP transform above.)
3217 * Credit information has been moved from source file headers into CREDITS.
3218 * Source file headers have been made more standard.
3219 * The CASE macro now compiles without complaining even when it has
3222 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3224 * (Note that the way that the PCL macroexpansions were rewritten
3225 to accommodate the change in DEFGENERIC below breaks binary
3226 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3227 not run under sbcl-0.6.2. Once we get out of alpha releases,
3228 i.e. hit release 1.0.0, we'll probably try to maintain binary
3229 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3230 and sbcl-1.4.4. Until then, however, it might be fairly common
3231 for maintenance releases to break binary compatibility.)
3232 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3234 * The reporting of unhandled signals has been changed to print some
3235 explanatory text as well as the report form. (Previously only
3236 the report form was printed.)
3237 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3238 it defines, so that the compiler no longer issues undefined function
3239 warnings for compiled-but-not-yet-loaded generic functions.
3240 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3241 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3242 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3243 section of the ANSI spec are supported.
3244 * The "" nickname for the "KEYWORD" package has been removed.
3245 The reader still handles symbol tokens which begin with a package marker
3246 as keywords, but it doesn't expose its mechanism for doing so in the
3247 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3248 * The system now issues STYLE-WARNINGs for contradictory TYPE
3249 proclamations. (Warnings for contradictory FTYPE proclamations would
3250 be nice too, but those can't be done usefully unless the type system
3251 is made smarter about FUNCTION types.)
3252 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3253 systematized, so that "*target-*.lisp is supposed to exist only on the
3254 target and imply that there's a related file which exists on the
3255 host, and *host-*.lisp is supposed to exist only on the host and imply
3256 that there's a related file which exists on the target. This involves a
3257 lot of renaming. Hopefully the acute confusion caused by the renaming
3258 will be justified by the reduction in chronic confusion..
3259 ** runtime-type.lisp -> early-target-type.lisp
3260 ** target-type.lisp -> late-target-type.lisp
3261 ** early-host-format.lisp -> early-format.lisp
3262 ** late-host-format.lisp -> late-format.lisp
3263 ** host-error.lisp -> misc-error.lisp
3264 ** early-error.lisp -> early-target-error.lisp
3265 ** late-error.lisp -> late-target-error.lisp
3266 ** host-defboot.lisp -> early-defboot.lisp
3267 ** code/misc.lisp -> code/target-misc.lisp
3268 ** code/host-misc.lisp -> code/misc.lisp
3269 ** code/numbers.lisp -> code/target-numbers.lisp
3270 ** code/early-numbers.lisp -> numbers.lisp
3271 ** early-host-type.lisp -> early-type.lisp
3272 ** late-host-type.lisp -> late-type.lisp
3273 ** host-typep.lisp -> typep.lisp
3274 ** load.lisp -> target-load.lisp
3275 ** host-load.lisp -> load.lisp
3276 ** host-disassem.lisp -> disassem.lisp
3277 ** host-insts.lisp -> insts.lisp
3278 ** byte-comp.lisp -> target-byte-comp.lisp
3279 ** host-byte-comp.lisp -> byte-comp.lisp
3280 ** host-signal.lisp -> signal.lisp
3281 ** host-defstruct.lisp -> defstruct.lisp
3282 ** late-target-type.lisp -> deftypes-for-target.lisp
3283 Furthermore, several other previously target-only files foo.lisp (e.g.
3284 hash-table.lisp and random.lisp) have been split into a target-and-host
3285 foo.lisp file and a target-only target-foo.lisp file, with their key type
3286 definitions in the target-and-host part, so that the cross-compiler will
3287 know more about target types.
3288 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3289 gone away. In their place are various *BACKEND-FOO* variables
3290 corresponding to the slots of the old structure.
3291 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3292 into the target SBCL has been fixed.
3293 * The chill.lisp system for loading cold code into a running SBCL
3295 * Support for the CMU CL "scavenger hook" extension has been removed.
3296 (It was undocumented and unused in the CMU CL sources that SBCL was
3297 derived from, and stale in sbcl-0.6.1.)
3298 * Various errors in the cross-compiler type system were detected
3299 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3300 (enabling various consistency checks). Many of them were fixed,
3301 but some hard problems remain, so the compiler is back to
3302 running without *TYPE-SYSTEM-INITIALIZED* for now.
3303 * As part of the cross-compiler type system cleanup, I implemented
3304 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3305 * I have started adding UNCROSS calls throughout the type system
3306 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3307 on cross-compiler input files will be able to go away, and various
3309 * CONSTANTP now returns true for quoted forms (as explicitly required
3312 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3314 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3315 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3316 (MAKE-LIST -1). In the longer term, it ought to become true
3317 that declarations are assertions even at SAFETY 1. For now, it's not
3318 quite true even at SAFETY 3, but it's at least more nearly true..
