1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
3 * minor incompatible change: (SETF CLASS-NAME) and (SETF
4 GENERIC-FUNCTION-NAME) are no longer generic functions, and
5 therefore are not customizeable by user code (as seems to be at
6 least permitted and maybe required by AMOP). As a consolation,
7 however, the SBCL implementation of these functions now calls
8 REINITIALIZE-INSTANCE as specified by AMOP.
9 * bug fix: it is now possible to have more than one subclass of
10 STANDARD-GENERIC-FUNCTION without causing stack overflow.
11 (reported by Bruno Haible, Pascal Costanza and others)
12 * bug fix: the dependent update protocol now works for generic
13 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
15 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
16 merged with *DEFAULT-PATHNAME-DEFAULTS*.
17 * optimization: performance improvements to IO on file streams of
18 :ELEMENT-TYPE CHARACTER
20 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
21 * bug fix: add a workaround to SBCL looping infinitely at startup on
22 Linux kernels with apparently buggy implementations of personality().
23 (thanks to Svein Ove Aas)
24 * bug fix: Unicode symbols are correctly printed in LDB backtraces
25 (thanks to David Lichteblau)
26 * bug fix: local bindings shadow global symbol macros for
27 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
28 * optimization: non-open coded uses of numeric comparison operators
29 (e.g. >) no longer cons when called with more than one parameter
30 on platforms supporting dynamic-extent allocation.
31 * enhancement: saving cores with foreign code loaded is now supported
32 on MIPS/Linux in addition to the previously supported platforms.
33 * bug fix: FIND-RESTART now tests for activity, not applicability when given
34 a restart object as identifier. (reported by Helmut Eller for CMUCL)
35 * bug fix: division by zero in sb-sprof when no samples were collected
36 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
37 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
38 (thanks to Alexander Kjeldaas)
39 * new feature: ignoring the return values of destructive functions
40 like NREVERSE signals a compile-time style-warning.
41 (thanks to Kevin Reid)
42 * new feature: the alignment of alien structure fields can be
43 explicitly specified. (thanks to Cyrus Harmon)
44 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
45 and dump core on SIGQUIT
47 ** incompatible change: threads do not inherit values of specials
48 from their parents (see manual)
49 ** bug fix: threads stacks belonging to dead threads are freed by the
50 next exiting thread, no need to gc to collect thread stacks anymore
51 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
52 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
53 ** bug fix: dynamic variable and thread start related gc lossage
54 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
56 * fixed some bugs revealed by Paul Dietz' test suite:
57 ** SUBTYPEP is slightly more accurate on heinously complicated
58 CONS types where some of the members have uncertain (in the
59 NIL, NIL sense) type relationships to each other.
61 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
62 no longer purified unless :PURIFY T is explicitly specified.
63 ** Non-purified cores are significantly smaller than before
65 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
66 * new feature: timers based on Zach Beane's excellent timer package
67 * added support for the following external formats: koi8-u,
68 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
69 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
70 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
71 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
72 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
73 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
74 * incompatible change: a threaded SBCL will no longer revert to
75 non-threaded mode on non-NPTL systems, but refuse to start entirely.
76 * bug fix: interrupts are disabled until startup is complete; no
77 more sigsegvs when receiving a signal to soon
78 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
80 * bug fix: add a workaround for the memory randomization features in
81 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
82 workaround will only be in effect on systems with the proc filesystem
84 * bug fix: printing objects of type HASH-TABLE signals a
85 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
87 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
88 have been running for over 50 days. (reported by Gilbert Baumann)
89 * bug fix: the logic for getting names of functions gets less
90 confused when confronded with alternate-metaclass
91 funcallable-instances. (reported by Cyrus Harmon)
92 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
93 consistent, even on internal alternate-metaclass objects.
94 * bug fix: flush closure information collected by physical
95 environment analysis prepass before the main pass. (bug reported
97 * bug fix: compiler pack phase does not modify a hashtable, which is
98 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
99 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
100 may be expanded. (reported by Alexander <malishev> on c.l.l)
101 * bug fix: no more style-warnings when compiling code using LAST
102 under high-SPEED low-SPACE settings. (reported by David Wallin
104 * fixed some incompatibilities between SBCL's MOP and the MOP
106 ** the METAOBJECT class is now implemented;
107 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
108 STANDARD-OBJECT, as required;
109 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
110 now compatible (as required by VALIDATE-SUPERCLASS); there
111 remains a consistency requirement over the presence or absence
112 of the FUNCTION class in the superclass list of finalized
113 classes; see the manual for more details;
114 ** the :AROUND method for COMPUTE-SLOTS on
115 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
116 requested slot ordering.
118 ** bug fix: parent thread now can be gc'ed even with a live
120 ** bug fix: binding a special with PROGV to no value is not
121 the same as that symbol not having been bound (thanks to
123 ** bug fix: binding specials is thread safe (thanks to
125 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
126 and SB-SPROF do not die with 'no handler for signal XX in
127 interrupt_handle_now(..)' anymore
128 ** bug fix: WITH-TIMEOUT works with multiple running threads
129 * fixed some bugs revealed by Paul Dietz' test suite:
130 ** ENSURE-GENERIC-FUNCTION should take a method class object for
131 the :method-class keyword argument.
133 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
134 * new port: the Solaris operating system on x86 processors is now
135 mostly supported, though some rough edges in the environment
136 remain. (thanks to Daisuke Homma)
137 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
138 space for its heap. (thanks to Thiemo Seufer)
139 * optimization: SBCL on MIPS platforms now supports dynamic-extent
140 closures. (thanks to Thiemo Seufer)
141 * minor incompatible change: eof selects abort in the debugger.
142 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
143 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
145 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
146 deprecated, and will go away in a future revision of SBCL.
147 * minor incompatible change: GC-ON and GC-OFF are no longer
148 implemented with a counter, it does not matter how many times gc
149 is switched on or off
150 * bug fix: discriminating functions for generic function classes
151 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
152 make invalid assumptions about method precedence order. (reported
154 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
155 platforms. (thanks to Thiemo Seufer)
156 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
157 * bug fix: correct pathname printing with printer escaping is on.
158 (thanks to Kevin Reid)
159 * bug fix: complex VOP definitions in "user-space" no longer trigger
160 package locks. (reported by Zach Beane)
161 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
162 for funcallable-instance functions, and is no different from
164 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
165 characters in the current readtable. (reported by Nicholas Neuss)
166 * bug fix: SUBTYPEP on various CONS types returns more a more
167 accurate acknowledgment of its certainty.
168 * optimizations: REMOVE-DUPLICATES now runs in linear time on
169 lists in some cases. This partially fixes bug 384.
170 * flush all standard streams before prompting in the REPL and the
172 * bug fix: signal handling and triggering gc do not conflict
173 directly anymore, in particular a high frequency sb-sprof does
174 not prevent gc from running
175 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
176 approximation for timezone and DST information between the
177 universal time epoch and the smallest negative 32-bit time_t.
178 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
179 year 1899 so long as the encoded time is non-negative.
180 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
181 the correct answer for non-constant rotate amounts when performing
182 an inline 32-bit rotation.
184 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
185 there is only one thread in the session
186 ** bug fix: memory leak for streams created in one thread and
187 written to in another
188 ** bug fix: lockup when compiled with gcc4
189 ** bug fix: race that allows the gc to be triggered when gc is
191 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
192 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
194 ** bug fix: finished threads are gc'ed properly
195 * fixed some bugs revealed by Paul Dietz' test suite:
196 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
197 has arguments with a different set of applicable methods from
198 the orignal arguments.
199 ** The type error thrown by MAP now has a correct expected-type
201 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
202 name a compiled function.
203 ** Three MISC tests where a large bignum was improperly coerced to
204 a float that couldn't represent that bignum during type
205 derivation were fixed.
206 ** SUBTYPEP can now handle types involving the negation of a
207 list-form FUNCTION type.
208 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
209 as (COMPLEX (AND RATIO (NOT FIXNUM))).
210 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
212 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
213 * New feature: Experimental support for bivalent streams: streams
214 opened with :element-type :default now allow character and binary
215 (unsigned-byte 8) I/O
216 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
217 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
219 * bug fix: no more highly sporadic "couldn't check whether ~S is
220 readable" when reading a stream and an interrupt hits in the middle
221 of a select system call
222 * compiler better recognizes complex arrays (reported by Eduardo
224 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
225 for variables on SBCL built with linkage-tables. (reported by Luis
227 * various error reporting improvements.
228 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
229 (Thanks to Thiemo Seufer)
230 * bug fix: nested reader invokations work correctly
231 * bug fix: it is possible to have simultaneous references to foreign
232 code and foreign data with the same name.
234 ** added x86-64 support
235 ** incompatible change: the threading api now works with thread
236 objects instead of thread ids
237 ** bug fix: threads are protected from signals and interruption when
238 starting up or going down
239 ** bug fix: a race where an exiting thread could lose its stack to gc
240 ** bug fix: don't halt on infinite error in threads if possible
241 ** fixed numerous gc deadlocks introduced in the pthread merge
242 ** bug fix: fixed thread safety issues in read and print
243 ** bug fix: debugger doesn't hang on session lock if interrupted at
244 an inappropriate moment
245 ** bug fix: run-program is now thread safe(r)
246 ** bug fix: inner with-recursive-lock no longer releases the mutex
247 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
248 * fixed some bugs revealed by Paul Dietz' test suite:
249 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
250 EXPECTED-TYPE slots filled.
251 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
253 ** MAKE-CONDITION accepts classes as type-designators.
254 ** COMPILE may never return NIL.
255 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
256 range before calling Unix time functions
258 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
259 * numerous signal handling fixes to increase stability
260 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
261 * minor incompatible change: we now correctly canonize default
262 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
263 as per the MOP, rather than the historical (INITARG INITFUNCTION
264 INITFORM). (reported by Bruno Haible)
265 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
266 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
267 for more information.
268 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
269 pathname is a directory pathname.
270 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
271 * dynamic space size on PPC has been increased to 768Mb. (thanks to
273 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
274 :METACLASS argument in addition to a class name. (reported by
275 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
276 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
277 (reported by Bruno Haible; patch from Gerd Moellmann)
278 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
280 * bug fix: more cleanups to the floating point exception handling on
281 x86-64 (thanks to James Knight)
282 * bug fix: the compiler does not try to constant fold calls of
283 COERCE and BIT-* functions when they return freshly constructed
284 objects. (reported by Nikodemus Siivola and Paul Dietz)
285 * optimization: improved the MIPS versions of generic (in the
286 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
287 * optimization: direct conversion of (unsigned-byte 32) to floats on
288 the PowerPC platform.
289 * optimization: structure instances with raw slots now use less
290 memory, and probably show better memory locality. (thanks to
292 * optimization: DYNAMIC-EXTENT declarations for lists and closures
293 are treated as requests for stack allocation on the x86-64,
294 Alpha-32, PPC and SPARC platforms.
295 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
296 the wrong file descriptor; implementation of SOCKET-OPEN-P.
297 (thanks to Tony Martinez)
299 ** gcing a dead thread can no longer lead to lockups
300 ** threads block signals until they are set up properly
301 ** errno is no longer shared by threads
302 ** interrupt-thread restores the eflags register on x86
303 ** fixed some lockups due to gc/thread interaction
304 * fixed some bugs revealed by Paul Dietz' test suite:
305 ** invalid dotted lists no longer raise a read error when
307 ** signal an error if a symbol that names a declaration is used
308 as the name of a type, or vice versa
309 ** allow using the (declare (typespec var*)) abbreviation for
310 (declare (type typespec var*)) with all type specifiers
311 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
312 ** always return NIL from PPRINT-POP when OBJECT is NIL
313 ** don't signal errors when pretty-printing malformed LABELS,
314 FLET or MACROLET forms
315 ** declarations in a DOLIST body are also in scope for the
317 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
319 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
322 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
323 * fixed cross-compiler leakages that prevented building a 32-bit
324 target with a 64-bit host compiler.
325 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
326 opened with :IF-EXISTS :OVERWRITE.
327 * fixed bug 281: error for an invalid qualifier in a short-form method
328 combination method is not signalled until the faulty method is called.
329 * bug fix: iteration variable type inferrer failed to deal with open
330 intervals. (reported by Alan Shields)
331 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
332 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
333 or more runtime options were provided to the sbcl binary.
334 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
335 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
336 * optimization: there's now a fast-path for fixnum arguments in the
337 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
338 * optimization: the code generated on x86-64 is more compact thanks
339 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
340 * bug fix: floating-point exception handling now partly works on
341 x86-64. (Thanks to James Knight)
342 * improvement to the MIPS backend from Thiemo Seufer: C-style
343 64-bit long long arguments and return values to alien functions
345 * fixed some bugs revealed by Paul Dietz' test suite:
346 ** the type-error signalled from WARN has a filled-in DATUM slot.
347 ** the type-error required when a stream is not associated with
348 a file has the stream as its datum.
349 ** type-errors on single-floats on x86-64 no longer have
350 :INVALID-OBJECT as the datum
351 ** the type-errors signalled for invalid function names now have
352 a correct expected type
353 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
354 for structures defined with a :TYPE. Documentation strings for
355 typed structures are no longer immediately discarded
356 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
357 broadcast streams and synonym streams. FILE-LENGTH now also works
358 on broadcast streams.
360 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
361 * incompatible change: the --noprogrammer option, deprecated since
362 version 0.7.5, has been removed. Please use the equivalent
363 --disable-debugger option instead.
364 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
366 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
367 inoperational for a while now) has been completely removed.
368 * null lexical environments are now printed as #<NULL-LEXENV>,
369 significantly reducing the amount of clutter in typical backtraces.
370 * documentation on weak pointers, finalization, and after GC hooks
371 has been added to the manual.
372 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
373 as other specialized array types, is much faster. SUBSEQ and
374 COPY-SEQ on such arrays have also been sped up.
375 * optimization: EQL is now more efficient when at least other argument
376 is known to be of type (OR FIXNUM (NOT NUMBER)).
377 * fixed bug: the runtime is now less vulnerable to changes in the
378 size of the SBCL object on OS X, and virtual memory is reserved for
379 all spaces that need to be at a fixed address.
380 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
381 * fixed bug: finalizers and after GC hooks that cause consing are now
383 * fixed bug: compiler error messages and summaries are now printed to
384 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
385 * fixed inference of the upper bound of an iteration variable.
386 (reported by Rajat Datta).
387 * fixed bug 211e: calling local functions with duplicated constant
388 keyword argument no longer causes a bogus style warning about an
390 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
391 ftype declaration to be disregarded. (reported by Dave Roberts)
392 * fixed bug 373: caused by erronous compilation of references to alien
393 variables in the runtime on ppc/darwin.
394 * fixed bug 376: CONJUGATE type deriver.
395 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
396 (reported by Baughn on #lisp)
397 * a cross-compiler bug on non-x86 platforms has been identified and
398 fixed. (thanks to Bruno Haible)
399 * improvements to the MIPS runtime code for increased stability.
400 (thanks to Thiemo Seufer)
401 * increased the maximimum compact environment size to allow
402 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
403 * partial workaround for bug 135: don't name the function we're
404 calling for hairy cases of EVAL, so as not to accumulate one environment
405 entry per call to EVAL. (reported by Kevin Reid)
406 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
407 * optimization: type testing for non-vector arrays should be faster.
408 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
410 * bug fix: setting 31st element of a bit vector to zero did not work
412 * bug fix: redefining a class definition which failed due to a
413 previous accessor / function clash now works (but see BUGS entry
414 #380 for more problems in this area). (thanks to Zach Beane)
415 * the long form of DEFINE-METHOD-COMBINATION disables method group
416 checking when given a single method group with pattern *. (thanks
418 * on x86 compiler supports stack allocation of results of simple
419 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
420 * support for building and running on Mac OS X 10.4 "Tiger" has been added
421 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
422 * fixed some bugs related to Unicode integration:
423 ** the restarts for recovering from input and output encoding
424 errors only appear when there is in fact such an error to
426 * fixed some bugs revealed by Paul Dietz' test suite:
427 ** MISC.549 and similar: late transformation of unsafe type
428 assertions into derived types caused unexpected code
430 ** SCALE-FLOAT type deriver is less wrong.
431 ** type derivers for EXP, LOG and similar functions compute result
432 types for complex arguments better.
433 ** (MISC.563) CONJUGATE type deriver works for very restricted
435 ** out-of-line type testers for character strings are available.
436 ** EQUAL compiler transform understands specialness of objects
438 ** accessing double-floats stored on the stack now works on x86-64.
439 ** debugger internals could sometimes create invalid lispobjs,
440 resulting in GC crashes.
441 ** MISC.548: type check weakening can convert required type into
443 ** initialization forms for bindings are not in scope of free special
446 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
447 * incompatible change: thread support for non-NPTL systems has
448 been removed - locking is buggy and unreliable. A threaded
449 SBCL build will now warn at startup and refuse to create
450 new threads, unless futex support is detected in the kernel
451 * incompatible change: the top level REPL now has only an ABORT
452 restart associated with it, not TOPLEVEL and ABORT as it used to.
453 TOP and TOPLEVEL are now available as debugger commands for
454 returning to the top level.
455 * incompatible change: forms evaluated in the REPL now use the
456 global optimization policy.
457 * incompatible change: user- and system-initialization files are
458 no longer processed with LOAD, but by READ and EVAL; hence the
459 global optimization policy, startup package, readtable, etc,
461 * The .fasl file format number has been incremented because of
462 various incompatible changes.
463 * internal entry point details and argument counts no longer appear
464 in backtraces unless explicitly requested by setting
465 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
466 * built-in and standard functions no longer have names like "top
467 level local call to FOO".
468 * fixed bug 32: functions defined in non-null lexical environments
469 now have more legible printed representation
470 * fixed bug 33: functions defined in non-null lexical environemnts
471 are now more amenable to inspection by INSPECT.
472 * workaround for bug 354: XEPs no longer appear in backtraces unless
473 explicitly requested.
474 * fixed bug: receiving the signal which results from attempting to
475 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
476 SIGBUS on other *BSDs) on architectures where the C stack is also
477 the Lisp stack (x86 and x86-64) from foreign code no longer leads
478 to debugger confusion or wild instability. (reported by Cheuksan
480 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
481 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
482 (reported by Lutz Euler)
483 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
484 compiled code. (reported by Paul Dietz)
485 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
486 specialized parameter to a method to something that is not TYPEP
487 the specializer is now possible.
488 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
489 face of package deletion.
490 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
491 pathname. (reported by David Tolpin on c.l.l)
492 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
493 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
494 STRING-STREAM and FILE-STREAM, respectively.
495 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
496 than previously; in addition, its test suite is now run on build.
497 (thanks to Luke Gorrie)
498 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
499 * optimization: added a immediate representation for single-floats
501 * fixed some bugs revealed by Paul Dietz' test suite:
502 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
503 ** attempting to create a package with a colliding nickname causes
504 correctable errors to be signalled.
505 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
506 ** several x86-64 backend bugs related to sign-extension of immediate
509 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
510 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
511 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
513 * fixed loading of multiply forward-referenced layouts.
514 (thanks to Cheuksan Wang)
515 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
516 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
517 (see COMPILE-FILE documentation for details of :PRINT option)
518 * fixed bugs 19 and 317: fixed-format floating point printing is
519 more accurate. This also fixes a bug reported by Adam Warner
520 related to the ~@F format directive.
521 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
523 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
524 dispatch table if the from-char is a dispatch macro character.
525 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
526 by the wordsize no longer ignore the last word. (reported by Lutz
528 * fixed bug in type checking of dynamic-extent variables. (reported
530 * optimization: sequence traversal functions use their freedom to
531 coerce function designators to functions.
532 * optimization: code with many calls to CLOS methods specialized on
533 CLOS classes has had redundant type checks removed.
534 * optimization: added declarations to speed up operations that access
535 the internal character database (for example STRING-UPCASE)
536 * optimization: comparison operations between floats and sufficiently small
537 fixnums no longer create extra rationals
538 * fixed some bugs related to Unicode integration:
539 ** portions of multibyte characters at the end of buffers for
540 character-based file input are correctly transferred to the
541 start of the buffer at the next read.
542 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
543 passing it through to OPEN.
544 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
545 argument given to internal calls to OPEN.
546 ** the built-in comment readers (introduced by character sequences
547 ";" and "#|") are more forgiving to encoding errors; they will
548 STYLE-WARN and then attempt to resync the stream at a character
549 boundary. (thanks to Teemu Kalvas)
550 * fixed some bugs in the x86-64 port:
551 ** Negative short int return values from c-calls are sign-extended
553 ** The stack is aligned to 16-bytes for c-calls, as required by
554 the ABI. (thanks to Cheuksan Wang)
555 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
556 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
557 for enums. (thanks to Vincent Arkesteijn)
558 ** Multiple small optimizations and bugfixes for floating point
560 * fixed some bugs revealed by Paul Dietz' test suite:
561 ** Space, Tab, Linefeed, Return and Page have the invalid
562 secondary constituent character trait.
563 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
565 ** WITH-INPUT-FROM-STRING should only update the index place on
567 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
569 ** Bit-array operations (BIT-AND and similar) worked incorrectly
570 with one-dimensional arrays with fill pointers.
571 ** TYPE-OF failed on a complex with an integer realpart and a
573 ** compiler failure during type inference for the code of form
574 (IF (EQL X (THE ...)) ...) (MISC.535).
576 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
577 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
578 platform. The port supports 61-bit fixnums, large memory spaces
579 and reloading shared object files.
580 * enhancement: saving cores with foreign code loaded is now
581 supported on ppc/Darwin in addition to the previously supported
583 * enhancement: the statistical profiler now walks deeper into the
584 call stack for more meaningful call-graphs and accrued time
585 reports (x86/x86-64 only). It also now reports time spent in
587 * enhancement: it is now possible to trace most individual methods
588 of a generic function in addition to tracing the generic function
590 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
592 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
593 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
594 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
595 produces an error. (thanks to Vincent Arkesteijn)
596 * bug fix: NAMESTRING on pathnames with :WILD components in their
597 directories works correctly. (thanks to Artem V. Andreev)
598 * fixed bug 125: compiler preserves identity of closures. (reported
600 * bug fixed: functions with &REST arguments sometimes failed with
601 "Undefined function" when compiled with (DEBUG 3). (reported by
603 * bug fix: overflow during compiling of setting element of a bit
604 vector with constant index and value. (reported by Timmy Douglas)
605 * build fix: fixed the dependence on *LOAD-PATHNAME* and
606 *COMPILE-FILE-PATHNAME* being absolute pathnames.
607 * on x86 compiler partially supports stack allocation of dynamic-extent
609 * GO and RETURN-FROM do not check the extent of their exit points
610 when compiled with SAFETY 0.
611 * fixed some bugs related to Unicode integration:
612 ** encoding and decoding errors are now much more robustly
613 handled; it should now be possible to recover even from invalid
614 input or output to the terminal. (thanks to Teemu Kalvas)
615 ** provided a first cut at implementing STRING-TO-OCTETS and
616 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
617 ** altered the SB-MD5 contributed module slightly, changing the
618 interface just enough for it to be supportable for builds where
619 lisp characters are not eight bits.
620 * fixed some bugs revealed by Paul Dietz' test suite:
621 ** the FORMATTER-generated functions for ~V[ conditionals require
622 the correct number of arguments.
623 ** READ-FROM-STRING returns the mandated second value when applied
624 to displaced strings.
625 ** the #\Rubout and #\Backspace characters are treated as invalid
626 constituent characters by the tokenizer.
628 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
629 * new feature: reloading changed shared object files with
630 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
631 * new feature: references to foreign variables and functions
632 can now be compiled and loaded before the corresponding shared
633 object file is loaded, as long as the foreign definitions are
634 available at runtime.
635 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
636 supported. (thanks to Dan Debertin)
637 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
638 just keywords. (thanks to Vincent Arkesteijn)
639 * fixed bug #331: structure-class instances corresponding to
640 DEFSTRUCT forms are now created eagerly.
641 * fixed bug #345: backtraces from calls to undefined functions work
642 on x86 as well. Related bug #61 is now also partially fixed on x86
643 (backtraces from throws to unknown catch tags.)
644 * bug fix: lambda-list parsing is now stricter vrt. order and number
645 of lambda-list keywords.
646 * bug fix: as specified by AMOP, an error is signalled if a
647 class-option appears multiple times in a DEFCLASS form. (reported
649 * bug fix: RANDOM can be compiled when the compiler derives the type
650 of its numeric argument as a disjoint set of small integers.
651 (reported by Paul Dietz)
652 * bug fix: starting a core saved with shared objects loaded when
653 those objects are not available no longer causes threaded SBCL to
654 hang. (reported by Sean Ross)
655 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
657 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
658 argument, which must always be NIL. (reported by Kalle Niemitalo)
659 * bug fix: printing 1.0d+23 no longer results in an error.
660 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
661 * bug fix: structure slot setters preserve evaluation order. (thanks
663 * fixed some bugs related to Unicode integration:
664 ** RUN-PROGRAM can allow its child to take input from a Lisp
665 stream. (reported by Stefan Scholl)
666 ** ASDF-INSTALL successfully downloads .tar.gz files in all
667 locales. (reported by Ken Causey)
668 * fixed some bugs revealed by Paul Dietz' test suite:
669 ** INCF, DECF and REMF evaluate their place form as specified in
671 ** FORMATTER expands ~{ ~} iteration directives with V or #
672 parameters correctly.
673 ** FORMATTER deals with the ~@[ ~] conditional directive where the
674 consequent uses no arguments correctly.
675 ** the system has a partial understanding of the (COMPLEX RATIO)
677 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
680 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
681 * new feature: a build-time option (controlled by the :SB-UNICODE
682 keyword feature, enabled by default) for building the system with
683 support for the entire 21-bit character space defined by the
685 * new feature: the system now has rudimentary external-format
686 support; the primary user-visible change at this time is that
687 characters with the high bit set (such as lower-case-e-acute) will
688 print correctly to a terminal in a UTF-8 environment.
689 * minor incompatible change: BASE-CHAR no longer names a class;
690 however, CHARACTER continues to do so, as required by ANSI.
691 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
692 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
693 has been moved to the SB-EXT package (temporarily re-exported from
695 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
696 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
698 * fixed bug #351: better error-handlind and reporting for malformed
700 * fixed bug #350: bignum-printing is now more memory-efficient,
701 allowing printing of very large bignums, eg. (expt 2 10000000).
702 (reported by Bruno Haible)
703 * fixed bug #302: better primitive-type selection for intersection
705 * fixed bug #308: non-graphic characters now all have names, as
706 required. (reported by Bruno Haible)
707 * bug fix: redefining a class with different superclasses now correctly
708 removes it from the direct-subclasses of its previous superclasses.
709 (reported by David Morse)
710 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
711 new value now works. (reported by Bruno Haible)
712 * bug fix: correct canonicalization of multiple non-standard slot
713 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
714 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
715 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
716 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
717 now exists, an signals an error.
718 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
719 during macroexpansion for non-iteration variables as well. (reported
720 by Bruno Haible for CMUCL)
721 * bug fix: Cyclic structures and unprintable objects in compiler
722 messages no longer cause errors. (reported by Bruno Haible)
723 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
724 and READ-FROM-STRING all now return a primary value of NIL if
725 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
726 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
727 (reported by Bruno Haible for CMUCL)
728 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
729 do not designate a condition. (reported by Bruno Haible for
731 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
732 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
733 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
734 permitting resolution in favour of any of the conflicting symbols.
735 (reported by Bruno Haible for CMUCL)
736 * bug fix: EQUAL compiler optimizations is less aggressive on
737 strings which can potentially compare true despite having distinct
738 specialized array element types.
739 * bug fix: unit enumerations can be defined without dividing by
740 zero. (thanks to Vincent Arkesteijn)
741 * FORMAT compile-time argument count checking has been enhanced.
742 (report from Bruno Haible for CMUCL)
743 * a partial workaround for the bug 262: the compiler does not try to
744 inline-expand a local function doing RETURN-FROM from a deleted
745 BLOCK. (thanks to Peter Denno for the bug report and to David
746 Wragg for the simple test case)
747 * fixed some bugs revealed by Paul Dietz' test suite:
748 ** RENAME-PACKAGE allows all package designators as new package
750 ** constraint propagation and lambda variable substitution are
751 more cautious in dealing with partially deleted code.
752 ** compiler sometimes lost reoptimization passes.
753 ** CERROR, given a condition as condition designator, passes its
754 remaining arguments to the continue format control without
756 ** Case-altering FORMAT directives work correctly on non-ASCII
758 ** The REAL type specifier handles bounds outside the floating
759 point ranges without signalling FLOATING-POINT-OVERFLOW.
760 ** Functions with IR1-transformations can create intercomponent
761 references to global functions.
762 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
764 ** FORMAT ~R treats a nil value for its first parameter correctly.
766 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
767 * enhancement: saving cores with foreign code loaded is now
768 supported on x86/NetBSD and sparc/Linux in addition to the previously
770 * bug fix: on some platforms repeated installations caused multiple
771 copies of HTML documentation to be installed -- should not happen
772 any more. (reported by Stefan Scholl)
773 * bug fix: parsing self-recursive alien record types multiple times
774 no longer causes infinite recursion. (reported by Thomas F. Burdick,
775 original patch by Helmut Eller for CMUCL)
776 * bug fix: stack-exhaustion detection works now on NetBSD as well.
777 (thanks to Richard Kreuter)
778 * bug fix: defining classes whose accessors are methods on existing
779 generic functions in other (locked) packages no longer signals
780 bogus package lock violations. (reported by François-René Rideau)
781 * bug fix: special variables as DEFMETHOD parameters no longer have
782 associated bogus type declarations. (reported by David Wragg and
784 * bug fix: read-write consistency on streams of element-type
785 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
786 * bug fix: redefiniton of the only method of a generic function with
787 no DEFGENERIC no longer emits a full WARNING. In addition,
788 redefinition of generic functions with no DEFGENERIC to an
789 incompatible lambda list now signals an error. (thanks to Zach
791 * bug fix: DEFGENERIC now works even when there's a function of the
792 same name in an enclosing lexical environment. (thanks to Zach
794 * fixed compiler failure, caused by instrumenting code during
795 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
796 * optimization: added loop analysis and improved register allocation
797 * fixed some bugs revealed by Paul Dietz' test suite:
798 ** POSITION on displaced vectors with non-zero displacement
799 returns the right answer.
800 ** (SIMPLE-STRING) is a valid type specifier for sequence
802 ** *PRINT-LEVEL* handling for slotless structures is pedantically
804 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
805 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
807 ** The justification version of the FORMAT ~< directive treats
808 non-zero minpad parameter correctly.
810 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
811 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
812 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
813 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
814 the supported interface.
815 * new feature: Single-stepping of code compiled with DEBUG 2 or
816 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
817 * new feature: saving cores with foreign code loaded is now
818 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
819 Timothy Moore's work for CMUCL)
820 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
821 parameters to * instead of NIL if no initform is supplied.
822 (reported by Johan Bockgaard)
823 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
824 correctly when FUNCALL appears as the car of the form. Note:
825 despite this FUNCALL forms are not currently subject to
826 compiler-macro expansion. (port of Raymond Toy's fix for the
827 same from CMUCL, reported by Johan Bockgaard)
828 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
829 (thanks for Teemu Kalvas)
830 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
831 case instead of terminating the iteration (thanks for Julian
832 Squires, Sean Champ and Raymond Toy)
833 * bug fix: incorrect expansion of defgeneric that caused a style
834 warning. (thanks for Zach Beane)
835 * bug fix: slot accessor effective method computation works properly
836 for classes with multiple non-standard applicable methods on
837 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
838 * on x86 compiler supports stack allocation of results of LIST and
839 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
840 CMUCL implementation by Gerd Moellmann)
841 * fixed some bugs revealed by Paul Dietz' test suite:
842 ** FORMAT strings with both the ~<~:;~> form of the justification
843 directive and pretty-printing directives cause an error.
845 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
846 * incompatible change: the internal functions
847 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
848 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
849 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
850 instead of the old functions.
851 * new feature: on platforms where "dladdr" is available foreign
852 function names now appear in backtraces. (based on Helmut Eller's
854 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
855 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
857 * documentation: ASDF manual is now distributed with SBCL.
858 * bug fix: SBCL can now load its contributed modules with REQUIRE
859 even if the system-provided entries have been removed from
860 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
862 * bug fix: backtraces involving undefined functions or assembly
863 routines are more informative. (thanks to Brian Downing)
864 * bug fix: mutually referent alien structures now work correctly.
865 (reported by Rick Taube)
866 * bug fix: structures defined by WITH-ALIEN can be referred to
867 within other definitions in the same WITH-ALIEN.
868 * bug fix: division operators (MOD, TRUNCATE and the like) with
869 constant zero divisors and integer dividends no longer generate
871 * bug fix: provide default methods for INPUT-STREAM-P and
872 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
873 * bug fix: improve the detection and resolution of MOP metacycles
874 (where effective method computation depends on the generic function
875 having its effective method computed). (reported by Bruno Haible)
876 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
877 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
878 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
879 * bug fix: pathnames with relative directory components can be
880 represented relative to default pathnames.
881 * optimization: in taking the GCD of bignums, reduce the two bignums
882 to approximately the same size (using Euclid's algorithm) before
883 applying the more sophisticated binary GCD. (thanks to Juho
885 * optimization: COUNT on bitvectors now operates word-at-a-time.
886 * optimization: ASH with a positive, but not necessarily constant,
887 (leftwards) shift, when in a modular context, is compiled to a
889 * fixed some bugs revealed by Paul Dietz' test suite:
890 ** FORMAT variable parameters ("~V<char>") are defaulted properly
891 if the corresponding argument is NIL.
892 ** FORMAT directives accepting a minpad parameter treat negative
894 ** literal commas as character parameters to FORMAT directives are
896 ** literal spaces directly after ~<Newline> directives within a
897 format-logical-block (~:< ~@:>) do not induce :FILL-style
898 conditional newlines.
899 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
900 as specified: it no longer includes conditional newlines).
901 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
903 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
904 :PER-LINE-PREFIX argument does not evaluate to a string.
906 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
907 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
908 the manual for details; for now, package locks can be disabled by
909 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
910 if no major problems are found then it is likely that they will be
911 compiled in unconditionally.
912 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
913 unsupported operators on all platforms. To load a shared library
914 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
915 object file, link it into a shared library outside of SBCL and
916 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
918 * bug workaround: ROOM T and the GC were not working together
919 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
920 reporting facility aren't preserved. That reporting has been
921 disabled, so now until and unless someone figures out how to make it
922 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
923 * minor incompatible change: as threatened around sbcl-0.8.0, the
924 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
925 The symbols are also exported from SB-PCL for backwards
926 compatibility, but more so than before SB-PCL should be treated as
927 an implementation-internal package.
928 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
930 * the SB-SPROF contrib now works on (most) non-x86 architectures.
931 It is known as of this release not to work on the Alpha, however.
932 * fixed bug #167: errors signalled due to illegal syntax in method
933 bodies are now more legible.
934 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
935 designators and can hence be used with TYPEP.
936 * fixed bug #333: CHECK-TYPE now ensures that the type error
937 signalled, if any, has the right object to be accessed by
938 TYPE-ERROR-DATUM. (reported by Tony Martinez)
939 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
940 5.1.2.3 for multiple-value place subforms. (reported by Kalle
942 * fixed bug #334: programmatic addition of slots using specialized
943 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
944 and :ALLOCATION :CLASS slots.
945 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
946 float exponents. (rereported by Peter Seibel)
947 * fixed a bug: #\Space (and other whitespace characters) are no
948 longer considered to be macro characters in standard syntax by
950 * fixed bug: initialization of condition class metaobjects no longer
951 causes an instance of the condition to be created. (reported by
953 * fixed bug: it is now possible to have slots such that
954 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
955 description is neither :INSTANCE nor :CLASS.
956 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
957 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
958 SLOT-VALUE-USING-CLASS), and no longer causes errors with
959 non-standard SLOT-DEFINITION-ALLOCATION values.
960 * fixed bugs: various race conditions handling exiting threads.
961 CL-PPCRE's thread torture test now passes.
962 * fixed arguably-a-bug: GC time proportional to number of threads in
963 system even when most of them are idle
964 * optimization: improved performance of BIT and SBIT on bit-vectors.
965 * .fasl file incompatibility: The fasl file version number has
966 been incremented because of changes associated with package locks.
968 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
969 * minor incompatible change: the system no longer provides
970 optimization hints (conditions which are TYPEP
971 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
972 mode; these hints are still emitted when higher SPEED optimization
974 * new contrib module: a sampling profiler (profiling by statistical
975 sampling, rather than by instrumenting functions) is available as
976 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
977 * the behaviour of the standard function ED is now customizeable by
978 third parties through a hook variable: see ED's documentation
979 string for information on the protocol.
980 * the compiler no longer emits efficiency notes for (FUNCALL X)
981 when the type of X is uncertain under default optimization
983 * fixed bug 276: mutating a binding of a specialized parameter to a
984 method to something that is not TYPEP the specializer is now
986 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
987 exhibit the required behaviour on the x86 platform. (thanks to
988 Peter van Eynde, Eric Marsden and Bruno Haible)
989 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
990 even for difficult arguments. (reported by Peter Graves)
991 * fixed bug 141a: the backquote printer now descends quoted
993 * fixed another bug in backquote printing: no more destructive
994 modification of the form's list structure. (reported by Brian
996 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
997 move between its address being taken and the call to
998 interrupt_thread, fixing a crashing race condition.
999 * the SB-POSIX contrib implementation has been adjusted so that it
1000 no longer exhibits ridiculously poor performance when constructing
1001 instances corresponding to C structs.
1003 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1004 * minor incompatible change: the sb-grovel contrib now treats C
1005 structures as alien (in the sense of SB-ALIEN) objects rather than
1006 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1007 has implications for memory management of client code
1008 (sb-grovel-returned objects must now be manually managed) and for
1009 type safety (alien objects now have full types).
1010 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1011 used to control emission of compiler diagnostics, rather than the
1012 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1013 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1014 quality should be considered deprecated.
1015 * install.sh now installs the user manual as well
1016 * (not quite a new documentable feature, but worth considering in
1017 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1018 beginnings of a semantically meaningful condition hierarchy is
1019 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1021 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1022 designator as the defaults argument.
1023 * fixed bug: Displaced arrays whose displaced-to array has become
1024 too small now cause ARRAY-DIMENSION to signal an error, providing
1025 for safer bounds-checking. (reported by Bruno Haible)
1026 * fixed bug: DEFCLASS slot definitions with identical :READER and
1027 :WRITER names now signal a reasonable error. (reported by Thomas
1029 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1031 * fixed bug: Invalid slot specification errors now print correctly.
1032 (thanks to Zach Beane)
1033 * fixed bug 320: Shared to local slot value transfers in class
1034 redefinitions now happen corrently from superclasses as
1035 well. (reported by Bruno Haible)
1036 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1038 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1039 improper lists correctly. (reported by Bruno Haible)
1040 * fixed bug 313: source-transform for <fun-name> was erroneously
1041 applied to a call of a value of a variable with name <fun-name>.
1042 (reported by Antonio Menezes Leitao)
1043 * fixed bug 307: The obsolete instance protocol ensures that
1044 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1045 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1046 remove dead unknown-values globs from the middle of the stack.
1047 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1048 easier to resume long computations after using *BREAK-ON-SIGNALS*
1049 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1050 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1051 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1052 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1053 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1054 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1056 * fixed a (fairly theoretical) bug in string printing: if
1057 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1058 does not have array-element-type equal to the most general string
1060 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1061 function name. (thanks to Nikodemus Siivola)
1062 * fixed bug: compile-time format string checker failed on
1063 non-closed ~{. (reported by Thomas F Burdick)
1064 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1065 don't warn on legal loop code involving more than one aggregate
1066 boolean. (thanks to Nikodemus Siivola)
1067 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1068 in keyword position to MAKE-PACKAGE was not being reported as
1070 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1071 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1073 * fixed bugs 280 and 312: the checking for multiple definitions in a
1074 file is less likely to become confused by uses of inline
1076 * fixed bug: the #S reader macro performs the keyword coercion
1077 specified for slot names. (reported by Kalle Niemitalo)
1078 * fixed bug: lambda lists may contain symbols whose names start with
1079 & but are not lambda-list-keywords; their occurrence triggers a
1081 * fixed bug 321: define-method-combination argument lambda lists do
1082 not cause invalid code to be generated when &optional or &aux
1083 variables have default values. (reported by Bruno Haible)
1084 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1085 have CLOS classes; this appears also to have fixed reported
1086 instability in stack exhaustion detection.
1087 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1088 MUFFLE-WARNING when no associated restart is present is now
1090 * optimization: rearranged the expansion of various defining macros
1091 so that each expands into only one top-level form in a
1092 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1094 * optimization: used a previously little-used slot in symbols to
1095 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1097 * fixed some bugs revealed by Paul Dietz' test suite:
1098 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1100 ** Fixed an optimization bug related to inheritance of initforms
1101 from local to shared slots.
1102 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1103 ** CAST optimizer forgot to flush argument derived type.
1104 ** print/read consistency on floats is now orders of magnitude
1105 more likely. (thanks also to Bruno Haible for a similar report
1107 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1108 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1109 for objects of type REAL. Make it so.
1110 ** SXHASH is specified (infelicitously) to respect similarity,
1111 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1112 it so. (thanks to Markus Ziegler)
1113 ** on the Alpha, the compiler succeeds in compiling functions
1114 returning a known number of arguments greater than 63.
1115 ** fixed handling of invalid NIL arguments in keyword position
1117 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1119 ** provided a readably-printable representation for RANDOM-STATE
1121 ** ensured that pathnames, where they have a namestring, always
1122 print using #P"..." syntax.
1124 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1125 * Support for the forthcoming 2.0 version of the NetBSD kernel
1126 running on x86 hardware has been added. (thanks to Perry
1127 E. Metzger most immediately, and others for their past work)
1128 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1129 current release version; SBCL's OpenBSD support had been broken
1130 since about the time of OpenBSD's switch to ELF binary format.)
1131 * [placeholder for DX summary]
1132 ** user code with &REST lists declared dynamic-extent, under high
1133 speed or space and low safety and debug optimization policy.
1134 * The manual has been converted to Texinfo format and the debugger
1135 chapter from the cmucl manual has been added.
1136 * A facility has been added to extract documentation strings from
1137 sbcl and store them as Texinfo-formatted snippets for inclusion in
1138 the manual (via Texinfo's @include directive)
1139 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1140 the test case to Dave Roberts)
1141 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1142 pointers. (reported by Sean Ross)
1143 * bug fix: PROFILE output is printed nicely even for large numerical
1144 values. (thanks to Zach Beane)
1145 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1146 greater than 32 handle EOF correctly.
1147 * bug fix: on X86 an immediate argument of the IMUL instruction is
1148 correctly printed in disassembly. (thanks to Lutz Euler)
1149 * bug fix: class slots in redefined classes preserve their old
1150 values. (thanks to Bruno Haible and Nikodemus Siivola)
1151 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1152 succeeds. (reported by Marco Baringer)
1153 * fixed some bugs revealed by Paul Dietz' test suite:
1154 ** READ-SEQUENCE now works on ECHO-STREAMs.
1155 ** RATIONALIZE works more according to its specification. (thanks
1158 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1159 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1160 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1161 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1162 should matter to you only if you rebind the printer control
1163 variables and then find you want different bindings in the
1164 debugger than in the ordinary execution of your program.)
1165 * The runtime build system has been tweaked to support building
1166 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1167 assembler and linker. (thanks to Nikodemus Siivola)
1168 * Unbound, undefined, undeclared variables now trigger full
1169 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1170 more often programmer error than deliberate exploitation of undefined
1172 * optimization: the hash algorithm for strings has changed to one
1173 that is less vulnerable to spurious collisions. (thanks to Juho
1175 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1176 needless bounds checking. (thanks to Juho Snellman)
1177 * optimization: implemented multiplication as a modular
1178 (UNSIGNED-BYTE 32) operation on the PPC backend.
1179 * fixed some bugs revealed by Paul Dietz' test suite:
1180 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1182 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1183 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1184 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1186 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1188 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1190 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1191 their output stream on EOF from read.
1192 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1193 have been read to end-of-file.
1194 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1196 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1197 description of determination of which consecutive characters
1199 ** Printing the "Space" character with escaping on now yields "#\\ ",
1200 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1201 ** Reading floating-point numbers with *READ-BASE* set to a number
1202 less than 10 works correctly.
1203 ** Reading floating-point numbers with *READ-BASE* set to a number
1204 more than 10 works correctly.
1205 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1206 the readtable currently in effect.
1208 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1209 * minor incompatible change: parsing of namestrings on a physical
1210 (Unix) host has changed; numbers after the final #\. in a
1211 namestring are no longer interpreted as a version field. This is
1212 intented to be largely invisible to the user, except that the
1213 meaning of the namestring "*.*.*" has changed: it now refers to a
1214 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1215 should usually be replaced by
1216 (make-pathname :name :wild :type :wild :version :wild)
1217 with the added benefit that this is more likely to be portable.
1218 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1219 option to OPEN now signals an error if the file being opened
1220 exists; this may have an impact on existing code.
1221 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1222 signals received, on the PowerPC platforms (both Linux and
1223 Darwin). (thanks to Pierre Mai for pointing out the location of
1225 * several fixes on OS X: The system now builds and runs cleanly on
1226 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1227 (thanks to Brian Mastenbrook)
1228 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1229 second-resolution: integer multiples of 1/3600 between -24 and 24.
1230 (thanks to Vincent Arkesteijn)
1231 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1232 type when called with 1 argument; PEEK-CHAR checked type of
1233 PEEK-TYPE only after having read first character from a
1234 stream. (reported by Peter Graves)
1235 * bug fix: the garbage collector now has much better locality
1236 behaviour, and in particular no longer treats all memory as being
1237 exhausted when many small objects point to each other in a deeply
1239 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1240 recognized as being TYPEP their class.
1241 * bug fix: the PUSHNEW documentation string has been corrected.
1242 (thanks to Vincent Arkesteijn)
1243 * bug fix: defaulting of the value for the last of an atypically
1244 large number of multiple values being bound was not being
1245 performed correctly on the Alpha or PPC platforms
1246 * optimization: implemented multiplication as a modular
1247 (UNSIGNED-BYTE 32) operation on the x86 backend.
1248 * optimization: SEARCH on simple-base-strings can now be open-coded.
1249 (see also contrib/compiler-extras.lisp for inspiration for
1250 teaching the compiler about the Boyer-Moore algorithm).
1251 * value, returned by MAX (and MIN) called with several EQUALP, but
1252 not EQL, arguments now does not depend on compiler settings.
1253 * fixed some bugs revealed by Paul Dietz' test suite:
1254 ** in stack analysis liveness information is propagated from
1255 non-local entry points.
1256 ** pathwise CAST removing failed when the CAST node did not start
1258 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1259 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1261 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1262 host is already defined.
1263 ** RENAME-FILE works on streams instead of signalling an internal
1265 ** PEEK-CHAR uses the current readtable when determining whether
1266 or not a character is whitespace.
1267 ** MERGE-PATHNAMES handles the case when the pathname does not
1268 specify a name while the default-pathname specifies a version.
1269 ** Pathnames now stand a better chance of respecting print/read
1271 ** Attempting to use standardized file system operators with a
1272 pathname with invalid :DIRECTORY components signals a
1274 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1276 ** DIRECTORY on logical pathnames is more correct.
1277 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1278 signal a TYPE-ERROR if their argument is not a stream.
1279 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1280 designator argument does not designate a stream.
1281 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1282 examining the synonym.
1283 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1285 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1286 element-type larger than ([UN]SIGNED-BYTE 32).
1288 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1289 * When built with the :SB-FUTEX feature, threaded builds now take
1290 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1291 for faster/more reliable mutex and condition variable support.
1292 * Incompatible change (but one you probably shouldn't have been using
1293 anyway): the interface and code for arbitrating between multiple
1294 threads in the same user session has been redesigned.
1295 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1296 error when called without an explicit environment argument.
1297 (thanks to Nikodemus Siivola)
1298 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1299 confused about their position. (thanks to Adam Warner and Gerd
1301 * bug fix: Pretty printing backquoted forms with unquotations in the
1302 argument list position of various code constructs such as LAMBDA
1303 now works correctly. (reported by Paul Dietz)
1304 * bug fix: Pretty printing unquotations no longer loses all
1305 stream position information.
1306 * optimization: performance of string output streams is now less
1307 poor for multiple small sequence writes.
1308 * optimization: performance of CSUBTYPEP in the presence of complex
1309 expressions involving CONS and NOT many times has been improved.
1310 (reported by Paul Dietz)
1311 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1313 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1314 streams for the REPL, for the debugger, and so on. (thanks to
1316 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1317 * fixed some bugs revealed by Paul Dietz' test suite:
1318 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1319 is ignored if it is too small, rather than propagating through
1320 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1321 ** extremely complex negations of CONS types were not being
1322 sufficiently canonicalized, leading to inconsistencies in
1324 ** VALUES tranformer lost derived type.
1326 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1327 * fixed a bootstrapping bug: the build process no longer assumes
1328 that the various BOOLE-related constants have the same value in
1329 host and target lisps. (noted by Paul Dietz' test suite on an
1330 SBCL binary built from CLISP)
1331 * The system can now be dynamically linked on the MIPS platform,
1332 which enables dynamic loading of foreign code from Lisp. (thanks
1333 to Ralf Baechle for discussions on the MIPS ABI)
1334 * The system now records debugging information for its own source
1335 files in a filesystem-position-independent manner, relative to
1336 the "SYS" logical host.
1337 * fixed a compiler bug: MV-LET convertion did not check references
1338 to the "max args" entry point. (reported by Brian Downing)
1339 * tweaked disassembly notes to be less confident about proclaiming
1340 some instruction as an LRA. (thanks to Brian Downing)
1341 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1342 now each have their own history, command character, and other
1343 characteristics. (thanks to David Lichteblau)
1344 * fixed some bugs revealed by Paul Dietz' test suite:
1345 ** compiler failure in compiling LOGAND expressions including a
1347 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1348 shift greater than 32.
1349 ** FLUSH-DEST did not mark blocks for type check regeneration.
1350 ** HANDLER-CASE failed to accept declarations in handler clauses
1351 in some circumstances.
1353 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1354 * New code in contrib/sb-introspect (still probably not entirely
1355 stable yet) provides some support for smart Lisp development
1356 environments like SLIME.
1357 * The conditions signalled for errors occurring when loading .fasl
1358 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1359 in a way which should help ASDF recover gracefully.
1360 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1361 stuff has been cleaned up. If you code contrib/ stuff, this might
1362 affect you, and you can look at contrib/README, contrib/STANDARDS,
1363 and/or the 0.8.4.27 diff to check.
1364 * In full calls the compiler now does not generate checks for declared
1365 argument types for all arguments.
1366 * various threading fixes
1367 ** and some experimental patches which didn't make it into
1368 the main tree for this release, but which are shipped in
1369 contrib/experimental-thread.patch as a possible fix for some
1370 failures (deadlock, spinning...) in GC-intensive multithreaded
1372 * fixed PPC build problem (source code incompatibility of different
1373 library versions): added offsetof() hackery which attempts to divine
1374 where glibc maintainers put uc_mcontext today
1375 * fixed bug 282: compiler does not trust type assertions while passing
1376 arguments to a full call.
1377 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1378 &OPTIONAL VALUES type parameter.
1379 * fix bug 214: algorithm for noting rejected templates is now more
1380 similar to that of template seletion. (also reported by rydis on
1382 * fixed bug 141b: printing backquoted information readably and prettily
1383 inserts a space where necessary.
1384 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1385 (CHARACTER 10) are now reported as errors, rather than propagated
1386 as unknown types. (reported by piso on #lisp)
1387 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1388 with values NIL and :ERROR. (thanks to Milan Zamazal)
1389 * fixed bug 191c: CLOS now does proper keyword argument checking as
1390 described in CLHS 7.6.5 and 7.6.5.1.
1391 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1392 longer raise an error; further, using a list as a for-as-arithmetic
1393 counter now raises a meaningful error.
1394 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1395 now understood by sequence creation functions such as MAKE-SEQUENCE
1397 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1398 a STRING-INPUT-STREAM.
1399 * compiler enhancement: SIGNUM is now better able to derive the type
1401 * type declarations inside WITH-SLOTS are checked. (reported by
1403 * fixed some bugs revealed by Paul Dietz' test suite:
1404 ** incorrect optimization of TRUNCATE for a positive first
1405 argument and negative second.
1406 ** compiler failure in let-convertion during flushing dead code.
1407 ** compiler failure while deriving type of TRUNCATE on an
1408 interval, containing 0.
1409 ** ASH of a negative bignum by a negative bignum count now returns
1411 ** intersection of CONS types now canonicalizes properly, fixing
1412 inconsistencies in SUBTYPEP.
1414 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1415 * incompatible change: The --disable-debugger command line
1416 option now clobbers the debugger at a more fundamental
1417 level, by redefining #'INVOKE-DEBUGGER instead of by
1418 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1419 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1420 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1421 enter the debugger REPL and then suffer recursive errors
1422 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1423 while under the new system, BREAK in --disable-debugger mode
1424 terminates the system just as an unhandled error would.
1425 * fixed compiler performance when processing loops with a step >1;
1426 * bug fix: DOCUMENTATION now retrieves generic function
1427 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1428 support has been systematized, and now supports the methods
1429 specified by ANSI, along with a default method and a method for
1430 slot documentation. (reported by Nathan Froyd)
1431 * bug fix: effective methods associated with a generic function are
1432 no longer cached over a change of that generic function's method
1433 combination. (reported by Andreas Fuchs)
1434 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1435 names a non-existent directory. (thanks to Andreas Fuchs)
1436 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1437 the PPC platform, lead to stack corruption; this has been fixed.
1438 (reported by Rainer Joswig)
1439 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1440 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1441 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1442 arithmetic in general, is now much more reliable on the PPC
1444 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1445 platform now returns the right answer.
1446 * optimization: restored some effective method precomputation in
1447 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1448 precomputation is now tunable.
1449 * optimization: compiler-internal data structure use has been
1450 reviewed, and changes have been made that should improve the
1451 performance of the compiler by about 20%.
1452 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1453 simple-base-strings and simple-bit-vectors is improved.
1454 * optimization: the optimization of 32-bit logical and arithmetic
1455 functions introduced in version 0.8.3 on the x86 has been
1456 implemented on the mips, ppc and sparc platforms; an
1457 implementation of the same facility, but for 64-bit arithmetic,
1458 has been added for the alpha.
1459 * microoptimization: the compiler is better able to make use of the
1460 x86 LEA instruction for multiplication by constants.
1461 * bug fix: in some situations compiler did not report usage of
1462 generic arithmetic in (SPEED 3) policy.
1463 * bug 145b fix: compiler used wrong type specifier while converting
1464 MEMBER-types to numeric.
1465 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1467 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1468 index argument now works properly on the Alpha platform.
1469 * bug fix: floating point exception treatment on the Alpha platform
1471 * bug fix: FILE-POSITION works much better on string input and
1472 output streams. (thanks to Nikodemus Siivola)
1473 * bug fix: many threading/garbage collection symptoms sorted.
1474 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1475 might be pseudo-atomic.
1476 * internal change: Stopping for GC is now done with signals not
1477 ptrace. GC is now done in whichever thread wanted it, instead of
1479 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1481 * bug fix: result form in DO is not contained in the implicit
1483 * incompatible change: ICR structure is changed; the value part of
1484 CONTINUATION is now called LVAR; corresponding functions are
1485 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1486 * added type deriver for ISQRT (thanks to Robert E. Brown).
1487 * bug fix: better support for loading from the command line when an
1488 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1490 * fixed some bugs revealed by Paul Dietz' test suite:
1491 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1492 ** ROUND and FROUND now give the right answer when given very
1493 small float arguments.
1494 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1496 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1497 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1498 ** LCM should return a non-negative integer.
1499 ** PARSE-INTEGER returned the index of a terminator instead of the
1500 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1501 ** PARSE-INTEGER returned an incorrect index being applied to a
1503 ** LCM with two arguments of 0 returns 0 rather than signalling
1505 ** unsigned addition of a 32-bit constant with the high bit set no
1506 longer causes an internal compiler error on the x86.
1507 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1508 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1509 with negative last argument.
1510 ** byte specifiers with zero size and position no longer cause
1511 an error during type derivation.
1512 ** bignum multiplication on the Alpha platform now returns the
1514 * porting: The system now builds on SuSE AMD64, although it still
1515 generates a 32-bit binary.
1516 * .fasl file incompatibility: The fasl file version number has
1517 been incremented (because of the changes to internal compiler
1518 data structures referred to above).
1520 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1521 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1522 more accurately, on the Darwin kernel running on PowerPC hardware.
1523 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1524 * Compiler code deletion notes now signal a condition of type
1525 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1526 an associated MUFFLE-WARNING restart.
1527 * The compiler now performs limited argument count validation of
1528 constant format strings in FORMAT, and where appropriate in ERROR,
1529 CERROR and WARN. (thanks to Gerd Moellmann)
1530 * New ASDF-INSTALL contrib can be used for automatic download and
1531 installation of third-party Lisp code from CCLAN or other sites
1533 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1534 which forces another thread to execute a function supplied by the
1536 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1537 now accept and act upon their :ELEMENT-TYPE keyword argument.
1538 (reported by Martin Atzmueller, Edi Weitz)
1539 * bug fix: FILE-POSITION now accepts position designators up to
1540 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1541 is the greater. (thanks to Patrik Nordebo)
1542 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1543 Kalle Olavi Niemitalo)
1544 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1545 printable. (reported by Eric Marsden)
1546 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1547 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1548 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1549 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1550 (thanks to Stig Sandoe)
1551 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1552 UNREAD-CHAR now correctly includes the unread character in the
1553 target sequence. (thanks to Gerd Moellmann)
1554 * bug fix in threaded builds: the system can now be suspended and
1555 resumed by shell job control with minimal disruption.
1556 * bug fixes in times and timezones >2038AD
1557 * better handling of "where is GNU make?" problem in build scripts
1558 (thanks to Nikodemus Siivola)
1559 * new optimization: inside a named function any reference to a
1560 function with the same name is considered to be a self-reference;
1561 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1562 optimization quality.
1563 * new optimization on x86: logical functions and + now have
1564 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1565 used when the result is truncated to 32 bits.
1566 * VALUES declaration is partially enabled.
1567 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1568 * fixed some bugs revealed by Paul Dietz' test suite:
1569 ** The system now obeys the constraint imposed by
1570 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1571 types form a lattice under type intersection.
1572 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1573 ** ASSOC now ignores NIL elements in an alist.
1574 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1575 and (1+ MOST-POSITIVE-FIXNUM) answers.
1576 ** The addition of a method with invalid qualifiers to a generic
1577 function does not cause an error to be signalled immediately;
1578 a warning is signalled, and the error is generated only on
1579 calling the generic function.
1580 * changed .fasl file version number, in part to add type codes for
1581 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1582 obscure ANSI requirements
1584 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1585 * fixed bug 148: failure to inline-expand a local function left
1586 garbage, confusing the compiler.
1587 * fixed bugs 3cd: structure slot readers perform type check if the
1588 slot can have an invalid value (i.e. it is either not initialized
1589 or can be written with a less specific slot writer).
1590 * bug fix: the compiler now traps array references to elements off
1591 the end of an array; previously, the bounds checking in some
1592 circumstances could go off-by-one.
1593 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1595 * type declarations for array element types now obey the description
1596 on the CLHS page "Declaration TYPE", as per discussions on
1597 sbcl-help around 2003-05-08. This means that a declaration
1598 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1599 declaration, all references to BAR will be asserted or assumed
1600 (with THE, so dependent on compiler policy) to involve objects of
1601 type FOO. Note that no such declaration is implied in
1602 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1603 * declared types of functions from the "Conditions"
1604 chapter. (reported by Paul Dietz)
1605 * bug fix: CERROR accepts a function as its first argument.
1606 * bug fix: NTH an NTHCDR accept a bignum as index
1607 arguments. (reported by Adam Warner)
1608 * optimization: character compare routines now optimize comparing
1609 against a constant character. (reported by Gilbert Baumann)
1610 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1611 argument now works properly on the MIPS platform.
1612 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1613 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1614 declared to have a functional type, does not cause an error
1616 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1617 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1618 * fixed bug: if last continuation of a deleted block has a
1619 destination, this destination should be deleted too. (reported by
1621 * fixed a bug in the bootstrap process: the host compiler's values
1622 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1623 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1624 test case from Patrik Nordebo)
1625 * improved the ability of the disassembler on the PPC platform to
1626 provide helpful disassembly notes.
1627 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1628 the class in more cases than previously.
1629 * bug fix: FILE-POSITION now understands :START and :END for
1630 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1631 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1632 8) data. (thanks to David Lichteblau)
1633 * bug fix: it is possible to add a method to a generic function
1634 without lambda list.
1635 * bug fix: reader failed to signal END-OF-FILE inside an
1636 object representation. (reported by Nikodemus Siivola)
1637 * fixed some bugs revealed by Paul Dietz' test suite:
1638 ** LAST and [N]BUTLAST should accept a bignum.
1639 ** condition slot accessors are methods.
1640 ** (VECTOR NIL) is a subtype of STRING.
1642 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1643 * minor incompatible change: some nonsensical specialized lambda
1644 lists (used in DEFMETHOD) which were previously ignored now signal
1646 * minor incompatible change: the system is now aware of the types of
1647 variables in the COMMON-LISP package, and will signal errors for
1648 most violations of these type constraints (where previously they
1649 were silently accepted).
1650 * minor incompatible change: COMPILE-FILE now uses the freedom
1651 afforded (ANSI 3.2.2.3) to use derived function types for
1652 functions defined in the same file. This also permits the system
1653 to warn on static type mismatches and function
1654 redefinition. (Currently it does not work with high DEBUG level.)
1655 * minor incompatible change: VALUES declaration is disabled.
1656 * When issuing notes, the compiler now signals a condition of type
1657 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1658 restart for use in user handlers. It is expected that the
1659 COMPILER-NOTE condition will eventually become a condition
1660 supertype to a hierarchy of note types, which will then be
1661 handleable in a similar fashion. However, at the moment, no such
1662 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1663 but it's an implementation detail, not a classification for the
1665 * Changes in type checking closed the following bugs:
1666 ** type checking of unused values (192b, 194d, 203);
1667 ** template selection based on unsafe type assertions (192c, 236);
1668 ** type checking in branches (194bc).
1669 * A short form of VALUES type specifier has ANSI meaning (it has
1670 increased the number of situations when SBCL cannot perform type
1672 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1673 keywords or constants is permissible.
1674 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1675 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1676 classes. (thanks to Antonio Martinez)
1677 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1678 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1679 argument to a pretty-print stream. Instead, it leaves any such
1680 operations to DESCRIBE-OBJECT methods.
1681 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1683 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1684 lambda lists. (thanks to David Lichteblau)
1685 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1686 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1687 (thanks to Gerd Moellmann)
1688 * fixed bug 246: increased compilation speed of long
1689 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1691 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1693 * DEFCONSTANT now throws a condition of type
1694 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1695 constant to a non-EQL value; CONTINUE and ABORT restarts
1696 respectively change and preserve the value.
1697 * fixed bug 63: The code walker, part of the implementation of CLOS,
1698 is now better at handling symbol macros.
1699 * bug fix: There is no longer an internal implementation type named
1700 CL:LENGTH. (reported by Raymond Toy)
1701 * bug fix: In macro-like defining macros/special operators the
1702 implicit block does not enclose the lambda list.
1703 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1704 enclosed in parentheses) are not suitable as type specifiers, and
1705 their use properly signals an error now.
1706 * bug fix: An argument count mismatch for a type specifier in code
1707 being compiled no longer causes an unhandled error at compile
1708 time, but signals a compile-time warning.
1709 * fixed simple vector readable printing
1710 * bug fix: DESCRIBE takes more care over whether the class
1711 precedence list slot of a class is bound before accessing it.
1712 (reported by Markus Krummenacker)
1713 * bug fix: FORMATTER can successfully compile pretty-printer format
1714 strings which use variants of the ~* directive inside.
1715 * bug fix: SEARCH now applies its TEST predicate to the elements of
1716 the arguments in the correct order. (thanks to Wolfhard Buss)
1717 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1718 in type inference. (thanks to Robert E. Brown)
1719 * bug fix: Reading in symbols with an explicit package name of ""
1720 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1721 * fixed some bugs revealed by Paul Dietz' test suite:
1722 ** NIL is now allowed as a structure slot name.
1723 ** Arbitrary numbers, not just REALs, are allowed in certain
1724 circumstances in LOOP for-as-arithmetic clauses.
1725 ** Multiple class redefinitions before slot access no longer
1726 causes a type error.
1727 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1728 association between the name and a class.
1729 ** Generic functions with non-standard method-combination and over
1730 five methods all of which return constants no longer return NIL
1731 after the first few invocations. (thanks to Gerd Moellmann)
1732 ** CALL-NEXT-METHOD with no arguments now passes the original
1733 values of the arguments, even in the presence of assignment.
1734 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1736 ** DISASSEMBLE works with closures and funcallable instances.
1737 ** ADD-METHOD now returns the generic function, not the new method.
1738 ** FIND-METHOD signals an error if the lengths of the specializers
1739 is incompatible with the generic function, even if the ERRORP
1741 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1742 which its argument is a member.
1743 ** DEFCLASS only redefines the class named by its class-name
1744 argument if that name is the proper name of the class;
1745 otherwise, it creates a new class.
1746 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1747 of the UNBOUND-SLOT condition to the name of the slot.
1748 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1749 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1750 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1751 SLOT-MAKUNBOUND in the specified fashion.
1753 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1754 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1755 cross-compilation host. As a consequence, we can now bootstrap our
1756 way up to SBCL starting with a bare gcc toolchain and human-readable
1757 source code (first the source to CLISP, then the source to SBCL).
1758 * A contributed module containing a partial implementation of the
1759 simple-streams interface has been included. (thanks to Rudi
1761 * A contributed module implementing the RFC1321 Message Digest
1762 Algorithm, known as MD5, has been included.
1763 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1764 no longer has any effect, as the code controlled by this feature
1765 has been deleted. (As far as we know, no-one has ever built using
1766 this feature, and its semantics were confused in any case).
1767 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1768 work on conditions (as required by the ANSI specification),
1769 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1770 expected behaviour on conditions. Users should note, however,
1771 that such behaviour is not required by the ANSI specification,
1772 and so use of this behaviour may render their code unportable.
1773 * fixed some bugs revealed by Paul Dietz' test suite:
1774 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1776 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1777 ** SXHASH on condition objects no longer returns NIL.
1778 ** :ALLOCATION :CLASS slots are better treated; their values are
1779 updated on class redefinition, and initforms inherited from
1780 superclasses are applied.
1781 ** REMOVE-METHOD returns its generic function argument even when
1782 no method was removed.
1783 ** SHARED-INITIALIZE now initializes the values of the requested
1784 slots, including those with :ALLOCATION :CLASS.
1785 ** ALLOCATE-INSTANCE now works on structure classes defined via
1786 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1788 ** SLOT-EXISTS-P now works on conditions, as well as structures
1790 ** MAKE-LOAD-FORM now has the required methods on
1791 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1792 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1793 keyword, and now implements the SLOT-NAMES argument.
1794 ** methods with &OPTIONAL arguments no longer allow too many
1795 arguments to be passed in the call without error.
1796 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1797 option is consistent with the required arguments of the generic
1798 function lambda list.
1799 * bug fix: REQUIRE accepts a string designator. (Thanks to
1801 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1802 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1803 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1804 * bug fix: The compiler now checks for duplicated variables in macro
1806 * bug fix: SETQ on globals returns the correct value.
1807 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1808 not a COMPILER-ERROR (followed by some other strange error on
1809 choosing the CONTINUE restart).
1810 * bug fix: make.sh and friends are now more consistent in the way that
1811 they look for GNU "make".
1813 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1814 * experimental native threads support (on x86 Linux >=2.4 only).
1815 This is not compiled in by default: you need to add :SB-THREAD to
1816 the target features. See the "Beyond ANSI" chapter of the manual
1818 * fix for longstanding nonANSIism: The old distinction between
1819 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1820 The return value from CL:FIND-CLASS is now a CLOS class, and
1821 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1822 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1823 * An interface to the MetaObject Protocol, as described in Kiczales,
1824 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1825 MIT Press, 1991, is available from the SB-MOP package.
1826 * incompatible change: the SB-PCL package should now be considered
1827 a private implementation detail, and no longer a semi-private MOP
1829 * minor incompatible change: due to rearrangement for threads, the
1830 control stack and binding stack are now allocated at arbitrary
1831 addresses instead of being hardcoded per-port. Users affected by
1832 this probably have to be doing advanced things with shared
1833 libraries, and will know who they are.
1834 * minor incompatible change: Previously, all --eval forms used were
1835 processed with READ before any of them were processed with EVAL.
1836 Now each --eval form is processed with both READ and EVAL before
1837 the next --eval form is processed. (Thus package operations like
1838 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1839 work as the user might reasonably expect.)
1840 * minor incompatible change: *STANDARD-INPUT* is now only an
1841 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1843 * minor incompatible change: Y-OR-N-P is now character-oriented,
1844 not line oriented. Also, YES-OR-NO-P now works without errors.
1845 (thanks to Antonio Martinez)
1846 * sb-aclrepl module improvements: an integrated inspector, added
1847 repl features, and a bug fix to :trace command.
1848 * Known functions, which cannot be open coded by the backend, are
1849 considered to be able to check types of their arguments. (fixing
1850 a bug report by Nathan J. Froyd)
1851 * fixed a bug in computing method discriminating functions: It is
1852 now possible to define methods specialized on classes which have
1853 forward-referenced superclasses. (thanks to Gerd Moellmann)
1854 * fixed evaluation order in optional entries (reported by Gilbert
1856 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1857 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1858 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1859 required-by-AMOP NAME argument, as well as CLASS and
1860 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1861 * fixed bug 20: DEFMETHOD can define methods using names that are
1862 not the proper names of classes to designate class specializers.
1863 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1864 Unix file descriptors, instead of blowing up. (thanks to Antonio
1866 * Garbage collection refactoring: user-visible change is that a
1867 call to the GC function during WITHOUT-GCING will not do garbage
1868 collection until the end of the WITHOUT-GCING. If you were doing
1869 this you were probably losing anyway.
1870 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1871 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1872 TYPEP the latter but not the former.
1873 * The compiler issues a full WARNING for calls to undefined functions
1874 with names from the CL package.
1875 * MAP-INTO for a vector destination is open coded. (reported by
1876 Brian Downing on c.l.l)
1877 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1878 documentation string.
1879 * fixed some bugs revealed by Paul Dietz' test suite:
1880 ** COPY-ALIST now signals an error if its argument is a dotted
1882 ** Condition slots are now accessed more correctly in the presence
1883 of multiple initargs for a given slot.
1884 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1885 exclude restarts of the same name associated with a different
1887 ** DEFCLASS of forward-referenced classes with another
1888 forward-referenced class in the superclasses list no longer
1890 ** Condition slots are now initialized once each, not multiple
1891 times. (thanks to Gerd Moellmann)
1892 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1894 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1895 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1896 arguments contain duplicated elements.
1897 ** RESTART-CASE understands local macros.
1898 ** RESTART-CASE associates exactly its own restarts with a condition.
1899 ** ENDP in safe mode checks its argument to be of type LIST.
1900 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1901 in question is unbound.
1902 ** Optimized MAKE-INSTANCE functions no longer cause internal
1903 assertion failures in the presence of duplicate initargs.
1904 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1905 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1906 propagated even in the ctor optimized implementation of
1908 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1910 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1911 * a better implementation of SXHASH on (simple) bit vectors,
1912 measured both in execution speed and in distribution of results
1913 over the positive fixnums, has been installed. Likewise, a better
1914 implementation of EQUAL for simple bit vectors is now available.
1915 * fixed CEILING optimization for a divisor of form 2^k.
1916 * fixed bug 240 (emitting extra style warnings "using the lexical
1917 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1918 by Antonio Martinez)
1919 * fixed SXHASH, giving different results for NIL depending on type
1920 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1921 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1922 variable. (found by Rolf Wester)
1923 * SBCL does not ignore type declarations for special
1924 variables. (reported by rif on c.l.l 2003-03-05)
1925 * some bug fixes in contrib/sb-aclrepl/
1926 * fixed some bugs revealed by Paul Dietz' test suite:
1927 ** a bug in the CONS type specifier, whereby the CAR and CDR
1928 types got intertwined, has been fixed;
1929 ** the type system is now able to reason about the interaction
1930 between INTEGER and RATIO types more completely;
1931 ** APPEND, [N]REVERSE and NRECONC check that those their
1932 arguments, which must be proper lists, are really so;
1933 ** An array specialized to be unable to hold elements has been
1934 implemented, as required -- yes, really -- by ANSI;
1935 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1936 on malformed property lists;
1938 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1939 * incompatible packaging change: in line with Unix convention,
1940 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1941 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1942 * REQUIRE and PROVIDE are now optionally capable of doing something
1943 useful. See the documentation string for REQUIRE.
1944 * infrastructure for a managed SBCL contrib system: contributed
1945 modules in this release include:
1946 ** the ASDF system definition facility;
1947 ** an interface to the BSD Sockets API;
1948 ** an ACL-like convenience interface to the repl;
1949 (thanks to Kevin Rosenberg)
1950 ** an implementation of ROTATE-BYTE, with efficient implementation
1952 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1953 gives a closer approximation to the right answer than previously.
1954 (thanks to Raymond Toy)
1955 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1956 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1957 optional environment arguments, as required by ANSI.
1958 * fixed bugs in other functions taking environment objects, allowing
1959 calls with an explicit NIL environment argument to be compiled
1961 * fixed bug 228: primary return values from
1962 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1963 COMPILE or FUNCTION.
1964 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1965 :TYPE will no longer signal errors on innocuous objects.
1966 * fixed bug 231b: SETQ is better at respecting type declarations in
1967 the lexical environment.
1968 * fixed a bug in DEFCLASS: classes named by symbols with no or
1969 unprintable packages can now be defined.
1970 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1971 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1972 * fixed compiler failure related to checking types of functions.
1973 (reported by Robert E. Brown)
1974 * the compiler is now much more consistent in its error-checking
1975 treatment of bounding index arguments to sequence functions: in
1976 (SAFETY 3) code, errors will be signalled in almost all cases if
1977 invalid sequence bounding indices are passed to functions defined
1978 by ANSI to operate on sequences.
1979 * fixed a bug in the build procedure: documentation of SBCL-specific
1980 packages is now preserved and available in the final Lisp image.
1981 * lifted FDEFINITION lookup out of loops in the implementation of
1982 many list operations. (thanks to Robert E. Brown)
1983 * fixed a bug in the reader: the #n# reader macro now works for
1984 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1985 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1986 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1987 E. Brown and Tony Martinez respectively)
1988 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1989 SLOT-BOUNDP: in particular, it is now able to optimize them much
1990 better, and is now not vulnerable to having packages renamed.
1991 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1992 always called when a slot is not present in an instance. (thanks
1994 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1995 longer signal an error on unreading a character following EOF on
1996 the previous constituent stream. (thanks to Tony Martinez)
1997 * fixed some bugs revealed by Paul Dietz' test suite:
1998 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1999 not just nonnegative fixnums;
2000 ** the logical bit-array operators such as BIT-AND now accept an
2001 explicit NIL for their "opt-arg" argument (to indicate a
2002 freshly-consed result bit-array);
2003 ** ELT now signals an error on an invalid sequence index in safe
2005 ** the type system is now cleverer about negations of numeric
2006 types, and consequently understands the BIGNUM and RATIO types
2008 ** the type system is now cleverer about the interaction between
2009 INTEGER and RATIO types: while bugs still remain, many more
2010 cases are accurately computed;
2011 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2012 if it is in the last clause;
2013 ** CONSTANTLY now correctly returns a side-effect-free function in
2015 ** DECLARE is no longer treated as a special-operator; in
2016 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2017 * incremented fasl file version number due to the change in the
2018 DEFSTRUCT-SLOT-DESCRIPTION structure.
2020 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2021 * minor incompatible change: code processed by the "interpreter" or
2022 EVAL now has a compilation optimization policy of (DEBUG 2)
2023 (changed from (DEBUG 1)) to improve debuggability of interactive
2024 development, and to allow the use of the debug RETURN command in
2026 * an experimental implementation of the RETURN command for the
2027 debugger has been included. (thanks to Frederik Kuivinen)
2028 * fixed bug 62: constraints were not propagated into a loop.
2029 * fixed bug in embedded calls of SORT (reported and investigated by
2031 * fixed some bugs revealed by Paul Dietz' test suite:
2032 ** printing and reading of arrays with some dimensions having
2033 length 0 (thanks to Gerd Moellmann);
2034 ** BOA constructor with &AUX argument without a default value does
2035 not cause a type error;
2036 ** CONSTANTP now returns true for all self-evaluating objects.
2038 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2039 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2040 accessors that are related by inheritance, as specified in the
2041 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2042 Valtteri Vuorikoski)
2043 * The compiler is now able to inline functions that were defined in
2044 a complex lexical environment (e.g. inside a MACROLET).
2045 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2047 * Support for the upcoming FreeBSD-5.0 release has been included.
2048 (thanks to Dag-Erling Smorgrav)
2049 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2050 effect when it is not in a toplevel context.
2051 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2052 stand a better chance of being correct. (thanks to Gerd
2054 * fixed bug in COERCE, which now signals an error on coercing a
2055 rational to a bounded real type which excludes the expected
2057 * The compiler is now able to derive types more accurately from the
2058 COERCE and COMPILE functions.
2059 * fixed bug 223: functional binding is considered to be constant
2060 only for symbols in the CL package.
2061 * fixed bug 231: SETQ did not check the type of a variable being set
2062 (reported by Robert E. Brown)
2063 * A new optimization for MAKE-INSTANCE has been included, fixing
2064 various bugs (including relating to :ALLOCATION :CLASS slots and
2065 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2067 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2068 ** As required by ANSI, LOOP now disallows anonymous collection
2069 clauses such as COLLECT I in conjunction with aggregate boolean
2070 clauses such as THEREIS (= I 1);
2071 ** LOOP now signals an error when any variable is reused in the
2072 same loop (including the potentially useful construct analogous
2073 to WITH A = 1 WITH A = (1+ A);
2074 ** IT is only a special loop symbol within the first clause of a
2075 conditional loop clause;
2076 ** LOOP with a typed iteration variable over a hashtable now
2077 signals a type error iff it should.
2078 * fixed some other bugs revealed by Paul Dietz' test suite:
2079 ** FILE-STREAM now names the class previously known as FD-STREAM;
2080 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2081 argument) no longer signals an error;
2082 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2083 of the slot symbol, rather than using the current package
2084 ((:CONC-NAME "") continues to intern the slot's name in the
2086 * incremented fasl file version number, because of the incompatible
2087 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2088 of the new implementation of DEFINE-COMPILER-MACRO.
2090 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2091 * Support for building SBCL for MIPS platforms running in
2092 little-endian mode has now been checked in, and basic
2093 functionality on said platforms verified.
2094 * minor incompatible change: PCL now records the pathname of a file
2095 in which methods and the like are defined, rather than its
2097 * minor incompatible change: TRUENAME now considers the truename of
2098 a file naming a directory to be the pathname with :DIRECTORY
2099 component indicating that directory.
2100 * minor incompatible change: a NAMED clause in the extended form of
2101 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2102 reason for the previous behaviour is unclear.
2103 * more systematization and improvement of CLOS and MOP conformance
2104 in PCL (thanks to Gerd Moellman and Pierre Mai):
2105 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2107 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2108 primary methods with no specializers;
2109 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2111 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2112 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2113 FORWARD-REFERENCED-CLASSes; error reporting on
2114 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2116 ** SXHASH on CLOS instances now uses a slot internal to the
2117 instance to return different numbers on distinct instances,
2118 while preserving the same return value through invocations of
2120 ** DEFMETHOD signals errors when methods with longer incongruent
2121 lambda lists are added to generic functions;
2122 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2123 CLASS, as specified in AMOP;
2124 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2125 on the order returned by the primary method for classes of
2126 class STANDARD-CLASS;
2127 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2128 * fixed some bugs shown by Paul Dietz' test suite:
2129 ** DOLIST puts its body in TAGBODY;
2130 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2132 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2133 value producing form;
2134 ** if more variables are given to PROGV than values, extra
2135 variables are bound and made to have no value;
2136 ** NSUBSTITUTE on list arguments gets the right answer with
2138 ** ELT signals an error of type TYPE-ERROR when the index argument
2139 is not a valid sequence index;
2140 ** LOOP signals (at macroexpansion time) an error of type
2141 PROGRAM-ERROR when duplicate variable names are found;
2142 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2143 ** FUNCALL of special-operators now cause an error of type
2145 ** PSETQ now works as required in the presence of side-effecting
2146 symbol-macro places;
2147 ** NCONC accepts any object as its last argument;
2148 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2150 ** loop-for-as-package does not require a package to be explicitely
2152 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2154 * fixed bug 166: compiler preserves "there is a way to go"
2155 invariant when deleting code.
2156 * fixed bug 172: macro lambda lists with required arguments after
2157 &REST arguments now cause an error to be signalled. (thanks to
2159 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2160 bug. (thanks to Matthew Danish)
2161 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2163 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2164 arguments are lexically rebound. (thanks to Gerd Moellmann and
2166 * fixed bug 194: error messages are now more informative when there
2167 is no primary method applicable in a call to a generic
2168 function. (thanks to Gerd Moellmann)
2169 * fixed bug in command line argument checking (thanks to Julian
2171 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2172 (thanks to Matthew Danish)
2173 * incremented fasl file version number, because of the
2174 SXHASH-related changes in the layout of CLOS data structures
2176 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2177 * minor incompatible change: The runtime (the Unix executable named
2178 "sbcl") is now much pickier about the .core files it will load.
2179 Essentially it now requires .core files to descend from the same
2180 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2181 as the runtime does. (The intent is to prevent the crashes which
2182 can occur, and which can even be reported as mysterious failures,
2183 when people patch the sources or change the build parameters
2184 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2185 sbcl and .core files.)
2186 * fixed bug: VALUES-LIST is no longer optimized away.
2187 * fixed bug 142: The FFI conversion of C string values to Lisp
2188 string values no longer conses excessively. (thanks to Nathan
2189 Froyd porting Raymond Toy's fix to CMU CL)
2190 * began to systematize and improve MOP conformance in PCL (thanks to
2191 Nathan Froyd, Gerd Moellman and Pierre Mai):
2192 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2194 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2195 ** FINALIZE-INHERITANCE is now called on class finalization;
2196 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2197 argument precedence order.
2198 * fixed bug 202: The compiler no longer fails on functions whose
2199 derived types contradict their declared type.
2200 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2201 so it can be non-toplevel.
2202 * The fasl file version number has changed (because of the new
2203 implementation of DEFMACRO).
2204 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2205 safe code, that any length requirement by their type-specifier
2206 argument is valid. The exceptions to this are described in bug
2208 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2209 functions now check the types of their inputs as required by ANSI.
2210 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2211 introduced symbol is DECLAREd to be SPECIAL.
2212 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2213 * fixed bug: :COUNT argument to sequence functions may be negative
2214 * fixed bug: body of DO-SYMBOLS may contain declarations
2215 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2216 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2217 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2218 symbol macro only once
2219 * fixed printing of call frame when argument list is unavailable
2220 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2221 * compiler no longer signals WARNING on unknown keyword
2224 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2225 * A beta-quality port to the mips architecture running Linux,
2226 based on the old CMUCL backend, has been made. It has been tested
2227 on a big-endian kernel, and works sufficiently well to be able to
2228 rebuild itself; it has not been tested in little-endian mode.
2229 * fixed an inconsistency between gencgc.c and purify.c which made
2230 dumping/loading .core files unreliable
2231 * fixed bug 120a: The compiler now deals correctly with IFs where
2232 the consequent is the same as the alternative, instead of
2233 misderiving the return type. (thanks to Alexey Dejneka)
2234 * fixed bug 113: Logical pathnames are now dumpable (the logical
2235 host is resolved at load-time, throwing an error if it is not
2237 * fixed bug 174: FORMAT's error message is slightly clearer when a
2238 non-printing character is used in a format directive.
2239 * fixed several bugs in compiler checking of type declarations, i.e.
2240 violations of the Python "declarations are assertions" principle
2241 (thanks to Alexey Dejneka)
2242 * fixed several bugs in PCL's error checking (thanks to Gerd
2244 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2246 * fixed bug in compilation of functions as first class values
2247 (thanks to Antonio Martinez-Shotton)
2248 * The compiler's handling TYPE-ERRORs which it can prove will
2249 inevitably happen at runtime has been cleaned up and corrected
2250 in several ways. (thanks to Alexey Dejneka)
2251 * improved argument type checking for various basic arithmetic
2252 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2253 lavished on them in the past that they can be compiled in many
2254 ways in different special cases
2255 * fixed bug 181: compiler checks validity of user supplied type
2257 * cleaned up code flushing in optimization: Function calls which
2258 should signal errors for safety purposes (e.g. which ANSI says
2259 should signal errors when their arguments are of incorrect type)
2260 are no longer optimized away.
2261 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2262 * incremented fasl file version number, because changes in the
2263 implementation of sequence functions like COERCE caused
2264 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2265 in old inline expansions) to become undefined. (Actually these
2266 changes were later undone, so we might very well be binary
2267 compatible with 0.7.7 after all, but leaving the version number
2268 incremented seemed like the simplest and most conservative
2271 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2272 * An alpha-quality port to the parisc architecture running Linux,
2273 based on the old CMUCL backend, has been made. This, even more so
2274 than the other backends, should be considered still a work in
2275 progress; known problems include that the Linux kernel in 64-bit
2276 mode does not propagate the correct sigcontext structure to
2277 userspace, and consequently SBCL on a parisc64 kernel will not
2279 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2280 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2281 suppressed.) Also now that the compiler is looking at declarations
2282 in the environment, it checks optimization declarations as well,
2283 and suppresses inlining when (> DEBUG SPEED).
2284 * More fixes have been made to treatment of floating point exception
2285 treatment and other Unix signals. In particular, floating point
2286 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2287 * The detection and handling of control stack exhaustion (infinite
2288 or very deeply nested recursion) has changed. Stack exhaustion
2289 detection is now done by write-protecting pages at the OS level
2290 and applies at all optimization settings; when found, a
2291 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2292 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2293 be caught using IGNORE-ERRORS.
2294 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2295 correctly when transforming recursive LABELS functions to LETs.
2296 (thanks to Alexey Dejneka)
2297 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2298 that are names of constants or global variables.
2299 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2300 alien routines with docstrings.
2301 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2302 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2304 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2305 (thanks to Eric Marsden)
2306 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2307 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2308 * Bug fix: The SPARC backend can now compile functions involving
2309 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2310 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2311 to a functions expecting a PATHNAME-DESIGNATOR.
2312 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2313 lambda lists. (thanks to Alexey Dejneka)
2314 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2315 functions so that EXPORTing the name of the class doesn't cause
2316 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2317 to Antonio Martinez for reporting this)
2318 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2319 constant-folded without failing an assertion. (thanks to Einar
2320 Floystad Dorum for reporting this)
2321 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2322 types are now dealt with more correctly by the compiler.
2323 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2324 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2325 behaviour analogous to RENAME-FILE. This puts its behaviour more
2326 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2328 * The fasl file version number has changed again. (because of the
2329 bug fix involving the names of PCL MAKE-INSTANCE functions)
2331 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2332 * bug fix: Floating point exceptions are treated much more
2333 consistently on the x86/Linux and PPC/Linux platforms.
2334 * Array initialization with :INITIAL-ELEMENT is now much faster for
2335 cases when the compiler cannot open code the array creation, but
2336 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2337 array accesses have also seen a speed increase.
2338 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2339 specified. (This was at the root of some bad interactions between
2340 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2342 * bug fix: Internal error arguments for undefined functions are now
2343 computed correctly on the PPC/Linux platform.
2344 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2345 Raymond Toy's patch for CMU CL)
2346 * Support for the Solaris 9 operating environment has been included
2347 (thanks to Daniel Merritt)
2348 * A very ugly but hopefully complete draft of the missing FFI chapter
2349 of the manual has been created by reformatting the corresponding
2350 CMU CL manual chapter into (currently very ugly and incoherent)
2351 DocBook and bringing it up to date for SBCL behavior. Thus, the
2352 manual is now essentially complete, at least by my extreme
2353 once-and-only-once standards, whereby it's acceptable to refer to
2354 the doc strings of SB-EXT functions as the primary documentation.
2355 * The fasl file version number has changed again, due to cleanup of
2356 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2357 is no longer a static symbol.)
2359 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2360 * SBCL now builds with OpenMCL (version 0.12) as the
2361 cross-compilation host; also, more progress has been made toward
2362 bootstrapping under CLISP.
2363 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2365 * bug 158 fixed: The compiler can now deal with integer loop
2366 increments different from 1; fixing this turned out also to fix
2368 * bug 169 fixed: no more bogus warnings about using lexical bindings
2369 despite the presence of perfectly good SPECIAL declarations (thanks
2370 to David Lichteblau)
2371 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2372 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2373 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2375 * bug 180 fixed: Method combination specifications no longer ignore
2376 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2377 * bug fix: Structure type predicate functions now check their argument
2378 count as they should.
2379 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2380 correctly. (thanks to Pierre Mai)
2381 * minor incompatible change: The --noprogrammer option is deprecated
2382 in favor of the new --disable-debugger option, which is very similar.
2383 (The major difference is that it takes effect at a slightly different
2384 time at startup, causing handling of errors in --sysinit and
2385 --userinit files will be affected differently.) The
2386 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2387 been added to allow this functionality to be controlled from ordinary
2388 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2389 maintainers, who might want to run non-interactive scripts to
2390 build SBCL cores which will later be used interactively.)
2391 * minor incompatible change: The LOAD function no longer, when given
2392 a wild pathname to load, loads all files matching that pathname.
2393 Instead, an error of type FILE-ERROR is signalled.
2395 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2396 * bug 147 fixed: The compiler preserves its block link/count
2397 invariants more correctly now so that it doesn't crash. (thanks
2399 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2401 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2402 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2403 no-value-for-this-character as NIL (as specified by ANSI).
2404 * HOST-NAMESTRING on physical pathnames now returns a string that is
2405 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2406 (thanks to Christophe Rhodes)
2407 * The Alpha port handles icache flushing more correctly. (thanks to
2409 * More progress has been made toward bootstrapping under CLISP. (thanks
2410 to Christophe Rhodes)
2411 * The fasl file format has changed again, because dynamic loading
2412 on OpenBSD (which has non-ELF object files) motivated some cleanups
2413 in the way that foreign symbols are transformed and passed around.
2414 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2415 is no longer treated as whitespace by the reader, but instead as
2416 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2417 |A
\7fB|, instead of A as it used to.
2419 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2420 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2421 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2422 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2423 this as of 0.7.1.45, but was left out of the previous news section
2424 (thanks to Dan Barlow)
2425 * SBCL now runs on the Solaris operating system on SPARC architectures
2426 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2427 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2428 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2429 * SPARC backend cleanups, allowing builds of cores optimized for V8
2430 and V9 SPARCS, and also emission of code targeted to a particular
2431 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2433 * SBCL is closer to bootstrapping under CLISP, thanks to various
2434 fixes by Christophe Rhodes.
2435 * The fasl file format has changed again, to allow the compiler's
2436 INFO database to support symbol macros.
2437 * The user manual (in doc/) is formatted into HTML more nicely.
2438 (thanks to coreythomas)
2439 * The system is smarter about SUBTYPEP relationships, especially
2440 those involving NOT types (including types such as ATOM which are
2441 represented internally using NOT types). Thus SUBTYPEP is less
2442 likely to return (VALUES NIL NIL) in general, and in particular
2443 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2445 * The fasl file format has changed again, because the internal
2446 representation of types now includes a new slot to support the new
2447 SUBTYPEP-of-NOT-types logic.
2448 * (not a change in the main branch of SBCL, but a related prototype
2449 which can hopefully be merged into the main branch of SBCL in the
2450 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2451 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2452 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2453 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2454 dispatch macro character combinations. (thanks to Alexey Dejneka)
2455 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2456 that superficially look like logical namestrings correctly.
2457 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2459 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2460 default supplied-p) for &optional and &key arguments. (thanks to
2463 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2464 * incompatible change: The compiler is now less aggressive about
2465 tail call optimization, doing it only when (> SPACE DEBUG) or
2466 (> SPEED DEBUG). (This is an incompatible change because there are
2467 programs which relied on the old CMU-CL-style behavior to optimize
2468 away their unbounded recursion which will now die of stack overflow.)
2469 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2470 for non-GENCGC systems has been increased to 20M (since that
2471 seems much closer to the likely performance optimum for modern
2472 systems than the old 4M value was)
2473 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2474 and *DEBUG-PRINT-LEVEL*
2475 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2476 of CMU CL's support for SPARC, and various endianness and other
2477 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2478 * new syntactic sugar for the Unix command line: --load foo.bar is now
2479 an alternate notation for --eval '(load "foo.bar")'.
2481 ** The system now detects stack overflow and handles it gracefully,
2482 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2483 optimization settings. (This is a good thing in general, and
2484 its introduction in this version should be particularly timely
2485 for anyone whose code fails because of suppression of tail
2487 ** The system now hunts for the C variable "environ" in a more
2488 devious way, to avoid segfaults when the C library version differs
2489 between compile time and run time. (thanks to Christophe Rhodes)
2490 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2491 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2492 ** The compiler no longer issues bogus style warnings for undefined
2493 classes in the same source file as the DEFCLASSes which defined
2494 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2496 ** fixes in CONDITION class precedence list for undefined function
2497 errors (thanks to Alexei Dejneka)
2498 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2499 correctly. (thanks to Dan Barlow)
2500 ** portability fixes aiming at bootstrapping under CLISP (thanks
2501 to Dave McDonald and Christophe Rhodes)
2502 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2503 ** fixes in type translation and and type inference (thanks to
2505 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2506 ** various fixes in the new ports (thanks to Dan Barlow)
2507 * several changes related to debugging:
2508 ** suppression of tail recursion, as noted above
2509 ** stack overflow detection, as noted above
2510 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2511 is now the default. (For some time encapsulation has been more
2512 reliable than the breakpoint-based :ENCAPSULATE NIL
2513 implementation, at least on X86 systems; and I just noticed that
2514 encapsulation also seems closer to the spirit of the ANSI
2517 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2519 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2520 up properly again. (There was a packaging bug in 0.7.0 which
2521 left their definitions in SB-SYS::LOAD-FOREIGN and
2522 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2523 vital for most things which interface to C-level interfaces,
2524 like extensions working with sockets or databases or
2525 Perl-compatible regexes or whatever, and the need to fix
2526 this bug is the main reason that 0.7.1 was released so
2528 ** DEFGENERIC is now choosier about the methods it redefines, so that
2529 reLOADing a previously-LOADed file containing DEFGENERICs does
2530 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2531 cycle now works as it should. (thanks to Alexey Dejneka)
2532 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2533 fixed by Christophe Rhodes. (He actually submitted this patch
2534 months ago, and I delayed until after 0.7.0.)
2535 ** Bug 111 (internal compiler confusion about runtime checks on
2536 FUNCTION types) was fixed by Alexey Dejneka.
2537 * Some internal cleanups (getting rid of variables which aren't
2538 needed now that the byte interpreter is gone) caused the fasl
2539 file format number to change again.
2541 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2542 * major incompatible change: The default fasl file extension, i.e. the
2543 default extension for files produced by COMPILE-FILE, has changed
2544 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2546 ** There are many changes in the implementation of the compiler.
2547 SBCL is now essentially a compiler-only implementation of ANSI
2548 Common Lisp. EVAL still "interprets" a few special cases, but
2549 almost all the interesting cases are handled by creating
2550 a LAMBDA expression, calling COMPILE on it, then calling
2551 FUNCALL on the result.
2552 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2553 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2554 Since the code is newer, there might still be some new bugs
2555 (though not as many as before Martin Atzmueller's fixes:-). But
2556 the new code is substantially simpler and clearer, and hopefully
2557 any remaining bugs will be simpler, less fundamental, and more
2558 fixable then the bugs in the old code.
2559 ** The revised compiler is still a little unsteady on its feet.
2561 *** The debugging information it produces (particularly the names
2562 of FUNCTION objects) is sometimes much less useful than what
2563 the old compiler produced.
2564 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2565 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2566 within a MACROLET) has been temporarily weakened.
2567 ** There are new compiler optimizations for various functions:
2568 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2569 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2570 *** the math functions TRUNCATE, FLOOR, and CEILING
2571 *** the function-of-all-trades COERCE
2572 Mostly these should be transparent, but there's one
2573 potentially-annoying problem (bug 117): when the compiler
2574 inline-expands a function and does type analysis on the result,
2575 it can create control paths which have type mismatches, and
2576 when it can't prove that those control paths aren't taken,
2577 it will issue WARNINGs about the type mismatches. This is
2578 a particular problem in practice for the new sequence functions.
2579 It's not clear how this should be fixed, and for now, a
2580 workaround is given in the entry for 117 in the BUGS file.
2581 ** (Because of the interaction between the two previous items --
2582 occasional inlining problems and new inline expansions -- some
2583 of the new sequence function optimizations won't really kick in
2584 completely until debugging information, and then inlining, are
2585 straightened out in some future version.)
2586 * minor incompatible changes:
2587 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2588 DIRECTORY no longer implicitly promotes NIL slots of its
2589 pathname argument to :WILD. In particular, when you ask for the
2590 contents of a directory (which you used to be able to do without
2591 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2592 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2593 ** changes in behavior that ANSI explicitly defines to be
2594 implementation dependent:
2595 *** The new compiler-only implementation still conforms with ANSI,
2596 but acts a little different than before. Besides the obvious
2597 changes in performance tradeoffs (that the cost per form passed
2598 to EVAL has gone up, and the cost per form executed by EVAL
2599 has gone down), the behavior of the system changes a little
2600 because there are no longer any interpreted function objects.
2601 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2602 e.g. doing COMPILE on the output of interactive DEFUN is
2604 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2606 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2607 has changed from (:CL) to NIL.
2608 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2609 e.g. #\Newline and #\Tab, don't have names specified in the
2610 ANSI Common Lisp standard, is now based on their ASCII symbolic
2611 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2612 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2613 are no longer used for output.
2614 ** changes in internal implementation constants:
2615 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2616 4 million. (If your application spends a lot of time GCing and
2617 you have a lot of RAM, you might want to experiment with
2618 increasing it even more.)
2619 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2620 However, almost all old code should still continue to work without
2621 immediate update, as SB-C-CALL is now a (deprecated) nickname
2623 ** Old operator names in the style DEF-FOO are now deprecated in
2624 favor of new corresponding names DEFINE-FOO, for consistency with
2625 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2626 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2627 internal symbols, but a few supported extensions like
2628 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2629 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2630 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2631 etc. as you get deeper into recursive calls to the debugger
2632 command loop, instead of the old "5]", "5]]", "5]]]"
2633 sequence. (I was motivated to do this when squabbles between
2634 ILISP and SBCL left me very deeply nested in the debugger. In the
2635 short term, this change will probably provoke more ILISP/SBCL
2636 squabbles, but hopefully it will be an improvement in the long run.)
2637 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2638 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2639 function, since declaiming return types involving aliens is
2640 (1) annoyingly messy to do by hand and (2) vital to efficient
2641 compilation of code which calls such functions.
2642 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2643 longer reexported by the SB-EXT package. They're solely useful
2644 for alien code, so it seems more logical that you should get
2645 them from the SB-ALIEN package, not in SB-EXT.
2646 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2647 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2648 features. Instead, the code that they used to control is always
2649 built into the system.
2650 * many other bug fixes
2651 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2652 advantage of the new EVAL-WHEN stuff and to clean them up in
2653 general, and they are now more ANSI-compliant in a number of
2654 ways. Martin Atzmueller is responsible for a lot of this.
2655 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2657 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2658 *** correct ERROR type for various file operations
2659 *** some fixes for Lisp streams
2660 *** DEFMETHOD syntax checking
2661 *** changing old weird representation of debug information as
2662 strings (which, among their other deficiencies, don't transform
2663 correctly when you rename packages, and don't change their
2664 print representation when you change things like *PACKAGE*
2665 and *PRINT-LENGTH*) to symbols and lists of symbols
2666 He also made several improvements and fixed several bugs in DESCRIBE.
2667 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2669 *** misbehavior of WRITE-STRING/WRITE-LINE
2670 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2671 and several other LOOP problems as well
2672 *** DIRECTORY when similar filenames are present
2673 *** DEFGENERIC with :METHOD options
2674 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2675 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2676 *** argument ordering in FIND with :TEST option
2677 *** mishandled package designator argument in APROPOS-LIST
2678 *** various problems in the backquote readmacro
2679 *** a bug in APROPOS
2680 *** probably some others that I'm not describing very well here,
2681 since the CVS log documents them by reference to sbcl-devel
2682 messages, and the SourceForge archives aren't working well.:-(
2683 ** Dan Barlow improved the Alpha port (and is making progress on the
2684 PPC port, for those of you who think different).
2685 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2686 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2687 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2688 continues to work on the SPARC port (for those of you in a position
2689 to look down upon our little PC-compatible boxes from a great height).
2690 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2691 and uses that copy, rather than the current dynamic value, when
2692 it's trying to decide whether to truncate output. Thus e.g.
2693 (let ((*print-lines* 50))
2694 (pprint-logical-block (stream nil)
2696 (let ((*print-lines* 8))
2697 (print (aref possiblybigthings i) stream)))))
2698 should now truncate the logical block only at 50 lines, instead of
2699 often truncating it at 8 lines, as it did before.
2700 * The doc/cmucl/ directory, containing old CMU CL documentation
2701 from the time of the fork, is no longer part of the base system.
2702 SourceForge has shut down its anonymous FTP service, and with it
2703 my original plan for distributing the old CMU CL documentation
2704 there. For now, if you need these files you can download an old
2705 SBCL source release and extract them from it.
2706 * The fasl file version number changed again, for dozens of reasons,
2707 some of which are apparent above.
2709 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2710 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2711 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2712 patch, so that DISASSEMBLE output is much nicer.
2713 * The code in the SB-PROFILE package now seems reasonably stable.
2714 I still haven't decided what the final interface should look like
2715 (I'd like PROFILE to interact cleanly with TRACE, since both
2716 facilities use function encapsulation) but if you have a need
2717 for profiling now, you can probably use it successfully with
2718 the current CMU-CL-style interface.
2719 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2720 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2721 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2722 link now returns the dangling link itself, and for similar
2723 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2724 link itself. (In these cases the old code signalled an error and
2725 looped endlessly, respectively.) Thus, DIRECTORY now works even
2726 in the presence of dangling and cyclic symbolic links.
2727 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2728 is now a supported extension again, since the consensus on
2729 sbcl-devel was that it can be useful for ordinary development
2730 work, not just for debugging SBCL itself.
2731 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2732 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2733 that currently-defined functions might be redefined later with
2734 different return types.
2735 * Hash tables can be printed readably, as inspired by CMU CL code
2736 of Eric Marsden and SBCL code of Martin Atzmueller.
2737 * better error handling in CLOS method combination, thanks to
2738 Martin Atzmueller porting Pierre Mai's CMU CL patches
2739 * more overflow fixes for >16Mbyte I/O buffers
2740 * A bug in READ has been fixed, so that now a single Ctrl-D
2741 character suffices to cause end-of-file on character streams.
2742 In particular, now you only need one Ctrl-D at the command
2743 line (not two) to exit SBCL.
2744 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2746 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2747 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2748 does the right thing.
2749 * The implementation of some type tests, especially for CONDITION
2750 types, is now tidier and maybe faster, due to CMU CL code
2751 originally by Douglas Crosher, ported by Martin Atzmueller.
2752 * Some math functions have been fixed, and there are new
2753 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2754 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2755 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2756 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2757 likely to be built into sbcl-0.7.0.)
2758 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2759 CL, ported to SBCL by Martin Atzmueller)
2760 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2761 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2762 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2763 * There's a new slam.sh hack to shorten the edit/compile/debug
2764 cycle for low-level changes to SBCL itself, and a new
2765 :SB-AFTER-XC-CORE target feature to control the generation of
2766 the after-xc.core file needed by slam.sh.
2767 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2768 COMPILE-FILE is no longer supported, so that now every function
2769 gets an entry point, so that block compilation looks a little
2770 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2771 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2772 returns the number of bytes consed since the system started,
2773 rather than the number consed since the first time the function
2774 was called. (The new definition parallels ANSI functions like
2775 CL:GET-INTERNAL-RUN-TIME.)
2776 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2777 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2778 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2779 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2780 * Fasl file version numbers are now independent of the target CPU,
2781 since historically most system changes which required version
2782 number changes have affected all CPUs equally. Similarly,
2783 the byte fasl file version is now equal to the ordinary
2786 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2787 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2788 is no longer recognized. I apologize for this, because it was
2789 listed in SB-EXT as a supported extension, but I found that
2790 its existing behavior was poorly specified, as well as incorrectly
2791 specified, and it looked like too much of a mess to straighten it
2792 out. I have enough on my hands trying to get ANSI stuff to work..
2793 * many patches ported from CMU CL by Martin Atzmueller, with
2794 half a dozen bug fixes in pretty-printing and the debugger, and
2795 half a dozen others elsewhere
2796 * fixed bug 13: Floating point infinities are now supported again.
2797 They might still be a little bit flaky, but thanks to bug reports
2798 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2799 as flaky as they were.
2800 * The --noprogrammer command line option is now supported. (Its
2801 behavior is slightly different in detail from what the old man
2802 page claimed it would do, but it's still appropriate under the
2803 same circumstances that the man page talks about.)
2804 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2805 are now supported, and enabled by default. Thus, the compiler can
2806 handle many floating point and complex operations much less
2807 inefficiently. (Thus e.g. you can implement a complex FFT
2809 * The compiler now detects type mismatches between DECLAIM FTYPE
2810 and DEFUN better, and implements CHECK-TYPE more correctly, and
2811 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2812 thanks to patches from Martin Atzmueller.
2813 * various fixes to make the cross-compiler more portable to
2814 ANSI-conforming-but-different cross-compilation hosts (notably
2815 Lispworks for Windows, following bug reports from Arthur Lemmens)
2816 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2817 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2818 fixed thanks to Pierre Mai's CMU CL patches.
2819 * fixes to keep the system from overflowing internal counters when
2820 it tries to use i/o buffers larger than 16M bytes
2821 * fixed bug 45a: Various internal functions required to support
2822 complex special functions have been merged from CMU CL sources.
2823 (When I was first setting up SBCL, I misunderstood a compile-time
2824 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2825 * improved support for type intersection and union, fixing bug 12
2826 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2827 more obscure bugs as well
2828 * some steps toward byte-compiling non-performance-critical
2829 parts of the system, courtesy of patches from Martin Atzmueller
2830 * Christophe Rhodes has made some debian packages of sbcl at
2831 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2832 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2833 stable, but are nonetheless usable. When he's ready, I'd be happy
2834 to add them to the SourceForge "File Releases" section. (And if
2835 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2836 * new fasl file format version number (because of changes in
2837 internal representation of (OR ..) types to accommodate the new
2838 support for (AND ..) types, among other things)
2840 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2841 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2843 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2845 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2846 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2847 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2848 * DESCRIBE now works on structure objects again.
2849 * Most function call argument type mismatches are now handled as
2850 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2851 know whether the function will be redefined before the call is
2852 executed. (The compiler could flag local calls with full WARNINGs,
2853 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2854 it doesn't keep track of enough information to know whether calls
2855 are local in this sense.)
2856 * Compiler output is now more verbose, with messages truncated
2857 later than before. (There should be some supported way for users
2858 to override the default verbosity, but I haven't decided how to
2859 provide it yet, so this behavior is still controlled by the internal
2860 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2861 src/compiler/ir1util.lisp.)
2862 * Fasl file format version numbers have increased again, because
2863 support for the Gray streams extension changes the layout of the
2864 system's STREAM objects.
2865 * The Gray subclassable streams extension now works, thanks to a
2866 patch from Martin Atzmueller.
2867 * The full LOAD-FOREIGN extension (not just the primitive
2868 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2869 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2870 but like most other programs, it defaults to copying the Unix
2871 environment from the original process instead of starting the
2872 new process in an empty environment.
2873 * Extensions which manipulate the Unix environment now support
2874 an :ENVIRONMENT keyword option which doesn't smash case or
2875 do other bad things. The CMU-CL-style :ENV option is retained
2876 for porting convenience.
2877 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2878 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2880 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2882 * A patch from Martin Atzmueller seems to have solved the SIGINT
2883 problem, and as far as we know, signal-handling now works cleanly.
2884 (If you find any new bugs, please report them!)
2885 * The system no longer defaults Lisp source file names to types
2886 ".l", ".cl", or ".lsp", but only to ".lisp".
2887 * The compiler no longer uses special default file extensions for
2888 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2889 seems to expect a single default extension for all compiled code,
2890 and there's no compelling reason to try to stretch the standard
2891 to allow two different extensions.) Instead, byte-compiled files
2892 default to the same extension as native-compiled files.
2893 * Fasl file format version numbers have increased again, because
2894 a rearrangement of internal implementation packages made some
2895 dumped symbols in old fasl files unreadable in new cores.
2896 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2897 many fewer weird special cases.
2898 * Bug #17 (differing COMPILE-FILE behavior between logical and
2899 physical pathnames) has been fixed, and some related misbehavior too,
2900 thanks to a patch from Martin Atzmueller.
2901 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2902 by Tim Moore, ported to SBCL by Martin Atzmueller.
2903 * Martin Atzmueller fixed several filesystem-related problems,
2904 including bug #36, in part by porting CMU CL patches, which were
2905 written in part by Paul Werkowski.
2906 * More compiler warnings in src/runtime/ are gone, thanks to
2907 more patches from Martin Atzmueller.
2908 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2911 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2913 * DESCRIBE now works on CONDITION objects.
2914 * The debugger now handles errors which arise when trying to print
2915 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2917 * The build system now uses an additional file, customize-target-features.lisp,
2918 to allow local modifications to the target *FEATURES* list. (The point of
2919 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2920 debugging features enabled, without having to worry about propagating it
2921 into everyone's system when I do a "cvs update".) When no
2922 customize-target-features.lisp file exists, the target *FEATURES* list
2923 should be constructed the same way as before.
2924 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2925 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2926 warn about behavior which is undefined under the ANSI spec. Note: This
2927 is specified by ANSI, but it's not very popular with programmers.
2928 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2929 macro in the SBCL sources for an example of a workaround which you
2930 might use to make portable ANSI-standard code which does what you want.
2931 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2932 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2933 when it's used as a non-toplevel form. (This is required in order
2934 to implement the DEFCONSTANT-EQX macro.)
2935 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2936 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2937 funny things, due to bugs in the implementation of EVAL-WHEN
2938 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2939 non-toplevel DEFCONSTANTs are uncommon.)
2940 * The core file version number and fasl file version number have been
2941 incremented, because the old noncompliant DEFCONSTANT behavior involved
2942 calling functions which no longer exist, and because I also took the
2943 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2944 * fixed bug 1 (error handling before read-eval-print loop starts), and
2945 redid debugger restarts and related debugger commands somewhat while
2947 ** The QUIT debugger command is gone, since it did something
2948 rather different than the SB-EXT:QUIT command, and since it never
2949 worked properly outside the main toplevel read/eval/print loop.
2950 Invoking the new TOPLEVEL restart provides the same functionality.
2951 ** The GO debugger command is also gone, since you can just invoke
2952 the CONTINUE restart directly instead.
2953 ** The TOP debugger command is also gone, since it's redundant with the
2954 FRAME 0 command, and since it interfered with abbreviations for the
2956 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2958 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2959 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2960 some porting work by Martin Atzmueller)
2961 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2962 with a boatload of patches which he ported to SBCL. Now that those
2964 ** The system tries to make sure that its low-priority messages
2965 are prefixed by semicolons, to help people who like to use
2966 syntax highlighting in their ILISP buffer. (This patch
2967 was originally due to Raymond Toy.)
2968 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2969 patches originally written by Raymond Toy.
2970 ** The compiler understands coercion between single-value and
2971 multiple-VALUES type expressions better, getting rid of some very
2972 weird behavior, thanks to patches originally by Robert MacLachlan
2973 and Douglas Crosher.
2974 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2975 lambda lists, thanks to a patch originally by Pierre Mai.
2976 ** The system no longer bogusly warns about "abbreviated type
2978 ** The compiler gets less confused by inlining and RETURN-FROM,
2979 thanks to some patches originally by Tim Moore.
2980 ** The system no longer hangs when dumping circular lists to fasl
2981 files, thanks to a patch originally from Douglas Crosher.
2982 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2983 undefined function error.
2984 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2985 in structure slot initforms) for now, documented workaround instead:-|
2986 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2987 * fixed bug 5: added stubs for various Gray stream functions called
2988 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2989 installed, at least appropriate type errors are generated
2990 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2991 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2992 consistently in DEFMETHOD forms.
2993 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2994 it doesn't seem to affect SBCL after all
2995 * The C runtime system now builds with better optimization and many
2996 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2998 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3000 * The system is now under CVS at SourceForge (instead of the
3001 CVS repository on my home machine).
3002 * The new signal handling code has been tweaked to treat register
3003 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3004 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3005 downstream, e.g. in debug-int.lisp, has implicit dependencies
3006 on the unsignedness of integer representation of machine words,
3007 and that caused the system to bomb out with infinite regress
3008 when trying to recover from type errors involving signed values,
3009 e.g. (BUTLAST '(1 2 3) -1).)
3010 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3011 (This was one of the bugs Peter Van Eynde reported back in July.)
3012 * The system now uses code inspired by Colin Walters' O(N)
3013 implementation of MAP (from the cmucl-imp@cons.org mailing
3014 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3015 inline the MAP operation, and there is more than one
3016 sequence argument to the MAP call (so that it can't just
3017 do ETYPECASE once and for all based on the type of the
3018 single sequence argument). (The old non-inline implementation
3019 of the general M-argument sequence-of-length-N case required
3020 O(M*N*N) time when any of the sequence arguments were LISTs.)
3021 * The QUIT :UNIX-CODE keyword argument has been renamed to
3022 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3023 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3025 * Raymond Wiker's patches to port dynamic loading from Linux to
3026 FreeBSD have been added.
3027 * The BUGS file is now more nearly up to date, thanks in large part
3028 to Martin Atzmueller's review of it.
3029 * The debugger now flushes standard output streams before it begins
3030 its output ("debugger invoked" and so forth).
3031 * The core version number and fasl file version number have both
3032 been incremented, because of incompatible changes in the layout
3034 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3035 used to not be called for a saved Lisp image.)
3036 * Martin Atzmueller's version of a patch to fix a compiler crash,
3037 as posted on sbcl-devel 13 September 2000, has been installed.
3038 * Instead of installing Martin Atzmueller's patch for the
3039 compiler transform for SUBSEQ, I deleted the compiler transform,
3040 and transforms for some similar consing operations.
3041 * A bug in signal handling which kept TRACE from working on OpenBSD
3043 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3044 away by constant folding
3045 * The system now defines its address space constants in one place
3046 (in the Lisp sources), and propagates them automatically elsewhere
3047 (through GENESIS and the sbcl.h file). Therefore, patching the
3048 address map is less unnecessarily tedious and error-prone. The
3049 Lisp names of address space constants have also been systematized.
3050 * CVS tags like dollar-Header-dollar have been removed from
3051 the sources, because they have never saved me trouble and
3052 they've been source of trouble working with patches and other
3053 diff-related operations.
3054 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3055 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3057 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3059 * The system has been ported to OpenBSD.
3060 * The system now compiles with a simple "sh make.sh" on the systems
3061 that it's supported on. I.e., now you no longer need to tweak
3062 text in the target-features.lisp-expr and symlinks in src/runtime/
3063 by hand, the make.sh takes care of it for you.
3064 * The system is no longer so grossly inefficient when compiling code
3065 involving vectors implemented as general (not simple) vectors (VECTOR T),
3066 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3067 which dares to use the various sequence functions on non-simple
3068 vectors, takes less of a performance hit.
3069 * There is now a primitive type predicate VECTOR-T-P
3070 to test for the (VECTOR T) type, so that e.g.
3071 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3072 can now be compiled with some semblance of efficiency. (The old code
3073 turned the type declaration into a full call to %TYPEP at runtime!)
3074 * AREF on (VECTOR T) is still not fast, since it's still compiled
3075 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3076 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3077 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3078 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3079 without making full calls to SUBTYPEP at runtime.
3080 (Some analogous efficiency issues for non-simple vectors specialized to
3081 element types other than T, or for non-simple multidimensional arrays,
3082 have not been addressed. They could almost certainly be handled the
3083 same way if anyone is motivated to do so.)
3084 * The changes in array handling break binary compatibility, so
3085 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3086 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3087 instead of (VALUES T T).
3088 * By following the instructions that Dan Barlow posted to sbcl-devel
3089 on 2 July 2000, I was able to enable primitive dynamic object
3090 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3091 functionality is not implemented (since it calls ld to resolve
3092 library references automatically, requiring RUN-PROGRAM for its
3093 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3094 try to resolve library references) is now supported.
3095 * The system now flushes the standard output streams when it terminates,
3096 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3097 them at several other probably-convenient times, e.g. in each pass of
3098 the toplevel read-eval-print loop, and after evaluating a form given
3099 as an "--eval" command-line option. (These changes were motivated by a
3100 discussion of stream flushing issues on cmucl-imp in August 2000.)
3101 * The source transform for TYPEP of array types no longer assumes
3102 that an array whose element type is a not-yet-defined type
3103 is implemented as an array of T, but instead punts, so that the
3104 type will be interpreted at runtime.
3105 * There is now some support for cross-compiling in make.sh: each of
3106 the phases of make.sh has its own script. (This should be transparent
3107 to people doing ordinary, non-cross-compile builds.)
3108 * Since my laptop doesn't have hundreds of megabytes of memory like
3109 my desktop machine, I became more motivated to do some items on
3110 my to-do list in order to reduce the size of the system a little:
3111 ** Arrange for various needed-only-at-cold-init things to be
3112 uninterned after cold init. To support this, those things have
3113 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3114 symbols with such names are now uninterned after cold init).
3115 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3116 fasl files for cold load.
3117 ** Remove the old compiler structure pooling code (which used to
3118 be conditional on the target feature :SB-ALLOC) completely.
3119 ** Redo the representation of some data in cold init to be more compact.
3120 (I also looked into supporting byte compiled code at bootstrap time,
3121 which would probably reduce the size of the system a lot, but that
3122 looked too complicated, so I punted for now.)
3123 * The maximum signal nesting depth in the src/runtime/ support code has
3124 been reduced from 4096 to 256. (I don't know any reason for the very
3125 large old value. If the new smaller value turns out to break something,
3126 I'll probably just bump it back up.)
3127 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3129 * Many, many bugs reported by Peter Van Eynde have been added to
3130 the BUGS list; some have even been fixed.
3131 * While enabling dynamic object file loading, I tried to make the
3132 code easier to understand, renaming various functions and variables
3133 with less ambiguous names, and changing some function calling
3134 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3135 * While trying to figure out how to do the OpenBSD port, I tried to
3136 clean up some of the code in src/runtime/. In particular, I dropped
3137 support for non-POSIX signal handling, added various comments,
3138 tweaked the code to reduce the number of compilation warnings, and
3139 renamed some files to increase consistency.
3140 * To support the new automatic configuration functionality in make.sh,
3141 the source file target-features.lisp-expr has been replaced with the
3142 source file base-target-features.lisp-expr and the machine-generated
3143 file local-target-features.lisp-expr.
3144 * fixed a stupid quoting error in make.sh so that using CMU CL
3145 "lisp -batch" as cross-compilation host works again
3147 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3149 * DESCRIBE no longer tries to call itself recursively to describe
3150 bound/fbound values, so that it no longer fails on symbols which are
3151 bound to themselves (like keywords, T, and NIL).
3152 * DESCRIBE now works on generic functions.
3153 * The printer now prints less-screwed-up representations of closures
3154 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3155 * A private symbol is used instead of the :EMPTY keyword previously
3156 used to mark empty slots in hash tables. Thus
3157 (DEFVAR *HT* (MAKE-HASH-TABLE))
3158 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3159 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3160 now does what ANSI says that it should. (You can still get
3161 similar noncompliant behavior if bang on the hash table
3162 implementation with all the symbols you get back from
3163 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3164 This breaks binary compatibility, since tests for equality to
3165 :EMPTY are wired into things like the macroexpansion of
3166 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3168 * There's now a minimal placeholder implementation for CL:STEP,
3169 as required by ANSI.
3170 * An obscure bug in the interaction of the normal compiler, the byte
3171 compiler, inlining, and structure predicates has been patched
3172 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3173 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3175 * Missing ordinary arguments in a macro call are now detected even
3176 when the macro lambda list contains &KEY or &REST.
3177 * The debugger no longer complains about encountering the top of the
3178 stack when you type "FRAME 0" to explicitly instruct it to go to
3179 the top of the stack. And it now prints the frame you request even
3180 if it's the current frame (instead of saying "You are here.").
3181 * As specified by ANSI, the system now always prints keywords
3182 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3184 * The default initial SIZE of HASH-TABLEs is now smaller.
3185 * Type information from CLOS class dispatch is now propagated
3186 into DEFMETHOD bodies, so that e.g.
3187 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3189 is now basically equivalent to
3190 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3191 (DECLARE (TYPE SINGLE-FLOAT X))
3193 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3194 operation, without having to do run-time type dispatch.
3195 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3196 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3197 * Since I don't seem to be making big file reorganizations very often
3198 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3199 is overflowing my ability to conveniently back them up), I've finally
3200 checked the system into CVS. (The CVS repository is on my home system,
3201 not at SourceForge -- putting it on SourceForge might come later.)
3202 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3203 high-level GC-NOTIFY-FOO functions send their output. (There's
3204 still very little control of where low-level verbose GC functions
3205 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3206 less than it used to -- the GC-NOTIFY-FOO functions are now under
3207 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3208 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3209 in only one place in the source code, and propagates it automatically
3210 everywhere that it's needed. Thus e.g. when I bump the version from
3211 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3212 * The C source files now include boilerplate legalese and documentation
3213 at the head of each file (just as the Lisp source files already did).
3214 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3215 to his page will be replaced with a link to his new CLiki service.
3217 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3219 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3220 * The build process now looks for GNU make under the default name "gmake",
3221 instead of "make" as it used to. If GNU make is not available as "gmake"
3222 on your system, you can change this default behavior by setting the
3223 GNUMAKE environment variable.
3224 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3225 can build without error under CMU CL.
3227 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3229 * There is now a partial SBCL user manual (with some new text and some
3230 text cribbed from the CMU CL manual).
3231 * The beginnings of a profiler have been added (starting with the
3232 CMU CL profiler and simplifying and cleaning up). Eventually the
3233 main interface should be through the TRACE macro, but for now,
3234 it's still accessed through vaguely CMU-CL-style functions and macros
3235 exported from the package SB-PROFILE.
3236 * Some problems left over from porting CMU CL to the new
3237 cross-compilation bootstrap process have been cleaned up:
3238 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3239 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3240 part when getting rid of the compiler *BACKEND* stuff.)
3241 ** The value of *NULL-TYPE* was screwed up, because it was
3242 being initialized before the type system knew the final
3243 definition of the 'NULL type. This screwed up several key
3244 optimizations in the compiler, causing inefficiency in all sorts
3245 of places. (I found it because I wanted to understand why
3246 GET-INTERNAL-RUN-TIME was consing.)
3247 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3248 PROCLAIM FTYPE information. Unfortunately this broke binary
3249 compatibility again, since now the forms output by DEFGENERIC
3250 to refer to functions which didn't exist in 0.6.3.
3251 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3252 can use the new (as of 0.6.3) transform for SOME into MAP into
3254 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3255 values to (INTEGER 0 1000000), so that the time code will no longer
3256 occasionally get blown up by Linux returning 1000000 microseconds
3257 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3258 its output conform to the ANSI spec. (Alas, this makes its output
3259 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3260 * A full call to MAP NIL with a single sequence argument no longer conses.
3261 * fixes to problems pointed out by Martin Atzmueller:
3262 * The manual page no longer talks about multiprocessing as though
3263 it were currently supported.
3264 * The ILISP support patches have been removed from the distribution,
3265 because as of version 5.10.1, ILISP now supports SBCL without us
3266 having to maintain patches.
3267 * added a modified version of Raymond Toy's recent CMU CL patch for
3268 EQUALP comparison of HASH-TABLE
3270 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3272 * The system still can't cross-compile itself with
3273 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3274 entails), but at least it can compile more of itself that way
3275 than it used to be able to, and various buglets which were uncovered
3276 by trying to cross-compile itself that way have now been fixed.
3277 * This release breaks binary compatibility again. This time
3278 at least I've incremented the FASL file format version to 2, so that the
3279 problem can be detected reliably instead of just causing weird errors.
3280 * various new style warnings:
3281 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3282 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3283 ** using the lexical binding of a variable named in the *FOO* style
3284 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3285 as specified by ANSI.
3286 * *RANDOM-STATE* is no longer automatically initialized from
3287 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3288 default behavior of the system is to repeat its behavior every time
3289 it's run. If you'd like to change this behavior, you can always
3290 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3291 old convention there was no comparably easy way to get the system to
3292 repeat its behavior every time it was run.
3293 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3294 FTYPE declarations has been removed, in favor of their ANSI
3295 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3296 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3297 the types of their sequence arguments can be determined at compile time.
3298 This is done through a new open code expansion for MAP which eliminates
3299 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3300 argument types can be determined at compile time.
3301 * The optimizer now transforms COERCE into an identity operation when it
3302 can prove that the coerced object is already of the correct type. (This
3303 can be a win for machine generated code, including the output of other
3304 optimization transforms, such as the MAP transform above.)
3305 * Credit information has been moved from source file headers into CREDITS.
3306 * Source file headers have been made more standard.
3307 * The CASE macro now compiles without complaining even when it has
3310 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3312 * (Note that the way that the PCL macroexpansions were rewritten
3313 to accommodate the change in DEFGENERIC below breaks binary
3314 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3315 not run under sbcl-0.6.2. Once we get out of alpha releases,
3316 i.e. hit release 1.0.0, we'll probably try to maintain binary
3317 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3318 and sbcl-1.4.4. Until then, however, it might be fairly common
3319 for maintenance releases to break binary compatibility.)
3320 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3322 * The reporting of unhandled signals has been changed to print some
3323 explanatory text as well as the report form. (Previously only
3324 the report form was printed.)
3325 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3326 it defines, so that the compiler no longer issues undefined function
3327 warnings for compiled-but-not-yet-loaded generic functions.
3328 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3329 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3330 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3331 section of the ANSI spec are supported.
3332 * The "" nickname for the "KEYWORD" package has been removed.
3333 The reader still handles symbol tokens which begin with a package marker
3334 as keywords, but it doesn't expose its mechanism for doing so in the
3335 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3336 * The system now issues STYLE-WARNINGs for contradictory TYPE
3337 proclamations. (Warnings for contradictory FTYPE proclamations would
3338 be nice too, but those can't be done usefully unless the type system
3339 is made smarter about FUNCTION types.)
3340 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3341 systematized, so that "*target-*.lisp is supposed to exist only on the
3342 target and imply that there's a related file which exists on the
3343 host, and *host-*.lisp is supposed to exist only on the host and imply
3344 that there's a related file which exists on the target. This involves a
3345 lot of renaming. Hopefully the acute confusion caused by the renaming
3346 will be justified by the reduction in chronic confusion..
3347 ** runtime-type.lisp -> early-target-type.lisp
3348 ** target-type.lisp -> late-target-type.lisp
3349 ** early-host-format.lisp -> early-format.lisp
3350 ** late-host-format.lisp -> late-format.lisp
3351 ** host-error.lisp -> misc-error.lisp
3352 ** early-error.lisp -> early-target-error.lisp
3353 ** late-error.lisp -> late-target-error.lisp
3354 ** host-defboot.lisp -> early-defboot.lisp
3355 ** code/misc.lisp -> code/target-misc.lisp
3356 ** code/host-misc.lisp -> code/misc.lisp
3357 ** code/numbers.lisp -> code/target-numbers.lisp
3358 ** code/early-numbers.lisp -> numbers.lisp
3359 ** early-host-type.lisp -> early-type.lisp
3360 ** late-host-type.lisp -> late-type.lisp
3361 ** host-typep.lisp -> typep.lisp
3362 ** load.lisp -> target-load.lisp
3363 ** host-load.lisp -> load.lisp
3364 ** host-disassem.lisp -> disassem.lisp
3365 ** host-insts.lisp -> insts.lisp
3366 ** byte-comp.lisp -> target-byte-comp.lisp
3367 ** host-byte-comp.lisp -> byte-comp.lisp
3368 ** host-signal.lisp -> signal.lisp
3369 ** host-defstruct.lisp -> defstruct.lisp
3370 ** late-target-type.lisp -> deftypes-for-target.lisp
3371 Furthermore, several other previously target-only files foo.lisp (e.g.
3372 hash-table.lisp and random.lisp) have been split into a target-and-host
3373 foo.lisp file and a target-only target-foo.lisp file, with their key type
3374 definitions in the target-and-host part, so that the cross-compiler will
3375 know more about target types.
3376 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3377 gone away. In their place are various *BACKEND-FOO* variables
3378 corresponding to the slots of the old structure.
3379 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3380 into the target SBCL has been fixed.
3381 * The chill.lisp system for loading cold code into a running SBCL
3383 * Support for the CMU CL "scavenger hook" extension has been removed.
3384 (It was undocumented and unused in the CMU CL sources that SBCL was
3385 derived from, and stale in sbcl-0.6.1.)
3386 * Various errors in the cross-compiler type system were detected
3387 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3388 (enabling various consistency checks). Many of them were fixed,
3389 but some hard problems remain, so the compiler is back to
3390 running without *TYPE-SYSTEM-INITIALIZED* for now.
3391 * As part of the cross-compiler type system cleanup, I implemented
3392 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3393 * I have started adding UNCROSS calls throughout the type system
3394 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3395 on cross-compiler input files will be able to go away, and various
3397 * CONSTANTP now returns true for quoted forms (as explicitly required
3400 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3402 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3403 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3404 (MAKE-LIST -1). In the longer term, it ought to become true
3405 that declarations are assertions even at SAFETY 1. For now, it's not
3406 quite true even at SAFETY 3, but it's at least more nearly true..
3407 (Note that this change seems to increases the size of the system by
3408 O(5%) and to decrease the speed of the compiler by 20% or more.)
3409 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3410 for the problem of printing dozens of lines of distracting information
3411 about low-level system machinery as part of the top stack frame
3412 on entry to the debugger when an undefined function was called.
3413 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3414 is not reset to COMMON-LISP-USER.
3415 * Compilation of stuff related to dyncount.lisp has been made conditional
3416 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3417 smaller. The various dyncount-related symbols have been moved into
3418 a new "SB-DYNCOUNT" package.
3419 * tty-inspect.lisp has been renamed to inspect.lisp.
3420 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3421 feature has gone away. (When we eventually port to other flavors of
3422 libc and/or Unix, we'll try to make the differences between flavors
3423 invisible at the user level.)
3424 * Various other *FEATURES* tags, and/or their associated conditionals,
3425 have been removed if obsolescent, or given better documentation, or
3426 sometimes given more-mnemonic names.
3428 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3430 * tidied up "make.sh" script
3431 * tidied up system directory structure
3432 * better "clean.sh" behavior
3433 * added doc/FOR-CMUCL-DEVELOPERS
3434 * many many small tweaks to output format, e.g. removing possibly-confusing
3435 trailing #\. character in DESCRIBE-INSTANCE
3436 * (EQUALP #\A 'A) no longer signals an error.
3437 * new hashing code, including EQUALP hashing
3438 * tidied up Lisp initialization and toplevel
3439 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3440 * command line argument processing
3441 * added POSIX-GETENV function to deal with Unix-ish environment variables
3442 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3443 terminating SBCL on EOF
3444 * non-verbose GC by default
3445 * There is no more "sbcl" shell script; the sbcl file is now the C
3446 runtime executable (just like CMU CL).
3447 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3449 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3450 symbol and package information as Lisp native symbol and package objects
3451 instead of strings naming symbols and strings naming packages. This way,
3452 whenever packages are renamed (as in warm init), debug information is
3453 transformed along with everything else.
3454 * tweaked the optimization policy declarations which control the building
3455 of SBCL itself. Now, among other things, the system no longer saves
3456 source location debugging information. (This helps two problems at once
3457 by reducing SBCL size and by keeping SBCL from trying to look for its
3458 sources -- which may not exist -- when reporting errors.)
3459 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3460 debugging and testing purposes
3461 * cleaned up printing, making the printer call PRINT-OBJECT for
3462 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3463 methods, giving nearly-ANSI behavior
3464 * converted almost all special variables to use *FOO* naming convention
3465 * deleted PARSE-TIME functionality, since it can be done portably
3466 * moved some files out of cold init into warm init
3467 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3469 * regularized formatting of source files
3470 * added an install.sh script
3471 * fixed ridiculous memory usage of cross-compiler by making
3472 compiler/alloc.lisp not try to do pooling unless it can hook
3473 itself into the GC of the cross-compilation host. Now the system
3474 builds nicely on my old laptop.
3475 * added :SB-ALLOC in target-features.lisp-expr
3476 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3477 was not implemented)
3478 * re-did condition handling and note reporting in the compiler. Notes
3479 are no longer handled by signalling conditions. Style warnings
3480 and warnings are handled more correctly and reported in such a way
3481 that it's easy to find one or the other in your output (so that you
3482 can e.g. figure out which of many problems caused COMPILE-FILE to
3484 * changed the severity of several compiler warnings from full WARNING
3485 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3486 compiler note reporting so that it doesn't use the condition system
3487 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3488 COMPILE-FILE command)
3489 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3490 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3491 no longer supported, since their implementation was deeply intertwingled
3492 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3493 * removed bogus "support" for compiler macros named (SETF FOO), and
3494 removed the compiler macro for SETF INFO (but only after making a fool
3495 of myself on the cmucl-imp mailing list by posting a bogus patch for
3496 DEFINE-COMPILER-MACRO..)
3497 * Compiled files containing forms which have side effects on the Lisp
3498 reader (such as DEFPACKAGE forms) are now handled more correctly.
3499 (Compiler queuing of top level lambdas has been suppressed by setting
3500 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3501 * deleted various currently-unused source files, e.g. gengc.lisp. They
3502 may be added back at some point e.g. when porting to other architectures,
3503 but until they are it's distracting to distribute them and to try to
3505 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3506 there were so many of them they're just distractions, and UNCROSS is
3507 known to be able to handle the current sources
3508 * moved PROFILE functionality into TRACE, so that it will be clear
3509 how the wrapping and unwrapping of functions when you profile them
3510 interacts with the wrapping and unwrapping of functions when you
3511 trace them. (Actually, the functionality isn't there yet, but at least
3512 the interface specification is there. Hopefully, the functionality will
3513 arrive with some future maintenance release.)
3514 * removed host-oops.lisp
3515 * changed signature of QUIT function to allow UNIX-CODE argument
3516 * fixed READ-SEQUENCE bug
3517 * tweaked verbose GC output so that it looks more like the progress
3518 output that ANSI specifies for functions like LOAD
3519 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3520 * added <http://sbcl.sourceforge.com> to the banner information printed by