1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
3 * bug fix: add a workaround to SBCL looping infinitely at startup on
4 Linux kernels with apparently buggy implementations of personality().
5 (thanks to Svein Ove Aas)
6 * bug fix: Unicode symbols are correctly printed in LDB backtraces
7 (thanks to David Lichteblau)
8 * optimization: non-open coded uses of numeric comparison operators
9 (e.g. >) no longer cons when called with more than one parameter
10 on platforms supporting dynamic-extent allocation.
11 * enhancement: saving cores with foreign code loaded is now supported
12 on MIPS/Linux in addition to the previously supported platforms.
13 * bug fix: threads stacks belonging to dead threads are freed by the
14 next exiting thread, no need to gc to collect thread stacks anymore
16 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
17 * new feature: timers based on Zach Beane's excellent timer package
18 * added support for the following external formats: koi8-u,
19 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
20 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
21 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
22 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
23 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
24 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
25 * incompatible change: a threaded SBCL will no longer revert to
26 non-threaded mode on non-NPTL systems, but refuse to start entirely.
27 * bug fix: interrupts are disabled until startup is complete; no
28 more sigsegvs when receiving a signal to soon
29 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
31 * bug fix: add a workaround for the memory randomization features in
32 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
33 workaround will only be in effect on systems with the proc filesystem
35 * bug fix: printing objects of type HASH-TABLE signals a
36 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
38 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
39 have been running for over 50 days. (reported by Gilbert Baumann)
40 * bug fix: the logic for getting names of functions gets less
41 confused when confronded with alternate-metaclass
42 funcallable-instances. (reported by Cyrus Harmon)
43 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
44 consistent, even on internal alternate-metaclass objects.
45 * bug fix: flush closure information collected by physical
46 environment analysis prepass before the main pass. (bug reported
48 * bug fix: compiler pack phase does not modify a hashtable, which is
49 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
50 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
51 may be expanded. (reported by Alexander <malishev> on c.l.l)
52 * bug fix: no more style-warnings when compiling code using LAST
53 under high-SPEED low-SPACE settings. (reported by David Wallin
55 * fixed some incompatibilities between SBCL's MOP and the MOP
57 ** the METAOBJECT class is now implemented;
58 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
59 STANDARD-OBJECT, as required;
60 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
61 now compatible (as required by VALIDATE-SUPERCLASS); there
62 remains a consistency requirement over the presence or absence
63 of the FUNCTION class in the superclass list of finalized
64 classes; see the manual for more details;
65 ** the :AROUND method for COMPUTE-SLOTS on
66 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
67 requested slot ordering.
69 ** bug fix: parent thread now can be gc'ed even with a live
71 ** bug fix: binding a special with PROGV to no value is not
72 the same as that symbol not having been bound (thanks to
74 ** bug fix: binding specials is thread safe (thanks to
76 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
77 and SB-SPROF do not die with 'no handler for signal XX in
78 interrupt_handle_now(..)' anymore
79 ** bug fix: WITH-TIMEOUT works with multiple running threads
80 * fixed some bugs revealed by Paul Dietz' test suite:
81 ** ENSURE-GENERIC-FUNCTION should take a method class object for
82 the :method-class keyword argument.
84 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
85 * new port: the Solaris operating system on x86 processors is now
86 mostly supported, though some rough edges in the environment
87 remain. (thanks to Daisuke Homma)
88 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
89 space for its heap. (thanks to Thiemo Seufer)
90 * optimization: SBCL on MIPS platforms now supports dynamic-extent
91 closures. (thanks to Thiemo Seufer)
92 * minor incompatible change: eof selects abort in the debugger.
93 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
94 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
96 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
97 deprecated, and will go away in a future revision of SBCL.
98 * minor incompatible change: GC-ON and GC-OFF are no longer
99 implemented with a counter, it does not matter how many times gc
100 is switched on or off
101 * bug fix: discriminating functions for generic function classes
102 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
103 make invalid assumptions about method precedence order. (reported
105 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
106 platforms. (thanks to Thiemo Seufer)
107 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
108 * bug fix: correct pathname printing with printer escaping is on.
109 (thanks to Kevin Reid)
110 * bug fix: complex VOP definitions in "user-space" no longer trigger
111 package locks. (reported by Zach Beane)
112 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
113 for funcallable-instance functions, and is no different from
115 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
116 characters in the current readtable. (reported by Nicholas Neuss)
117 * bug fix: SUBTYPEP on various CONS types returns more a more
118 accurate acknowledgment of its certainty.
119 * optimizations: REMOVE-DUPLICATES now runs in linear time on
120 lists in some cases. This partially fixes bug 384.
121 * flush all standard streams before prompting in the REPL and the
123 * bug fix: signal handling and triggering gc do not conflict
124 directly anymore, in particular a high frequency sb-sprof does
125 not prevent gc from running
126 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
127 approximation for timezone and DST information between the
128 universal time epoch and the smallest negative 32-bit time_t.
129 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
130 year 1899 so long as the encoded time is non-negative.
131 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
132 the correct answer for non-constant rotate amounts when performing
133 an inline 32-bit rotation.
135 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
136 there is only one thread in the session
137 ** bug fix: memory leak for streams created in one thread and
138 written to in another
139 ** bug fix: lockup when compiled with gcc4
140 ** bug fix: race that allows the gc to be triggered when gc is
142 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
143 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
145 ** bug fix: finished threads are gc'ed properly
146 * fixed some bugs revealed by Paul Dietz' test suite:
147 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
148 has arguments with a different set of applicable methods from
149 the orignal arguments.
150 ** The type error thrown by MAP now has a correct expected-type
152 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
153 name a compiled function.
154 ** Three MISC tests where a large bignum was improperly coerced to
155 a float that couldn't represent that bignum during type
156 derivation were fixed.
157 ** SUBTYPEP can now handle types involving the negation of a
158 list-form FUNCTION type.
159 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
160 as (COMPLEX (AND RATIO (NOT FIXNUM))).
161 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
163 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
164 * New feature: Experimental support for bivalent streams: streams
165 opened with :element-type :default now allow character and binary
166 (unsigned-byte 8) I/O
167 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
168 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
170 * bug fix: no more highly sporadic "couldn't check whether ~S is
171 readable" when reading a stream and an interrupt hits in the middle
172 of a select system call
173 * compiler better recognizes complex arrays (reported by Eduardo
175 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
176 for variables on SBCL built with linkage-tables. (reported by Luis
178 * various error reporting improvements.
179 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
180 (Thanks to Thiemo Seufer)
181 * bug fix: nested reader invokations work correctly
182 * bug fix: it is possible to have simultaneous references to foreign
183 code and foreign data with the same name.
185 ** added x86-64 support
186 ** incompatible change: the threading api now works with thread
187 objects instead of thread ids
188 ** bug fix: threads are protected from signals and interruption when
189 starting up or going down
190 ** bug fix: a race where an exiting thread could lose its stack to gc
191 ** bug fix: don't halt on infinite error in threads if possible
192 ** fixed numerous gc deadlocks introduced in the pthread merge
193 ** bug fix: fixed thread safety issues in read and print
194 ** bug fix: debugger doesn't hang on session lock if interrupted at
195 an inappropriate moment
196 ** bug fix: run-program is now thread safe(r)
197 ** bug fix: inner with-recursive-lock no longer releases the mutex
198 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
199 * fixed some bugs revealed by Paul Dietz' test suite:
200 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
201 EXPECTED-TYPE slots filled.
202 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
204 ** MAKE-CONDITION accepts classes as type-designators.
205 ** COMPILE may never return NIL.
206 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
207 range before calling Unix time functions
209 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
210 * numerous signal handling fixes to increase stability
211 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
212 * minor incompatible change: we now correctly canonize default
213 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
214 as per the MOP, rather than the historical (INITARG INITFUNCTION
215 INITFORM). (reported by Bruno Haible)
216 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
217 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
218 for more information.
219 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
220 pathname is a directory pathname.
221 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
222 * dynamic space size on PPC has been increased to 768Mb. (thanks to
224 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
225 :METACLASS argument in addition to a class name. (reported by
226 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
227 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
228 (reported by Bruno Haible; patch from Gerd Moellmann)
229 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
231 * bug fix: more cleanups to the floating point exception handling on
232 x86-64 (thanks to James Knight)
233 * bug fix: the compiler does not try to constant fold calls of
234 COERCE and BIT-* functions when they return freshly constructed
235 objects. (reported by Nikodemus Siivola and Paul Dietz)
236 * optimization: improved the MIPS versions of generic (in the
237 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
238 * optimization: direct conversion of (unsigned-byte 32) to floats on
239 the PowerPC platform.
240 * optimization: structure instances with raw slots now use less
241 memory, and probably show better memory locality. (thanks to
243 * optimization: DYNAMIC-EXTENT declarations for lists and closures
244 are treated as requests for stack allocation on the x86-64,
245 Alpha-32, PPC and SPARC platforms.
246 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
247 the wrong file descriptor; implementation of SOCKET-OPEN-P.
248 (thanks to Tony Martinez)
250 ** gcing a dead thread can no longer lead to lockups
251 ** threads block signals until they are set up properly
252 ** errno is no longer shared by threads
253 ** interrupt-thread restores the eflags register on x86
254 ** fixed some lockups due to gc/thread interaction
255 * fixed some bugs revealed by Paul Dietz' test suite:
256 ** invalid dotted lists no longer raise a read error when
258 ** signal an error if a symbol that names a declaration is used
259 as the name of a type, or vice versa
260 ** allow using the (declare (typespec var*)) abbreviation for
261 (declare (type typespec var*)) with all type specifiers
262 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
263 ** always return NIL from PPRINT-POP when OBJECT is NIL
264 ** don't signal errors when pretty-printing malformed LABELS,
265 FLET or MACROLET forms
266 ** declarations in a DOLIST body are also in scope for the
268 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
270 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
273 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
274 * fixed cross-compiler leakages that prevented building a 32-bit
275 target with a 64-bit host compiler.
276 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
277 opened with :IF-EXISTS :OVERWRITE.
278 * fixed bug 281: error for an invalid qualifier in a short-form method
279 combination method is not signalled until the faulty method is called.
280 * bug fix: iteration variable type inferrer failed to deal with open
281 intervals. (reported by Alan Shields)
282 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
283 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
284 or more runtime options were provided to the sbcl binary.
285 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
286 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
287 * optimization: there's now a fast-path for fixnum arguments in the
288 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
289 * optimization: the code generated on x86-64 is more compact thanks
290 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
291 * bug fix: floating-point exception handling now partly works on
292 x86-64. (Thanks to James Knight)
293 * improvement to the MIPS backend from Thiemo Seufer: C-style
294 64-bit long long arguments and return values to alien functions
296 * fixed some bugs revealed by Paul Dietz' test suite:
297 ** the type-error signalled from WARN has a filled-in DATUM slot.
298 ** the type-error required when a stream is not associated with
299 a file has the stream as its datum.
300 ** type-errors on single-floats on x86-64 no longer have
301 :INVALID-OBJECT as the datum
302 ** the type-errors signalled for invalid function names now have
303 a correct expected type
304 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
305 for structures defined with a :TYPE. Documentation strings for
306 typed structures are no longer immediately discarded
307 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
308 broadcast streams and synonym streams. FILE-LENGTH now also works
309 on broadcast streams.
311 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
312 * incompatible change: the --noprogrammer option, deprecated since
313 version 0.7.5, has been removed. Please use the equivalent
314 --disable-debugger option instead.
315 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
317 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
318 inoperational for a while now) has been completely removed.
319 * null lexical environments are now printed as #<NULL-LEXENV>,
320 significantly reducing the amount of clutter in typical backtraces.
321 * documentation on weak pointers, finalization, and after GC hooks
322 has been added to the manual.
323 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
324 as other specialized array types, is much faster. SUBSEQ and
325 COPY-SEQ on such arrays have also been sped up.
326 * optimization: EQL is now more efficient when at least other argument
327 is known to be of type (OR FIXNUM (NOT NUMBER)).
328 * fixed bug: the runtime is now less vulnerable to changes in the
329 size of the SBCL object on OS X, and virtual memory is reserved for
330 all spaces that need to be at a fixed address.
331 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
332 * fixed bug: finalizers and after GC hooks that cause consing are now
334 * fixed bug: compiler error messages and summaries are now printed to
335 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
336 * fixed inference of the upper bound of an iteration variable.
337 (reported by Rajat Datta).
338 * fixed bug 211e: calling local functions with duplicated constant
339 keyword argument no longer causes a bogus style warning about an
341 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
342 ftype declaration to be disregarded. (reported by Dave Roberts)
343 * fixed bug 373: caused by erronous compilation of references to alien
344 variables in the runtime on ppc/darwin.
345 * fixed bug 376: CONJUGATE type deriver.
346 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
347 (reported by Baughn on #lisp)
348 * a cross-compiler bug on non-x86 platforms has been identified and
349 fixed. (thanks to Bruno Haible)
350 * improvements to the MIPS runtime code for increased stability.
351 (thanks to Thiemo Seufer)
352 * increased the maximimum compact environment size to allow
353 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
354 * partial workaround for bug 135: don't name the function we're
355 calling for hairy cases of EVAL, so as not to accumulate one environment
356 entry per call to EVAL. (reported by Kevin Reid)
357 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
358 * optimization: type testing for non-vector arrays should be faster.
359 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
361 * bug fix: setting 31st element of a bit vector to zero did not work
363 * bug fix: redefining a class definition which failed due to a
364 previous accessor / function clash now works (but see BUGS entry
365 #380 for more problems in this area). (thanks to Zach Beane)
366 * the long form of DEFINE-METHOD-COMBINATION disables method group
367 checking when given a single method group with pattern *. (thanks
369 * on x86 compiler supports stack allocation of results of simple
370 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
371 * support for building and running on Mac OS X 10.4 "Tiger" has been added
372 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
373 * fixed some bugs related to Unicode integration:
374 ** the restarts for recovering from input and output encoding
375 errors only appear when there is in fact such an error to
377 * fixed some bugs revealed by Paul Dietz' test suite:
378 ** MISC.549 and similar: late transformation of unsafe type
379 assertions into derived types caused unexpected code
381 ** SCALE-FLOAT type deriver is less wrong.
382 ** type derivers for EXP, LOG and similar functions compute result
383 types for complex arguments better.
384 ** (MISC.563) CONJUGATE type deriver works for very restricted
386 ** out-of-line type testers for character strings are available.
387 ** EQUAL compiler transform understands specialness of objects
389 ** accessing double-floats stored on the stack now works on x86-64.
390 ** debugger internals could sometimes create invalid lispobjs,
391 resulting in GC crashes.
392 ** MISC.548: type check weakening can convert required type into
394 ** initialization forms for bindings are not in scope of free special
397 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
398 * incompatible change: thread support for non-NPTL systems has
399 been removed - locking is buggy and unreliable. A threaded
400 SBCL build will now warn at startup and refuse to create
401 new threads, unless futex support is detected in the kernel
402 * incompatible change: the top level REPL now has only an ABORT
403 restart associated with it, not TOPLEVEL and ABORT as it used to.
404 TOP and TOPLEVEL are now available as debugger commands for
405 returning to the top level.
406 * incompatible change: forms evaluated in the REPL now use the
407 global optimization policy.
408 * incompatible change: user- and system-initialization files are
409 no longer processed with LOAD, but by READ and EVAL; hence the
410 global optimization policy, startup package, readtable, etc,
412 * The .fasl file format number has been incremented because of
413 various incompatible changes.
414 * internal entry point details and argument counts no longer appear
415 in backtraces unless explicitly requested by setting
416 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
417 * built-in and standard functions no longer have names like "top
418 level local call to FOO".
419 * fixed bug 32: functions defined in non-null lexical environments
420 now have more legible printed representation
421 * fixed bug 33: functions defined in non-null lexical environemnts
422 are now more amenable to inspection by INSPECT.
423 * workaround for bug 354: XEPs no longer appear in backtraces unless
424 explicitly requested.
425 * fixed bug: receiving the signal which results from attempting to
426 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
427 SIGBUS on other *BSDs) on architectures where the C stack is also
428 the Lisp stack (x86 and x86-64) from foreign code no longer leads
429 to debugger confusion or wild instability. (reported by Cheuksan
431 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
432 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
433 (reported by Lutz Euler)
434 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
435 compiled code. (reported by Paul Dietz)
436 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
437 specialized parameter to a method to something that is not TYPEP
438 the specializer is now possible.
439 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
440 face of package deletion.
441 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
442 pathname. (reported by David Tolpin on c.l.l)
443 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
444 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
445 STRING-STREAM and FILE-STREAM, respectively.
446 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
447 than previously; in addition, its test suite is now run on build.
448 (thanks to Luke Gorrie)
449 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
450 * optimization: added a immediate representation for single-floats
452 * fixed some bugs revealed by Paul Dietz' test suite:
453 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
454 ** attempting to create a package with a colliding nickname causes
455 correctable errors to be signalled.
456 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
457 ** several x86-64 backend bugs related to sign-extension of immediate
460 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
461 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
462 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
464 * fixed loading of multiply forward-referenced layouts.
465 (thanks to Cheuksan Wang)
466 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
467 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
468 (see COMPILE-FILE documentation for details of :PRINT option)
469 * fixed bugs 19 and 317: fixed-format floating point printing is
470 more accurate. This also fixes a bug reported by Adam Warner
471 related to the ~@F format directive.
472 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
474 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
475 dispatch table if the from-char is a dispatch macro character.
476 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
477 by the wordsize no longer ignore the last word. (reported by Lutz
479 * fixed bug in type checking of dynamic-extent variables. (reported
481 * optimization: sequence traversal functions use their freedom to
482 coerce function designators to functions.
483 * optimization: code with many calls to CLOS methods specialized on
484 CLOS classes has had redundant type checks removed.
485 * optimization: added declarations to speed up operations that access
486 the internal character database (for example STRING-UPCASE)
487 * optimization: comparison operations between floats and sufficiently small
488 fixnums no longer create extra rationals
489 * fixed some bugs related to Unicode integration:
490 ** portions of multibyte characters at the end of buffers for
491 character-based file input are correctly transferred to the
492 start of the buffer at the next read.
493 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
494 passing it through to OPEN.
495 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
496 argument given to internal calls to OPEN.
497 ** the built-in comment readers (introduced by character sequences
498 ";" and "#|") are more forgiving to encoding errors; they will
499 STYLE-WARN and then attempt to resync the stream at a character
500 boundary. (thanks to Teemu Kalvas)
501 * fixed some bugs in the x86-64 port:
502 ** Negative short int return values from c-calls are sign-extended
504 ** The stack is aligned to 16-bytes for c-calls, as required by
505 the ABI. (thanks to Cheuksan Wang)
506 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
507 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
508 for enums. (thanks to Vincent Arkesteijn)
509 ** Multiple small optimizations and bugfixes for floating point
511 * fixed some bugs revealed by Paul Dietz' test suite:
512 ** Space, Tab, Linefeed, Return and Page have the invalid
513 secondary constituent character trait.
514 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
516 ** WITH-INPUT-FROM-STRING should only update the index place on
518 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
520 ** Bit-array operations (BIT-AND and similar) worked incorrectly
521 with one-dimensional arrays with fill pointers.
522 ** TYPE-OF failed on a complex with an integer realpart and a
524 ** compiler failure during type inference for the code of form
525 (IF (EQL X (THE ...)) ...) (MISC.535).
527 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
528 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
529 platform. The port supports 61-bit fixnums, large memory spaces
530 and reloading shared object files.
531 * enhancement: saving cores with foreign code loaded is now
532 supported on ppc/Darwin in addition to the previously supported
534 * enhancement: the statistical profiler now walks deeper into the
535 call stack for more meaningful call-graphs and accrued time
536 reports (x86/x86-64 only). It also now reports time spent in
538 * enhancement: it is now possible to trace most individual methods
539 of a generic function in addition to tracing the generic function
541 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
543 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
544 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
545 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
546 produces an error. (thanks to Vincent Arkesteijn)
547 * bug fix: NAMESTRING on pathnames with :WILD components in their
548 directories works correctly. (thanks to Artem V. Andreev)
549 * fixed bug 125: compiler preserves identity of closures. (reported
551 * bug fixed: functions with &REST arguments sometimes failed with
552 "Undefined function" when compiled with (DEBUG 3). (reported by
554 * bug fix: overflow during compiling of setting element of a bit
555 vector with constant index and value. (reported by Timmy Douglas)
556 * build fix: fixed the dependence on *LOAD-PATHNAME* and
557 *COMPILE-FILE-PATHNAME* being absolute pathnames.
558 * on x86 compiler partially supports stack allocation of dynamic-extent
560 * GO and RETURN-FROM do not check the extent of their exit points
561 when compiled with SAFETY 0.
562 * fixed some bugs related to Unicode integration:
563 ** encoding and decoding errors are now much more robustly
564 handled; it should now be possible to recover even from invalid
565 input or output to the terminal. (thanks to Teemu Kalvas)
566 ** provided a first cut at implementing STRING-TO-OCTETS and
567 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
568 ** altered the SB-MD5 contributed module slightly, changing the
569 interface just enough for it to be supportable for builds where
570 lisp characters are not eight bits.
571 * fixed some bugs revealed by Paul Dietz' test suite:
572 ** the FORMATTER-generated functions for ~V[ conditionals require
573 the correct number of arguments.
574 ** READ-FROM-STRING returns the mandated second value when applied
575 to displaced strings.
576 ** the #\Rubout and #\Backspace characters are treated as invalid
577 constituent characters by the tokenizer.
579 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
580 * new feature: reloading changed shared object files with
581 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
582 * new feature: references to foreign variables and functions
583 can now be compiled and loaded before the corresponding shared
584 object file is loaded, as long as the foreign definitions are
585 available at runtime.
586 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
587 supported. (thanks to Dan Debertin)
588 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
589 just keywords. (thanks to Vincent Arkesteijn)
590 * fixed bug #331: structure-class instances corresponding to
591 DEFSTRUCT forms are now created eagerly.
592 * fixed bug #345: backtraces from calls to undefined functions work
593 on x86 as well. Related bug #61 is now also partially fixed on x86
594 (backtraces from throws to unknown catch tags.)
595 * bug fix: lambda-list parsing is now stricter vrt. order and number
596 of lambda-list keywords.
597 * bug fix: as specified by AMOP, an error is signalled if a
598 class-option appears multiple times in a DEFCLASS form. (reported
600 * bug fix: RANDOM can be compiled when the compiler derives the type
601 of its numeric argument as a disjoint set of small integers.
602 (reported by Paul Dietz)
603 * bug fix: starting a core saved with shared objects loaded when
604 those objects are not available no longer causes threaded SBCL to
605 hang. (reported by Sean Ross)
606 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
608 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
609 argument, which must always be NIL. (reported by Kalle Niemitalo)
610 * bug fix: printing 1.0d+23 no longer results in an error.
611 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
612 * bug fix: structure slot setters preserve evaluation order. (thanks
614 * fixed some bugs related to Unicode integration:
615 ** RUN-PROGRAM can allow its child to take input from a Lisp
616 stream. (reported by Stefan Scholl)
617 ** ASDF-INSTALL successfully downloads .tar.gz files in all
618 locales. (reported by Ken Causey)
619 * fixed some bugs revealed by Paul Dietz' test suite:
620 ** INCF, DECF and REMF evaluate their place form as specified in
622 ** FORMATTER expands ~{ ~} iteration directives with V or #
623 parameters correctly.
624 ** FORMATTER deals with the ~@[ ~] conditional directive where the
625 consequent uses no arguments correctly.
626 ** the system has a partial understanding of the (COMPLEX RATIO)
628 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
631 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
632 * new feature: a build-time option (controlled by the :SB-UNICODE
633 keyword feature, enabled by default) for building the system with
634 support for the entire 21-bit character space defined by the
636 * new feature: the system now has rudimentary external-format
637 support; the primary user-visible change at this time is that
638 characters with the high bit set (such as lower-case-e-acute) will
639 print correctly to a terminal in a UTF-8 environment.
640 * minor incompatible change: BASE-CHAR no longer names a class;
641 however, CHARACTER continues to do so, as required by ANSI.
642 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
643 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
644 has been moved to the SB-EXT package (temporarily re-exported from
646 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
647 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
649 * fixed bug #351: better error-handlind and reporting for malformed
651 * fixed bug #350: bignum-printing is now more memory-efficient,
652 allowing printing of very large bignums, eg. (expt 2 10000000).
653 (reported by Bruno Haible)
654 * fixed bug #302: better primitive-type selection for intersection
656 * fixed bug #308: non-graphic characters now all have names, as
657 required. (reported by Bruno Haible)
658 * bug fix: redefining a class with different superclasses now correctly
659 removes it from the direct-subclasses of its previous superclasses.
660 (reported by David Morse)
661 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
662 new value now works. (reported by Bruno Haible)
663 * bug fix: correct canonicalization of multiple non-standard slot
664 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
665 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
666 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
667 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
668 now exists, an signals an error.
669 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
670 during macroexpansion for non-iteration variables as well. (reported
671 by Bruno Haible for CMUCL)
672 * bug fix: Cyclic structures and unprintable objects in compiler
673 messages no longer cause errors. (reported by Bruno Haible)
674 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
675 and READ-FROM-STRING all now return a primary value of NIL if
676 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
677 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
678 (reported by Bruno Haible for CMUCL)
679 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
680 do not designate a condition. (reported by Bruno Haible for
682 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
683 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
684 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
685 permitting resolution in favour of any of the conflicting symbols.
686 (reported by Bruno Haible for CMUCL)
687 * bug fix: EQUAL compiler optimizations is less aggressive on
688 strings which can potentially compare true despite having distinct
689 specialized array element types.
690 * bug fix: unit enumerations can be defined without dividing by
691 zero. (thanks to Vincent Arkesteijn)
692 * FORMAT compile-time argument count checking has been enhanced.
693 (report from Bruno Haible for CMUCL)
694 * a partial workaround for the bug 262: the compiler does not try to
695 inline-expand a local function doing RETURN-FROM from a deleted
696 BLOCK. (thanks to Peter Denno for the bug report and to David
697 Wragg for the simple test case)
698 * fixed some bugs revealed by Paul Dietz' test suite:
699 ** RENAME-PACKAGE allows all package designators as new package
701 ** constraint propagation and lambda variable substitution are
702 more cautious in dealing with partially deleted code.
703 ** compiler sometimes lost reoptimization passes.
704 ** CERROR, given a condition as condition designator, passes its
705 remaining arguments to the continue format control without
707 ** Case-altering FORMAT directives work correctly on non-ASCII
709 ** The REAL type specifier handles bounds outside the floating
710 point ranges without signalling FLOATING-POINT-OVERFLOW.
711 ** Functions with IR1-transformations can create intercomponent
712 references to global functions.
713 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
715 ** FORMAT ~R treats a nil value for its first parameter correctly.
717 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
718 * enhancement: saving cores with foreign code loaded is now
719 supported on x86/NetBSD and sparc/Linux in addition to the previously
721 * bug fix: on some platforms repeated installations caused multiple
722 copies of HTML documentation to be installed -- should not happen
723 any more. (reported by Stefan Scholl)
724 * bug fix: parsing self-recursive alien record types multiple times
725 no longer causes infinite recursion. (reported by Thomas F. Burdick,
726 original patch by Helmut Eller for CMUCL)
727 * bug fix: stack-exhaustion detection works now on NetBSD as well.
728 (thanks to Richard Kreuter)
729 * bug fix: defining classes whose accessors are methods on existing
730 generic functions in other (locked) packages no longer signals
731 bogus package lock violations. (reported by François-René Rideau)
732 * bug fix: special variables as DEFMETHOD parameters no longer have
733 associated bogus type declarations. (reported by David Wragg and
735 * bug fix: read-write consistency on streams of element-type
736 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
737 * bug fix: redefiniton of the only method of a generic function with
738 no DEFGENERIC no longer emits a full WARNING. In addition,
739 redefinition of generic functions with no DEFGENERIC to an
740 incompatible lambda list now signals an error. (thanks to Zach
742 * bug fix: DEFGENERIC now works even when there's a function of the
743 same name in an enclosing lexical environment. (thanks to Zach
745 * fixed compiler failure, caused by instrumenting code during
746 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
747 * optimization: added loop analysis and improved register allocation
748 * fixed some bugs revealed by Paul Dietz' test suite:
749 ** POSITION on displaced vectors with non-zero displacement
750 returns the right answer.
751 ** (SIMPLE-STRING) is a valid type specifier for sequence
753 ** *PRINT-LEVEL* handling for slotless structures is pedantically
755 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
756 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
758 ** The justification version of the FORMAT ~< directive treats
759 non-zero minpad parameter correctly.
761 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
762 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
763 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
764 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
765 the supported interface.
766 * new feature: Single-stepping of code compiled with DEBUG 2 or
767 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
768 * new feature: saving cores with foreign code loaded is now
769 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
770 Timothy Moore's work for CMUCL)
771 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
772 parameters to * instead of NIL if no initform is supplied.
773 (reported by Johan Bockgaard)
774 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
775 correctly when FUNCALL appears as the car of the form. Note:
776 despite this FUNCALL forms are not currently subject to
777 compiler-macro expansion. (port of Raymond Toy's fix for the
778 same from CMUCL, reported by Johan Bockgaard)
779 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
780 (thanks for Teemu Kalvas)
781 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
782 case instead of terminating the iteration (thanks for Julian
783 Squires, Sean Champ and Raymond Toy)
784 * bug fix: incorrect expansion of defgeneric that caused a style
785 warning. (thanks for Zach Beane)
786 * bug fix: slot accessor effective method computation works properly
787 for classes with multiple non-standard applicable methods on
788 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
789 * on x86 compiler supports stack allocation of results of LIST and
790 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
791 CMUCL implementation by Gerd Moellmann)
792 * fixed some bugs revealed by Paul Dietz' test suite:
793 ** FORMAT strings with both the ~<~:;~> form of the justification
794 directive and pretty-printing directives cause an error.
796 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
797 * incompatible change: the internal functions
798 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
799 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
800 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
801 instead of the old functions.
802 * new feature: on platforms where "dladdr" is available foreign
803 function names now appear in backtraces. (based on Helmut Eller's
805 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
806 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
808 * documentation: ASDF manual is now distributed with SBCL.
809 * bug fix: SBCL can now load its contributed modules with REQUIRE
810 even if the system-provided entries have been removed from
811 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
813 * bug fix: backtraces involving undefined functions or assembly
814 routines are more informative. (thanks to Brian Downing)
815 * bug fix: mutually referent alien structures now work correctly.
816 (reported by Rick Taube)
817 * bug fix: structures defined by WITH-ALIEN can be referred to
818 within other definitions in the same WITH-ALIEN.
819 * bug fix: division operators (MOD, TRUNCATE and the like) with
820 constant zero divisors and integer dividends no longer generate
822 * bug fix: provide default methods for INPUT-STREAM-P and
823 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
824 * bug fix: improve the detection and resolution of MOP metacycles
825 (where effective method computation depends on the generic function
826 having its effective method computed). (reported by Bruno Haible)
827 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
828 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
829 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
830 * bug fix: pathnames with relative directory components can be
831 represented relative to default pathnames.
832 * optimization: in taking the GCD of bignums, reduce the two bignums
833 to approximately the same size (using Euclid's algorithm) before
834 applying the more sophisticated binary GCD. (thanks to Juho
836 * optimization: COUNT on bitvectors now operates word-at-a-time.
837 * optimization: ASH with a positive, but not necessarily constant,
838 (leftwards) shift, when in a modular context, is compiled to a
840 * fixed some bugs revealed by Paul Dietz' test suite:
841 ** FORMAT variable parameters ("~V<char>") are defaulted properly
842 if the corresponding argument is NIL.
843 ** FORMAT directives accepting a minpad parameter treat negative
845 ** literal commas as character parameters to FORMAT directives are
847 ** literal spaces directly after ~<Newline> directives within a
848 format-logical-block (~:< ~@:>) do not induce :FILL-style
849 conditional newlines.
850 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
851 as specified: it no longer includes conditional newlines).
852 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
854 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
855 :PER-LINE-PREFIX argument does not evaluate to a string.
857 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
858 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
859 the manual for details; for now, package locks can be disabled by
860 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
861 if no major problems are found then it is likely that they will be
862 compiled in unconditionally.
863 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
864 unsupported operators on all platforms. To load a shared library
865 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
866 object file, link it into a shared library outside of SBCL and
867 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
869 * bug workaround: ROOM T and the GC were not working together
870 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
871 reporting facility aren't preserved. That reporting has been
872 disabled, so now until and unless someone figures out how to make it
873 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
874 * minor incompatible change: as threatened around sbcl-0.8.0, the
875 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
876 The symbols are also exported from SB-PCL for backwards
877 compatibility, but more so than before SB-PCL should be treated as
878 an implementation-internal package.
879 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
881 * the SB-SPROF contrib now works on (most) non-x86 architectures.
882 It is known as of this release not to work on the Alpha, however.
883 * fixed bug #167: errors signalled due to illegal syntax in method
884 bodies are now more legible.
885 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
886 designators and can hence be used with TYPEP.
887 * fixed bug #333: CHECK-TYPE now ensures that the type error
888 signalled, if any, has the right object to be accessed by
889 TYPE-ERROR-DATUM. (reported by Tony Martinez)
890 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
891 5.1.2.3 for multiple-value place subforms. (reported by Kalle
893 * fixed bug #334: programmatic addition of slots using specialized
894 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
895 and :ALLOCATION :CLASS slots.
896 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
897 float exponents. (rereported by Peter Seibel)
898 * fixed a bug: #\Space (and other whitespace characters) are no
899 longer considered to be macro characters in standard syntax by
901 * fixed bug: initialization of condition class metaobjects no longer
902 causes an instance of the condition to be created. (reported by
904 * fixed bug: it is now possible to have slots such that
905 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
906 description is neither :INSTANCE nor :CLASS.
907 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
908 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
909 SLOT-VALUE-USING-CLASS), and no longer causes errors with
910 non-standard SLOT-DEFINITION-ALLOCATION values.
911 * fixed bugs: various race conditions handling exiting threads.
912 CL-PPCRE's thread torture test now passes.
913 * fixed arguably-a-bug: GC time proportional to number of threads in
914 system even when most of them are idle
915 * optimization: improved performance of BIT and SBIT on bit-vectors.
916 * .fasl file incompatibility: The fasl file version number has
917 been incremented because of changes associated with package locks.
919 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
920 * minor incompatible change: the system no longer provides
921 optimization hints (conditions which are TYPEP
922 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
923 mode; these hints are still emitted when higher SPEED optimization
925 * new contrib module: a sampling profiler (profiling by statistical
926 sampling, rather than by instrumenting functions) is available as
927 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
928 * the behaviour of the standard function ED is now customizeable by
929 third parties through a hook variable: see ED's documentation
930 string for information on the protocol.
931 * the compiler no longer emits efficiency notes for (FUNCALL X)
932 when the type of X is uncertain under default optimization
934 * fixed bug 276: mutating a binding of a specialized parameter to a
935 method to something that is not TYPEP the specializer is now
937 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
938 exhibit the required behaviour on the x86 platform. (thanks to
939 Peter van Eynde, Eric Marsden and Bruno Haible)
940 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
941 even for difficult arguments. (reported by Peter Graves)
942 * fixed bug 141a: the backquote printer now descends quoted
944 * fixed another bug in backquote printing: no more destructive
945 modification of the form's list structure. (reported by Brian
947 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
948 move between its address being taken and the call to
949 interrupt_thread, fixing a crashing race condition.
950 * the SB-POSIX contrib implementation has been adjusted so that it
951 no longer exhibits ridiculously poor performance when constructing
952 instances corresponding to C structs.
954 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
955 * minor incompatible change: the sb-grovel contrib now treats C
956 structures as alien (in the sense of SB-ALIEN) objects rather than
957 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
958 has implications for memory management of client code
959 (sb-grovel-returned objects must now be manually managed) and for
960 type safety (alien objects now have full types).
961 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
962 used to control emission of compiler diagnostics, rather than the
963 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
964 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
965 quality should be considered deprecated.
966 * install.sh now installs the user manual as well
967 * (not quite a new documentable feature, but worth considering in
968 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
969 beginnings of a semantically meaningful condition hierarchy is
970 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
972 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
973 designator as the defaults argument.
974 * fixed bug: Displaced arrays whose displaced-to array has become
975 too small now cause ARRAY-DIMENSION to signal an error, providing
976 for safer bounds-checking. (reported by Bruno Haible)
977 * fixed bug: DEFCLASS slot definitions with identical :READER and
978 :WRITER names now signal a reasonable error. (reported by Thomas
980 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
982 * fixed bug: Invalid slot specification errors now print correctly.
983 (thanks to Zach Beane)
984 * fixed bug 320: Shared to local slot value transfers in class
985 redefinitions now happen corrently from superclasses as
986 well. (reported by Bruno Haible)
987 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
989 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
990 improper lists correctly. (reported by Bruno Haible)
991 * fixed bug 313: source-transform for <fun-name> was erroneously
992 applied to a call of a value of a variable with name <fun-name>.
993 (reported by Antonio Menezes Leitao)
994 * fixed bug 307: The obsolete instance protocol ensures that
995 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
996 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
997 remove dead unknown-values globs from the middle of the stack.
998 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
999 easier to resume long computations after using *BREAK-ON-SIGNALS*
1000 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1001 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1002 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1003 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1004 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1005 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1007 * fixed a (fairly theoretical) bug in string printing: if
1008 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1009 does not have array-element-type equal to the most general string
1011 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1012 function name. (thanks to Nikodemus Siivola)
1013 * fixed bug: compile-time format string checker failed on
1014 non-closed ~{. (reported by Thomas F Burdick)
1015 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1016 don't warn on legal loop code involving more than one aggregate
1017 boolean. (thanks to Nikodemus Siivola)
1018 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1019 in keyword position to MAKE-PACKAGE was not being reported as
1021 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1022 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1024 * fixed bugs 280 and 312: the checking for multiple definitions in a
1025 file is less likely to become confused by uses of inline
1027 * fixed bug: the #S reader macro performs the keyword coercion
1028 specified for slot names. (reported by Kalle Niemitalo)
1029 * fixed bug: lambda lists may contain symbols whose names start with
1030 & but are not lambda-list-keywords; their occurrence triggers a
1032 * fixed bug 321: define-method-combination argument lambda lists do
1033 not cause invalid code to be generated when &optional or &aux
1034 variables have default values. (reported by Bruno Haible)
1035 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1036 have CLOS classes; this appears also to have fixed reported
1037 instability in stack exhaustion detection.
1038 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1039 MUFFLE-WARNING when no associated restart is present is now
1041 * optimization: rearranged the expansion of various defining macros
1042 so that each expands into only one top-level form in a
1043 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1045 * optimization: used a previously little-used slot in symbols to
1046 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1048 * fixed some bugs revealed by Paul Dietz' test suite:
1049 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1051 ** Fixed an optimization bug related to inheritance of initforms
1052 from local to shared slots.
1053 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1054 ** CAST optimizer forgot to flush argument derived type.
1055 ** print/read consistency on floats is now orders of magnitude
1056 more likely. (thanks also to Bruno Haible for a similar report
1058 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1059 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1060 for objects of type REAL. Make it so.
1061 ** SXHASH is specified (infelicitously) to respect similarity,
1062 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1063 it so. (thanks to Markus Ziegler)
1064 ** on the Alpha, the compiler succeeds in compiling functions
1065 returning a known number of arguments greater than 63.
1066 ** fixed handling of invalid NIL arguments in keyword position
1068 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1070 ** provided a readably-printable representation for RANDOM-STATE
1072 ** ensured that pathnames, where they have a namestring, always
1073 print using #P"..." syntax.
1075 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1076 * Support for the forthcoming 2.0 version of the NetBSD kernel
1077 running on x86 hardware has been added. (thanks to Perry
1078 E. Metzger most immediately, and others for their past work)
1079 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1080 current release version; SBCL's OpenBSD support had been broken
1081 since about the time of OpenBSD's switch to ELF binary format.)
1082 * [placeholder for DX summary]
1083 ** user code with &REST lists declared dynamic-extent, under high
1084 speed or space and low safety and debug optimization policy.
1085 * The manual has been converted to Texinfo format and the debugger
1086 chapter from the cmucl manual has been added.
1087 * A facility has been added to extract documentation strings from
1088 sbcl and store them as Texinfo-formatted snippets for inclusion in
1089 the manual (via Texinfo's @include directive)
1090 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1091 the test case to Dave Roberts)
1092 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1093 pointers. (reported by Sean Ross)
1094 * bug fix: PROFILE output is printed nicely even for large numerical
1095 values. (thanks to Zach Beane)
1096 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1097 greater than 32 handle EOF correctly.
1098 * bug fix: on X86 an immediate argument of the IMUL instruction is
1099 correctly printed in disassembly. (thanks to Lutz Euler)
1100 * bug fix: class slots in redefined classes preserve their old
1101 values. (thanks to Bruno Haible and Nikodemus Siivola)
1102 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1103 succeeds. (reported by Marco Baringer)
1104 * fixed some bugs revealed by Paul Dietz' test suite:
1105 ** READ-SEQUENCE now works on ECHO-STREAMs.
1106 ** RATIONALIZE works more according to its specification. (thanks
1109 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1110 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1111 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1112 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1113 should matter to you only if you rebind the printer control
1114 variables and then find you want different bindings in the
1115 debugger than in the ordinary execution of your program.)
1116 * The runtime build system has been tweaked to support building
1117 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1118 assembler and linker. (thanks to Nikodemus Siivola)
1119 * Unbound, undefined, undeclared variables now trigger full
1120 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1121 more often programmer error than deliberate exploitation of undefined
1123 * optimization: the hash algorithm for strings has changed to one
1124 that is less vulnerable to spurious collisions. (thanks to Juho
1126 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1127 needless bounds checking. (thanks to Juho Snellman)
1128 * optimization: implemented multiplication as a modular
1129 (UNSIGNED-BYTE 32) operation on the PPC backend.
1130 * fixed some bugs revealed by Paul Dietz' test suite:
1131 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1133 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1134 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1135 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1137 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1139 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1141 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1142 their output stream on EOF from read.
1143 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1144 have been read to end-of-file.
1145 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1147 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1148 description of determination of which consecutive characters
1150 ** Printing the "Space" character with escaping on now yields "#\\ ",
1151 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1152 ** Reading floating-point numbers with *READ-BASE* set to a number
1153 less than 10 works correctly.
1154 ** Reading floating-point numbers with *READ-BASE* set to a number
1155 more than 10 works correctly.
1156 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1157 the readtable currently in effect.
1159 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1160 * minor incompatible change: parsing of namestrings on a physical
1161 (Unix) host has changed; numbers after the final #\. in a
1162 namestring are no longer interpreted as a version field. This is
1163 intented to be largely invisible to the user, except that the
1164 meaning of the namestring "*.*.*" has changed: it now refers to a
1165 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1166 should usually be replaced by
1167 (make-pathname :name :wild :type :wild :version :wild)
1168 with the added benefit that this is more likely to be portable.
1169 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1170 option to OPEN now signals an error if the file being opened
1171 exists; this may have an impact on existing code.
1172 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1173 signals received, on the PowerPC platforms (both Linux and
1174 Darwin). (thanks to Pierre Mai for pointing out the location of
1176 * several fixes on OS X: The system now builds and runs cleanly on
1177 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1178 (thanks to Brian Mastenbrook)
1179 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1180 second-resolution: integer multiples of 1/3600 between -24 and 24.
1181 (thanks to Vincent Arkesteijn)
1182 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1183 type when called with 1 argument; PEEK-CHAR checked type of
1184 PEEK-TYPE only after having read first character from a
1185 stream. (reported by Peter Graves)
1186 * bug fix: the garbage collector now has much better locality
1187 behaviour, and in particular no longer treats all memory as being
1188 exhausted when many small objects point to each other in a deeply
1190 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1191 recognized as being TYPEP their class.
1192 * bug fix: the PUSHNEW documentation string has been corrected.
1193 (thanks to Vincent Arkesteijn)
1194 * bug fix: defaulting of the value for the last of an atypically
1195 large number of multiple values being bound was not being
1196 performed correctly on the Alpha or PPC platforms
1197 * optimization: implemented multiplication as a modular
1198 (UNSIGNED-BYTE 32) operation on the x86 backend.
1199 * optimization: SEARCH on simple-base-strings can now be open-coded.
1200 (see also contrib/compiler-extras.lisp for inspiration for
1201 teaching the compiler about the Boyer-Moore algorithm).
1202 * value, returned by MAX (and MIN) called with several EQUALP, but
1203 not EQL, arguments now does not depend on compiler settings.
1204 * fixed some bugs revealed by Paul Dietz' test suite:
1205 ** in stack analysis liveness information is propagated from
1206 non-local entry points.
1207 ** pathwise CAST removing failed when the CAST node did not start
1209 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1210 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1212 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1213 host is already defined.
1214 ** RENAME-FILE works on streams instead of signalling an internal
1216 ** PEEK-CHAR uses the current readtable when determining whether
1217 or not a character is whitespace.
1218 ** MERGE-PATHNAMES handles the case when the pathname does not
1219 specify a name while the default-pathname specifies a version.
1220 ** Pathnames now stand a better chance of respecting print/read
1222 ** Attempting to use standardized file system operators with a
1223 pathname with invalid :DIRECTORY components signals a
1225 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1227 ** DIRECTORY on logical pathnames is more correct.
1228 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1229 signal a TYPE-ERROR if their argument is not a stream.
1230 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1231 designator argument does not designate a stream.
1232 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1233 examining the synonym.
1234 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1236 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1237 element-type larger than ([UN]SIGNED-BYTE 32).
1239 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1240 * When built with the :SB-FUTEX feature, threaded builds now take
1241 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1242 for faster/more reliable mutex and condition variable support.
1243 * Incompatible change (but one you probably shouldn't have been using
1244 anyway): the interface and code for arbitrating between multiple
1245 threads in the same user session has been redesigned.
1246 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1247 error when called without an explicit environment argument.
1248 (thanks to Nikodemus Siivola)
1249 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1250 confused about their position. (thanks to Adam Warner and Gerd
1252 * bug fix: Pretty printing backquoted forms with unquotations in the
1253 argument list position of various code constructs such as LAMBDA
1254 now works correctly. (reported by Paul Dietz)
1255 * bug fix: Pretty printing unquotations no longer loses all
1256 stream position information.
1257 * optimization: performance of string output streams is now less
1258 poor for multiple small sequence writes.
1259 * optimization: performance of CSUBTYPEP in the presence of complex
1260 expressions involving CONS and NOT many times has been improved.
1261 (reported by Paul Dietz)
1262 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1264 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1265 streams for the REPL, for the debugger, and so on. (thanks to
1267 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1268 * fixed some bugs revealed by Paul Dietz' test suite:
1269 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1270 is ignored if it is too small, rather than propagating through
1271 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1272 ** extremely complex negations of CONS types were not being
1273 sufficiently canonicalized, leading to inconsistencies in
1275 ** VALUES tranformer lost derived type.
1277 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1278 * fixed a bootstrapping bug: the build process no longer assumes
1279 that the various BOOLE-related constants have the same value in
1280 host and target lisps. (noted by Paul Dietz' test suite on an
1281 SBCL binary built from CLISP)
1282 * The system can now be dynamically linked on the MIPS platform,
1283 which enables dynamic loading of foreign code from Lisp. (thanks
1284 to Ralf Baechle for discussions on the MIPS ABI)
1285 * The system now records debugging information for its own source
1286 files in a filesystem-position-independent manner, relative to
1287 the "SYS" logical host.
1288 * fixed a compiler bug: MV-LET convertion did not check references
1289 to the "max args" entry point. (reported by Brian Downing)
1290 * tweaked disassembly notes to be less confident about proclaiming
1291 some instruction as an LRA. (thanks to Brian Downing)
1292 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1293 now each have their own history, command character, and other
1294 characteristics. (thanks to David Lichteblau)
1295 * fixed some bugs revealed by Paul Dietz' test suite:
1296 ** compiler failure in compiling LOGAND expressions including a
1298 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1299 shift greater than 32.
1300 ** FLUSH-DEST did not mark blocks for type check regeneration.
1301 ** HANDLER-CASE failed to accept declarations in handler clauses
1302 in some circumstances.
1304 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1305 * New code in contrib/sb-introspect (still probably not entirely
1306 stable yet) provides some support for smart Lisp development
1307 environments like SLIME.
1308 * The conditions signalled for errors occurring when loading .fasl
1309 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1310 in a way which should help ASDF recover gracefully.
1311 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1312 stuff has been cleaned up. If you code contrib/ stuff, this might
1313 affect you, and you can look at contrib/README, contrib/STANDARDS,
1314 and/or the 0.8.4.27 diff to check.
1315 * In full calls the compiler now does not generate checks for declared
1316 argument types for all arguments.
1317 * various threading fixes
1318 ** and some experimental patches which didn't make it into
1319 the main tree for this release, but which are shipped in
1320 contrib/experimental-thread.patch as a possible fix for some
1321 failures (deadlock, spinning...) in GC-intensive multithreaded
1323 * fixed PPC build problem (source code incompatibility of different
1324 library versions): added offsetof() hackery which attempts to divine
1325 where glibc maintainers put uc_mcontext today
1326 * fixed bug 282: compiler does not trust type assertions while passing
1327 arguments to a full call.
1328 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1329 &OPTIONAL VALUES type parameter.
1330 * fix bug 214: algorithm for noting rejected templates is now more
1331 similar to that of template seletion. (also reported by rydis on
1333 * fixed bug 141b: printing backquoted information readably and prettily
1334 inserts a space where necessary.
1335 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1336 (CHARACTER 10) are now reported as errors, rather than propagated
1337 as unknown types. (reported by piso on #lisp)
1338 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1339 with values NIL and :ERROR. (thanks to Milan Zamazal)
1340 * fixed bug 191c: CLOS now does proper keyword argument checking as
1341 described in CLHS 7.6.5 and 7.6.5.1.
1342 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1343 longer raise an error; further, using a list as a for-as-arithmetic
1344 counter now raises a meaningful error.
1345 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1346 now understood by sequence creation functions such as MAKE-SEQUENCE
1348 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1349 a STRING-INPUT-STREAM.
1350 * compiler enhancement: SIGNUM is now better able to derive the type
1352 * type declarations inside WITH-SLOTS are checked. (reported by
1354 * fixed some bugs revealed by Paul Dietz' test suite:
1355 ** incorrect optimization of TRUNCATE for a positive first
1356 argument and negative second.
1357 ** compiler failure in let-convertion during flushing dead code.
1358 ** compiler failure while deriving type of TRUNCATE on an
1359 interval, containing 0.
1360 ** ASH of a negative bignum by a negative bignum count now returns
1362 ** intersection of CONS types now canonicalizes properly, fixing
1363 inconsistencies in SUBTYPEP.
1365 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1366 * incompatible change: The --disable-debugger command line
1367 option now clobbers the debugger at a more fundamental
1368 level, by redefining #'INVOKE-DEBUGGER instead of by
1369 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1370 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1371 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1372 enter the debugger REPL and then suffer recursive errors
1373 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1374 while under the new system, BREAK in --disable-debugger mode
1375 terminates the system just as an unhandled error would.
1376 * fixed compiler performance when processing loops with a step >1;
1377 * bug fix: DOCUMENTATION now retrieves generic function
1378 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1379 support has been systematized, and now supports the methods
1380 specified by ANSI, along with a default method and a method for
1381 slot documentation. (reported by Nathan Froyd)
1382 * bug fix: effective methods associated with a generic function are
1383 no longer cached over a change of that generic function's method
1384 combination. (reported by Andreas Fuchs)
1385 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1386 names a non-existent directory. (thanks to Andreas Fuchs)
1387 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1388 the PPC platform, lead to stack corruption; this has been fixed.
1389 (reported by Rainer Joswig)
1390 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1391 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1392 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1393 arithmetic in general, is now much more reliable on the PPC
1395 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1396 platform now returns the right answer.
1397 * optimization: restored some effective method precomputation in
1398 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1399 precomputation is now tunable.
1400 * optimization: compiler-internal data structure use has been
1401 reviewed, and changes have been made that should improve the
1402 performance of the compiler by about 20%.
1403 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1404 simple-base-strings and simple-bit-vectors is improved.
1405 * optimization: the optimization of 32-bit logical and arithmetic
1406 functions introduced in version 0.8.3 on the x86 has been
1407 implemented on the mips, ppc and sparc platforms; an
1408 implementation of the same facility, but for 64-bit arithmetic,
1409 has been added for the alpha.
1410 * microoptimization: the compiler is better able to make use of the
1411 x86 LEA instruction for multiplication by constants.
1412 * bug fix: in some situations compiler did not report usage of
1413 generic arithmetic in (SPEED 3) policy.
1414 * bug 145b fix: compiler used wrong type specifier while converting
1415 MEMBER-types to numeric.
1416 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1418 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1419 index argument now works properly on the Alpha platform.
1420 * bug fix: floating point exception treatment on the Alpha platform
1422 * bug fix: FILE-POSITION works much better on string input and
1423 output streams. (thanks to Nikodemus Siivola)
1424 * bug fix: many threading/garbage collection symptoms sorted.
1425 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1426 might be pseudo-atomic.
1427 * internal change: Stopping for GC is now done with signals not
1428 ptrace. GC is now done in whichever thread wanted it, instead of
1430 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1432 * bug fix: result form in DO is not contained in the implicit
1434 * incompatible change: ICR structure is changed; the value part of
1435 CONTINUATION is now called LVAR; corresponding functions are
1436 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1437 * added type deriver for ISQRT (thanks to Robert E. Brown).
1438 * bug fix: better support for loading from the command line when an
1439 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1441 * fixed some bugs revealed by Paul Dietz' test suite:
1442 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1443 ** ROUND and FROUND now give the right answer when given very
1444 small float arguments.
1445 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1447 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1448 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1449 ** LCM should return a non-negative integer.
1450 ** PARSE-INTEGER returned the index of a terminator instead of the
1451 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1452 ** PARSE-INTEGER returned an incorrect index being applied to a
1454 ** LCM with two arguments of 0 returns 0 rather than signalling
1456 ** unsigned addition of a 32-bit constant with the high bit set no
1457 longer causes an internal compiler error on the x86.
1458 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1459 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1460 with negative last argument.
1461 ** byte specifiers with zero size and position no longer cause
1462 an error during type derivation.
1463 ** bignum multiplication on the Alpha platform now returns the
1465 * porting: The system now builds on SuSE AMD64, although it still
1466 generates a 32-bit binary.
1467 * .fasl file incompatibility: The fasl file version number has
1468 been incremented (because of the changes to internal compiler
1469 data structures referred to above).
1471 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1472 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1473 more accurately, on the Darwin kernel running on PowerPC hardware.
1474 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1475 * Compiler code deletion notes now signal a condition of type
1476 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1477 an associated MUFFLE-WARNING restart.
1478 * The compiler now performs limited argument count validation of
1479 constant format strings in FORMAT, and where appropriate in ERROR,
1480 CERROR and WARN. (thanks to Gerd Moellmann)
1481 * New ASDF-INSTALL contrib can be used for automatic download and
1482 installation of third-party Lisp code from CCLAN or other sites
1484 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1485 which forces another thread to execute a function supplied by the
1487 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1488 now accept and act upon their :ELEMENT-TYPE keyword argument.
1489 (reported by Martin Atzmueller, Edi Weitz)
1490 * bug fix: FILE-POSITION now accepts position designators up to
1491 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1492 is the greater. (thanks to Patrik Nordebo)
1493 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1494 Kalle Olavi Niemitalo)
1495 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1496 printable. (reported by Eric Marsden)
1497 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1498 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1499 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1500 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1501 (thanks to Stig Sandoe)
1502 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1503 UNREAD-CHAR now correctly includes the unread character in the
1504 target sequence. (thanks to Gerd Moellmann)
1505 * bug fix in threaded builds: the system can now be suspended and
1506 resumed by shell job control with minimal disruption.
1507 * bug fixes in times and timezones >2038AD
1508 * better handling of "where is GNU make?" problem in build scripts
1509 (thanks to Nikodemus Siivola)
1510 * new optimization: inside a named function any reference to a
1511 function with the same name is considered to be a self-reference;
1512 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1513 optimization quality.
1514 * new optimization on x86: logical functions and + now have
1515 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1516 used when the result is truncated to 32 bits.
1517 * VALUES declaration is partially enabled.
1518 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1519 * fixed some bugs revealed by Paul Dietz' test suite:
1520 ** The system now obeys the constraint imposed by
1521 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1522 types form a lattice under type intersection.
1523 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1524 ** ASSOC now ignores NIL elements in an alist.
1525 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1526 and (1+ MOST-POSITIVE-FIXNUM) answers.
1527 ** The addition of a method with invalid qualifiers to a generic
1528 function does not cause an error to be signalled immediately;
1529 a warning is signalled, and the error is generated only on
1530 calling the generic function.
1531 * changed .fasl file version number, in part to add type codes for
1532 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1533 obscure ANSI requirements
1535 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1536 * fixed bug 148: failure to inline-expand a local function left
1537 garbage, confusing the compiler.
1538 * fixed bugs 3cd: structure slot readers perform type check if the
1539 slot can have an invalid value (i.e. it is either not initialized
1540 or can be written with a less specific slot writer).
1541 * bug fix: the compiler now traps array references to elements off
1542 the end of an array; previously, the bounds checking in some
1543 circumstances could go off-by-one.
1544 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1546 * type declarations for array element types now obey the description
1547 on the CLHS page "Declaration TYPE", as per discussions on
1548 sbcl-help around 2003-05-08. This means that a declaration
1549 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1550 declaration, all references to BAR will be asserted or assumed
1551 (with THE, so dependent on compiler policy) to involve objects of
1552 type FOO. Note that no such declaration is implied in
1553 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1554 * declared types of functions from the "Conditions"
1555 chapter. (reported by Paul Dietz)
1556 * bug fix: CERROR accepts a function as its first argument.
1557 * bug fix: NTH an NTHCDR accept a bignum as index
1558 arguments. (reported by Adam Warner)
1559 * optimization: character compare routines now optimize comparing
1560 against a constant character. (reported by Gilbert Baumann)
1561 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1562 argument now works properly on the MIPS platform.
1563 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1564 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1565 declared to have a functional type, does not cause an error
1567 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1568 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1569 * fixed bug: if last continuation of a deleted block has a
1570 destination, this destination should be deleted too. (reported by
1572 * fixed a bug in the bootstrap process: the host compiler's values
1573 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1574 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1575 test case from Patrik Nordebo)
1576 * improved the ability of the disassembler on the PPC platform to
1577 provide helpful disassembly notes.
1578 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1579 the class in more cases than previously.
1580 * bug fix: FILE-POSITION now understands :START and :END for
1581 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1582 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1583 8) data. (thanks to David Lichteblau)
1584 * bug fix: it is possible to add a method to a generic function
1585 without lambda list.
1586 * bug fix: reader failed to signal END-OF-FILE inside an
1587 object representation. (reported by Nikodemus Siivola)
1588 * fixed some bugs revealed by Paul Dietz' test suite:
1589 ** LAST and [N]BUTLAST should accept a bignum.
1590 ** condition slot accessors are methods.
1591 ** (VECTOR NIL) is a subtype of STRING.
1593 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1594 * minor incompatible change: some nonsensical specialized lambda
1595 lists (used in DEFMETHOD) which were previously ignored now signal
1597 * minor incompatible change: the system is now aware of the types of
1598 variables in the COMMON-LISP package, and will signal errors for
1599 most violations of these type constraints (where previously they
1600 were silently accepted).
1601 * minor incompatible change: COMPILE-FILE now uses the freedom
1602 afforded (ANSI 3.2.2.3) to use derived function types for
1603 functions defined in the same file. This also permits the system
1604 to warn on static type mismatches and function
1605 redefinition. (Currently it does not work with high DEBUG level.)
1606 * minor incompatible change: VALUES declaration is disabled.
1607 * When issuing notes, the compiler now signals a condition of type
1608 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1609 restart for use in user handlers. It is expected that the
1610 COMPILER-NOTE condition will eventually become a condition
1611 supertype to a hierarchy of note types, which will then be
1612 handleable in a similar fashion. However, at the moment, no such
1613 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1614 but it's an implementation detail, not a classification for the
1616 * Changes in type checking closed the following bugs:
1617 ** type checking of unused values (192b, 194d, 203);
1618 ** template selection based on unsafe type assertions (192c, 236);
1619 ** type checking in branches (194bc).
1620 * A short form of VALUES type specifier has ANSI meaning (it has
1621 increased the number of situations when SBCL cannot perform type
1623 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1624 keywords or constants is permissible.
1625 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1626 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1627 classes. (thanks to Antonio Martinez)
1628 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1629 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1630 argument to a pretty-print stream. Instead, it leaves any such
1631 operations to DESCRIBE-OBJECT methods.
1632 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1634 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1635 lambda lists. (thanks to David Lichteblau)
1636 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1637 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1638 (thanks to Gerd Moellmann)
1639 * fixed bug 246: increased compilation speed of long
1640 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1642 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1644 * DEFCONSTANT now throws a condition of type
1645 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1646 constant to a non-EQL value; CONTINUE and ABORT restarts
1647 respectively change and preserve the value.
1648 * fixed bug 63: The code walker, part of the implementation of CLOS,
1649 is now better at handling symbol macros.
1650 * bug fix: There is no longer an internal implementation type named
1651 CL:LENGTH. (reported by Raymond Toy)
1652 * bug fix: In macro-like defining macros/special operators the
1653 implicit block does not enclose the lambda list.
1654 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1655 enclosed in parentheses) are not suitable as type specifiers, and
1656 their use properly signals an error now.
1657 * bug fix: An argument count mismatch for a type specifier in code
1658 being compiled no longer causes an unhandled error at compile
1659 time, but signals a compile-time warning.
1660 * fixed simple vector readable printing
1661 * bug fix: DESCRIBE takes more care over whether the class
1662 precedence list slot of a class is bound before accessing it.
1663 (reported by Markus Krummenacker)
1664 * bug fix: FORMATTER can successfully compile pretty-printer format
1665 strings which use variants of the ~* directive inside.
1666 * bug fix: SEARCH now applies its TEST predicate to the elements of
1667 the arguments in the correct order. (thanks to Wolfhard Buss)
1668 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1669 in type inference. (thanks to Robert E. Brown)
1670 * bug fix: Reading in symbols with an explicit package name of ""
1671 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1672 * fixed some bugs revealed by Paul Dietz' test suite:
1673 ** NIL is now allowed as a structure slot name.
1674 ** Arbitrary numbers, not just REALs, are allowed in certain
1675 circumstances in LOOP for-as-arithmetic clauses.
1676 ** Multiple class redefinitions before slot access no longer
1677 causes a type error.
1678 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1679 association between the name and a class.
1680 ** Generic functions with non-standard method-combination and over
1681 five methods all of which return constants no longer return NIL
1682 after the first few invocations. (thanks to Gerd Moellmann)
1683 ** CALL-NEXT-METHOD with no arguments now passes the original
1684 values of the arguments, even in the presence of assignment.
1685 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1687 ** DISASSEMBLE works with closures and funcallable instances.
1688 ** ADD-METHOD now returns the generic function, not the new method.
1689 ** FIND-METHOD signals an error if the lengths of the specializers
1690 is incompatible with the generic function, even if the ERRORP
1692 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1693 which its argument is a member.
1694 ** DEFCLASS only redefines the class named by its class-name
1695 argument if that name is the proper name of the class;
1696 otherwise, it creates a new class.
1697 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1698 of the UNBOUND-SLOT condition to the name of the slot.
1699 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1700 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1701 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1702 SLOT-MAKUNBOUND in the specified fashion.
1704 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1705 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1706 cross-compilation host. As a consequence, we can now bootstrap our
1707 way up to SBCL starting with a bare gcc toolchain and human-readable
1708 source code (first the source to CLISP, then the source to SBCL).
1709 * A contributed module containing a partial implementation of the
1710 simple-streams interface has been included. (thanks to Rudi
1712 * A contributed module implementing the RFC1321 Message Digest
1713 Algorithm, known as MD5, has been included.
1714 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1715 no longer has any effect, as the code controlled by this feature
1716 has been deleted. (As far as we know, no-one has ever built using
1717 this feature, and its semantics were confused in any case).
1718 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1719 work on conditions (as required by the ANSI specification),
1720 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1721 expected behaviour on conditions. Users should note, however,
1722 that such behaviour is not required by the ANSI specification,
1723 and so use of this behaviour may render their code unportable.
1724 * fixed some bugs revealed by Paul Dietz' test suite:
1725 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1727 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1728 ** SXHASH on condition objects no longer returns NIL.
1729 ** :ALLOCATION :CLASS slots are better treated; their values are
1730 updated on class redefinition, and initforms inherited from
1731 superclasses are applied.
1732 ** REMOVE-METHOD returns its generic function argument even when
1733 no method was removed.
1734 ** SHARED-INITIALIZE now initializes the values of the requested
1735 slots, including those with :ALLOCATION :CLASS.
1736 ** ALLOCATE-INSTANCE now works on structure classes defined via
1737 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1739 ** SLOT-EXISTS-P now works on conditions, as well as structures
1741 ** MAKE-LOAD-FORM now has the required methods on
1742 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1743 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1744 keyword, and now implements the SLOT-NAMES argument.
1745 ** methods with &OPTIONAL arguments no longer allow too many
1746 arguments to be passed in the call without error.
1747 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1748 option is consistent with the required arguments of the generic
1749 function lambda list.
1750 * bug fix: REQUIRE accepts a string designator. (Thanks to
1752 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1753 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1754 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1755 * bug fix: The compiler now checks for duplicated variables in macro
1757 * bug fix: SETQ on globals returns the correct value.
1758 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1759 not a COMPILER-ERROR (followed by some other strange error on
1760 choosing the CONTINUE restart).
1761 * bug fix: make.sh and friends are now more consistent in the way that
1762 they look for GNU "make".
1764 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1765 * experimental native threads support (on x86 Linux >=2.4 only).
1766 This is not compiled in by default: you need to add :SB-THREAD to
1767 the target features. See the "Beyond ANSI" chapter of the manual
1769 * fix for longstanding nonANSIism: The old distinction between
1770 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1771 The return value from CL:FIND-CLASS is now a CLOS class, and
1772 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1773 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1774 * An interface to the MetaObject Protocol, as described in Kiczales,
1775 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1776 MIT Press, 1991, is available from the SB-MOP package.
1777 * incompatible change: the SB-PCL package should now be considered
1778 a private implementation detail, and no longer a semi-private MOP
1780 * minor incompatible change: due to rearrangement for threads, the
1781 control stack and binding stack are now allocated at arbitrary
1782 addresses instead of being hardcoded per-port. Users affected by
1783 this probably have to be doing advanced things with shared
1784 libraries, and will know who they are.
1785 * minor incompatible change: Previously, all --eval forms used were
1786 processed with READ before any of them were processed with EVAL.
1787 Now each --eval form is processed with both READ and EVAL before
1788 the next --eval form is processed. (Thus package operations like
1789 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1790 work as the user might reasonably expect.)
1791 * minor incompatible change: *STANDARD-INPUT* is now only an
1792 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1794 * minor incompatible change: Y-OR-N-P is now character-oriented,
1795 not line oriented. Also, YES-OR-NO-P now works without errors.
1796 (thanks to Antonio Martinez)
1797 * sb-aclrepl module improvements: an integrated inspector, added
1798 repl features, and a bug fix to :trace command.
1799 * Known functions, which cannot be open coded by the backend, are
1800 considered to be able to check types of their arguments. (fixing
1801 a bug report by Nathan J. Froyd)
1802 * fixed a bug in computing method discriminating functions: It is
1803 now possible to define methods specialized on classes which have
1804 forward-referenced superclasses. (thanks to Gerd Moellmann)
1805 * fixed evaluation order in optional entries (reported by Gilbert
1807 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1808 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1809 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1810 required-by-AMOP NAME argument, as well as CLASS and
1811 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1812 * fixed bug 20: DEFMETHOD can define methods using names that are
1813 not the proper names of classes to designate class specializers.
1814 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1815 Unix file descriptors, instead of blowing up. (thanks to Antonio
1817 * Garbage collection refactoring: user-visible change is that a
1818 call to the GC function during WITHOUT-GCING will not do garbage
1819 collection until the end of the WITHOUT-GCING. If you were doing
1820 this you were probably losing anyway.
1821 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1822 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1823 TYPEP the latter but not the former.
1824 * The compiler issues a full WARNING for calls to undefined functions
1825 with names from the CL package.
1826 * MAP-INTO for a vector destination is open coded. (reported by
1827 Brian Downing on c.l.l)
1828 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1829 documentation string.
1830 * fixed some bugs revealed by Paul Dietz' test suite:
1831 ** COPY-ALIST now signals an error if its argument is a dotted
1833 ** Condition slots are now accessed more correctly in the presence
1834 of multiple initargs for a given slot.
1835 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1836 exclude restarts of the same name associated with a different
1838 ** DEFCLASS of forward-referenced classes with another
1839 forward-referenced class in the superclasses list no longer
1841 ** Condition slots are now initialized once each, not multiple
1842 times. (thanks to Gerd Moellmann)
1843 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1845 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1846 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1847 arguments contain duplicated elements.
1848 ** RESTART-CASE understands local macros.
1849 ** RESTART-CASE associates exactly its own restarts with a condition.
1850 ** ENDP in safe mode checks its argument to be of type LIST.
1851 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1852 in question is unbound.
1853 ** Optimized MAKE-INSTANCE functions no longer cause internal
1854 assertion failures in the presence of duplicate initargs.
1855 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1856 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1857 propagated even in the ctor optimized implementation of
1859 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1861 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1862 * a better implementation of SXHASH on (simple) bit vectors,
1863 measured both in execution speed and in distribution of results
1864 over the positive fixnums, has been installed. Likewise, a better
1865 implementation of EQUAL for simple bit vectors is now available.
1866 * fixed CEILING optimization for a divisor of form 2^k.
1867 * fixed bug 240 (emitting extra style warnings "using the lexical
1868 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1869 by Antonio Martinez)
1870 * fixed SXHASH, giving different results for NIL depending on type
1871 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1872 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1873 variable. (found by Rolf Wester)
1874 * SBCL does not ignore type declarations for special
1875 variables. (reported by rif on c.l.l 2003-03-05)
1876 * some bug fixes in contrib/sb-aclrepl/
1877 * fixed some bugs revealed by Paul Dietz' test suite:
1878 ** a bug in the CONS type specifier, whereby the CAR and CDR
1879 types got intertwined, has been fixed;
1880 ** the type system is now able to reason about the interaction
1881 between INTEGER and RATIO types more completely;
1882 ** APPEND, [N]REVERSE and NRECONC check that those their
1883 arguments, which must be proper lists, are really so;
1884 ** An array specialized to be unable to hold elements has been
1885 implemented, as required -- yes, really -- by ANSI;
1886 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1887 on malformed property lists;
1889 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1890 * incompatible packaging change: in line with Unix convention,
1891 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1892 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1893 * REQUIRE and PROVIDE are now optionally capable of doing something
1894 useful. See the documentation string for REQUIRE.
1895 * infrastructure for a managed SBCL contrib system: contributed
1896 modules in this release include:
1897 ** the ASDF system definition facility;
1898 ** an interface to the BSD Sockets API;
1899 ** an ACL-like convenience interface to the repl;
1900 (thanks to Kevin Rosenberg)
1901 ** an implementation of ROTATE-BYTE, with efficient implementation
1903 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1904 gives a closer approximation to the right answer than previously.
1905 (thanks to Raymond Toy)
1906 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1907 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1908 optional environment arguments, as required by ANSI.
1909 * fixed bugs in other functions taking environment objects, allowing
1910 calls with an explicit NIL environment argument to be compiled
1912 * fixed bug 228: primary return values from
1913 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1914 COMPILE or FUNCTION.
1915 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1916 :TYPE will no longer signal errors on innocuous objects.
1917 * fixed bug 231b: SETQ is better at respecting type declarations in
1918 the lexical environment.
1919 * fixed a bug in DEFCLASS: classes named by symbols with no or
1920 unprintable packages can now be defined.
1921 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1922 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1923 * fixed compiler failure related to checking types of functions.
1924 (reported by Robert E. Brown)
1925 * the compiler is now much more consistent in its error-checking
1926 treatment of bounding index arguments to sequence functions: in
1927 (SAFETY 3) code, errors will be signalled in almost all cases if
1928 invalid sequence bounding indices are passed to functions defined
1929 by ANSI to operate on sequences.
1930 * fixed a bug in the build procedure: documentation of SBCL-specific
1931 packages is now preserved and available in the final Lisp image.
1932 * lifted FDEFINITION lookup out of loops in the implementation of
1933 many list operations. (thanks to Robert E. Brown)
1934 * fixed a bug in the reader: the #n# reader macro now works for
1935 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1936 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1937 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1938 E. Brown and Tony Martinez respectively)
1939 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1940 SLOT-BOUNDP: in particular, it is now able to optimize them much
1941 better, and is now not vulnerable to having packages renamed.
1942 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1943 always called when a slot is not present in an instance. (thanks
1945 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1946 longer signal an error on unreading a character following EOF on
1947 the previous constituent stream. (thanks to Tony Martinez)
1948 * fixed some bugs revealed by Paul Dietz' test suite:
1949 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1950 not just nonnegative fixnums;
1951 ** the logical bit-array operators such as BIT-AND now accept an
1952 explicit NIL for their "opt-arg" argument (to indicate a
1953 freshly-consed result bit-array);
1954 ** ELT now signals an error on an invalid sequence index in safe
1956 ** the type system is now cleverer about negations of numeric
1957 types, and consequently understands the BIGNUM and RATIO types
1959 ** the type system is now cleverer about the interaction between
1960 INTEGER and RATIO types: while bugs still remain, many more
1961 cases are accurately computed;
1962 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1963 if it is in the last clause;
1964 ** CONSTANTLY now correctly returns a side-effect-free function in
1966 ** DECLARE is no longer treated as a special-operator; in
1967 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1968 * incremented fasl file version number due to the change in the
1969 DEFSTRUCT-SLOT-DESCRIPTION structure.
1971 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1972 * minor incompatible change: code processed by the "interpreter" or
1973 EVAL now has a compilation optimization policy of (DEBUG 2)
1974 (changed from (DEBUG 1)) to improve debuggability of interactive
1975 development, and to allow the use of the debug RETURN command in
1977 * an experimental implementation of the RETURN command for the
1978 debugger has been included. (thanks to Frederik Kuivinen)
1979 * fixed bug 62: constraints were not propagated into a loop.
1980 * fixed bug in embedded calls of SORT (reported and investigated by
1982 * fixed some bugs revealed by Paul Dietz' test suite:
1983 ** printing and reading of arrays with some dimensions having
1984 length 0 (thanks to Gerd Moellmann);
1985 ** BOA constructor with &AUX argument without a default value does
1986 not cause a type error;
1987 ** CONSTANTP now returns true for all self-evaluating objects.
1989 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1990 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1991 accessors that are related by inheritance, as specified in the
1992 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1993 Valtteri Vuorikoski)
1994 * The compiler is now able to inline functions that were defined in
1995 a complex lexical environment (e.g. inside a MACROLET).
1996 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1998 * Support for the upcoming FreeBSD-5.0 release has been included.
1999 (thanks to Dag-Erling Smorgrav)
2000 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2001 effect when it is not in a toplevel context.
2002 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2003 stand a better chance of being correct. (thanks to Gerd
2005 * fixed bug in COERCE, which now signals an error on coercing a
2006 rational to a bounded real type which excludes the expected
2008 * The compiler is now able to derive types more accurately from the
2009 COERCE and COMPILE functions.
2010 * fixed bug 223: functional binding is considered to be constant
2011 only for symbols in the CL package.
2012 * fixed bug 231: SETQ did not check the type of a variable being set
2013 (reported by Robert E. Brown)
2014 * A new optimization for MAKE-INSTANCE has been included, fixing
2015 various bugs (including relating to :ALLOCATION :CLASS slots and
2016 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2018 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2019 ** As required by ANSI, LOOP now disallows anonymous collection
2020 clauses such as COLLECT I in conjunction with aggregate boolean
2021 clauses such as THEREIS (= I 1);
2022 ** LOOP now signals an error when any variable is reused in the
2023 same loop (including the potentially useful construct analogous
2024 to WITH A = 1 WITH A = (1+ A);
2025 ** IT is only a special loop symbol within the first clause of a
2026 conditional loop clause;
2027 ** LOOP with a typed iteration variable over a hashtable now
2028 signals a type error iff it should.
2029 * fixed some other bugs revealed by Paul Dietz' test suite:
2030 ** FILE-STREAM now names the class previously known as FD-STREAM;
2031 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2032 argument) no longer signals an error;
2033 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2034 of the slot symbol, rather than using the current package
2035 ((:CONC-NAME "") continues to intern the slot's name in the
2037 * incremented fasl file version number, because of the incompatible
2038 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2039 of the new implementation of DEFINE-COMPILER-MACRO.
2041 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2042 * Support for building SBCL for MIPS platforms running in
2043 little-endian mode has now been checked in, and basic
2044 functionality on said platforms verified.
2045 * minor incompatible change: PCL now records the pathname of a file
2046 in which methods and the like are defined, rather than its
2048 * minor incompatible change: TRUENAME now considers the truename of
2049 a file naming a directory to be the pathname with :DIRECTORY
2050 component indicating that directory.
2051 * minor incompatible change: a NAMED clause in the extended form of
2052 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2053 reason for the previous behaviour is unclear.
2054 * more systematization and improvement of CLOS and MOP conformance
2055 in PCL (thanks to Gerd Moellman and Pierre Mai):
2056 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2058 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2059 primary methods with no specializers;
2060 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2062 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2063 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2064 FORWARD-REFERENCED-CLASSes; error reporting on
2065 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2067 ** SXHASH on CLOS instances now uses a slot internal to the
2068 instance to return different numbers on distinct instances,
2069 while preserving the same return value through invocations of
2071 ** DEFMETHOD signals errors when methods with longer incongruent
2072 lambda lists are added to generic functions;
2073 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2074 CLASS, as specified in AMOP;
2075 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2076 on the order returned by the primary method for classes of
2077 class STANDARD-CLASS;
2078 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2079 * fixed some bugs shown by Paul Dietz' test suite:
2080 ** DOLIST puts its body in TAGBODY;
2081 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2083 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2084 value producing form;
2085 ** if more variables are given to PROGV than values, extra
2086 variables are bound and made to have no value;
2087 ** NSUBSTITUTE on list arguments gets the right answer with
2089 ** ELT signals an error of type TYPE-ERROR when the index argument
2090 is not a valid sequence index;
2091 ** LOOP signals (at macroexpansion time) an error of type
2092 PROGRAM-ERROR when duplicate variable names are found;
2093 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2094 ** FUNCALL of special-operators now cause an error of type
2096 ** PSETQ now works as required in the presence of side-effecting
2097 symbol-macro places;
2098 ** NCONC accepts any object as its last argument;
2099 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2101 ** loop-for-as-package does not require a package to be explicitely
2103 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2105 * fixed bug 166: compiler preserves "there is a way to go"
2106 invariant when deleting code.
2107 * fixed bug 172: macro lambda lists with required arguments after
2108 &REST arguments now cause an error to be signalled. (thanks to
2110 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2111 bug. (thanks to Matthew Danish)
2112 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2114 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2115 arguments are lexically rebound. (thanks to Gerd Moellmann and
2117 * fixed bug 194: error messages are now more informative when there
2118 is no primary method applicable in a call to a generic
2119 function. (thanks to Gerd Moellmann)
2120 * fixed bug in command line argument checking (thanks to Julian
2122 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2123 (thanks to Matthew Danish)
2124 * incremented fasl file version number, because of the
2125 SXHASH-related changes in the layout of CLOS data structures
2127 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2128 * minor incompatible change: The runtime (the Unix executable named
2129 "sbcl") is now much pickier about the .core files it will load.
2130 Essentially it now requires .core files to descend from the same
2131 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2132 as the runtime does. (The intent is to prevent the crashes which
2133 can occur, and which can even be reported as mysterious failures,
2134 when people patch the sources or change the build parameters
2135 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2136 sbcl and .core files.)
2137 * fixed bug: VALUES-LIST is no longer optimized away.
2138 * fixed bug 142: The FFI conversion of C string values to Lisp
2139 string values no longer conses excessively. (thanks to Nathan
2140 Froyd porting Raymond Toy's fix to CMU CL)
2141 * began to systematize and improve MOP conformance in PCL (thanks to
2142 Nathan Froyd, Gerd Moellman and Pierre Mai):
2143 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2145 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2146 ** FINALIZE-INHERITANCE is now called on class finalization;
2147 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2148 argument precedence order.
2149 * fixed bug 202: The compiler no longer fails on functions whose
2150 derived types contradict their declared type.
2151 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2152 so it can be non-toplevel.
2153 * The fasl file version number has changed (because of the new
2154 implementation of DEFMACRO).
2155 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2156 safe code, that any length requirement by their type-specifier
2157 argument is valid. The exceptions to this are described in bug
2159 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2160 functions now check the types of their inputs as required by ANSI.
2161 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2162 introduced symbol is DECLAREd to be SPECIAL.
2163 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2164 * fixed bug: :COUNT argument to sequence functions may be negative
2165 * fixed bug: body of DO-SYMBOLS may contain declarations
2166 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2167 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2168 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2169 symbol macro only once
2170 * fixed printing of call frame when argument list is unavailable
2171 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2172 * compiler no longer signals WARNING on unknown keyword
2175 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2176 * A beta-quality port to the mips architecture running Linux,
2177 based on the old CMUCL backend, has been made. It has been tested
2178 on a big-endian kernel, and works sufficiently well to be able to
2179 rebuild itself; it has not been tested in little-endian mode.
2180 * fixed an inconsistency between gencgc.c and purify.c which made
2181 dumping/loading .core files unreliable
2182 * fixed bug 120a: The compiler now deals correctly with IFs where
2183 the consequent is the same as the alternative, instead of
2184 misderiving the return type. (thanks to Alexey Dejneka)
2185 * fixed bug 113: Logical pathnames are now dumpable (the logical
2186 host is resolved at load-time, throwing an error if it is not
2188 * fixed bug 174: FORMAT's error message is slightly clearer when a
2189 non-printing character is used in a format directive.
2190 * fixed several bugs in compiler checking of type declarations, i.e.
2191 violations of the Python "declarations are assertions" principle
2192 (thanks to Alexey Dejneka)
2193 * fixed several bugs in PCL's error checking (thanks to Gerd
2195 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2197 * fixed bug in compilation of functions as first class values
2198 (thanks to Antonio Martinez-Shotton)
2199 * The compiler's handling TYPE-ERRORs which it can prove will
2200 inevitably happen at runtime has been cleaned up and corrected
2201 in several ways. (thanks to Alexey Dejneka)
2202 * improved argument type checking for various basic arithmetic
2203 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2204 lavished on them in the past that they can be compiled in many
2205 ways in different special cases
2206 * fixed bug 181: compiler checks validity of user supplied type
2208 * cleaned up code flushing in optimization: Function calls which
2209 should signal errors for safety purposes (e.g. which ANSI says
2210 should signal errors when their arguments are of incorrect type)
2211 are no longer optimized away.
2212 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2213 * incremented fasl file version number, because changes in the
2214 implementation of sequence functions like COERCE caused
2215 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2216 in old inline expansions) to become undefined. (Actually these
2217 changes were later undone, so we might very well be binary
2218 compatible with 0.7.7 after all, but leaving the version number
2219 incremented seemed like the simplest and most conservative
2222 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2223 * An alpha-quality port to the parisc architecture running Linux,
2224 based on the old CMUCL backend, has been made. This, even more so
2225 than the other backends, should be considered still a work in
2226 progress; known problems include that the Linux kernel in 64-bit
2227 mode does not propagate the correct sigcontext structure to
2228 userspace, and consequently SBCL on a parisc64 kernel will not
2230 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2231 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2232 suppressed.) Also now that the compiler is looking at declarations
2233 in the environment, it checks optimization declarations as well,
2234 and suppresses inlining when (> DEBUG SPEED).
2235 * More fixes have been made to treatment of floating point exception
2236 treatment and other Unix signals. In particular, floating point
2237 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2238 * The detection and handling of control stack exhaustion (infinite
2239 or very deeply nested recursion) has changed. Stack exhaustion
2240 detection is now done by write-protecting pages at the OS level
2241 and applies at all optimization settings; when found, a
2242 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2243 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2244 be caught using IGNORE-ERRORS.
2245 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2246 correctly when transforming recursive LABELS functions to LETs.
2247 (thanks to Alexey Dejneka)
2248 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2249 that are names of constants or global variables.
2250 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2251 alien routines with docstrings.
2252 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2253 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2255 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2256 (thanks to Eric Marsden)
2257 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2258 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2259 * Bug fix: The SPARC backend can now compile functions involving
2260 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2261 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2262 to a functions expecting a PATHNAME-DESIGNATOR.
2263 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2264 lambda lists. (thanks to Alexey Dejneka)
2265 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2266 functions so that EXPORTing the name of the class doesn't cause
2267 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2268 to Antonio Martinez for reporting this)
2269 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2270 constant-folded without failing an assertion. (thanks to Einar
2271 Floystad Dorum for reporting this)
2272 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2273 types are now dealt with more correctly by the compiler.
2274 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2275 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2276 behaviour analogous to RENAME-FILE. This puts its behaviour more
2277 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2279 * The fasl file version number has changed again. (because of the
2280 bug fix involving the names of PCL MAKE-INSTANCE functions)
2282 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2283 * bug fix: Floating point exceptions are treated much more
2284 consistently on the x86/Linux and PPC/Linux platforms.
2285 * Array initialization with :INITIAL-ELEMENT is now much faster for
2286 cases when the compiler cannot open code the array creation, but
2287 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2288 array accesses have also seen a speed increase.
2289 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2290 specified. (This was at the root of some bad interactions between
2291 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2293 * bug fix: Internal error arguments for undefined functions are now
2294 computed correctly on the PPC/Linux platform.
2295 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2296 Raymond Toy's patch for CMU CL)
2297 * Support for the Solaris 9 operating environment has been included
2298 (thanks to Daniel Merritt)
2299 * A very ugly but hopefully complete draft of the missing FFI chapter
2300 of the manual has been created by reformatting the corresponding
2301 CMU CL manual chapter into (currently very ugly and incoherent)
2302 DocBook and bringing it up to date for SBCL behavior. Thus, the
2303 manual is now essentially complete, at least by my extreme
2304 once-and-only-once standards, whereby it's acceptable to refer to
2305 the doc strings of SB-EXT functions as the primary documentation.
2306 * The fasl file version number has changed again, due to cleanup of
2307 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2308 is no longer a static symbol.)
2310 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2311 * SBCL now builds with OpenMCL (version 0.12) as the
2312 cross-compilation host; also, more progress has been made toward
2313 bootstrapping under CLISP.
2314 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2316 * bug 158 fixed: The compiler can now deal with integer loop
2317 increments different from 1; fixing this turned out also to fix
2319 * bug 169 fixed: no more bogus warnings about using lexical bindings
2320 despite the presence of perfectly good SPECIAL declarations (thanks
2321 to David Lichteblau)
2322 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2323 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2324 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2326 * bug 180 fixed: Method combination specifications no longer ignore
2327 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2328 * bug fix: Structure type predicate functions now check their argument
2329 count as they should.
2330 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2331 correctly. (thanks to Pierre Mai)
2332 * minor incompatible change: The --noprogrammer option is deprecated
2333 in favor of the new --disable-debugger option, which is very similar.
2334 (The major difference is that it takes effect at a slightly different
2335 time at startup, causing handling of errors in --sysinit and
2336 --userinit files will be affected differently.) The
2337 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2338 been added to allow this functionality to be controlled from ordinary
2339 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2340 maintainers, who might want to run non-interactive scripts to
2341 build SBCL cores which will later be used interactively.)
2342 * minor incompatible change: The LOAD function no longer, when given
2343 a wild pathname to load, loads all files matching that pathname.
2344 Instead, an error of type FILE-ERROR is signalled.
2346 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2347 * bug 147 fixed: The compiler preserves its block link/count
2348 invariants more correctly now so that it doesn't crash. (thanks
2350 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2352 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2353 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2354 no-value-for-this-character as NIL (as specified by ANSI).
2355 * HOST-NAMESTRING on physical pathnames now returns a string that is
2356 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2357 (thanks to Christophe Rhodes)
2358 * The Alpha port handles icache flushing more correctly. (thanks to
2360 * More progress has been made toward bootstrapping under CLISP. (thanks
2361 to Christophe Rhodes)
2362 * The fasl file format has changed again, because dynamic loading
2363 on OpenBSD (which has non-ELF object files) motivated some cleanups
2364 in the way that foreign symbols are transformed and passed around.
2365 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2366 is no longer treated as whitespace by the reader, but instead as
2367 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2368 |A
\7fB|, instead of A as it used to.
2370 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2371 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2372 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2373 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2374 this as of 0.7.1.45, but was left out of the previous news section
2375 (thanks to Dan Barlow)
2376 * SBCL now runs on the Solaris operating system on SPARC architectures
2377 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2378 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2379 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2380 * SPARC backend cleanups, allowing builds of cores optimized for V8
2381 and V9 SPARCS, and also emission of code targeted to a particular
2382 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2384 * SBCL is closer to bootstrapping under CLISP, thanks to various
2385 fixes by Christophe Rhodes.
2386 * The fasl file format has changed again, to allow the compiler's
2387 INFO database to support symbol macros.
2388 * The user manual (in doc/) is formatted into HTML more nicely.
2389 (thanks to coreythomas)
2390 * The system is smarter about SUBTYPEP relationships, especially
2391 those involving NOT types (including types such as ATOM which are
2392 represented internally using NOT types). Thus SUBTYPEP is less
2393 likely to return (VALUES NIL NIL) in general, and in particular
2394 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2396 * The fasl file format has changed again, because the internal
2397 representation of types now includes a new slot to support the new
2398 SUBTYPEP-of-NOT-types logic.
2399 * (not a change in the main branch of SBCL, but a related prototype
2400 which can hopefully be merged into the main branch of SBCL in the
2401 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2402 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2403 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2404 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2405 dispatch macro character combinations. (thanks to Alexey Dejneka)
2406 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2407 that superficially look like logical namestrings correctly.
2408 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2410 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2411 default supplied-p) for &optional and &key arguments. (thanks to
2414 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2415 * incompatible change: The compiler is now less aggressive about
2416 tail call optimization, doing it only when (> SPACE DEBUG) or
2417 (> SPEED DEBUG). (This is an incompatible change because there are
2418 programs which relied on the old CMU-CL-style behavior to optimize
2419 away their unbounded recursion which will now die of stack overflow.)
2420 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2421 for non-GENCGC systems has been increased to 20M (since that
2422 seems much closer to the likely performance optimum for modern
2423 systems than the old 4M value was)
2424 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2425 and *DEBUG-PRINT-LEVEL*
2426 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2427 of CMU CL's support for SPARC, and various endianness and other
2428 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2429 * new syntactic sugar for the Unix command line: --load foo.bar is now
2430 an alternate notation for --eval '(load "foo.bar")'.
2432 ** The system now detects stack overflow and handles it gracefully,
2433 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2434 optimization settings. (This is a good thing in general, and
2435 its introduction in this version should be particularly timely
2436 for anyone whose code fails because of suppression of tail
2438 ** The system now hunts for the C variable "environ" in a more
2439 devious way, to avoid segfaults when the C library version differs
2440 between compile time and run time. (thanks to Christophe Rhodes)
2441 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2442 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2443 ** The compiler no longer issues bogus style warnings for undefined
2444 classes in the same source file as the DEFCLASSes which defined
2445 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2447 ** fixes in CONDITION class precedence list for undefined function
2448 errors (thanks to Alexei Dejneka)
2449 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2450 correctly. (thanks to Dan Barlow)
2451 ** portability fixes aiming at bootstrapping under CLISP (thanks
2452 to Dave McDonald and Christophe Rhodes)
2453 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2454 ** fixes in type translation and and type inference (thanks to
2456 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2457 ** various fixes in the new ports (thanks to Dan Barlow)
2458 * several changes related to debugging:
2459 ** suppression of tail recursion, as noted above
2460 ** stack overflow detection, as noted above
2461 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2462 is now the default. (For some time encapsulation has been more
2463 reliable than the breakpoint-based :ENCAPSULATE NIL
2464 implementation, at least on X86 systems; and I just noticed that
2465 encapsulation also seems closer to the spirit of the ANSI
2468 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2470 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2471 up properly again. (There was a packaging bug in 0.7.0 which
2472 left their definitions in SB-SYS::LOAD-FOREIGN and
2473 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2474 vital for most things which interface to C-level interfaces,
2475 like extensions working with sockets or databases or
2476 Perl-compatible regexes or whatever, and the need to fix
2477 this bug is the main reason that 0.7.1 was released so
2479 ** DEFGENERIC is now choosier about the methods it redefines, so that
2480 reLOADing a previously-LOADed file containing DEFGENERICs does
2481 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2482 cycle now works as it should. (thanks to Alexey Dejneka)
2483 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2484 fixed by Christophe Rhodes. (He actually submitted this patch
2485 months ago, and I delayed until after 0.7.0.)
2486 ** Bug 111 (internal compiler confusion about runtime checks on
2487 FUNCTION types) was fixed by Alexey Dejneka.
2488 * Some internal cleanups (getting rid of variables which aren't
2489 needed now that the byte interpreter is gone) caused the fasl
2490 file format number to change again.
2492 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2493 * major incompatible change: The default fasl file extension, i.e. the
2494 default extension for files produced by COMPILE-FILE, has changed
2495 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2497 ** There are many changes in the implementation of the compiler.
2498 SBCL is now essentially a compiler-only implementation of ANSI
2499 Common Lisp. EVAL still "interprets" a few special cases, but
2500 almost all the interesting cases are handled by creating
2501 a LAMBDA expression, calling COMPILE on it, then calling
2502 FUNCALL on the result.
2503 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2504 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2505 Since the code is newer, there might still be some new bugs
2506 (though not as many as before Martin Atzmueller's fixes:-). But
2507 the new code is substantially simpler and clearer, and hopefully
2508 any remaining bugs will be simpler, less fundamental, and more
2509 fixable then the bugs in the old code.
2510 ** The revised compiler is still a little unsteady on its feet.
2512 *** The debugging information it produces (particularly the names
2513 of FUNCTION objects) is sometimes much less useful than what
2514 the old compiler produced.
2515 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2516 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2517 within a MACROLET) has been temporarily weakened.
2518 ** There are new compiler optimizations for various functions:
2519 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2520 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2521 *** the math functions TRUNCATE, FLOOR, and CEILING
2522 *** the function-of-all-trades COERCE
2523 Mostly these should be transparent, but there's one
2524 potentially-annoying problem (bug 117): when the compiler
2525 inline-expands a function and does type analysis on the result,
2526 it can create control paths which have type mismatches, and
2527 when it can't prove that those control paths aren't taken,
2528 it will issue WARNINGs about the type mismatches. This is
2529 a particular problem in practice for the new sequence functions.
2530 It's not clear how this should be fixed, and for now, a
2531 workaround is given in the entry for 117 in the BUGS file.
2532 ** (Because of the interaction between the two previous items --
2533 occasional inlining problems and new inline expansions -- some
2534 of the new sequence function optimizations won't really kick in
2535 completely until debugging information, and then inlining, are
2536 straightened out in some future version.)
2537 * minor incompatible changes:
2538 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2539 DIRECTORY no longer implicitly promotes NIL slots of its
2540 pathname argument to :WILD. In particular, when you ask for the
2541 contents of a directory (which you used to be able to do without
2542 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2543 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2544 ** changes in behavior that ANSI explicitly defines to be
2545 implementation dependent:
2546 *** The new compiler-only implementation still conforms with ANSI,
2547 but acts a little different than before. Besides the obvious
2548 changes in performance tradeoffs (that the cost per form passed
2549 to EVAL has gone up, and the cost per form executed by EVAL
2550 has gone down), the behavior of the system changes a little
2551 because there are no longer any interpreted function objects.
2552 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2553 e.g. doing COMPILE on the output of interactive DEFUN is
2555 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2557 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2558 has changed from (:CL) to NIL.
2559 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2560 e.g. #\Newline and #\Tab, don't have names specified in the
2561 ANSI Common Lisp standard, is now based on their ASCII symbolic
2562 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2563 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2564 are no longer used for output.
2565 ** changes in internal implementation constants:
2566 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2567 4 million. (If your application spends a lot of time GCing and
2568 you have a lot of RAM, you might want to experiment with
2569 increasing it even more.)
2570 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2571 However, almost all old code should still continue to work without
2572 immediate update, as SB-C-CALL is now a (deprecated) nickname
2574 ** Old operator names in the style DEF-FOO are now deprecated in
2575 favor of new corresponding names DEFINE-FOO, for consistency with
2576 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2577 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2578 internal symbols, but a few supported extensions like
2579 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2580 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2581 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2582 etc. as you get deeper into recursive calls to the debugger
2583 command loop, instead of the old "5]", "5]]", "5]]]"
2584 sequence. (I was motivated to do this when squabbles between
2585 ILISP and SBCL left me very deeply nested in the debugger. In the
2586 short term, this change will probably provoke more ILISP/SBCL
2587 squabbles, but hopefully it will be an improvement in the long run.)
2588 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2589 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2590 function, since declaiming return types involving aliens is
2591 (1) annoyingly messy to do by hand and (2) vital to efficient
2592 compilation of code which calls such functions.
2593 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2594 longer reexported by the SB-EXT package. They're solely useful
2595 for alien code, so it seems more logical that you should get
2596 them from the SB-ALIEN package, not in SB-EXT.
2597 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2598 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2599 features. Instead, the code that they used to control is always
2600 built into the system.
2601 * many other bug fixes
2602 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2603 advantage of the new EVAL-WHEN stuff and to clean them up in
2604 general, and they are now more ANSI-compliant in a number of
2605 ways. Martin Atzmueller is responsible for a lot of this.
2606 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2608 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2609 *** correct ERROR type for various file operations
2610 *** some fixes for Lisp streams
2611 *** DEFMETHOD syntax checking
2612 *** changing old weird representation of debug information as
2613 strings (which, among their other deficiencies, don't transform
2614 correctly when you rename packages, and don't change their
2615 print representation when you change things like *PACKAGE*
2616 and *PRINT-LENGTH*) to symbols and lists of symbols
2617 He also made several improvements and fixed several bugs in DESCRIBE.
2618 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2620 *** misbehavior of WRITE-STRING/WRITE-LINE
2621 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2622 and several other LOOP problems as well
2623 *** DIRECTORY when similar filenames are present
2624 *** DEFGENERIC with :METHOD options
2625 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2626 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2627 *** argument ordering in FIND with :TEST option
2628 *** mishandled package designator argument in APROPOS-LIST
2629 *** various problems in the backquote readmacro
2630 *** a bug in APROPOS
2631 *** probably some others that I'm not describing very well here,
2632 since the CVS log documents them by reference to sbcl-devel
2633 messages, and the SourceForge archives aren't working well.:-(
2634 ** Dan Barlow improved the Alpha port (and is making progress on the
2635 PPC port, for those of you who think different).
2636 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2637 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2638 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2639 continues to work on the SPARC port (for those of you in a position
2640 to look down upon our little PC-compatible boxes from a great height).
2641 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2642 and uses that copy, rather than the current dynamic value, when
2643 it's trying to decide whether to truncate output. Thus e.g.
2644 (let ((*print-lines* 50))
2645 (pprint-logical-block (stream nil)
2647 (let ((*print-lines* 8))
2648 (print (aref possiblybigthings i) stream)))))
2649 should now truncate the logical block only at 50 lines, instead of
2650 often truncating it at 8 lines, as it did before.
2651 * The doc/cmucl/ directory, containing old CMU CL documentation
2652 from the time of the fork, is no longer part of the base system.
2653 SourceForge has shut down its anonymous FTP service, and with it
2654 my original plan for distributing the old CMU CL documentation
2655 there. For now, if you need these files you can download an old
2656 SBCL source release and extract them from it.
2657 * The fasl file version number changed again, for dozens of reasons,
2658 some of which are apparent above.
2660 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2661 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2662 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2663 patch, so that DISASSEMBLE output is much nicer.
2664 * The code in the SB-PROFILE package now seems reasonably stable.
2665 I still haven't decided what the final interface should look like
2666 (I'd like PROFILE to interact cleanly with TRACE, since both
2667 facilities use function encapsulation) but if you have a need
2668 for profiling now, you can probably use it successfully with
2669 the current CMU-CL-style interface.
2670 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2671 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2672 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2673 link now returns the dangling link itself, and for similar
2674 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2675 link itself. (In these cases the old code signalled an error and
2676 looped endlessly, respectively.) Thus, DIRECTORY now works even
2677 in the presence of dangling and cyclic symbolic links.
2678 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2679 is now a supported extension again, since the consensus on
2680 sbcl-devel was that it can be useful for ordinary development
2681 work, not just for debugging SBCL itself.
2682 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2683 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2684 that currently-defined functions might be redefined later with
2685 different return types.
2686 * Hash tables can be printed readably, as inspired by CMU CL code
2687 of Eric Marsden and SBCL code of Martin Atzmueller.
2688 * better error handling in CLOS method combination, thanks to
2689 Martin Atzmueller porting Pierre Mai's CMU CL patches
2690 * more overflow fixes for >16Mbyte I/O buffers
2691 * A bug in READ has been fixed, so that now a single Ctrl-D
2692 character suffices to cause end-of-file on character streams.
2693 In particular, now you only need one Ctrl-D at the command
2694 line (not two) to exit SBCL.
2695 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2697 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2698 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2699 does the right thing.
2700 * The implementation of some type tests, especially for CONDITION
2701 types, is now tidier and maybe faster, due to CMU CL code
2702 originally by Douglas Crosher, ported by Martin Atzmueller.
2703 * Some math functions have been fixed, and there are new
2704 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2705 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2706 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2707 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2708 likely to be built into sbcl-0.7.0.)
2709 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2710 CL, ported to SBCL by Martin Atzmueller)
2711 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2712 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2713 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2714 * There's a new slam.sh hack to shorten the edit/compile/debug
2715 cycle for low-level changes to SBCL itself, and a new
2716 :SB-AFTER-XC-CORE target feature to control the generation of
2717 the after-xc.core file needed by slam.sh.
2718 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2719 COMPILE-FILE is no longer supported, so that now every function
2720 gets an entry point, so that block compilation looks a little
2721 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2722 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2723 returns the number of bytes consed since the system started,
2724 rather than the number consed since the first time the function
2725 was called. (The new definition parallels ANSI functions like
2726 CL:GET-INTERNAL-RUN-TIME.)
2727 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2728 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2729 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2730 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2731 * Fasl file version numbers are now independent of the target CPU,
2732 since historically most system changes which required version
2733 number changes have affected all CPUs equally. Similarly,
2734 the byte fasl file version is now equal to the ordinary
2737 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2738 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2739 is no longer recognized. I apologize for this, because it was
2740 listed in SB-EXT as a supported extension, but I found that
2741 its existing behavior was poorly specified, as well as incorrectly
2742 specified, and it looked like too much of a mess to straighten it
2743 out. I have enough on my hands trying to get ANSI stuff to work..
2744 * many patches ported from CMU CL by Martin Atzmueller, with
2745 half a dozen bug fixes in pretty-printing and the debugger, and
2746 half a dozen others elsewhere
2747 * fixed bug 13: Floating point infinities are now supported again.
2748 They might still be a little bit flaky, but thanks to bug reports
2749 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2750 as flaky as they were.
2751 * The --noprogrammer command line option is now supported. (Its
2752 behavior is slightly different in detail from what the old man
2753 page claimed it would do, but it's still appropriate under the
2754 same circumstances that the man page talks about.)
2755 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2756 are now supported, and enabled by default. Thus, the compiler can
2757 handle many floating point and complex operations much less
2758 inefficiently. (Thus e.g. you can implement a complex FFT
2760 * The compiler now detects type mismatches between DECLAIM FTYPE
2761 and DEFUN better, and implements CHECK-TYPE more correctly, and
2762 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2763 thanks to patches from Martin Atzmueller.
2764 * various fixes to make the cross-compiler more portable to
2765 ANSI-conforming-but-different cross-compilation hosts (notably
2766 Lispworks for Windows, following bug reports from Arthur Lemmens)
2767 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2768 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2769 fixed thanks to Pierre Mai's CMU CL patches.
2770 * fixes to keep the system from overflowing internal counters when
2771 it tries to use i/o buffers larger than 16M bytes
2772 * fixed bug 45a: Various internal functions required to support
2773 complex special functions have been merged from CMU CL sources.
2774 (When I was first setting up SBCL, I misunderstood a compile-time
2775 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2776 * improved support for type intersection and union, fixing bug 12
2777 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2778 more obscure bugs as well
2779 * some steps toward byte-compiling non-performance-critical
2780 parts of the system, courtesy of patches from Martin Atzmueller
2781 * Christophe Rhodes has made some debian packages of sbcl at
2782 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2783 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2784 stable, but are nonetheless usable. When he's ready, I'd be happy
2785 to add them to the SourceForge "File Releases" section. (And if
2786 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2787 * new fasl file format version number (because of changes in
2788 internal representation of (OR ..) types to accommodate the new
2789 support for (AND ..) types, among other things)
2791 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2792 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2794 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2796 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2797 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2798 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2799 * DESCRIBE now works on structure objects again.
2800 * Most function call argument type mismatches are now handled as
2801 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2802 know whether the function will be redefined before the call is
2803 executed. (The compiler could flag local calls with full WARNINGs,
2804 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2805 it doesn't keep track of enough information to know whether calls
2806 are local in this sense.)
2807 * Compiler output is now more verbose, with messages truncated
2808 later than before. (There should be some supported way for users
2809 to override the default verbosity, but I haven't decided how to
2810 provide it yet, so this behavior is still controlled by the internal
2811 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2812 src/compiler/ir1util.lisp.)
2813 * Fasl file format version numbers have increased again, because
2814 support for the Gray streams extension changes the layout of the
2815 system's STREAM objects.
2816 * The Gray subclassable streams extension now works, thanks to a
2817 patch from Martin Atzmueller.
2818 * The full LOAD-FOREIGN extension (not just the primitive
2819 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2820 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2821 but like most other programs, it defaults to copying the Unix
2822 environment from the original process instead of starting the
2823 new process in an empty environment.
2824 * Extensions which manipulate the Unix environment now support
2825 an :ENVIRONMENT keyword option which doesn't smash case or
2826 do other bad things. The CMU-CL-style :ENV option is retained
2827 for porting convenience.
2828 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2829 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2831 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2833 * A patch from Martin Atzmueller seems to have solved the SIGINT
2834 problem, and as far as we know, signal-handling now works cleanly.
2835 (If you find any new bugs, please report them!)
2836 * The system no longer defaults Lisp source file names to types
2837 ".l", ".cl", or ".lsp", but only to ".lisp".
2838 * The compiler no longer uses special default file extensions for
2839 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2840 seems to expect a single default extension for all compiled code,
2841 and there's no compelling reason to try to stretch the standard
2842 to allow two different extensions.) Instead, byte-compiled files
2843 default to the same extension as native-compiled files.
2844 * Fasl file format version numbers have increased again, because
2845 a rearrangement of internal implementation packages made some
2846 dumped symbols in old fasl files unreadable in new cores.
2847 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2848 many fewer weird special cases.
2849 * Bug #17 (differing COMPILE-FILE behavior between logical and
2850 physical pathnames) has been fixed, and some related misbehavior too,
2851 thanks to a patch from Martin Atzmueller.
2852 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2853 by Tim Moore, ported to SBCL by Martin Atzmueller.
2854 * Martin Atzmueller fixed several filesystem-related problems,
2855 including bug #36, in part by porting CMU CL patches, which were
2856 written in part by Paul Werkowski.
2857 * More compiler warnings in src/runtime/ are gone, thanks to
2858 more patches from Martin Atzmueller.
2859 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2862 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2864 * DESCRIBE now works on CONDITION objects.
2865 * The debugger now handles errors which arise when trying to print
2866 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2868 * The build system now uses an additional file, customize-target-features.lisp,
2869 to allow local modifications to the target *FEATURES* list. (The point of
2870 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2871 debugging features enabled, without having to worry about propagating it
2872 into everyone's system when I do a "cvs update".) When no
2873 customize-target-features.lisp file exists, the target *FEATURES* list
2874 should be constructed the same way as before.
2875 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2876 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2877 warn about behavior which is undefined under the ANSI spec. Note: This
2878 is specified by ANSI, but it's not very popular with programmers.
2879 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2880 macro in the SBCL sources for an example of a workaround which you
2881 might use to make portable ANSI-standard code which does what you want.
2882 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2883 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2884 when it's used as a non-toplevel form. (This is required in order
2885 to implement the DEFCONSTANT-EQX macro.)
2886 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2887 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2888 funny things, due to bugs in the implementation of EVAL-WHEN
2889 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2890 non-toplevel DEFCONSTANTs are uncommon.)
2891 * The core file version number and fasl file version number have been
2892 incremented, because the old noncompliant DEFCONSTANT behavior involved
2893 calling functions which no longer exist, and because I also took the
2894 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2895 * fixed bug 1 (error handling before read-eval-print loop starts), and
2896 redid debugger restarts and related debugger commands somewhat while
2898 ** The QUIT debugger command is gone, since it did something
2899 rather different than the SB-EXT:QUIT command, and since it never
2900 worked properly outside the main toplevel read/eval/print loop.
2901 Invoking the new TOPLEVEL restart provides the same functionality.
2902 ** The GO debugger command is also gone, since you can just invoke
2903 the CONTINUE restart directly instead.
2904 ** The TOP debugger command is also gone, since it's redundant with the
2905 FRAME 0 command, and since it interfered with abbreviations for the
2907 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2909 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2910 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2911 some porting work by Martin Atzmueller)
2912 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2913 with a boatload of patches which he ported to SBCL. Now that those
2915 ** The system tries to make sure that its low-priority messages
2916 are prefixed by semicolons, to help people who like to use
2917 syntax highlighting in their ILISP buffer. (This patch
2918 was originally due to Raymond Toy.)
2919 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2920 patches originally written by Raymond Toy.
2921 ** The compiler understands coercion between single-value and
2922 multiple-VALUES type expressions better, getting rid of some very
2923 weird behavior, thanks to patches originally by Robert MacLachlan
2924 and Douglas Crosher.
2925 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2926 lambda lists, thanks to a patch originally by Pierre Mai.
2927 ** The system no longer bogusly warns about "abbreviated type
2929 ** The compiler gets less confused by inlining and RETURN-FROM,
2930 thanks to some patches originally by Tim Moore.
2931 ** The system no longer hangs when dumping circular lists to fasl
2932 files, thanks to a patch originally from Douglas Crosher.
2933 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2934 undefined function error.
2935 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2936 in structure slot initforms) for now, documented workaround instead:-|
2937 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2938 * fixed bug 5: added stubs for various Gray stream functions called
2939 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2940 installed, at least appropriate type errors are generated
2941 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2942 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2943 consistently in DEFMETHOD forms.
2944 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2945 it doesn't seem to affect SBCL after all
2946 * The C runtime system now builds with better optimization and many
2947 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2949 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2951 * The system is now under CVS at SourceForge (instead of the
2952 CVS repository on my home machine).
2953 * The new signal handling code has been tweaked to treat register
2954 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2955 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2956 downstream, e.g. in debug-int.lisp, has implicit dependencies
2957 on the unsignedness of integer representation of machine words,
2958 and that caused the system to bomb out with infinite regress
2959 when trying to recover from type errors involving signed values,
2960 e.g. (BUTLAST '(1 2 3) -1).)
2961 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2962 (This was one of the bugs Peter Van Eynde reported back in July.)
2963 * The system now uses code inspired by Colin Walters' O(N)
2964 implementation of MAP (from the cmucl-imp@cons.org mailing
2965 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2966 inline the MAP operation, and there is more than one
2967 sequence argument to the MAP call (so that it can't just
2968 do ETYPECASE once and for all based on the type of the
2969 single sequence argument). (The old non-inline implementation
2970 of the general M-argument sequence-of-length-N case required
2971 O(M*N*N) time when any of the sequence arguments were LISTs.)
2972 * The QUIT :UNIX-CODE keyword argument has been renamed to
2973 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2974 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2976 * Raymond Wiker's patches to port dynamic loading from Linux to
2977 FreeBSD have been added.
2978 * The BUGS file is now more nearly up to date, thanks in large part
2979 to Martin Atzmueller's review of it.
2980 * The debugger now flushes standard output streams before it begins
2981 its output ("debugger invoked" and so forth).
2982 * The core version number and fasl file version number have both
2983 been incremented, because of incompatible changes in the layout
2985 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2986 used to not be called for a saved Lisp image.)
2987 * Martin Atzmueller's version of a patch to fix a compiler crash,
2988 as posted on sbcl-devel 13 September 2000, has been installed.
2989 * Instead of installing Martin Atzmueller's patch for the
2990 compiler transform for SUBSEQ, I deleted the compiler transform,
2991 and transforms for some similar consing operations.
2992 * A bug in signal handling which kept TRACE from working on OpenBSD
2994 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2995 away by constant folding
2996 * The system now defines its address space constants in one place
2997 (in the Lisp sources), and propagates them automatically elsewhere
2998 (through GENESIS and the sbcl.h file). Therefore, patching the
2999 address map is less unnecessarily tedious and error-prone. The
3000 Lisp names of address space constants have also been systematized.
3001 * CVS tags like dollar-Header-dollar have been removed from
3002 the sources, because they have never saved me trouble and
3003 they've been source of trouble working with patches and other
3004 diff-related operations.
3005 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3006 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3008 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3010 * The system has been ported to OpenBSD.
3011 * The system now compiles with a simple "sh make.sh" on the systems
3012 that it's supported on. I.e., now you no longer need to tweak
3013 text in the target-features.lisp-expr and symlinks in src/runtime/
3014 by hand, the make.sh takes care of it for you.
3015 * The system is no longer so grossly inefficient when compiling code
3016 involving vectors implemented as general (not simple) vectors (VECTOR T),
3017 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3018 which dares to use the various sequence functions on non-simple
3019 vectors, takes less of a performance hit.
3020 * There is now a primitive type predicate VECTOR-T-P
3021 to test for the (VECTOR T) type, so that e.g.
3022 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3023 can now be compiled with some semblance of efficiency. (The old code
3024 turned the type declaration into a full call to %TYPEP at runtime!)
3025 * AREF on (VECTOR T) is still not fast, since it's still compiled
3026 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3027 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3028 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3029 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3030 without making full calls to SUBTYPEP at runtime.
3031 (Some analogous efficiency issues for non-simple vectors specialized to
3032 element types other than T, or for non-simple multidimensional arrays,
3033 have not been addressed. They could almost certainly be handled the
3034 same way if anyone is motivated to do so.)
3035 * The changes in array handling break binary compatibility, so
3036 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3037 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3038 instead of (VALUES T T).
3039 * By following the instructions that Dan Barlow posted to sbcl-devel
3040 on 2 July 2000, I was able to enable primitive dynamic object
3041 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3042 functionality is not implemented (since it calls ld to resolve
3043 library references automatically, requiring RUN-PROGRAM for its
3044 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3045 try to resolve library references) is now supported.
3046 * The system now flushes the standard output streams when it terminates,
3047 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3048 them at several other probably-convenient times, e.g. in each pass of
3049 the toplevel read-eval-print loop, and after evaluating a form given
3050 as an "--eval" command-line option. (These changes were motivated by a
3051 discussion of stream flushing issues on cmucl-imp in August 2000.)
3052 * The source transform for TYPEP of array types no longer assumes
3053 that an array whose element type is a not-yet-defined type
3054 is implemented as an array of T, but instead punts, so that the
3055 type will be interpreted at runtime.
3056 * There is now some support for cross-compiling in make.sh: each of
3057 the phases of make.sh has its own script. (This should be transparent
3058 to people doing ordinary, non-cross-compile builds.)
3059 * Since my laptop doesn't have hundreds of megabytes of memory like
3060 my desktop machine, I became more motivated to do some items on
3061 my to-do list in order to reduce the size of the system a little:
3062 ** Arrange for various needed-only-at-cold-init things to be
3063 uninterned after cold init. To support this, those things have
3064 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3065 symbols with such names are now uninterned after cold init).
3066 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3067 fasl files for cold load.
3068 ** Remove the old compiler structure pooling code (which used to
3069 be conditional on the target feature :SB-ALLOC) completely.
3070 ** Redo the representation of some data in cold init to be more compact.
3071 (I also looked into supporting byte compiled code at bootstrap time,
3072 which would probably reduce the size of the system a lot, but that
3073 looked too complicated, so I punted for now.)
3074 * The maximum signal nesting depth in the src/runtime/ support code has
3075 been reduced from 4096 to 256. (I don't know any reason for the very
3076 large old value. If the new smaller value turns out to break something,
3077 I'll probably just bump it back up.)
3078 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3080 * Many, many bugs reported by Peter Van Eynde have been added to
3081 the BUGS list; some have even been fixed.
3082 * While enabling dynamic object file loading, I tried to make the
3083 code easier to understand, renaming various functions and variables
3084 with less ambiguous names, and changing some function calling
3085 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3086 * While trying to figure out how to do the OpenBSD port, I tried to
3087 clean up some of the code in src/runtime/. In particular, I dropped
3088 support for non-POSIX signal handling, added various comments,
3089 tweaked the code to reduce the number of compilation warnings, and
3090 renamed some files to increase consistency.
3091 * To support the new automatic configuration functionality in make.sh,
3092 the source file target-features.lisp-expr has been replaced with the
3093 source file base-target-features.lisp-expr and the machine-generated
3094 file local-target-features.lisp-expr.
3095 * fixed a stupid quoting error in make.sh so that using CMU CL
3096 "lisp -batch" as cross-compilation host works again
3098 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3100 * DESCRIBE no longer tries to call itself recursively to describe
3101 bound/fbound values, so that it no longer fails on symbols which are
3102 bound to themselves (like keywords, T, and NIL).
3103 * DESCRIBE now works on generic functions.
3104 * The printer now prints less-screwed-up representations of closures
3105 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3106 * A private symbol is used instead of the :EMPTY keyword previously
3107 used to mark empty slots in hash tables. Thus
3108 (DEFVAR *HT* (MAKE-HASH-TABLE))
3109 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3110 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3111 now does what ANSI says that it should. (You can still get
3112 similar noncompliant behavior if bang on the hash table
3113 implementation with all the symbols you get back from
3114 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3115 This breaks binary compatibility, since tests for equality to
3116 :EMPTY are wired into things like the macroexpansion of
3117 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3119 * There's now a minimal placeholder implementation for CL:STEP,
3120 as required by ANSI.
3121 * An obscure bug in the interaction of the normal compiler, the byte
3122 compiler, inlining, and structure predicates has been patched
3123 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3124 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3126 * Missing ordinary arguments in a macro call are now detected even
3127 when the macro lambda list contains &KEY or &REST.
3128 * The debugger no longer complains about encountering the top of the
3129 stack when you type "FRAME 0" to explicitly instruct it to go to
3130 the top of the stack. And it now prints the frame you request even
3131 if it's the current frame (instead of saying "You are here.").
3132 * As specified by ANSI, the system now always prints keywords
3133 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3135 * The default initial SIZE of HASH-TABLEs is now smaller.
3136 * Type information from CLOS class dispatch is now propagated
3137 into DEFMETHOD bodies, so that e.g.
3138 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3140 is now basically equivalent to
3141 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3142 (DECLARE (TYPE SINGLE-FLOAT X))
3144 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3145 operation, without having to do run-time type dispatch.
3146 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3147 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3148 * Since I don't seem to be making big file reorganizations very often
3149 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3150 is overflowing my ability to conveniently back them up), I've finally
3151 checked the system into CVS. (The CVS repository is on my home system,
3152 not at SourceForge -- putting it on SourceForge might come later.)
3153 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3154 high-level GC-NOTIFY-FOO functions send their output. (There's
3155 still very little control of where low-level verbose GC functions
3156 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3157 less than it used to -- the GC-NOTIFY-FOO functions are now under
3158 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3159 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3160 in only one place in the source code, and propagates it automatically
3161 everywhere that it's needed. Thus e.g. when I bump the version from
3162 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3163 * The C source files now include boilerplate legalese and documentation
3164 at the head of each file (just as the Lisp source files already did).
3165 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3166 to his page will be replaced with a link to his new CLiki service.
3168 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3170 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3171 * The build process now looks for GNU make under the default name "gmake",
3172 instead of "make" as it used to. If GNU make is not available as "gmake"
3173 on your system, you can change this default behavior by setting the
3174 GNUMAKE environment variable.
3175 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3176 can build without error under CMU CL.
3178 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3180 * There is now a partial SBCL user manual (with some new text and some
3181 text cribbed from the CMU CL manual).
3182 * The beginnings of a profiler have been added (starting with the
3183 CMU CL profiler and simplifying and cleaning up). Eventually the
3184 main interface should be through the TRACE macro, but for now,
3185 it's still accessed through vaguely CMU-CL-style functions and macros
3186 exported from the package SB-PROFILE.
3187 * Some problems left over from porting CMU CL to the new
3188 cross-compilation bootstrap process have been cleaned up:
3189 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3190 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3191 part when getting rid of the compiler *BACKEND* stuff.)
3192 ** The value of *NULL-TYPE* was screwed up, because it was
3193 being initialized before the type system knew the final
3194 definition of the 'NULL type. This screwed up several key
3195 optimizations in the compiler, causing inefficiency in all sorts
3196 of places. (I found it because I wanted to understand why
3197 GET-INTERNAL-RUN-TIME was consing.)
3198 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3199 PROCLAIM FTYPE information. Unfortunately this broke binary
3200 compatibility again, since now the forms output by DEFGENERIC
3201 to refer to functions which didn't exist in 0.6.3.
3202 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3203 can use the new (as of 0.6.3) transform for SOME into MAP into
3205 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3206 values to (INTEGER 0 1000000), so that the time code will no longer
3207 occasionally get blown up by Linux returning 1000000 microseconds
3208 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3209 its output conform to the ANSI spec. (Alas, this makes its output
3210 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3211 * A full call to MAP NIL with a single sequence argument no longer conses.
3212 * fixes to problems pointed out by Martin Atzmueller:
3213 * The manual page no longer talks about multiprocessing as though
3214 it were currently supported.
3215 * The ILISP support patches have been removed from the distribution,
3216 because as of version 5.10.1, ILISP now supports SBCL without us
3217 having to maintain patches.
3218 * added a modified version of Raymond Toy's recent CMU CL patch for
3219 EQUALP comparison of HASH-TABLE
3221 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3223 * The system still can't cross-compile itself with
3224 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3225 entails), but at least it can compile more of itself that way
3226 than it used to be able to, and various buglets which were uncovered
3227 by trying to cross-compile itself that way have now been fixed.
3228 * This release breaks binary compatibility again. This time
3229 at least I've incremented the FASL file format version to 2, so that the
3230 problem can be detected reliably instead of just causing weird errors.
3231 * various new style warnings:
3232 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3233 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3234 ** using the lexical binding of a variable named in the *FOO* style
3235 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3236 as specified by ANSI.
3237 * *RANDOM-STATE* is no longer automatically initialized from
3238 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3239 default behavior of the system is to repeat its behavior every time
3240 it's run. If you'd like to change this behavior, you can always
3241 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3242 old convention there was no comparably easy way to get the system to
3243 repeat its behavior every time it was run.
3244 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3245 FTYPE declarations has been removed, in favor of their ANSI
3246 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3247 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3248 the types of their sequence arguments can be determined at compile time.
3249 This is done through a new open code expansion for MAP which eliminates
3250 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3251 argument types can be determined at compile time.
3252 * The optimizer now transforms COERCE into an identity operation when it
3253 can prove that the coerced object is already of the correct type. (This
3254 can be a win for machine generated code, including the output of other
3255 optimization transforms, such as the MAP transform above.)
3256 * Credit information has been moved from source file headers into CREDITS.
3257 * Source file headers have been made more standard.
3258 * The CASE macro now compiles without complaining even when it has
3261 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3263 * (Note that the way that the PCL macroexpansions were rewritten
3264 to accommodate the change in DEFGENERIC below breaks binary
3265 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3266 not run under sbcl-0.6.2. Once we get out of alpha releases,
3267 i.e. hit release 1.0.0, we'll probably try to maintain binary
3268 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3269 and sbcl-1.4.4. Until then, however, it might be fairly common
3270 for maintenance releases to break binary compatibility.)
3271 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3273 * The reporting of unhandled signals has been changed to print some
3274 explanatory text as well as the report form. (Previously only
3275 the report form was printed.)
3276 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3277 it defines, so that the compiler no longer issues undefined function
3278 warnings for compiled-but-not-yet-loaded generic functions.
3279 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3280 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3281 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3282 section of the ANSI spec are supported.
3283 * The "" nickname for the "KEYWORD" package has been removed.
3284 The reader still handles symbol tokens which begin with a package marker
3285 as keywords, but it doesn't expose its mechanism for doing so in the
3286 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3287 * The system now issues STYLE-WARNINGs for contradictory TYPE
3288 proclamations. (Warnings for contradictory FTYPE proclamations would
3289 be nice too, but those can't be done usefully unless the type system
3290 is made smarter about FUNCTION types.)
3291 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3292 systematized, so that "*target-*.lisp is supposed to exist only on the
3293 target and imply that there's a related file which exists on the
3294 host, and *host-*.lisp is supposed to exist only on the host and imply
3295 that there's a related file which exists on the target. This involves a
3296 lot of renaming. Hopefully the acute confusion caused by the renaming
3297 will be justified by the reduction in chronic confusion..
3298 ** runtime-type.lisp -> early-target-type.lisp
3299 ** target-type.lisp -> late-target-type.lisp
3300 ** early-host-format.lisp -> early-format.lisp
3301 ** late-host-format.lisp -> late-format.lisp
3302 ** host-error.lisp -> misc-error.lisp
3303 ** early-error.lisp -> early-target-error.lisp
3304 ** late-error.lisp -> late-target-error.lisp
3305 ** host-defboot.lisp -> early-defboot.lisp
3306 ** code/misc.lisp -> code/target-misc.lisp
3307 ** code/host-misc.lisp -> code/misc.lisp
3308 ** code/numbers.lisp -> code/target-numbers.lisp
3309 ** code/early-numbers.lisp -> numbers.lisp
3310 ** early-host-type.lisp -> early-type.lisp
3311 ** late-host-type.lisp -> late-type.lisp
3312 ** host-typep.lisp -> typep.lisp
3313 ** load.lisp -> target-load.lisp
3314 ** host-load.lisp -> load.lisp
3315 ** host-disassem.lisp -> disassem.lisp
3316 ** host-insts.lisp -> insts.lisp
3317 ** byte-comp.lisp -> target-byte-comp.lisp
3318 ** host-byte-comp.lisp -> byte-comp.lisp
3319 ** host-signal.lisp -> signal.lisp
3320 ** host-defstruct.lisp -> defstruct.lisp
3321 ** late-target-type.lisp -> deftypes-for-target.lisp
3322 Furthermore, several other previously target-only files foo.lisp (e.g.
3323 hash-table.lisp and random.lisp) have been split into a target-and-host
3324 foo.lisp file and a target-only target-foo.lisp file, with their key type
3325 definitions in the target-and-host part, so that the cross-compiler will
3326 know more about target types.
3327 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3328 gone away. In their place are various *BACKEND-FOO* variables
3329 corresponding to the slots of the old structure.
3330 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3331 into the target SBCL has been fixed.
3332 * The chill.lisp system for loading cold code into a running SBCL
3334 * Support for the CMU CL "scavenger hook" extension has been removed.
3335 (It was undocumented and unused in the CMU CL sources that SBCL was
3336 derived from, and stale in sbcl-0.6.1.)
3337 * Various errors in the cross-compiler type system were detected
3338 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3339 (enabling various consistency checks). Many of them were fixed,
3340 but some hard problems remain, so the compiler is back to
3341 running without *TYPE-SYSTEM-INITIALIZED* for now.
3342 * As part of the cross-compiler type system cleanup, I implemented
3343 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3344 * I have started adding UNCROSS calls throughout the type system
3345 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3346 on cross-compiler input files will be able to go away, and various
3348 * CONSTANTP now returns true for quoted forms (as explicitly required
3351 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3353 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3354 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3355 (MAKE-LIST -1). In the longer term, it ought to become true
3356 that declarations are assertions even at SAFETY 1. For now, it's not
3357 quite true even at SAFETY 3, but it's at least more nearly true..
3358 (Note that this change seems to increases the size of the system by
3359 O(5%) and to decrease the speed of the compiler by 20% or more.)
3360 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3361 for the problem of printing dozens of lines of distracting information
3362 about low-level system machinery as part of the top stack frame
3363 on entry to the debugger when an undefined function was called.
3364 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3365 is not reset to COMMON-LISP-USER.
3366 * Compilation of stuff related to dyncount.lisp has been made conditional
3367 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3368 smaller. The various dyncount-related symbols have been moved into
3369 a new "SB-DYNCOUNT" package.
3370 * tty-inspect.lisp has been renamed to inspect.lisp.
3371 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3372 feature has gone away. (When we eventually port to other flavors of
3373 libc and/or Unix, we'll try to make the differences between flavors
3374 invisible at the user level.)
3375 * Various other *FEATURES* tags, and/or their associated conditionals,
3376 have been removed if obsolescent, or given better documentation, or
3377 sometimes given more-mnemonic names.
3379 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3381 * tidied up "make.sh" script
3382 * tidied up system directory structure
3383 * better "clean.sh" behavior
3384 * added doc/FOR-CMUCL-DEVELOPERS
3385 * many many small tweaks to output format, e.g. removing possibly-confusing
3386 trailing #\. character in DESCRIBE-INSTANCE
3387 * (EQUALP #\A 'A) no longer signals an error.
3388 * new hashing code, including EQUALP hashing
3389 * tidied up Lisp initialization and toplevel
3390 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3391 * command line argument processing
3392 * added POSIX-GETENV function to deal with Unix-ish environment variables
3393 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3394 terminating SBCL on EOF
3395 * non-verbose GC by default
3396 * There is no more "sbcl" shell script; the sbcl file is now the C
3397 runtime executable (just like CMU CL).
3398 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3400 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3401 symbol and package information as Lisp native symbol and package objects
3402 instead of strings naming symbols and strings naming packages. This way,
3403 whenever packages are renamed (as in warm init), debug information is
3404 transformed along with everything else.
3405 * tweaked the optimization policy declarations which control the building
3406 of SBCL itself. Now, among other things, the system no longer saves
3407 source location debugging information. (This helps two problems at once
3408 by reducing SBCL size and by keeping SBCL from trying to look for its
3409 sources -- which may not exist -- when reporting errors.)
3410 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3411 debugging and testing purposes
3412 * cleaned up printing, making the printer call PRINT-OBJECT for
3413 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3414 methods, giving nearly-ANSI behavior
3415 * converted almost all special variables to use *FOO* naming convention
3416 * deleted PARSE-TIME functionality, since it can be done portably
3417 * moved some files out of cold init into warm init
3418 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3420 * regularized formatting of source files
3421 * added an install.sh script
3422 * fixed ridiculous memory usage of cross-compiler by making
3423 compiler/alloc.lisp not try to do pooling unless it can hook
3424 itself into the GC of the cross-compilation host. Now the system
3425 builds nicely on my old laptop.
3426 * added :SB-ALLOC in target-features.lisp-expr
3427 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3428 was not implemented)
3429 * re-did condition handling and note reporting in the compiler. Notes
3430 are no longer handled by signalling conditions. Style warnings
3431 and warnings are handled more correctly and reported in such a way
3432 that it's easy to find one or the other in your output (so that you
3433 can e.g. figure out which of many problems caused COMPILE-FILE to
3435 * changed the severity of several compiler warnings from full WARNING
3436 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3437 compiler note reporting so that it doesn't use the condition system
3438 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3439 COMPILE-FILE command)
3440 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3441 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3442 no longer supported, since their implementation was deeply intertwingled
3443 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3444 * removed bogus "support" for compiler macros named (SETF FOO), and
3445 removed the compiler macro for SETF INFO (but only after making a fool
3446 of myself on the cmucl-imp mailing list by posting a bogus patch for
3447 DEFINE-COMPILER-MACRO..)
3448 * Compiled files containing forms which have side effects on the Lisp
3449 reader (such as DEFPACKAGE forms) are now handled more correctly.
3450 (Compiler queuing of top level lambdas has been suppressed by setting
3451 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3452 * deleted various currently-unused source files, e.g. gengc.lisp. They
3453 may be added back at some point e.g. when porting to other architectures,
3454 but until they are it's distracting to distribute them and to try to
3456 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3457 there were so many of them they're just distractions, and UNCROSS is
3458 known to be able to handle the current sources
3459 * moved PROFILE functionality into TRACE, so that it will be clear
3460 how the wrapping and unwrapping of functions when you profile them
3461 interacts with the wrapping and unwrapping of functions when you
3462 trace them. (Actually, the functionality isn't there yet, but at least
3463 the interface specification is there. Hopefully, the functionality will
3464 arrive with some future maintenance release.)
3465 * removed host-oops.lisp
3466 * changed signature of QUIT function to allow UNIX-CODE argument
3467 * fixed READ-SEQUENCE bug
3468 * tweaked verbose GC output so that it looks more like the progress
3469 output that ANSI specifies for functions like LOAD
3470 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3471 * added <http://sbcl.sourceforge.com> to the banner information printed by