3319 (Note that this change seems to increases the size of the system by
3320 O(5%) and to decrease the speed of the compiler by 20% or more.)
3321 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3322 for the problem of printing dozens of lines of distracting information
3323 about low-level system machinery as part of the top stack frame
3324 on entry to the debugger when an undefined function was called.
3325 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3326 is not reset to COMMON-LISP-USER.
3327 * Compilation of stuff related to dyncount.lisp has been made conditional
3328 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3329 smaller. The various dyncount-related symbols have been moved into
3330 a new "SB-DYNCOUNT" package.
3331 * tty-inspect.lisp has been renamed to inspect.lisp.
3332 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3333 feature has gone away. (When we eventually port to other flavors of
3334 libc and/or Unix, we'll try to make the differences between flavors
3335 invisible at the user level.)
3336 * Various other *FEATURES* tags, and/or their associated conditionals,
3337 have been removed if obsolescent, or given better documentation, or
3338 sometimes given more-mnemonic names.
3340 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3342 * tidied up "make.sh" script
3343 * tidied up system directory structure
3344 * better "clean.sh" behavior
3345 * added doc/FOR-CMUCL-DEVELOPERS
3346 * many many small tweaks to output format, e.g. removing possibly-confusing
3347 trailing #\. character in DESCRIBE-INSTANCE
3348 * (EQUALP #\A 'A) no longer signals an error.
3349 * new hashing code, including EQUALP hashing
3350 * tidied up Lisp initialization and toplevel
3351 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3352 * command line argument processing
3353 * added POSIX-GETENV function to deal with Unix-ish environment variables
3354 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3355 terminating SBCL on EOF
3356 * non-verbose GC by default
3357 * There is no more "sbcl" shell script; the sbcl file is now the C
3358 runtime executable (just like CMU CL).
3359 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3361 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3362 symbol and package information as Lisp native symbol and package objects
3363 instead of strings naming symbols and strings naming packages. This way,
3364 whenever packages are renamed (as in warm init), debug information is
3365 transformed along with everything else.
3366 * tweaked the optimization policy declarations which control the building
3367 of SBCL itself. Now, among other things, the system no longer saves
3368 source location debugging information. (This helps two problems at once
3369 by reducing SBCL size and by keeping SBCL from trying to look for its
3370 sources -- which may not exist -- when reporting errors.)
3371 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3372 debugging and testing purposes
3373 * cleaned up printing, making the printer call PRINT-OBJECT for
3374 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3375 methods, giving nearly-ANSI behavior
3376 * converted almost all special variables to use *FOO* naming convention
3377 * deleted PARSE-TIME functionality, since it can be done portably
3378 * moved some files out of cold init into warm init
3379 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3381 * regularized formatting of source files
3382 * added an install.sh script
3383 * fixed ridiculous memory usage of cross-compiler by making
3384 compiler/alloc.lisp not try to do pooling unless it can hook
3385 itself into the GC of the cross-compilation host. Now the system
3386 builds nicely on my old laptop.
3387 * added :SB-ALLOC in target-features.lisp-expr
3388 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3389 was not implemented)
3390 * re-did condition handling and note reporting in the compiler. Notes
3391 are no longer handled by signalling conditions. Style warnings
3392 and warnings are handled more correctly and reported in such a way
3393 that it's easy to find one or the other in your output (so that you
3394 can e.g. figure out which of many problems caused COMPILE-FILE to
3396 * changed the severity of several compiler warnings from full WARNING
3397 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3398 compiler note reporting so that it doesn't use the condition system
3399 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3400 COMPILE-FILE command)
3401 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3402 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3403 no longer supported, since their implementation was deeply intertwingled
3404 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3405 * removed bogus "support" for compiler macros named (SETF FOO), and
3406 removed the compiler macro for SETF INFO (but only after making a fool
3407 of myself on the cmucl-imp mailing list by posting a bogus patch for
3408 DEFINE-COMPILER-MACRO..)
3409 * Compiled files containing forms which have side effects on the Lisp
3410 reader (such as DEFPACKAGE forms) are now handled more correctly.
3411 (Compiler queuing of top level lambdas has been suppressed by setting
3412 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3413 * deleted various currently-unused source files, e.g. gengc.lisp. They
3414 may be added back at some point e.g. when porting to other architectures,
3415 but until they are it's distracting to distribute them and to try to
3417 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3418 there were so many of them they're just distractions, and UNCROSS is
3419 known to be able to handle the current sources
3420 * moved PROFILE functionality into TRACE, so that it will be clear
3421 how the wrapping and unwrapping of functions when you profile them
3422 interacts with the wrapping and unwrapping of functions when you
3423 trace them. (Actually, the functionality isn't there yet, but at least
3424 the interface specification is there. Hopefully, the functionality will
3425 arrive with some future maintenance release.)
3426 * removed host-oops.lisp
3427 * changed signature of QUIT function to allow UNIX-CODE argument
3428 * fixed READ-SEQUENCE bug
3429 * tweaked verbose GC output so that it looks more like the progress
3430 output that ANSI specifies for functions like LOAD
3431 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3432 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob