1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
3 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
4 merged with *DEFAULT-PATHNAME-DEFAULTS*.
5 * optimization: performance improvements to IO on file streams of
6 :ELEMENT-TYPE CHARACTER
8 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
9 * bug fix: add a workaround to SBCL looping infinitely at startup on
10 Linux kernels with apparently buggy implementations of personality().
11 (thanks to Svein Ove Aas)
12 * bug fix: Unicode symbols are correctly printed in LDB backtraces
13 (thanks to David Lichteblau)
14 * bug fix: local bindings shadow global symbol macros for
15 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
16 * optimization: non-open coded uses of numeric comparison operators
17 (e.g. >) no longer cons when called with more than one parameter
18 on platforms supporting dynamic-extent allocation.
19 * enhancement: saving cores with foreign code loaded is now supported
20 on MIPS/Linux in addition to the previously supported platforms.
21 * bug fix: FIND-RESTART now tests for activity, not applicability when given
22 a restart object as identifier. (reported by Helmut Eller for CMUCL)
23 * bug fix: division by zero in sb-sprof when no samples were collected
24 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
25 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
26 (thanks to Alexander Kjeldaas)
27 * new feature: ignoring the return values of destructive functions
28 like NREVERSE signals a compile-time style-warning.
29 (thanks to Kevin Reid)
30 * new feature: the alignment of alien structure fields can be
31 explicitly specified. (thanks to Cyrus Harmon)
32 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
33 and dump core on SIGQUIT
35 ** incompatible change: threads do not inherit values of specials
36 from their parents (see manual)
37 ** bug fix: threads stacks belonging to dead threads are freed by the
38 next exiting thread, no need to gc to collect thread stacks anymore
39 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
40 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
41 ** bug fix: dynamic variable and thread start related gc lossage
42 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
44 * fixed some bugs revealed by Paul Dietz' test suite:
45 ** SUBTYPEP is slightly more accurate on heinously complicated
46 CONS types where some of the members have uncertain (in the
47 NIL, NIL sense) type relationships to each other.
49 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
50 no longer purified unless :PURIFY T is explicitly specified.
51 ** Non-purified cores are significantly smaller than before
53 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
54 * new feature: timers based on Zach Beane's excellent timer package
55 * added support for the following external formats: koi8-u,
56 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
57 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
58 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
59 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
60 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
61 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
62 * incompatible change: a threaded SBCL will no longer revert to
63 non-threaded mode on non-NPTL systems, but refuse to start entirely.
64 * bug fix: interrupts are disabled until startup is complete; no
65 more sigsegvs when receiving a signal to soon
66 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
68 * bug fix: add a workaround for the memory randomization features in
69 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
70 workaround will only be in effect on systems with the proc filesystem
72 * bug fix: printing objects of type HASH-TABLE signals a
73 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
75 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
76 have been running for over 50 days. (reported by Gilbert Baumann)
77 * bug fix: the logic for getting names of functions gets less
78 confused when confronded with alternate-metaclass
79 funcallable-instances. (reported by Cyrus Harmon)
80 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
81 consistent, even on internal alternate-metaclass objects.
82 * bug fix: flush closure information collected by physical
83 environment analysis prepass before the main pass. (bug reported
85 * bug fix: compiler pack phase does not modify a hashtable, which is
86 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
87 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
88 may be expanded. (reported by Alexander <malishev> on c.l.l)
89 * bug fix: no more style-warnings when compiling code using LAST
90 under high-SPEED low-SPACE settings. (reported by David Wallin
92 * fixed some incompatibilities between SBCL's MOP and the MOP
94 ** the METAOBJECT class is now implemented;
95 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
96 STANDARD-OBJECT, as required;
97 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
98 now compatible (as required by VALIDATE-SUPERCLASS); there
99 remains a consistency requirement over the presence or absence
100 of the FUNCTION class in the superclass list of finalized
101 classes; see the manual for more details;
102 ** the :AROUND method for COMPUTE-SLOTS on
103 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
104 requested slot ordering.
106 ** bug fix: parent thread now can be gc'ed even with a live
108 ** bug fix: binding a special with PROGV to no value is not
109 the same as that symbol not having been bound (thanks to
111 ** bug fix: binding specials is thread safe (thanks to
113 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
114 and SB-SPROF do not die with 'no handler for signal XX in
115 interrupt_handle_now(..)' anymore
116 ** bug fix: WITH-TIMEOUT works with multiple running threads
117 * fixed some bugs revealed by Paul Dietz' test suite:
118 ** ENSURE-GENERIC-FUNCTION should take a method class object for
119 the :method-class keyword argument.
121 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
122 * new port: the Solaris operating system on x86 processors is now
123 mostly supported, though some rough edges in the environment
124 remain. (thanks to Daisuke Homma)
125 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
126 space for its heap. (thanks to Thiemo Seufer)
127 * optimization: SBCL on MIPS platforms now supports dynamic-extent
128 closures. (thanks to Thiemo Seufer)
129 * minor incompatible change: eof selects abort in the debugger.
130 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
131 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
133 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
134 deprecated, and will go away in a future revision of SBCL.
135 * minor incompatible change: GC-ON and GC-OFF are no longer
136 implemented with a counter, it does not matter how many times gc
137 is switched on or off
138 * bug fix: discriminating functions for generic function classes
139 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
140 make invalid assumptions about method precedence order. (reported
142 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
143 platforms. (thanks to Thiemo Seufer)
144 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
145 * bug fix: correct pathname printing with printer escaping is on.
146 (thanks to Kevin Reid)
147 * bug fix: complex VOP definitions in "user-space" no longer trigger
148 package locks. (reported by Zach Beane)
149 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
150 for funcallable-instance functions, and is no different from
152 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
153 characters in the current readtable. (reported by Nicholas Neuss)
154 * bug fix: SUBTYPEP on various CONS types returns more a more
155 accurate acknowledgment of its certainty.
156 * optimizations: REMOVE-DUPLICATES now runs in linear time on
157 lists in some cases. This partially fixes bug 384.
158 * flush all standard streams before prompting in the REPL and the
160 * bug fix: signal handling and triggering gc do not conflict
161 directly anymore, in particular a high frequency sb-sprof does
162 not prevent gc from running
163 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
164 approximation for timezone and DST information between the
165 universal time epoch and the smallest negative 32-bit time_t.
166 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
167 year 1899 so long as the encoded time is non-negative.
168 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
169 the correct answer for non-constant rotate amounts when performing
170 an inline 32-bit rotation.
172 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
173 there is only one thread in the session
174 ** bug fix: memory leak for streams created in one thread and
175 written to in another
176 ** bug fix: lockup when compiled with gcc4
177 ** bug fix: race that allows the gc to be triggered when gc is
179 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
180 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
182 ** bug fix: finished threads are gc'ed properly
183 * fixed some bugs revealed by Paul Dietz' test suite:
184 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
185 has arguments with a different set of applicable methods from
186 the orignal arguments.
187 ** The type error thrown by MAP now has a correct expected-type
189 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
190 name a compiled function.
191 ** Three MISC tests where a large bignum was improperly coerced to
192 a float that couldn't represent that bignum during type
193 derivation were fixed.
194 ** SUBTYPEP can now handle types involving the negation of a
195 list-form FUNCTION type.
196 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
197 as (COMPLEX (AND RATIO (NOT FIXNUM))).
198 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
200 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
201 * New feature: Experimental support for bivalent streams: streams
202 opened with :element-type :default now allow character and binary
203 (unsigned-byte 8) I/O
204 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
205 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
207 * bug fix: no more highly sporadic "couldn't check whether ~S is
208 readable" when reading a stream and an interrupt hits in the middle
209 of a select system call
210 * compiler better recognizes complex arrays (reported by Eduardo
212 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
213 for variables on SBCL built with linkage-tables. (reported by Luis
215 * various error reporting improvements.
216 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
217 (Thanks to Thiemo Seufer)
218 * bug fix: nested reader invokations work correctly
219 * bug fix: it is possible to have simultaneous references to foreign
220 code and foreign data with the same name.
222 ** added x86-64 support
223 ** incompatible change: the threading api now works with thread
224 objects instead of thread ids
225 ** bug fix: threads are protected from signals and interruption when
226 starting up or going down
227 ** bug fix: a race where an exiting thread could lose its stack to gc
228 ** bug fix: don't halt on infinite error in threads if possible
229 ** fixed numerous gc deadlocks introduced in the pthread merge
230 ** bug fix: fixed thread safety issues in read and print
231 ** bug fix: debugger doesn't hang on session lock if interrupted at
232 an inappropriate moment
233 ** bug fix: run-program is now thread safe(r)
234 ** bug fix: inner with-recursive-lock no longer releases the mutex
235 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
236 * fixed some bugs revealed by Paul Dietz' test suite:
237 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
238 EXPECTED-TYPE slots filled.
239 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
241 ** MAKE-CONDITION accepts classes as type-designators.
242 ** COMPILE may never return NIL.
243 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
244 range before calling Unix time functions
246 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
247 * numerous signal handling fixes to increase stability
248 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
249 * minor incompatible change: we now correctly canonize default
250 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
251 as per the MOP, rather than the historical (INITARG INITFUNCTION
252 INITFORM). (reported by Bruno Haible)
253 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
254 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
255 for more information.
256 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
257 pathname is a directory pathname.
258 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
259 * dynamic space size on PPC has been increased to 768Mb. (thanks to
261 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
262 :METACLASS argument in addition to a class name. (reported by
263 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
264 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
265 (reported by Bruno Haible; patch from Gerd Moellmann)
266 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
268 * bug fix: more cleanups to the floating point exception handling on
269 x86-64 (thanks to James Knight)
270 * bug fix: the compiler does not try to constant fold calls of
271 COERCE and BIT-* functions when they return freshly constructed
272 objects. (reported by Nikodemus Siivola and Paul Dietz)
273 * optimization: improved the MIPS versions of generic (in the
274 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
275 * optimization: direct conversion of (unsigned-byte 32) to floats on
276 the PowerPC platform.
277 * optimization: structure instances with raw slots now use less
278 memory, and probably show better memory locality. (thanks to
280 * optimization: DYNAMIC-EXTENT declarations for lists and closures
281 are treated as requests for stack allocation on the x86-64,
282 Alpha-32, PPC and SPARC platforms.
283 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
284 the wrong file descriptor; implementation of SOCKET-OPEN-P.
285 (thanks to Tony Martinez)
287 ** gcing a dead thread can no longer lead to lockups
288 ** threads block signals until they are set up properly
289 ** errno is no longer shared by threads
290 ** interrupt-thread restores the eflags register on x86
291 ** fixed some lockups due to gc/thread interaction
292 * fixed some bugs revealed by Paul Dietz' test suite:
293 ** invalid dotted lists no longer raise a read error when
295 ** signal an error if a symbol that names a declaration is used
296 as the name of a type, or vice versa
297 ** allow using the (declare (typespec var*)) abbreviation for
298 (declare (type typespec var*)) with all type specifiers
299 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
300 ** always return NIL from PPRINT-POP when OBJECT is NIL
301 ** don't signal errors when pretty-printing malformed LABELS,
302 FLET or MACROLET forms
303 ** declarations in a DOLIST body are also in scope for the
305 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
307 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
310 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
311 * fixed cross-compiler leakages that prevented building a 32-bit
312 target with a 64-bit host compiler.
313 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
314 opened with :IF-EXISTS :OVERWRITE.
315 * fixed bug 281: error for an invalid qualifier in a short-form method
316 combination method is not signalled until the faulty method is called.
317 * bug fix: iteration variable type inferrer failed to deal with open
318 intervals. (reported by Alan Shields)
319 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
320 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
321 or more runtime options were provided to the sbcl binary.
322 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
323 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
324 * optimization: there's now a fast-path for fixnum arguments in the
325 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
326 * optimization: the code generated on x86-64 is more compact thanks
327 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
328 * bug fix: floating-point exception handling now partly works on
329 x86-64. (Thanks to James Knight)
330 * improvement to the MIPS backend from Thiemo Seufer: C-style
331 64-bit long long arguments and return values to alien functions
333 * fixed some bugs revealed by Paul Dietz' test suite:
334 ** the type-error signalled from WARN has a filled-in DATUM slot.
335 ** the type-error required when a stream is not associated with
336 a file has the stream as its datum.
337 ** type-errors on single-floats on x86-64 no longer have
338 :INVALID-OBJECT as the datum
339 ** the type-errors signalled for invalid function names now have
340 a correct expected type
341 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
342 for structures defined with a :TYPE. Documentation strings for
343 typed structures are no longer immediately discarded
344 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
345 broadcast streams and synonym streams. FILE-LENGTH now also works
346 on broadcast streams.
348 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
349 * incompatible change: the --noprogrammer option, deprecated since
350 version 0.7.5, has been removed. Please use the equivalent
351 --disable-debugger option instead.
352 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
354 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
355 inoperational for a while now) has been completely removed.
356 * null lexical environments are now printed as #<NULL-LEXENV>,
357 significantly reducing the amount of clutter in typical backtraces.
358 * documentation on weak pointers, finalization, and after GC hooks
359 has been added to the manual.
360 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
361 as other specialized array types, is much faster. SUBSEQ and
362 COPY-SEQ on such arrays have also been sped up.
363 * optimization: EQL is now more efficient when at least other argument
364 is known to be of type (OR FIXNUM (NOT NUMBER)).
365 * fixed bug: the runtime is now less vulnerable to changes in the
366 size of the SBCL object on OS X, and virtual memory is reserved for
367 all spaces that need to be at a fixed address.
368 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
369 * fixed bug: finalizers and after GC hooks that cause consing are now
371 * fixed bug: compiler error messages and summaries are now printed to
372 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
373 * fixed inference of the upper bound of an iteration variable.
374 (reported by Rajat Datta).
375 * fixed bug 211e: calling local functions with duplicated constant
376 keyword argument no longer causes a bogus style warning about an
378 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
379 ftype declaration to be disregarded. (reported by Dave Roberts)
380 * fixed bug 373: caused by erronous compilation of references to alien
381 variables in the runtime on ppc/darwin.
382 * fixed bug 376: CONJUGATE type deriver.
383 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
384 (reported by Baughn on #lisp)
385 * a cross-compiler bug on non-x86 platforms has been identified and
386 fixed. (thanks to Bruno Haible)
387 * improvements to the MIPS runtime code for increased stability.
388 (thanks to Thiemo Seufer)
389 * increased the maximimum compact environment size to allow
390 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
391 * partial workaround for bug 135: don't name the function we're
392 calling for hairy cases of EVAL, so as not to accumulate one environment
393 entry per call to EVAL. (reported by Kevin Reid)
394 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
395 * optimization: type testing for non-vector arrays should be faster.
396 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
398 * bug fix: setting 31st element of a bit vector to zero did not work
400 * bug fix: redefining a class definition which failed due to a
401 previous accessor / function clash now works (but see BUGS entry
402 #380 for more problems in this area). (thanks to Zach Beane)
403 * the long form of DEFINE-METHOD-COMBINATION disables method group
404 checking when given a single method group with pattern *. (thanks
406 * on x86 compiler supports stack allocation of results of simple
407 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
408 * support for building and running on Mac OS X 10.4 "Tiger" has been added
409 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
410 * fixed some bugs related to Unicode integration:
411 ** the restarts for recovering from input and output encoding
412 errors only appear when there is in fact such an error to
414 * fixed some bugs revealed by Paul Dietz' test suite:
415 ** MISC.549 and similar: late transformation of unsafe type
416 assertions into derived types caused unexpected code
418 ** SCALE-FLOAT type deriver is less wrong.
419 ** type derivers for EXP, LOG and similar functions compute result
420 types for complex arguments better.
421 ** (MISC.563) CONJUGATE type deriver works for very restricted
423 ** out-of-line type testers for character strings are available.
424 ** EQUAL compiler transform understands specialness of objects
426 ** accessing double-floats stored on the stack now works on x86-64.
427 ** debugger internals could sometimes create invalid lispobjs,
428 resulting in GC crashes.
429 ** MISC.548: type check weakening can convert required type into
431 ** initialization forms for bindings are not in scope of free special
434 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
435 * incompatible change: thread support for non-NPTL systems has
436 been removed - locking is buggy and unreliable. A threaded
437 SBCL build will now warn at startup and refuse to create
438 new threads, unless futex support is detected in the kernel
439 * incompatible change: the top level REPL now has only an ABORT
440 restart associated with it, not TOPLEVEL and ABORT as it used to.
441 TOP and TOPLEVEL are now available as debugger commands for
442 returning to the top level.
443 * incompatible change: forms evaluated in the REPL now use the
444 global optimization policy.
445 * incompatible change: user- and system-initialization files are
446 no longer processed with LOAD, but by READ and EVAL; hence the
447 global optimization policy, startup package, readtable, etc,
449 * The .fasl file format number has been incremented because of
450 various incompatible changes.
451 * internal entry point details and argument counts no longer appear
452 in backtraces unless explicitly requested by setting
453 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
454 * built-in and standard functions no longer have names like "top
455 level local call to FOO".
456 * fixed bug 32: functions defined in non-null lexical environments
457 now have more legible printed representation
458 * fixed bug 33: functions defined in non-null lexical environemnts
459 are now more amenable to inspection by INSPECT.
460 * workaround for bug 354: XEPs no longer appear in backtraces unless
461 explicitly requested.
462 * fixed bug: receiving the signal which results from attempting to
463 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
464 SIGBUS on other *BSDs) on architectures where the C stack is also
465 the Lisp stack (x86 and x86-64) from foreign code no longer leads
466 to debugger confusion or wild instability. (reported by Cheuksan
468 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
469 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
470 (reported by Lutz Euler)
471 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
472 compiled code. (reported by Paul Dietz)
473 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
474 specialized parameter to a method to something that is not TYPEP
475 the specializer is now possible.
476 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
477 face of package deletion.
478 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
479 pathname. (reported by David Tolpin on c.l.l)
480 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
481 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
482 STRING-STREAM and FILE-STREAM, respectively.
483 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
484 than previously; in addition, its test suite is now run on build.
485 (thanks to Luke Gorrie)
486 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
487 * optimization: added a immediate representation for single-floats
489 * fixed some bugs revealed by Paul Dietz' test suite:
490 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
491 ** attempting to create a package with a colliding nickname causes
492 correctable errors to be signalled.
493 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
494 ** several x86-64 backend bugs related to sign-extension of immediate
497 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
498 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
499 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
501 * fixed loading of multiply forward-referenced layouts.
502 (thanks to Cheuksan Wang)
503 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
504 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
505 (see COMPILE-FILE documentation for details of :PRINT option)
506 * fixed bugs 19 and 317: fixed-format floating point printing is
507 more accurate. This also fixes a bug reported by Adam Warner
508 related to the ~@F format directive.
509 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
511 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
512 dispatch table if the from-char is a dispatch macro character.
513 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
514 by the wordsize no longer ignore the last word. (reported by Lutz
516 * fixed bug in type checking of dynamic-extent variables. (reported
518 * optimization: sequence traversal functions use their freedom to
519 coerce function designators to functions.
520 * optimization: code with many calls to CLOS methods specialized on
521 CLOS classes has had redundant type checks removed.
522 * optimization: added declarations to speed up operations that access
523 the internal character database (for example STRING-UPCASE)
524 * optimization: comparison operations between floats and sufficiently small
525 fixnums no longer create extra rationals
526 * fixed some bugs related to Unicode integration:
527 ** portions of multibyte characters at the end of buffers for
528 character-based file input are correctly transferred to the
529 start of the buffer at the next read.
530 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
531 passing it through to OPEN.
532 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
533 argument given to internal calls to OPEN.
534 ** the built-in comment readers (introduced by character sequences
535 ";" and "#|") are more forgiving to encoding errors; they will
536 STYLE-WARN and then attempt to resync the stream at a character
537 boundary. (thanks to Teemu Kalvas)
538 * fixed some bugs in the x86-64 port:
539 ** Negative short int return values from c-calls are sign-extended
541 ** The stack is aligned to 16-bytes for c-calls, as required by
542 the ABI. (thanks to Cheuksan Wang)
543 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
544 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
545 for enums. (thanks to Vincent Arkesteijn)
546 ** Multiple small optimizations and bugfixes for floating point
548 * fixed some bugs revealed by Paul Dietz' test suite:
549 ** Space, Tab, Linefeed, Return and Page have the invalid
550 secondary constituent character trait.
551 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
553 ** WITH-INPUT-FROM-STRING should only update the index place on
555 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
557 ** Bit-array operations (BIT-AND and similar) worked incorrectly
558 with one-dimensional arrays with fill pointers.
559 ** TYPE-OF failed on a complex with an integer realpart and a
561 ** compiler failure during type inference for the code of form
562 (IF (EQL X (THE ...)) ...) (MISC.535).
564 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
565 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
566 platform. The port supports 61-bit fixnums, large memory spaces
567 and reloading shared object files.
568 * enhancement: saving cores with foreign code loaded is now
569 supported on ppc/Darwin in addition to the previously supported
571 * enhancement: the statistical profiler now walks deeper into the
572 call stack for more meaningful call-graphs and accrued time
573 reports (x86/x86-64 only). It also now reports time spent in
575 * enhancement: it is now possible to trace most individual methods
576 of a generic function in addition to tracing the generic function
578 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
580 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
581 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
582 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
583 produces an error. (thanks to Vincent Arkesteijn)
584 * bug fix: NAMESTRING on pathnames with :WILD components in their
585 directories works correctly. (thanks to Artem V. Andreev)
586 * fixed bug 125: compiler preserves identity of closures. (reported
588 * bug fixed: functions with &REST arguments sometimes failed with
589 "Undefined function" when compiled with (DEBUG 3). (reported by
591 * bug fix: overflow during compiling of setting element of a bit
592 vector with constant index and value. (reported by Timmy Douglas)
593 * build fix: fixed the dependence on *LOAD-PATHNAME* and
594 *COMPILE-FILE-PATHNAME* being absolute pathnames.
595 * on x86 compiler partially supports stack allocation of dynamic-extent
597 * GO and RETURN-FROM do not check the extent of their exit points
598 when compiled with SAFETY 0.
599 * fixed some bugs related to Unicode integration:
600 ** encoding and decoding errors are now much more robustly
601 handled; it should now be possible to recover even from invalid
602 input or output to the terminal. (thanks to Teemu Kalvas)
603 ** provided a first cut at implementing STRING-TO-OCTETS and
604 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
605 ** altered the SB-MD5 contributed module slightly, changing the
606 interface just enough for it to be supportable for builds where
607 lisp characters are not eight bits.
608 * fixed some bugs revealed by Paul Dietz' test suite:
609 ** the FORMATTER-generated functions for ~V[ conditionals require
610 the correct number of arguments.
611 ** READ-FROM-STRING returns the mandated second value when applied
612 to displaced strings.
613 ** the #\Rubout and #\Backspace characters are treated as invalid
614 constituent characters by the tokenizer.
616 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
617 * new feature: reloading changed shared object files with
618 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
619 * new feature: references to foreign variables and functions
620 can now be compiled and loaded before the corresponding shared
621 object file is loaded, as long as the foreign definitions are
622 available at runtime.
623 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
624 supported. (thanks to Dan Debertin)
625 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
626 just keywords. (thanks to Vincent Arkesteijn)
627 * fixed bug #331: structure-class instances corresponding to
628 DEFSTRUCT forms are now created eagerly.
629 * fixed bug #345: backtraces from calls to undefined functions work
630 on x86 as well. Related bug #61 is now also partially fixed on x86
631 (backtraces from throws to unknown catch tags.)
632 * bug fix: lambda-list parsing is now stricter vrt. order and number
633 of lambda-list keywords.
634 * bug fix: as specified by AMOP, an error is signalled if a
635 class-option appears multiple times in a DEFCLASS form. (reported
637 * bug fix: RANDOM can be compiled when the compiler derives the type
638 of its numeric argument as a disjoint set of small integers.
639 (reported by Paul Dietz)
640 * bug fix: starting a core saved with shared objects loaded when
641 those objects are not available no longer causes threaded SBCL to
642 hang. (reported by Sean Ross)
643 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
645 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
646 argument, which must always be NIL. (reported by Kalle Niemitalo)
647 * bug fix: printing 1.0d+23 no longer results in an error.
648 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
649 * bug fix: structure slot setters preserve evaluation order. (thanks
651 * fixed some bugs related to Unicode integration:
652 ** RUN-PROGRAM can allow its child to take input from a Lisp
653 stream. (reported by Stefan Scholl)
654 ** ASDF-INSTALL successfully downloads .tar.gz files in all
655 locales. (reported by Ken Causey)
656 * fixed some bugs revealed by Paul Dietz' test suite:
657 ** INCF, DECF and REMF evaluate their place form as specified in
659 ** FORMATTER expands ~{ ~} iteration directives with V or #
660 parameters correctly.
661 ** FORMATTER deals with the ~@[ ~] conditional directive where the
662 consequent uses no arguments correctly.
663 ** the system has a partial understanding of the (COMPLEX RATIO)
665 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
668 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
669 * new feature: a build-time option (controlled by the :SB-UNICODE
670 keyword feature, enabled by default) for building the system with
671 support for the entire 21-bit character space defined by the
673 * new feature: the system now has rudimentary external-format
674 support; the primary user-visible change at this time is that
675 characters with the high bit set (such as lower-case-e-acute) will
676 print correctly to a terminal in a UTF-8 environment.
677 * minor incompatible change: BASE-CHAR no longer names a class;
678 however, CHARACTER continues to do so, as required by ANSI.
679 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
680 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
681 has been moved to the SB-EXT package (temporarily re-exported from
683 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
684 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
686 * fixed bug #351: better error-handlind and reporting for malformed
688 * fixed bug #350: bignum-printing is now more memory-efficient,
689 allowing printing of very large bignums, eg. (expt 2 10000000).
690 (reported by Bruno Haible)
691 * fixed bug #302: better primitive-type selection for intersection
693 * fixed bug #308: non-graphic characters now all have names, as
694 required. (reported by Bruno Haible)
695 * bug fix: redefining a class with different superclasses now correctly
696 removes it from the direct-subclasses of its previous superclasses.
697 (reported by David Morse)
698 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
699 new value now works. (reported by Bruno Haible)
700 * bug fix: correct canonicalization of multiple non-standard slot
701 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
702 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
703 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
704 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
705 now exists, an signals an error.
706 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
707 during macroexpansion for non-iteration variables as well. (reported
708 by Bruno Haible for CMUCL)
709 * bug fix: Cyclic structures and unprintable objects in compiler
710 messages no longer cause errors. (reported by Bruno Haible)
711 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
712 and READ-FROM-STRING all now return a primary value of NIL if
713 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
714 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
715 (reported by Bruno Haible for CMUCL)
716 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
717 do not designate a condition. (reported by Bruno Haible for
719 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
720 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
721 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
722 permitting resolution in favour of any of the conflicting symbols.
723 (reported by Bruno Haible for CMUCL)
724 * bug fix: EQUAL compiler optimizations is less aggressive on
725 strings which can potentially compare true despite having distinct
726 specialized array element types.
727 * bug fix: unit enumerations can be defined without dividing by
728 zero. (thanks to Vincent Arkesteijn)
729 * FORMAT compile-time argument count checking has been enhanced.
730 (report from Bruno Haible for CMUCL)
731 * a partial workaround for the bug 262: the compiler does not try to
732 inline-expand a local function doing RETURN-FROM from a deleted
733 BLOCK. (thanks to Peter Denno for the bug report and to David
734 Wragg for the simple test case)
735 * fixed some bugs revealed by Paul Dietz' test suite:
736 ** RENAME-PACKAGE allows all package designators as new package
738 ** constraint propagation and lambda variable substitution are
739 more cautious in dealing with partially deleted code.
740 ** compiler sometimes lost reoptimization passes.
741 ** CERROR, given a condition as condition designator, passes its
742 remaining arguments to the continue format control without
744 ** Case-altering FORMAT directives work correctly on non-ASCII
746 ** The REAL type specifier handles bounds outside the floating
747 point ranges without signalling FLOATING-POINT-OVERFLOW.
748 ** Functions with IR1-transformations can create intercomponent
749 references to global functions.
750 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
752 ** FORMAT ~R treats a nil value for its first parameter correctly.
754 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
755 * enhancement: saving cores with foreign code loaded is now
756 supported on x86/NetBSD and sparc/Linux in addition to the previously
758 * bug fix: on some platforms repeated installations caused multiple
759 copies of HTML documentation to be installed -- should not happen
760 any more. (reported by Stefan Scholl)
761 * bug fix: parsing self-recursive alien record types multiple times
762 no longer causes infinite recursion. (reported by Thomas F. Burdick,
763 original patch by Helmut Eller for CMUCL)
764 * bug fix: stack-exhaustion detection works now on NetBSD as well.
765 (thanks to Richard Kreuter)
766 * bug fix: defining classes whose accessors are methods on existing
767 generic functions in other (locked) packages no longer signals
768 bogus package lock violations. (reported by François-René Rideau)
769 * bug fix: special variables as DEFMETHOD parameters no longer have
770 associated bogus type declarations. (reported by David Wragg and
772 * bug fix: read-write consistency on streams of element-type
773 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
774 * bug fix: redefiniton of the only method of a generic function with
775 no DEFGENERIC no longer emits a full WARNING. In addition,
776 redefinition of generic functions with no DEFGENERIC to an
777 incompatible lambda list now signals an error. (thanks to Zach
779 * bug fix: DEFGENERIC now works even when there's a function of the
780 same name in an enclosing lexical environment. (thanks to Zach
782 * fixed compiler failure, caused by instrumenting code during
783 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
784 * optimization: added loop analysis and improved register allocation
785 * fixed some bugs revealed by Paul Dietz' test suite:
786 ** POSITION on displaced vectors with non-zero displacement
787 returns the right answer.
788 ** (SIMPLE-STRING) is a valid type specifier for sequence
790 ** *PRINT-LEVEL* handling for slotless structures is pedantically
792 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
793 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
795 ** The justification version of the FORMAT ~< directive treats
796 non-zero minpad parameter correctly.
798 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
799 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
800 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
801 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
802 the supported interface.
803 * new feature: Single-stepping of code compiled with DEBUG 2 or
804 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
805 * new feature: saving cores with foreign code loaded is now
806 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
807 Timothy Moore's work for CMUCL)
808 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
809 parameters to * instead of NIL if no initform is supplied.
810 (reported by Johan Bockgaard)
811 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
812 correctly when FUNCALL appears as the car of the form. Note:
813 despite this FUNCALL forms are not currently subject to
814 compiler-macro expansion. (port of Raymond Toy's fix for the
815 same from CMUCL, reported by Johan Bockgaard)
816 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
817 (thanks for Teemu Kalvas)
818 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
819 case instead of terminating the iteration (thanks for Julian
820 Squires, Sean Champ and Raymond Toy)
821 * bug fix: incorrect expansion of defgeneric that caused a style
822 warning. (thanks for Zach Beane)
823 * bug fix: slot accessor effective method computation works properly
824 for classes with multiple non-standard applicable methods on
825 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
826 * on x86 compiler supports stack allocation of results of LIST and
827 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
828 CMUCL implementation by Gerd Moellmann)
829 * fixed some bugs revealed by Paul Dietz' test suite:
830 ** FORMAT strings with both the ~<~:;~> form of the justification
831 directive and pretty-printing directives cause an error.
833 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
834 * incompatible change: the internal functions
835 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
836 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
837 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
838 instead of the old functions.
839 * new feature: on platforms where "dladdr" is available foreign
840 function names now appear in backtraces. (based on Helmut Eller's
842 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
843 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
845 * documentation: ASDF manual is now distributed with SBCL.
846 * bug fix: SBCL can now load its contributed modules with REQUIRE
847 even if the system-provided entries have been removed from
848 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
850 * bug fix: backtraces involving undefined functions or assembly
851 routines are more informative. (thanks to Brian Downing)
852 * bug fix: mutually referent alien structures now work correctly.
853 (reported by Rick Taube)
854 * bug fix: structures defined by WITH-ALIEN can be referred to
855 within other definitions in the same WITH-ALIEN.
856 * bug fix: division operators (MOD, TRUNCATE and the like) with
857 constant zero divisors and integer dividends no longer generate
859 * bug fix: provide default methods for INPUT-STREAM-P and
860 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
861 * bug fix: improve the detection and resolution of MOP metacycles
862 (where effective method computation depends on the generic function
863 having its effective method computed). (reported by Bruno Haible)
864 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
865 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
866 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
867 * bug fix: pathnames with relative directory components can be
868 represented relative to default pathnames.
869 * optimization: in taking the GCD of bignums, reduce the two bignums
870 to approximately the same size (using Euclid's algorithm) before
871 applying the more sophisticated binary GCD. (thanks to Juho
873 * optimization: COUNT on bitvectors now operates word-at-a-time.
874 * optimization: ASH with a positive, but not necessarily constant,
875 (leftwards) shift, when in a modular context, is compiled to a
877 * fixed some bugs revealed by Paul Dietz' test suite:
878 ** FORMAT variable parameters ("~V<char>") are defaulted properly
879 if the corresponding argument is NIL.
880 ** FORMAT directives accepting a minpad parameter treat negative
882 ** literal commas as character parameters to FORMAT directives are
884 ** literal spaces directly after ~<Newline> directives within a
885 format-logical-block (~:< ~@:>) do not induce :FILL-style
886 conditional newlines.
887 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
888 as specified: it no longer includes conditional newlines).
889 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
891 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
892 :PER-LINE-PREFIX argument does not evaluate to a string.
894 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
895 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
896 the manual for details; for now, package locks can be disabled by
897 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
898 if no major problems are found then it is likely that they will be
899 compiled in unconditionally.
900 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
901 unsupported operators on all platforms. To load a shared library
902 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
903 object file, link it into a shared library outside of SBCL and
904 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
906 * bug workaround: ROOM T and the GC were not working together
907 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
908 reporting facility aren't preserved. That reporting has been
909 disabled, so now until and unless someone figures out how to make it
910 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
911 * minor incompatible change: as threatened around sbcl-0.8.0, the
912 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
913 The symbols are also exported from SB-PCL for backwards
914 compatibility, but more so than before SB-PCL should be treated as
915 an implementation-internal package.
916 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
918 * the SB-SPROF contrib now works on (most) non-x86 architectures.
919 It is known as of this release not to work on the Alpha, however.
920 * fixed bug #167: errors signalled due to illegal syntax in method
921 bodies are now more legible.
922 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
923 designators and can hence be used with TYPEP.
924 * fixed bug #333: CHECK-TYPE now ensures that the type error
925 signalled, if any, has the right object to be accessed by
926 TYPE-ERROR-DATUM. (reported by Tony Martinez)
927 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
928 5.1.2.3 for multiple-value place subforms. (reported by Kalle
930 * fixed bug #334: programmatic addition of slots using specialized
931 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
932 and :ALLOCATION :CLASS slots.
933 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
934 float exponents. (rereported by Peter Seibel)
935 * fixed a bug: #\Space (and other whitespace characters) are no
936 longer considered to be macro characters in standard syntax by
938 * fixed bug: initialization of condition class metaobjects no longer
939 causes an instance of the condition to be created. (reported by
941 * fixed bug: it is now possible to have slots such that
942 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
943 description is neither :INSTANCE nor :CLASS.
944 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
945 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
946 SLOT-VALUE-USING-CLASS), and no longer causes errors with
947 non-standard SLOT-DEFINITION-ALLOCATION values.
948 * fixed bugs: various race conditions handling exiting threads.
949 CL-PPCRE's thread torture test now passes.
950 * fixed arguably-a-bug: GC time proportional to number of threads in
951 system even when most of them are idle
952 * optimization: improved performance of BIT and SBIT on bit-vectors.
953 * .fasl file incompatibility: The fasl file version number has
954 been incremented because of changes associated with package locks.
956 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
957 * minor incompatible change: the system no longer provides
958 optimization hints (conditions which are TYPEP
959 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
960 mode; these hints are still emitted when higher SPEED optimization
962 * new contrib module: a sampling profiler (profiling by statistical
963 sampling, rather than by instrumenting functions) is available as
964 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
965 * the behaviour of the standard function ED is now customizeable by
966 third parties through a hook variable: see ED's documentation
967 string for information on the protocol.
968 * the compiler no longer emits efficiency notes for (FUNCALL X)
969 when the type of X is uncertain under default optimization
971 * fixed bug 276: mutating a binding of a specialized parameter to a
972 method to something that is not TYPEP the specializer is now
974 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
975 exhibit the required behaviour on the x86 platform. (thanks to
976 Peter van Eynde, Eric Marsden and Bruno Haible)
977 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
978 even for difficult arguments. (reported by Peter Graves)
979 * fixed bug 141a: the backquote printer now descends quoted
981 * fixed another bug in backquote printing: no more destructive
982 modification of the form's list structure. (reported by Brian
984 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
985 move between its address being taken and the call to
986 interrupt_thread, fixing a crashing race condition.
987 * the SB-POSIX contrib implementation has been adjusted so that it
988 no longer exhibits ridiculously poor performance when constructing
989 instances corresponding to C structs.
991 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
992 * minor incompatible change: the sb-grovel contrib now treats C
993 structures as alien (in the sense of SB-ALIEN) objects rather than
994 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
995 has implications for memory management of client code
996 (sb-grovel-returned objects must now be manually managed) and for
997 type safety (alien objects now have full types).
998 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
999 used to control emission of compiler diagnostics, rather than the
1000 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1001 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1002 quality should be considered deprecated.
1003 * install.sh now installs the user manual as well
1004 * (not quite a new documentable feature, but worth considering in
1005 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1006 beginnings of a semantically meaningful condition hierarchy is
1007 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1009 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1010 designator as the defaults argument.
1011 * fixed bug: Displaced arrays whose displaced-to array has become
1012 too small now cause ARRAY-DIMENSION to signal an error, providing
1013 for safer bounds-checking. (reported by Bruno Haible)
1014 * fixed bug: DEFCLASS slot definitions with identical :READER and
1015 :WRITER names now signal a reasonable error. (reported by Thomas
1017 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1019 * fixed bug: Invalid slot specification errors now print correctly.
1020 (thanks to Zach Beane)
1021 * fixed bug 320: Shared to local slot value transfers in class
1022 redefinitions now happen corrently from superclasses as
1023 well. (reported by Bruno Haible)
1024 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1026 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1027 improper lists correctly. (reported by Bruno Haible)
1028 * fixed bug 313: source-transform for <fun-name> was erroneously
1029 applied to a call of a value of a variable with name <fun-name>.
1030 (reported by Antonio Menezes Leitao)
1031 * fixed bug 307: The obsolete instance protocol ensures that
1032 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1033 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1034 remove dead unknown-values globs from the middle of the stack.
1035 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1036 easier to resume long computations after using *BREAK-ON-SIGNALS*
1037 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1038 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1039 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1040 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1041 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1042 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1044 * fixed a (fairly theoretical) bug in string printing: if
1045 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1046 does not have array-element-type equal to the most general string
1048 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1049 function name. (thanks to Nikodemus Siivola)
1050 * fixed bug: compile-time format string checker failed on
1051 non-closed ~{. (reported by Thomas F Burdick)
1052 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1053 don't warn on legal loop code involving more than one aggregate
1054 boolean. (thanks to Nikodemus Siivola)
1055 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1056 in keyword position to MAKE-PACKAGE was not being reported as
1058 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1059 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1061 * fixed bugs 280 and 312: the checking for multiple definitions in a
1062 file is less likely to become confused by uses of inline
1064 * fixed bug: the #S reader macro performs the keyword coercion
1065 specified for slot names. (reported by Kalle Niemitalo)
1066 * fixed bug: lambda lists may contain symbols whose names start with
1067 & but are not lambda-list-keywords; their occurrence triggers a
1069 * fixed bug 321: define-method-combination argument lambda lists do
1070 not cause invalid code to be generated when &optional or &aux
1071 variables have default values. (reported by Bruno Haible)
1072 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1073 have CLOS classes; this appears also to have fixed reported
1074 instability in stack exhaustion detection.
1075 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1076 MUFFLE-WARNING when no associated restart is present is now
1078 * optimization: rearranged the expansion of various defining macros
1079 so that each expands into only one top-level form in a
1080 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1082 * optimization: used a previously little-used slot in symbols to
1083 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1085 * fixed some bugs revealed by Paul Dietz' test suite:
1086 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1088 ** Fixed an optimization bug related to inheritance of initforms
1089 from local to shared slots.
1090 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1091 ** CAST optimizer forgot to flush argument derived type.
1092 ** print/read consistency on floats is now orders of magnitude
1093 more likely. (thanks also to Bruno Haible for a similar report
1095 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1096 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1097 for objects of type REAL. Make it so.
1098 ** SXHASH is specified (infelicitously) to respect similarity,
1099 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1100 it so. (thanks to Markus Ziegler)
1101 ** on the Alpha, the compiler succeeds in compiling functions
1102 returning a known number of arguments greater than 63.
1103 ** fixed handling of invalid NIL arguments in keyword position
1105 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1107 ** provided a readably-printable representation for RANDOM-STATE
1109 ** ensured that pathnames, where they have a namestring, always
1110 print using #P"..." syntax.
1112 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1113 * Support for the forthcoming 2.0 version of the NetBSD kernel
1114 running on x86 hardware has been added. (thanks to Perry
1115 E. Metzger most immediately, and others for their past work)
1116 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1117 current release version; SBCL's OpenBSD support had been broken
1118 since about the time of OpenBSD's switch to ELF binary format.)
1119 * [placeholder for DX summary]
1120 ** user code with &REST lists declared dynamic-extent, under high
1121 speed or space and low safety and debug optimization policy.
1122 * The manual has been converted to Texinfo format and the debugger
1123 chapter from the cmucl manual has been added.
1124 * A facility has been added to extract documentation strings from
1125 sbcl and store them as Texinfo-formatted snippets for inclusion in
1126 the manual (via Texinfo's @include directive)
1127 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1128 the test case to Dave Roberts)
1129 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1130 pointers. (reported by Sean Ross)
1131 * bug fix: PROFILE output is printed nicely even for large numerical
1132 values. (thanks to Zach Beane)
1133 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1134 greater than 32 handle EOF correctly.
1135 * bug fix: on X86 an immediate argument of the IMUL instruction is
1136 correctly printed in disassembly. (thanks to Lutz Euler)
1137 * bug fix: class slots in redefined classes preserve their old
1138 values. (thanks to Bruno Haible and Nikodemus Siivola)
1139 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1140 succeeds. (reported by Marco Baringer)
1141 * fixed some bugs revealed by Paul Dietz' test suite:
1142 ** READ-SEQUENCE now works on ECHO-STREAMs.
1143 ** RATIONALIZE works more according to its specification. (thanks
1146 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1147 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1148 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1149 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1150 should matter to you only if you rebind the printer control
1151 variables and then find you want different bindings in the
1152 debugger than in the ordinary execution of your program.)
1153 * The runtime build system has been tweaked to support building
1154 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1155 assembler and linker. (thanks to Nikodemus Siivola)
1156 * Unbound, undefined, undeclared variables now trigger full
1157 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1158 more often programmer error than deliberate exploitation of undefined
1160 * optimization: the hash algorithm for strings has changed to one
1161 that is less vulnerable to spurious collisions. (thanks to Juho
1163 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1164 needless bounds checking. (thanks to Juho Snellman)
1165 * optimization: implemented multiplication as a modular
1166 (UNSIGNED-BYTE 32) operation on the PPC backend.
1167 * fixed some bugs revealed by Paul Dietz' test suite:
1168 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1170 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1171 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1172 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1174 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1176 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1178 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1179 their output stream on EOF from read.
1180 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1181 have been read to end-of-file.
1182 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1184 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1185 description of determination of which consecutive characters
1187 ** Printing the "Space" character with escaping on now yields "#\\ ",
1188 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1189 ** Reading floating-point numbers with *READ-BASE* set to a number
1190 less than 10 works correctly.
1191 ** Reading floating-point numbers with *READ-BASE* set to a number
1192 more than 10 works correctly.
1193 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1194 the readtable currently in effect.
1196 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1197 * minor incompatible change: parsing of namestrings on a physical
1198 (Unix) host has changed; numbers after the final #\. in a
1199 namestring are no longer interpreted as a version field. This is
1200 intented to be largely invisible to the user, except that the
1201 meaning of the namestring "*.*.*" has changed: it now refers to a
1202 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1203 should usually be replaced by
1204 (make-pathname :name :wild :type :wild :version :wild)
1205 with the added benefit that this is more likely to be portable.
1206 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1207 option to OPEN now signals an error if the file being opened
1208 exists; this may have an impact on existing code.
1209 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1210 signals received, on the PowerPC platforms (both Linux and
1211 Darwin). (thanks to Pierre Mai for pointing out the location of
1213 * several fixes on OS X: The system now builds and runs cleanly on
1214 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1215 (thanks to Brian Mastenbrook)
1216 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1217 second-resolution: integer multiples of 1/3600 between -24 and 24.
1218 (thanks to Vincent Arkesteijn)
1219 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1220 type when called with 1 argument; PEEK-CHAR checked type of
1221 PEEK-TYPE only after having read first character from a
1222 stream. (reported by Peter Graves)
1223 * bug fix: the garbage collector now has much better locality
1224 behaviour, and in particular no longer treats all memory as being
1225 exhausted when many small objects point to each other in a deeply
1227 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1228 recognized as being TYPEP their class.
1229 * bug fix: the PUSHNEW documentation string has been corrected.
1230 (thanks to Vincent Arkesteijn)
1231 * bug fix: defaulting of the value for the last of an atypically
1232 large number of multiple values being bound was not being
1233 performed correctly on the Alpha or PPC platforms
1234 * optimization: implemented multiplication as a modular
1235 (UNSIGNED-BYTE 32) operation on the x86 backend.
1236 * optimization: SEARCH on simple-base-strings can now be open-coded.
1237 (see also contrib/compiler-extras.lisp for inspiration for
1238 teaching the compiler about the Boyer-Moore algorithm).
1239 * value, returned by MAX (and MIN) called with several EQUALP, but
1240 not EQL, arguments now does not depend on compiler settings.
1241 * fixed some bugs revealed by Paul Dietz' test suite:
1242 ** in stack analysis liveness information is propagated from
1243 non-local entry points.
1244 ** pathwise CAST removing failed when the CAST node did not start
1246 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1247 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1249 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1250 host is already defined.
1251 ** RENAME-FILE works on streams instead of signalling an internal
1253 ** PEEK-CHAR uses the current readtable when determining whether
1254 or not a character is whitespace.
1255 ** MERGE-PATHNAMES handles the case when the pathname does not
1256 specify a name while the default-pathname specifies a version.
1257 ** Pathnames now stand a better chance of respecting print/read
1259 ** Attempting to use standardized file system operators with a
1260 pathname with invalid :DIRECTORY components signals a
1262 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1264 ** DIRECTORY on logical pathnames is more correct.
1265 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1266 signal a TYPE-ERROR if their argument is not a stream.
1267 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1268 designator argument does not designate a stream.
1269 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1270 examining the synonym.
1271 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1273 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1274 element-type larger than ([UN]SIGNED-BYTE 32).
1276 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1277 * When built with the :SB-FUTEX feature, threaded builds now take
1278 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1279 for faster/more reliable mutex and condition variable support.
1280 * Incompatible change (but one you probably shouldn't have been using
1281 anyway): the interface and code for arbitrating between multiple
1282 threads in the same user session has been redesigned.
1283 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1284 error when called without an explicit environment argument.
1285 (thanks to Nikodemus Siivola)
1286 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1287 confused about their position. (thanks to Adam Warner and Gerd
1289 * bug fix: Pretty printing backquoted forms with unquotations in the
1290 argument list position of various code constructs such as LAMBDA
1291 now works correctly. (reported by Paul Dietz)
1292 * bug fix: Pretty printing unquotations no longer loses all
1293 stream position information.
1294 * optimization: performance of string output streams is now less
1295 poor for multiple small sequence writes.
1296 * optimization: performance of CSUBTYPEP in the presence of complex
1297 expressions involving CONS and NOT many times has been improved.
1298 (reported by Paul Dietz)
1299 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1301 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1302 streams for the REPL, for the debugger, and so on. (thanks to
1304 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1305 * fixed some bugs revealed by Paul Dietz' test suite:
1306 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1307 is ignored if it is too small, rather than propagating through
1308 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1309 ** extremely complex negations of CONS types were not being
1310 sufficiently canonicalized, leading to inconsistencies in
1312 ** VALUES tranformer lost derived type.
1314 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1315 * fixed a bootstrapping bug: the build process no longer assumes
1316 that the various BOOLE-related constants have the same value in
1317 host and target lisps. (noted by Paul Dietz' test suite on an
1318 SBCL binary built from CLISP)
1319 * The system can now be dynamically linked on the MIPS platform,
1320 which enables dynamic loading of foreign code from Lisp. (thanks
1321 to Ralf Baechle for discussions on the MIPS ABI)
1322 * The system now records debugging information for its own source
1323 files in a filesystem-position-independent manner, relative to
1324 the "SYS" logical host.
1325 * fixed a compiler bug: MV-LET convertion did not check references
1326 to the "max args" entry point. (reported by Brian Downing)
1327 * tweaked disassembly notes to be less confident about proclaiming
1328 some instruction as an LRA. (thanks to Brian Downing)
1329 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1330 now each have their own history, command character, and other
1331 characteristics. (thanks to David Lichteblau)
1332 * fixed some bugs revealed by Paul Dietz' test suite:
1333 ** compiler failure in compiling LOGAND expressions including a
1335 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1336 shift greater than 32.
1337 ** FLUSH-DEST did not mark blocks for type check regeneration.
1338 ** HANDLER-CASE failed to accept declarations in handler clauses
1339 in some circumstances.
1341 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1342 * New code in contrib/sb-introspect (still probably not entirely
1343 stable yet) provides some support for smart Lisp development
1344 environments like SLIME.
1345 * The conditions signalled for errors occurring when loading .fasl
1346 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1347 in a way which should help ASDF recover gracefully.
1348 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1349 stuff has been cleaned up. If you code contrib/ stuff, this might
1350 affect you, and you can look at contrib/README, contrib/STANDARDS,
1351 and/or the 0.8.4.27 diff to check.
1352 * In full calls the compiler now does not generate checks for declared
1353 argument types for all arguments.
1354 * various threading fixes
1355 ** and some experimental patches which didn't make it into
1356 the main tree for this release, but which are shipped in
1357 contrib/experimental-thread.patch as a possible fix for some
1358 failures (deadlock, spinning...) in GC-intensive multithreaded
1360 * fixed PPC build problem (source code incompatibility of different
1361 library versions): added offsetof() hackery which attempts to divine
1362 where glibc maintainers put uc_mcontext today
1363 * fixed bug 282: compiler does not trust type assertions while passing
1364 arguments to a full call.
1365 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1366 &OPTIONAL VALUES type parameter.
1367 * fix bug 214: algorithm for noting rejected templates is now more
1368 similar to that of template seletion. (also reported by rydis on
1370 * fixed bug 141b: printing backquoted information readably and prettily
1371 inserts a space where necessary.
1372 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1373 (CHARACTER 10) are now reported as errors, rather than propagated
1374 as unknown types. (reported by piso on #lisp)
1375 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1376 with values NIL and :ERROR. (thanks to Milan Zamazal)
1377 * fixed bug 191c: CLOS now does proper keyword argument checking as
1378 described in CLHS 7.6.5 and 7.6.5.1.
1379 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1380 longer raise an error; further, using a list as a for-as-arithmetic
1381 counter now raises a meaningful error.
1382 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1383 now understood by sequence creation functions such as MAKE-SEQUENCE
1385 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1386 a STRING-INPUT-STREAM.
1387 * compiler enhancement: SIGNUM is now better able to derive the type
1389 * type declarations inside WITH-SLOTS are checked. (reported by
1391 * fixed some bugs revealed by Paul Dietz' test suite:
1392 ** incorrect optimization of TRUNCATE for a positive first
1393 argument and negative second.
1394 ** compiler failure in let-convertion during flushing dead code.
1395 ** compiler failure while deriving type of TRUNCATE on an
1396 interval, containing 0.
1397 ** ASH of a negative bignum by a negative bignum count now returns
1399 ** intersection of CONS types now canonicalizes properly, fixing
1400 inconsistencies in SUBTYPEP.
1402 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1403 * incompatible change: The --disable-debugger command line
1404 option now clobbers the debugger at a more fundamental
1405 level, by redefining #'INVOKE-DEBUGGER instead of by
1406 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1407 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1408 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1409 enter the debugger REPL and then suffer recursive errors
1410 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1411 while under the new system, BREAK in --disable-debugger mode
1412 terminates the system just as an unhandled error would.
1413 * fixed compiler performance when processing loops with a step >1;
1414 * bug fix: DOCUMENTATION now retrieves generic function
1415 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1416 support has been systematized, and now supports the methods
1417 specified by ANSI, along with a default method and a method for
1418 slot documentation. (reported by Nathan Froyd)
1419 * bug fix: effective methods associated with a generic function are
1420 no longer cached over a change of that generic function's method
1421 combination. (reported by Andreas Fuchs)
1422 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1423 names a non-existent directory. (thanks to Andreas Fuchs)
1424 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1425 the PPC platform, lead to stack corruption; this has been fixed.
1426 (reported by Rainer Joswig)
1427 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1428 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1429 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1430 arithmetic in general, is now much more reliable on the PPC
1432 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1433 platform now returns the right answer.
1434 * optimization: restored some effective method precomputation in
1435 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1436 precomputation is now tunable.
1437 * optimization: compiler-internal data structure use has been
1438 reviewed, and changes have been made that should improve the
1439 performance of the compiler by about 20%.
1440 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1441 simple-base-strings and simple-bit-vectors is improved.
1442 * optimization: the optimization of 32-bit logical and arithmetic
1443 functions introduced in version 0.8.3 on the x86 has been
1444 implemented on the mips, ppc and sparc platforms; an
1445 implementation of the same facility, but for 64-bit arithmetic,
1446 has been added for the alpha.
1447 * microoptimization: the compiler is better able to make use of the
1448 x86 LEA instruction for multiplication by constants.
1449 * bug fix: in some situations compiler did not report usage of
1450 generic arithmetic in (SPEED 3) policy.
1451 * bug 145b fix: compiler used wrong type specifier while converting
1452 MEMBER-types to numeric.
1453 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1455 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1456 index argument now works properly on the Alpha platform.
1457 * bug fix: floating point exception treatment on the Alpha platform
1459 * bug fix: FILE-POSITION works much better on string input and
1460 output streams. (thanks to Nikodemus Siivola)
1461 * bug fix: many threading/garbage collection symptoms sorted.
1462 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1463 might be pseudo-atomic.
1464 * internal change: Stopping for GC is now done with signals not
1465 ptrace. GC is now done in whichever thread wanted it, instead of
1467 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1469 * bug fix: result form in DO is not contained in the implicit
1471 * incompatible change: ICR structure is changed; the value part of
1472 CONTINUATION is now called LVAR; corresponding functions are
1473 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1474 * added type deriver for ISQRT (thanks to Robert E. Brown).
1475 * bug fix: better support for loading from the command line when an
1476 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1478 * fixed some bugs revealed by Paul Dietz' test suite:
1479 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1480 ** ROUND and FROUND now give the right answer when given very
1481 small float arguments.
1482 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1484 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1485 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1486 ** LCM should return a non-negative integer.
1487 ** PARSE-INTEGER returned the index of a terminator instead of the
1488 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1489 ** PARSE-INTEGER returned an incorrect index being applied to a
1491 ** LCM with two arguments of 0 returns 0 rather than signalling
1493 ** unsigned addition of a 32-bit constant with the high bit set no
1494 longer causes an internal compiler error on the x86.
1495 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1496 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1497 with negative last argument.
1498 ** byte specifiers with zero size and position no longer cause
1499 an error during type derivation.
1500 ** bignum multiplication on the Alpha platform now returns the
1502 * porting: The system now builds on SuSE AMD64, although it still
1503 generates a 32-bit binary.
1504 * .fasl file incompatibility: The fasl file version number has
1505 been incremented (because of the changes to internal compiler
1506 data structures referred to above).
1508 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1509 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1510 more accurately, on the Darwin kernel running on PowerPC hardware.
1511 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1512 * Compiler code deletion notes now signal a condition of type
1513 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1514 an associated MUFFLE-WARNING restart.
1515 * The compiler now performs limited argument count validation of
1516 constant format strings in FORMAT, and where appropriate in ERROR,
1517 CERROR and WARN. (thanks to Gerd Moellmann)
1518 * New ASDF-INSTALL contrib can be used for automatic download and
1519 installation of third-party Lisp code from CCLAN or other sites
1521 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1522 which forces another thread to execute a function supplied by the
1524 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1525 now accept and act upon their :ELEMENT-TYPE keyword argument.
1526 (reported by Martin Atzmueller, Edi Weitz)
1527 * bug fix: FILE-POSITION now accepts position designators up to
1528 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1529 is the greater. (thanks to Patrik Nordebo)
1530 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1531 Kalle Olavi Niemitalo)
1532 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1533 printable. (reported by Eric Marsden)
1534 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1535 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1536 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1537 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1538 (thanks to Stig Sandoe)
1539 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1540 UNREAD-CHAR now correctly includes the unread character in the
1541 target sequence. (thanks to Gerd Moellmann)
1542 * bug fix in threaded builds: the system can now be suspended and
1543 resumed by shell job control with minimal disruption.
1544 * bug fixes in times and timezones >2038AD
1545 * better handling of "where is GNU make?" problem in build scripts
1546 (thanks to Nikodemus Siivola)
1547 * new optimization: inside a named function any reference to a
1548 function with the same name is considered to be a self-reference;
1549 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1550 optimization quality.
1551 * new optimization on x86: logical functions and + now have
1552 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1553 used when the result is truncated to 32 bits.
1554 * VALUES declaration is partially enabled.
1555 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1556 * fixed some bugs revealed by Paul Dietz' test suite:
1557 ** The system now obeys the constraint imposed by
1558 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1559 types form a lattice under type intersection.
1560 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1561 ** ASSOC now ignores NIL elements in an alist.
1562 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1563 and (1+ MOST-POSITIVE-FIXNUM) answers.
1564 ** The addition of a method with invalid qualifiers to a generic
1565 function does not cause an error to be signalled immediately;
1566 a warning is signalled, and the error is generated only on
1567 calling the generic function.
1568 * changed .fasl file version number, in part to add type codes for
1569 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1570 obscure ANSI requirements
1572 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1573 * fixed bug 148: failure to inline-expand a local function left
1574 garbage, confusing the compiler.
1575 * fixed bugs 3cd: structure slot readers perform type check if the
1576 slot can have an invalid value (i.e. it is either not initialized
1577 or can be written with a less specific slot writer).
1578 * bug fix: the compiler now traps array references to elements off
1579 the end of an array; previously, the bounds checking in some
1580 circumstances could go off-by-one.
1581 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1583 * type declarations for array element types now obey the description
1584 on the CLHS page "Declaration TYPE", as per discussions on
1585 sbcl-help around 2003-05-08. This means that a declaration
1586 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1587 declaration, all references to BAR will be asserted or assumed
1588 (with THE, so dependent on compiler policy) to involve objects of
1589 type FOO. Note that no such declaration is implied in
1590 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1591 * declared types of functions from the "Conditions"
1592 chapter. (reported by Paul Dietz)
1593 * bug fix: CERROR accepts a function as its first argument.
1594 * bug fix: NTH an NTHCDR accept a bignum as index
1595 arguments. (reported by Adam Warner)
1596 * optimization: character compare routines now optimize comparing
1597 against a constant character. (reported by Gilbert Baumann)
1598 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1599 argument now works properly on the MIPS platform.
1600 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1601 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1602 declared to have a functional type, does not cause an error
1604 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1605 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1606 * fixed bug: if last continuation of a deleted block has a
1607 destination, this destination should be deleted too. (reported by
1609 * fixed a bug in the bootstrap process: the host compiler's values
1610 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1611 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1612 test case from Patrik Nordebo)
1613 * improved the ability of the disassembler on the PPC platform to
1614 provide helpful disassembly notes.
1615 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1616 the class in more cases than previously.
1617 * bug fix: FILE-POSITION now understands :START and :END for
1618 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1619 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1620 8) data. (thanks to David Lichteblau)
1621 * bug fix: it is possible to add a method to a generic function
1622 without lambda list.
1623 * bug fix: reader failed to signal END-OF-FILE inside an
1624 object representation. (reported by Nikodemus Siivola)
1625 * fixed some bugs revealed by Paul Dietz' test suite:
1626 ** LAST and [N]BUTLAST should accept a bignum.
1627 ** condition slot accessors are methods.
1628 ** (VECTOR NIL) is a subtype of STRING.
1630 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1631 * minor incompatible change: some nonsensical specialized lambda
1632 lists (used in DEFMETHOD) which were previously ignored now signal
1634 * minor incompatible change: the system is now aware of the types of
1635 variables in the COMMON-LISP package, and will signal errors for
1636 most violations of these type constraints (where previously they
1637 were silently accepted).
1638 * minor incompatible change: COMPILE-FILE now uses the freedom
1639 afforded (ANSI 3.2.2.3) to use derived function types for
1640 functions defined in the same file. This also permits the system
1641 to warn on static type mismatches and function
1642 redefinition. (Currently it does not work with high DEBUG level.)
1643 * minor incompatible change: VALUES declaration is disabled.
1644 * When issuing notes, the compiler now signals a condition of type
1645 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1646 restart for use in user handlers. It is expected that the
1647 COMPILER-NOTE condition will eventually become a condition
1648 supertype to a hierarchy of note types, which will then be
1649 handleable in a similar fashion. However, at the moment, no such
1650 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1651 but it's an implementation detail, not a classification for the
1653 * Changes in type checking closed the following bugs:
1654 ** type checking of unused values (192b, 194d, 203);
1655 ** template selection based on unsafe type assertions (192c, 236);
1656 ** type checking in branches (194bc).
1657 * A short form of VALUES type specifier has ANSI meaning (it has
1658 increased the number of situations when SBCL cannot perform type
1660 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1661 keywords or constants is permissible.
1662 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1663 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1664 classes. (thanks to Antonio Martinez)
1665 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1666 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1667 argument to a pretty-print stream. Instead, it leaves any such
1668 operations to DESCRIBE-OBJECT methods.
1669 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1671 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1672 lambda lists. (thanks to David Lichteblau)
1673 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1674 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1675 (thanks to Gerd Moellmann)
1676 * fixed bug 246: increased compilation speed of long
1677 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1679 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1681 * DEFCONSTANT now throws a condition of type
1682 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1683 constant to a non-EQL value; CONTINUE and ABORT restarts
1684 respectively change and preserve the value.
1685 * fixed bug 63: The code walker, part of the implementation of CLOS,
1686 is now better at handling symbol macros.
1687 * bug fix: There is no longer an internal implementation type named
1688 CL:LENGTH. (reported by Raymond Toy)
1689 * bug fix: In macro-like defining macros/special operators the
1690 implicit block does not enclose the lambda list.
1691 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1692 enclosed in parentheses) are not suitable as type specifiers, and
1693 their use properly signals an error now.
1694 * bug fix: An argument count mismatch for a type specifier in code
1695 being compiled no longer causes an unhandled error at compile
1696 time, but signals a compile-time warning.
1697 * fixed simple vector readable printing
1698 * bug fix: DESCRIBE takes more care over whether the class
1699 precedence list slot of a class is bound before accessing it.
1700 (reported by Markus Krummenacker)
1701 * bug fix: FORMATTER can successfully compile pretty-printer format
1702 strings which use variants of the ~* directive inside.
1703 * bug fix: SEARCH now applies its TEST predicate to the elements of
1704 the arguments in the correct order. (thanks to Wolfhard Buss)
1705 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1706 in type inference. (thanks to Robert E. Brown)
1707 * bug fix: Reading in symbols with an explicit package name of ""
1708 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1709 * fixed some bugs revealed by Paul Dietz' test suite:
1710 ** NIL is now allowed as a structure slot name.
1711 ** Arbitrary numbers, not just REALs, are allowed in certain
1712 circumstances in LOOP for-as-arithmetic clauses.
1713 ** Multiple class redefinitions before slot access no longer
1714 causes a type error.
1715 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1716 association between the name and a class.
1717 ** Generic functions with non-standard method-combination and over
1718 five methods all of which return constants no longer return NIL
1719 after the first few invocations. (thanks to Gerd Moellmann)
1720 ** CALL-NEXT-METHOD with no arguments now passes the original
1721 values of the arguments, even in the presence of assignment.
1722 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1724 ** DISASSEMBLE works with closures and funcallable instances.
1725 ** ADD-METHOD now returns the generic function, not the new method.
1726 ** FIND-METHOD signals an error if the lengths of the specializers
1727 is incompatible with the generic function, even if the ERRORP
1729 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1730 which its argument is a member.
1731 ** DEFCLASS only redefines the class named by its class-name
1732 argument if that name is the proper name of the class;
1733 otherwise, it creates a new class.
1734 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1735 of the UNBOUND-SLOT condition to the name of the slot.
1736 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1737 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1738 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1739 SLOT-MAKUNBOUND in the specified fashion.
1741 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1742 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1743 cross-compilation host. As a consequence, we can now bootstrap our
1744 way up to SBCL starting with a bare gcc toolchain and human-readable
1745 source code (first the source to CLISP, then the source to SBCL).
1746 * A contributed module containing a partial implementation of the
1747 simple-streams interface has been included. (thanks to Rudi
1749 * A contributed module implementing the RFC1321 Message Digest
1750 Algorithm, known as MD5, has been included.
1751 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1752 no longer has any effect, as the code controlled by this feature
1753 has been deleted. (As far as we know, no-one has ever built using
1754 this feature, and its semantics were confused in any case).
1755 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1756 work on conditions (as required by the ANSI specification),
1757 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1758 expected behaviour on conditions. Users should note, however,
1759 that such behaviour is not required by the ANSI specification,
1760 and so use of this behaviour may render their code unportable.
1761 * fixed some bugs revealed by Paul Dietz' test suite:
1762 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1764 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1765 ** SXHASH on condition objects no longer returns NIL.
1766 ** :ALLOCATION :CLASS slots are better treated; their values are
1767 updated on class redefinition, and initforms inherited from
1768 superclasses are applied.
1769 ** REMOVE-METHOD returns its generic function argument even when
1770 no method was removed.
1771 ** SHARED-INITIALIZE now initializes the values of the requested
1772 slots, including those with :ALLOCATION :CLASS.
1773 ** ALLOCATE-INSTANCE now works on structure classes defined via
1774 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1776 ** SLOT-EXISTS-P now works on conditions, as well as structures
1778 ** MAKE-LOAD-FORM now has the required methods on
1779 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1780 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1781 keyword, and now implements the SLOT-NAMES argument.
1782 ** methods with &OPTIONAL arguments no longer allow too many
1783 arguments to be passed in the call without error.
1784 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1785 option is consistent with the required arguments of the generic
1786 function lambda list.
1787 * bug fix: REQUIRE accepts a string designator. (Thanks to
1789 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1790 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1791 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1792 * bug fix: The compiler now checks for duplicated variables in macro
1794 * bug fix: SETQ on globals returns the correct value.
1795 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1796 not a COMPILER-ERROR (followed by some other strange error on
1797 choosing the CONTINUE restart).
1798 * bug fix: make.sh and friends are now more consistent in the way that
1799 they look for GNU "make".
1801 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1802 * experimental native threads support (on x86 Linux >=2.4 only).
1803 This is not compiled in by default: you need to add :SB-THREAD to
1804 the target features. See the "Beyond ANSI" chapter of the manual
1806 * fix for longstanding nonANSIism: The old distinction between
1807 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1808 The return value from CL:FIND-CLASS is now a CLOS class, and
1809 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1810 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1811 * An interface to the MetaObject Protocol, as described in Kiczales,
1812 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1813 MIT Press, 1991, is available from the SB-MOP package.
1814 * incompatible change: the SB-PCL package should now be considered
1815 a private implementation detail, and no longer a semi-private MOP
1817 * minor incompatible change: due to rearrangement for threads, the
1818 control stack and binding stack are now allocated at arbitrary
1819 addresses instead of being hardcoded per-port. Users affected by
1820 this probably have to be doing advanced things with shared
1821 libraries, and will know who they are.
1822 * minor incompatible change: Previously, all --eval forms used were
1823 processed with READ before any of them were processed with EVAL.
1824 Now each --eval form is processed with both READ and EVAL before
1825 the next --eval form is processed. (Thus package operations like
1826 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1827 work as the user might reasonably expect.)
1828 * minor incompatible change: *STANDARD-INPUT* is now only an
1829 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1831 * minor incompatible change: Y-OR-N-P is now character-oriented,
1832 not line oriented. Also, YES-OR-NO-P now works without errors.
1833 (thanks to Antonio Martinez)
1834 * sb-aclrepl module improvements: an integrated inspector, added
1835 repl features, and a bug fix to :trace command.
1836 * Known functions, which cannot be open coded by the backend, are
1837 considered to be able to check types of their arguments. (fixing
1838 a bug report by Nathan J. Froyd)
1839 * fixed a bug in computing method discriminating functions: It is
1840 now possible to define methods specialized on classes which have
1841 forward-referenced superclasses. (thanks to Gerd Moellmann)
1842 * fixed evaluation order in optional entries (reported by Gilbert
1844 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1845 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1846 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1847 required-by-AMOP NAME argument, as well as CLASS and
1848 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1849 * fixed bug 20: DEFMETHOD can define methods using names that are
1850 not the proper names of classes to designate class specializers.
1851 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1852 Unix file descriptors, instead of blowing up. (thanks to Antonio
1854 * Garbage collection refactoring: user-visible change is that a
1855 call to the GC function during WITHOUT-GCING will not do garbage
1856 collection until the end of the WITHOUT-GCING. If you were doing
1857 this you were probably losing anyway.
1858 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1859 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1860 TYPEP the latter but not the former.
1861 * The compiler issues a full WARNING for calls to undefined functions
1862 with names from the CL package.
1863 * MAP-INTO for a vector destination is open coded. (reported by
1864 Brian Downing on c.l.l)
1865 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1866 documentation string.
1867 * fixed some bugs revealed by Paul Dietz' test suite:
1868 ** COPY-ALIST now signals an error if its argument is a dotted
1870 ** Condition slots are now accessed more correctly in the presence
1871 of multiple initargs for a given slot.
1872 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1873 exclude restarts of the same name associated with a different
1875 ** DEFCLASS of forward-referenced classes with another
1876 forward-referenced class in the superclasses list no longer
1878 ** Condition slots are now initialized once each, not multiple
1879 times. (thanks to Gerd Moellmann)
1880 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1882 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1883 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1884 arguments contain duplicated elements.
1885 ** RESTART-CASE understands local macros.
1886 ** RESTART-CASE associates exactly its own restarts with a condition.
1887 ** ENDP in safe mode checks its argument to be of type LIST.
1888 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1889 in question is unbound.
1890 ** Optimized MAKE-INSTANCE functions no longer cause internal
1891 assertion failures in the presence of duplicate initargs.
1892 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1893 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1894 propagated even in the ctor optimized implementation of
1896 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1898 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1899 * a better implementation of SXHASH on (simple) bit vectors,
1900 measured both in execution speed and in distribution of results
1901 over the positive fixnums, has been installed. Likewise, a better
1902 implementation of EQUAL for simple bit vectors is now available.
1903 * fixed CEILING optimization for a divisor of form 2^k.
1904 * fixed bug 240 (emitting extra style warnings "using the lexical
1905 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1906 by Antonio Martinez)
1907 * fixed SXHASH, giving different results for NIL depending on type
1908 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1909 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1910 variable. (found by Rolf Wester)
1911 * SBCL does not ignore type declarations for special
1912 variables. (reported by rif on c.l.l 2003-03-05)
1913 * some bug fixes in contrib/sb-aclrepl/
1914 * fixed some bugs revealed by Paul Dietz' test suite:
1915 ** a bug in the CONS type specifier, whereby the CAR and CDR
1916 types got intertwined, has been fixed;
1917 ** the type system is now able to reason about the interaction
1918 between INTEGER and RATIO types more completely;
1919 ** APPEND, [N]REVERSE and NRECONC check that those their
1920 arguments, which must be proper lists, are really so;
1921 ** An array specialized to be unable to hold elements has been
1922 implemented, as required -- yes, really -- by ANSI;
1923 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1924 on malformed property lists;
1926 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1927 * incompatible packaging change: in line with Unix convention,
1928 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1929 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1930 * REQUIRE and PROVIDE are now optionally capable of doing something
1931 useful. See the documentation string for REQUIRE.
1932 * infrastructure for a managed SBCL contrib system: contributed
1933 modules in this release include:
1934 ** the ASDF system definition facility;
1935 ** an interface to the BSD Sockets API;
1936 ** an ACL-like convenience interface to the repl;
1937 (thanks to Kevin Rosenberg)
1938 ** an implementation of ROTATE-BYTE, with efficient implementation
1940 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1941 gives a closer approximation to the right answer than previously.
1942 (thanks to Raymond Toy)
1943 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1944 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1945 optional environment arguments, as required by ANSI.
1946 * fixed bugs in other functions taking environment objects, allowing
1947 calls with an explicit NIL environment argument to be compiled
1949 * fixed bug 228: primary return values from
1950 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1951 COMPILE or FUNCTION.
1952 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1953 :TYPE will no longer signal errors on innocuous objects.
1954 * fixed bug 231b: SETQ is better at respecting type declarations in
1955 the lexical environment.
1956 * fixed a bug in DEFCLASS: classes named by symbols with no or
1957 unprintable packages can now be defined.
1958 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1959 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1960 * fixed compiler failure related to checking types of functions.
1961 (reported by Robert E. Brown)
1962 * the compiler is now much more consistent in its error-checking
1963 treatment of bounding index arguments to sequence functions: in
1964 (SAFETY 3) code, errors will be signalled in almost all cases if
1965 invalid sequence bounding indices are passed to functions defined
1966 by ANSI to operate on sequences.
1967 * fixed a bug in the build procedure: documentation of SBCL-specific
1968 packages is now preserved and available in the final Lisp image.
1969 * lifted FDEFINITION lookup out of loops in the implementation of
1970 many list operations. (thanks to Robert E. Brown)
1971 * fixed a bug in the reader: the #n# reader macro now works for
1972 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1973 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1974 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1975 E. Brown and Tony Martinez respectively)
1976 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1977 SLOT-BOUNDP: in particular, it is now able to optimize them much
1978 better, and is now not vulnerable to having packages renamed.
1979 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1980 always called when a slot is not present in an instance. (thanks
1982 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1983 longer signal an error on unreading a character following EOF on
1984 the previous constituent stream. (thanks to Tony Martinez)
1985 * fixed some bugs revealed by Paul Dietz' test suite:
1986 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1987 not just nonnegative fixnums;
1988 ** the logical bit-array operators such as BIT-AND now accept an
1989 explicit NIL for their "opt-arg" argument (to indicate a
1990 freshly-consed result bit-array);
1991 ** ELT now signals an error on an invalid sequence index in safe
1993 ** the type system is now cleverer about negations of numeric
1994 types, and consequently understands the BIGNUM and RATIO types
1996 ** the type system is now cleverer about the interaction between
1997 INTEGER and RATIO types: while bugs still remain, many more
1998 cases are accurately computed;
1999 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2000 if it is in the last clause;
2001 ** CONSTANTLY now correctly returns a side-effect-free function in
2003 ** DECLARE is no longer treated as a special-operator; in
2004 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2005 * incremented fasl file version number due to the change in the
2006 DEFSTRUCT-SLOT-DESCRIPTION structure.
2008 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2009 * minor incompatible change: code processed by the "interpreter" or
2010 EVAL now has a compilation optimization policy of (DEBUG 2)
2011 (changed from (DEBUG 1)) to improve debuggability of interactive
2012 development, and to allow the use of the debug RETURN command in
2014 * an experimental implementation of the RETURN command for the
2015 debugger has been included. (thanks to Frederik Kuivinen)
2016 * fixed bug 62: constraints were not propagated into a loop.
2017 * fixed bug in embedded calls of SORT (reported and investigated by
2019 * fixed some bugs revealed by Paul Dietz' test suite:
2020 ** printing and reading of arrays with some dimensions having
2021 length 0 (thanks to Gerd Moellmann);
2022 ** BOA constructor with &AUX argument without a default value does
2023 not cause a type error;
2024 ** CONSTANTP now returns true for all self-evaluating objects.
2026 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2027 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2028 accessors that are related by inheritance, as specified in the
2029 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2030 Valtteri Vuorikoski)
2031 * The compiler is now able to inline functions that were defined in
2032 a complex lexical environment (e.g. inside a MACROLET).
2033 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2035 * Support for the upcoming FreeBSD-5.0 release has been included.
2036 (thanks to Dag-Erling Smorgrav)
2037 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2038 effect when it is not in a toplevel context.
2039 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2040 stand a better chance of being correct. (thanks to Gerd
2042 * fixed bug in COERCE, which now signals an error on coercing a
2043 rational to a bounded real type which excludes the expected
2045 * The compiler is now able to derive types more accurately from the
2046 COERCE and COMPILE functions.
2047 * fixed bug 223: functional binding is considered to be constant
2048 only for symbols in the CL package.
2049 * fixed bug 231: SETQ did not check the type of a variable being set
2050 (reported by Robert E. Brown)
2051 * A new optimization for MAKE-INSTANCE has been included, fixing
2052 various bugs (including relating to :ALLOCATION :CLASS slots and
2053 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2055 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2056 ** As required by ANSI, LOOP now disallows anonymous collection
2057 clauses such as COLLECT I in conjunction with aggregate boolean
2058 clauses such as THEREIS (= I 1);
2059 ** LOOP now signals an error when any variable is reused in the
2060 same loop (including the potentially useful construct analogous
2061 to WITH A = 1 WITH A = (1+ A);
2062 ** IT is only a special loop symbol within the first clause of a
2063 conditional loop clause;
2064 ** LOOP with a typed iteration variable over a hashtable now
2065 signals a type error iff it should.
2066 * fixed some other bugs revealed by Paul Dietz' test suite:
2067 ** FILE-STREAM now names the class previously known as FD-STREAM;
2068 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2069 argument) no longer signals an error;
2070 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2071 of the slot symbol, rather than using the current package
2072 ((:CONC-NAME "") continues to intern the slot's name in the
2074 * incremented fasl file version number, because of the incompatible
2075 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2076 of the new implementation of DEFINE-COMPILER-MACRO.
2078 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2079 * Support for building SBCL for MIPS platforms running in
2080 little-endian mode has now been checked in, and basic
2081 functionality on said platforms verified.
2082 * minor incompatible change: PCL now records the pathname of a file
2083 in which methods and the like are defined, rather than its
2085 * minor incompatible change: TRUENAME now considers the truename of
2086 a file naming a directory to be the pathname with :DIRECTORY
2087 component indicating that directory.
2088 * minor incompatible change: a NAMED clause in the extended form of
2089 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2090 reason for the previous behaviour is unclear.
2091 * more systematization and improvement of CLOS and MOP conformance
2092 in PCL (thanks to Gerd Moellman and Pierre Mai):
2093 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2095 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2096 primary methods with no specializers;
2097 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2099 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2100 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2101 FORWARD-REFERENCED-CLASSes; error reporting on
2102 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2104 ** SXHASH on CLOS instances now uses a slot internal to the
2105 instance to return different numbers on distinct instances,
2106 while preserving the same return value through invocations of
2108 ** DEFMETHOD signals errors when methods with longer incongruent
2109 lambda lists are added to generic functions;
2110 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2111 CLASS, as specified in AMOP;
2112 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2113 on the order returned by the primary method for classes of
2114 class STANDARD-CLASS;
2115 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2116 * fixed some bugs shown by Paul Dietz' test suite:
2117 ** DOLIST puts its body in TAGBODY;
2118 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2120 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2121 value producing form;
2122 ** if more variables are given to PROGV than values, extra
2123 variables are bound and made to have no value;
2124 ** NSUBSTITUTE on list arguments gets the right answer with
2126 ** ELT signals an error of type TYPE-ERROR when the index argument
2127 is not a valid sequence index;
2128 ** LOOP signals (at macroexpansion time) an error of type
2129 PROGRAM-ERROR when duplicate variable names are found;
2130 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2131 ** FUNCALL of special-operators now cause an error of type
2133 ** PSETQ now works as required in the presence of side-effecting
2134 symbol-macro places;
2135 ** NCONC accepts any object as its last argument;
2136 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2138 ** loop-for-as-package does not require a package to be explicitely
2140 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2142 * fixed bug 166: compiler preserves "there is a way to go"
2143 invariant when deleting code.
2144 * fixed bug 172: macro lambda lists with required arguments after
2145 &REST arguments now cause an error to be signalled. (thanks to
2147 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2148 bug. (thanks to Matthew Danish)
2149 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2151 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2152 arguments are lexically rebound. (thanks to Gerd Moellmann and
2154 * fixed bug 194: error messages are now more informative when there
2155 is no primary method applicable in a call to a generic
2156 function. (thanks to Gerd Moellmann)
2157 * fixed bug in command line argument checking (thanks to Julian
2159 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2160 (thanks to Matthew Danish)
2161 * incremented fasl file version number, because of the
2162 SXHASH-related changes in the layout of CLOS data structures
2164 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2165 * minor incompatible change: The runtime (the Unix executable named
2166 "sbcl") is now much pickier about the .core files it will load.
2167 Essentially it now requires .core files to descend from the same
2168 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2169 as the runtime does. (The intent is to prevent the crashes which
2170 can occur, and which can even be reported as mysterious failures,
2171 when people patch the sources or change the build parameters
2172 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2173 sbcl and .core files.)
2174 * fixed bug: VALUES-LIST is no longer optimized away.
2175 * fixed bug 142: The FFI conversion of C string values to Lisp
2176 string values no longer conses excessively. (thanks to Nathan
2177 Froyd porting Raymond Toy's fix to CMU CL)
2178 * began to systematize and improve MOP conformance in PCL (thanks to
2179 Nathan Froyd, Gerd Moellman and Pierre Mai):
2180 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2182 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2183 ** FINALIZE-INHERITANCE is now called on class finalization;
2184 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2185 argument precedence order.
2186 * fixed bug 202: The compiler no longer fails on functions whose
2187 derived types contradict their declared type.
2188 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2189 so it can be non-toplevel.
2190 * The fasl file version number has changed (because of the new
2191 implementation of DEFMACRO).
2192 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2193 safe code, that any length requirement by their type-specifier
2194 argument is valid. The exceptions to this are described in bug
2196 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2197 functions now check the types of their inputs as required by ANSI.
2198 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2199 introduced symbol is DECLAREd to be SPECIAL.
2200 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2201 * fixed bug: :COUNT argument to sequence functions may be negative
2202 * fixed bug: body of DO-SYMBOLS may contain declarations
2203 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2204 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2205 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2206 symbol macro only once
2207 * fixed printing of call frame when argument list is unavailable
2208 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2209 * compiler no longer signals WARNING on unknown keyword
2212 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2213 * A beta-quality port to the mips architecture running Linux,
2214 based on the old CMUCL backend, has been made. It has been tested
2215 on a big-endian kernel, and works sufficiently well to be able to
2216 rebuild itself; it has not been tested in little-endian mode.
2217 * fixed an inconsistency between gencgc.c and purify.c which made
2218 dumping/loading .core files unreliable
2219 * fixed bug 120a: The compiler now deals correctly with IFs where
2220 the consequent is the same as the alternative, instead of
2221 misderiving the return type. (thanks to Alexey Dejneka)
2222 * fixed bug 113: Logical pathnames are now dumpable (the logical
2223 host is resolved at load-time, throwing an error if it is not
2225 * fixed bug 174: FORMAT's error message is slightly clearer when a
2226 non-printing character is used in a format directive.
2227 * fixed several bugs in compiler checking of type declarations, i.e.
2228 violations of the Python "declarations are assertions" principle
2229 (thanks to Alexey Dejneka)
2230 * fixed several bugs in PCL's error checking (thanks to Gerd
2232 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2234 * fixed bug in compilation of functions as first class values
2235 (thanks to Antonio Martinez-Shotton)
2236 * The compiler's handling TYPE-ERRORs which it can prove will
2237 inevitably happen at runtime has been cleaned up and corrected
2238 in several ways. (thanks to Alexey Dejneka)
2239 * improved argument type checking for various basic arithmetic
2240 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2241 lavished on them in the past that they can be compiled in many
2242 ways in different special cases
2243 * fixed bug 181: compiler checks validity of user supplied type
2245 * cleaned up code flushing in optimization: Function calls which
2246 should signal errors for safety purposes (e.g. which ANSI says
2247 should signal errors when their arguments are of incorrect type)
2248 are no longer optimized away.
2249 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2250 * incremented fasl file version number, because changes in the
2251 implementation of sequence functions like COERCE caused
2252 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2253 in old inline expansions) to become undefined. (Actually these
2254 changes were later undone, so we might very well be binary
2255 compatible with 0.7.7 after all, but leaving the version number
2256 incremented seemed like the simplest and most conservative
2259 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2260 * An alpha-quality port to the parisc architecture running Linux,
2261 based on the old CMUCL backend, has been made. This, even more so
2262 than the other backends, should be considered still a work in
2263 progress; known problems include that the Linux kernel in 64-bit
2264 mode does not propagate the correct sigcontext structure to
2265 userspace, and consequently SBCL on a parisc64 kernel will not
2267 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2268 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2269 suppressed.) Also now that the compiler is looking at declarations
2270 in the environment, it checks optimization declarations as well,
2271 and suppresses inlining when (> DEBUG SPEED).
2272 * More fixes have been made to treatment of floating point exception
2273 treatment and other Unix signals. In particular, floating point
2274 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2275 * The detection and handling of control stack exhaustion (infinite
2276 or very deeply nested recursion) has changed. Stack exhaustion
2277 detection is now done by write-protecting pages at the OS level
2278 and applies at all optimization settings; when found, a
2279 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2280 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2281 be caught using IGNORE-ERRORS.
2282 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2283 correctly when transforming recursive LABELS functions to LETs.
2284 (thanks to Alexey Dejneka)
2285 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2286 that are names of constants or global variables.
2287 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2288 alien routines with docstrings.
2289 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2290 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2292 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2293 (thanks to Eric Marsden)
2294 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2295 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2296 * Bug fix: The SPARC backend can now compile functions involving
2297 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2298 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2299 to a functions expecting a PATHNAME-DESIGNATOR.
2300 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2301 lambda lists. (thanks to Alexey Dejneka)
2302 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2303 functions so that EXPORTing the name of the class doesn't cause
2304 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2305 to Antonio Martinez for reporting this)
2306 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2307 constant-folded without failing an assertion. (thanks to Einar
2308 Floystad Dorum for reporting this)
2309 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2310 types are now dealt with more correctly by the compiler.
2311 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2312 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2313 behaviour analogous to RENAME-FILE. This puts its behaviour more
2314 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2316 * The fasl file version number has changed again. (because of the
2317 bug fix involving the names of PCL MAKE-INSTANCE functions)
2319 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2320 * bug fix: Floating point exceptions are treated much more
2321 consistently on the x86/Linux and PPC/Linux platforms.
2322 * Array initialization with :INITIAL-ELEMENT is now much faster for
2323 cases when the compiler cannot open code the array creation, but
2324 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2325 array accesses have also seen a speed increase.
2326 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2327 specified. (This was at the root of some bad interactions between
2328 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2330 * bug fix: Internal error arguments for undefined functions are now
2331 computed correctly on the PPC/Linux platform.
2332 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2333 Raymond Toy's patch for CMU CL)
2334 * Support for the Solaris 9 operating environment has been included
2335 (thanks to Daniel Merritt)
2336 * A very ugly but hopefully complete draft of the missing FFI chapter
2337 of the manual has been created by reformatting the corresponding
2338 CMU CL manual chapter into (currently very ugly and incoherent)
2339 DocBook and bringing it up to date for SBCL behavior. Thus, the
2340 manual is now essentially complete, at least by my extreme
2341 once-and-only-once standards, whereby it's acceptable to refer to
2342 the doc strings of SB-EXT functions as the primary documentation.
2343 * The fasl file version number has changed again, due to cleanup of
2344 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2345 is no longer a static symbol.)
2347 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2348 * SBCL now builds with OpenMCL (version 0.12) as the
2349 cross-compilation host; also, more progress has been made toward
2350 bootstrapping under CLISP.
2351 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2353 * bug 158 fixed: The compiler can now deal with integer loop
2354 increments different from 1; fixing this turned out also to fix
2356 * bug 169 fixed: no more bogus warnings about using lexical bindings
2357 despite the presence of perfectly good SPECIAL declarations (thanks
2358 to David Lichteblau)
2359 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2360 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2361 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2363 * bug 180 fixed: Method combination specifications no longer ignore
2364 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2365 * bug fix: Structure type predicate functions now check their argument
2366 count as they should.
2367 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2368 correctly. (thanks to Pierre Mai)
2369 * minor incompatible change: The --noprogrammer option is deprecated
2370 in favor of the new --disable-debugger option, which is very similar.
2371 (The major difference is that it takes effect at a slightly different
2372 time at startup, causing handling of errors in --sysinit and
2373 --userinit files will be affected differently.) The
2374 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2375 been added to allow this functionality to be controlled from ordinary
2376 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2377 maintainers, who might want to run non-interactive scripts to
2378 build SBCL cores which will later be used interactively.)
2379 * minor incompatible change: The LOAD function no longer, when given
2380 a wild pathname to load, loads all files matching that pathname.
2381 Instead, an error of type FILE-ERROR is signalled.
2383 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2384 * bug 147 fixed: The compiler preserves its block link/count
2385 invariants more correctly now so that it doesn't crash. (thanks
2387 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2389 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2390 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2391 no-value-for-this-character as NIL (as specified by ANSI).
2392 * HOST-NAMESTRING on physical pathnames now returns a string that is
2393 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2394 (thanks to Christophe Rhodes)
2395 * The Alpha port handles icache flushing more correctly. (thanks to
2397 * More progress has been made toward bootstrapping under CLISP. (thanks
2398 to Christophe Rhodes)
2399 * The fasl file format has changed again, because dynamic loading
2400 on OpenBSD (which has non-ELF object files) motivated some cleanups
2401 in the way that foreign symbols are transformed and passed around.
2402 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2403 is no longer treated as whitespace by the reader, but instead as
2404 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2405 |A
\7fB|, instead of A as it used to.
2407 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2408 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2409 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2410 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2411 this as of 0.7.1.45, but was left out of the previous news section
2412 (thanks to Dan Barlow)
2413 * SBCL now runs on the Solaris operating system on SPARC architectures
2414 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2415 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2416 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2417 * SPARC backend cleanups, allowing builds of cores optimized for V8
2418 and V9 SPARCS, and also emission of code targeted to a particular
2419 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2421 * SBCL is closer to bootstrapping under CLISP, thanks to various
2422 fixes by Christophe Rhodes.
2423 * The fasl file format has changed again, to allow the compiler's
2424 INFO database to support symbol macros.
2425 * The user manual (in doc/) is formatted into HTML more nicely.
2426 (thanks to coreythomas)
2427 * The system is smarter about SUBTYPEP relationships, especially
2428 those involving NOT types (including types such as ATOM which are
2429 represented internally using NOT types). Thus SUBTYPEP is less
2430 likely to return (VALUES NIL NIL) in general, and in particular
2431 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2433 * The fasl file format has changed again, because the internal
2434 representation of types now includes a new slot to support the new
2435 SUBTYPEP-of-NOT-types logic.
2436 * (not a change in the main branch of SBCL, but a related prototype
2437 which can hopefully be merged into the main branch of SBCL in the
2438 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2439 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2440 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2441 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2442 dispatch macro character combinations. (thanks to Alexey Dejneka)
2443 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2444 that superficially look like logical namestrings correctly.
2445 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2447 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2448 default supplied-p) for &optional and &key arguments. (thanks to
2451 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2452 * incompatible change: The compiler is now less aggressive about
2453 tail call optimization, doing it only when (> SPACE DEBUG) or
2454 (> SPEED DEBUG). (This is an incompatible change because there are
2455 programs which relied on the old CMU-CL-style behavior to optimize
2456 away their unbounded recursion which will now die of stack overflow.)
2457 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2458 for non-GENCGC systems has been increased to 20M (since that
2459 seems much closer to the likely performance optimum for modern
2460 systems than the old 4M value was)
2461 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2462 and *DEBUG-PRINT-LEVEL*
2463 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2464 of CMU CL's support for SPARC, and various endianness and other
2465 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2466 * new syntactic sugar for the Unix command line: --load foo.bar is now
2467 an alternate notation for --eval '(load "foo.bar")'.
2469 ** The system now detects stack overflow and handles it gracefully,
2470 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2471 optimization settings. (This is a good thing in general, and
2472 its introduction in this version should be particularly timely
2473 for anyone whose code fails because of suppression of tail
2475 ** The system now hunts for the C variable "environ" in a more
2476 devious way, to avoid segfaults when the C library version differs
2477 between compile time and run time. (thanks to Christophe Rhodes)
2478 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2479 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2480 ** The compiler no longer issues bogus style warnings for undefined
2481 classes in the same source file as the DEFCLASSes which defined
2482 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2484 ** fixes in CONDITION class precedence list for undefined function
2485 errors (thanks to Alexei Dejneka)
2486 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2487 correctly. (thanks to Dan Barlow)
2488 ** portability fixes aiming at bootstrapping under CLISP (thanks
2489 to Dave McDonald and Christophe Rhodes)
2490 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2491 ** fixes in type translation and and type inference (thanks to
2493 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2494 ** various fixes in the new ports (thanks to Dan Barlow)
2495 * several changes related to debugging:
2496 ** suppression of tail recursion, as noted above
2497 ** stack overflow detection, as noted above
2498 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2499 is now the default. (For some time encapsulation has been more
2500 reliable than the breakpoint-based :ENCAPSULATE NIL
2501 implementation, at least on X86 systems; and I just noticed that
2502 encapsulation also seems closer to the spirit of the ANSI
2505 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2507 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2508 up properly again. (There was a packaging bug in 0.7.0 which
2509 left their definitions in SB-SYS::LOAD-FOREIGN and
2510 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2511 vital for most things which interface to C-level interfaces,
2512 like extensions working with sockets or databases or
2513 Perl-compatible regexes or whatever, and the need to fix
2514 this bug is the main reason that 0.7.1 was released so
2516 ** DEFGENERIC is now choosier about the methods it redefines, so that
2517 reLOADing a previously-LOADed file containing DEFGENERICs does
2518 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2519 cycle now works as it should. (thanks to Alexey Dejneka)
2520 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2521 fixed by Christophe Rhodes. (He actually submitted this patch
2522 months ago, and I delayed until after 0.7.0.)
2523 ** Bug 111 (internal compiler confusion about runtime checks on
2524 FUNCTION types) was fixed by Alexey Dejneka.
2525 * Some internal cleanups (getting rid of variables which aren't
2526 needed now that the byte interpreter is gone) caused the fasl
2527 file format number to change again.
2529 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2530 * major incompatible change: The default fasl file extension, i.e. the
2531 default extension for files produced by COMPILE-FILE, has changed
2532 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2534 ** There are many changes in the implementation of the compiler.
2535 SBCL is now essentially a compiler-only implementation of ANSI
2536 Common Lisp. EVAL still "interprets" a few special cases, but
2537 almost all the interesting cases are handled by creating
2538 a LAMBDA expression, calling COMPILE on it, then calling
2539 FUNCALL on the result.
2540 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2541 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2542 Since the code is newer, there might still be some new bugs
2543 (though not as many as before Martin Atzmueller's fixes:-). But
2544 the new code is substantially simpler and clearer, and hopefully
2545 any remaining bugs will be simpler, less fundamental, and more
2546 fixable then the bugs in the old code.
2547 ** The revised compiler is still a little unsteady on its feet.
2549 *** The debugging information it produces (particularly the names
2550 of FUNCTION objects) is sometimes much less useful than what
2551 the old compiler produced.
2552 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2553 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2554 within a MACROLET) has been temporarily weakened.
2555 ** There are new compiler optimizations for various functions:
2556 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2557 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2558 *** the math functions TRUNCATE, FLOOR, and CEILING
2559 *** the function-of-all-trades COERCE
2560 Mostly these should be transparent, but there's one
2561 potentially-annoying problem (bug 117): when the compiler
2562 inline-expands a function and does type analysis on the result,
2563 it can create control paths which have type mismatches, and
2564 when it can't prove that those control paths aren't taken,
2565 it will issue WARNINGs about the type mismatches. This is
2566 a particular problem in practice for the new sequence functions.
2567 It's not clear how this should be fixed, and for now, a
2568 workaround is given in the entry for 117 in the BUGS file.
2569 ** (Because of the interaction between the two previous items --
2570 occasional inlining problems and new inline expansions -- some
2571 of the new sequence function optimizations won't really kick in
2572 completely until debugging information, and then inlining, are
2573 straightened out in some future version.)
2574 * minor incompatible changes:
2575 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2576 DIRECTORY no longer implicitly promotes NIL slots of its
2577 pathname argument to :WILD. In particular, when you ask for the
2578 contents of a directory (which you used to be able to do without
2579 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2580 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2581 ** changes in behavior that ANSI explicitly defines to be
2582 implementation dependent:
2583 *** The new compiler-only implementation still conforms with ANSI,
2584 but acts a little different than before. Besides the obvious
2585 changes in performance tradeoffs (that the cost per form passed
2586 to EVAL has gone up, and the cost per form executed by EVAL
2587 has gone down), the behavior of the system changes a little
2588 because there are no longer any interpreted function objects.
2589 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2590 e.g. doing COMPILE on the output of interactive DEFUN is
2592 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2594 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2595 has changed from (:CL) to NIL.
2596 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2597 e.g. #\Newline and #\Tab, don't have names specified in the
2598 ANSI Common Lisp standard, is now based on their ASCII symbolic
2599 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2600 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2601 are no longer used for output.
2602 ** changes in internal implementation constants:
2603 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2604 4 million. (If your application spends a lot of time GCing and
2605 you have a lot of RAM, you might want to experiment with
2606 increasing it even more.)
2607 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2608 However, almost all old code should still continue to work without
2609 immediate update, as SB-C-CALL is now a (deprecated) nickname
2611 ** Old operator names in the style DEF-FOO are now deprecated in
2612 favor of new corresponding names DEFINE-FOO, for consistency with
2613 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2614 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2615 internal symbols, but a few supported extensions like
2616 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2617 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2618 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2619 etc. as you get deeper into recursive calls to the debugger
2620 command loop, instead of the old "5]", "5]]", "5]]]"
2621 sequence. (I was motivated to do this when squabbles between
2622 ILISP and SBCL left me very deeply nested in the debugger. In the
2623 short term, this change will probably provoke more ILISP/SBCL
2624 squabbles, but hopefully it will be an improvement in the long run.)
2625 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2626 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2627 function, since declaiming return types involving aliens is
2628 (1) annoyingly messy to do by hand and (2) vital to efficient
2629 compilation of code which calls such functions.
2630 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2631 longer reexported by the SB-EXT package. They're solely useful
2632 for alien code, so it seems more logical that you should get
2633 them from the SB-ALIEN package, not in SB-EXT.
2634 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2635 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2636 features. Instead, the code that they used to control is always
2637 built into the system.
2638 * many other bug fixes
2639 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2640 advantage of the new EVAL-WHEN stuff and to clean them up in
2641 general, and they are now more ANSI-compliant in a number of
2642 ways. Martin Atzmueller is responsible for a lot of this.
2643 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2645 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2646 *** correct ERROR type for various file operations
2647 *** some fixes for Lisp streams
2648 *** DEFMETHOD syntax checking
2649 *** changing old weird representation of debug information as
2650 strings (which, among their other deficiencies, don't transform
2651 correctly when you rename packages, and don't change their
2652 print representation when you change things like *PACKAGE*
2653 and *PRINT-LENGTH*) to symbols and lists of symbols
2654 He also made several improvements and fixed several bugs in DESCRIBE.
2655 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2657 *** misbehavior of WRITE-STRING/WRITE-LINE
2658 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2659 and several other LOOP problems as well
2660 *** DIRECTORY when similar filenames are present
2661 *** DEFGENERIC with :METHOD options
2662 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2663 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2664 *** argument ordering in FIND with :TEST option
2665 *** mishandled package designator argument in APROPOS-LIST
2666 *** various problems in the backquote readmacro
2667 *** a bug in APROPOS
2668 *** probably some others that I'm not describing very well here,
2669 since the CVS log documents them by reference to sbcl-devel
2670 messages, and the SourceForge archives aren't working well.:-(
2671 ** Dan Barlow improved the Alpha port (and is making progress on the
2672 PPC port, for those of you who think different).
2673 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2674 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2675 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2676 continues to work on the SPARC port (for those of you in a position
2677 to look down upon our little PC-compatible boxes from a great height).
2678 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2679 and uses that copy, rather than the current dynamic value, when
2680 it's trying to decide whether to truncate output. Thus e.g.
2681 (let ((*print-lines* 50))
2682 (pprint-logical-block (stream nil)
2684 (let ((*print-lines* 8))
2685 (print (aref possiblybigthings i) stream)))))
2686 should now truncate the logical block only at 50 lines, instead of
2687 often truncating it at 8 lines, as it did before.
2688 * The doc/cmucl/ directory, containing old CMU CL documentation
2689 from the time of the fork, is no longer part of the base system.
2690 SourceForge has shut down its anonymous FTP service, and with it
2691 my original plan for distributing the old CMU CL documentation
2692 there. For now, if you need these files you can download an old
2693 SBCL source release and extract them from it.
2694 * The fasl file version number changed again, for dozens of reasons,
2695 some of which are apparent above.
2697 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2698 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2699 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2700 patch, so that DISASSEMBLE output is much nicer.
2701 * The code in the SB-PROFILE package now seems reasonably stable.
2702 I still haven't decided what the final interface should look like
2703 (I'd like PROFILE to interact cleanly with TRACE, since both
2704 facilities use function encapsulation) but if you have a need
2705 for profiling now, you can probably use it successfully with
2706 the current CMU-CL-style interface.
2707 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2708 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2709 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2710 link now returns the dangling link itself, and for similar
2711 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2712 link itself. (In these cases the old code signalled an error and
2713 looped endlessly, respectively.) Thus, DIRECTORY now works even
2714 in the presence of dangling and cyclic symbolic links.
2715 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2716 is now a supported extension again, since the consensus on
2717 sbcl-devel was that it can be useful for ordinary development
2718 work, not just for debugging SBCL itself.
2719 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2720 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2721 that currently-defined functions might be redefined later with
2722 different return types.
2723 * Hash tables can be printed readably, as inspired by CMU CL code
2724 of Eric Marsden and SBCL code of Martin Atzmueller.
2725 * better error handling in CLOS method combination, thanks to
2726 Martin Atzmueller porting Pierre Mai's CMU CL patches
2727 * more overflow fixes for >16Mbyte I/O buffers
2728 * A bug in READ has been fixed, so that now a single Ctrl-D
2729 character suffices to cause end-of-file on character streams.
2730 In particular, now you only need one Ctrl-D at the command
2731 line (not two) to exit SBCL.
2732 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2734 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2735 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2736 does the right thing.
2737 * The implementation of some type tests, especially for CONDITION
2738 types, is now tidier and maybe faster, due to CMU CL code
2739 originally by Douglas Crosher, ported by Martin Atzmueller.
2740 * Some math functions have been fixed, and there are new
2741 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2742 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2743 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2744 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2745 likely to be built into sbcl-0.7.0.)
2746 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2747 CL, ported to SBCL by Martin Atzmueller)
2748 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2749 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2750 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2751 * There's a new slam.sh hack to shorten the edit/compile/debug
2752 cycle for low-level changes to SBCL itself, and a new
2753 :SB-AFTER-XC-CORE target feature to control the generation of
2754 the after-xc.core file needed by slam.sh.
2755 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2756 COMPILE-FILE is no longer supported, so that now every function
2757 gets an entry point, so that block compilation looks a little
2758 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2759 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2760 returns the number of bytes consed since the system started,
2761 rather than the number consed since the first time the function
2762 was called. (The new definition parallels ANSI functions like
2763 CL:GET-INTERNAL-RUN-TIME.)
2764 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2765 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2766 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2767 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2768 * Fasl file version numbers are now independent of the target CPU,
2769 since historically most system changes which required version
2770 number changes have affected all CPUs equally. Similarly,
2771 the byte fasl file version is now equal to the ordinary
2774 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2775 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2776 is no longer recognized. I apologize for this, because it was
2777 listed in SB-EXT as a supported extension, but I found that
2778 its existing behavior was poorly specified, as well as incorrectly
2779 specified, and it looked like too much of a mess to straighten it
2780 out. I have enough on my hands trying to get ANSI stuff to work..
2781 * many patches ported from CMU CL by Martin Atzmueller, with
2782 half a dozen bug fixes in pretty-printing and the debugger, and
2783 half a dozen others elsewhere
2784 * fixed bug 13: Floating point infinities are now supported again.
2785 They might still be a little bit flaky, but thanks to bug reports
2786 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2787 as flaky as they were.
2788 * The --noprogrammer command line option is now supported. (Its
2789 behavior is slightly different in detail from what the old man
2790 page claimed it would do, but it's still appropriate under the
2791 same circumstances that the man page talks about.)
2792 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2793 are now supported, and enabled by default. Thus, the compiler can
2794 handle many floating point and complex operations much less
2795 inefficiently. (Thus e.g. you can implement a complex FFT
2797 * The compiler now detects type mismatches between DECLAIM FTYPE
2798 and DEFUN better, and implements CHECK-TYPE more correctly, and
2799 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2800 thanks to patches from Martin Atzmueller.
2801 * various fixes to make the cross-compiler more portable to
2802 ANSI-conforming-but-different cross-compilation hosts (notably
2803 Lispworks for Windows, following bug reports from Arthur Lemmens)
2804 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2805 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2806 fixed thanks to Pierre Mai's CMU CL patches.
2807 * fixes to keep the system from overflowing internal counters when
2808 it tries to use i/o buffers larger than 16M bytes
2809 * fixed bug 45a: Various internal functions required to support
2810 complex special functions have been merged from CMU CL sources.
2811 (When I was first setting up SBCL, I misunderstood a compile-time
2812 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2813 * improved support for type intersection and union, fixing bug 12
2814 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2815 more obscure bugs as well
2816 * some steps toward byte-compiling non-performance-critical
2817 parts of the system, courtesy of patches from Martin Atzmueller
2818 * Christophe Rhodes has made some debian packages of sbcl at
2819 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2820 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2821 stable, but are nonetheless usable. When he's ready, I'd be happy
2822 to add them to the SourceForge "File Releases" section. (And if
2823 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2824 * new fasl file format version number (because of changes in
2825 internal representation of (OR ..) types to accommodate the new
2826 support for (AND ..) types, among other things)
2828 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2829 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2831 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2833 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2834 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2835 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2836 * DESCRIBE now works on structure objects again.
2837 * Most function call argument type mismatches are now handled as
2838 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2839 know whether the function will be redefined before the call is
2840 executed. (The compiler could flag local calls with full WARNINGs,
2841 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2842 it doesn't keep track of enough information to know whether calls
2843 are local in this sense.)
2844 * Compiler output is now more verbose, with messages truncated
2845 later than before. (There should be some supported way for users
2846 to override the default verbosity, but I haven't decided how to
2847 provide it yet, so this behavior is still controlled by the internal
2848 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2849 src/compiler/ir1util.lisp.)
2850 * Fasl file format version numbers have increased again, because
2851 support for the Gray streams extension changes the layout of the
2852 system's STREAM objects.
2853 * The Gray subclassable streams extension now works, thanks to a
2854 patch from Martin Atzmueller.
2855 * The full LOAD-FOREIGN extension (not just the primitive
2856 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2857 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2858 but like most other programs, it defaults to copying the Unix
2859 environment from the original process instead of starting the
2860 new process in an empty environment.
2861 * Extensions which manipulate the Unix environment now support
2862 an :ENVIRONMENT keyword option which doesn't smash case or
2863 do other bad things. The CMU-CL-style :ENV option is retained
2864 for porting convenience.
2865 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2866 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2868 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2870 * A patch from Martin Atzmueller seems to have solved the SIGINT
2871 problem, and as far as we know, signal-handling now works cleanly.
2872 (If you find any new bugs, please report them!)
2873 * The system no longer defaults Lisp source file names to types
2874 ".l", ".cl", or ".lsp", but only to ".lisp".
2875 * The compiler no longer uses special default file extensions for
2876 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2877 seems to expect a single default extension for all compiled code,
2878 and there's no compelling reason to try to stretch the standard
2879 to allow two different extensions.) Instead, byte-compiled files
2880 default to the same extension as native-compiled files.
2881 * Fasl file format version numbers have increased again, because
2882 a rearrangement of internal implementation packages made some
2883 dumped symbols in old fasl files unreadable in new cores.
2884 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2885 many fewer weird special cases.
2886 * Bug #17 (differing COMPILE-FILE behavior between logical and
2887 physical pathnames) has been fixed, and some related misbehavior too,
2888 thanks to a patch from Martin Atzmueller.
2889 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2890 by Tim Moore, ported to SBCL by Martin Atzmueller.
2891 * Martin Atzmueller fixed several filesystem-related problems,
2892 including bug #36, in part by porting CMU CL patches, which were
2893 written in part by Paul Werkowski.
2894 * More compiler warnings in src/runtime/ are gone, thanks to
2895 more patches from Martin Atzmueller.
2896 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2899 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2901 * DESCRIBE now works on CONDITION objects.
2902 * The debugger now handles errors which arise when trying to print
2903 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2905 * The build system now uses an additional file, customize-target-features.lisp,
2906 to allow local modifications to the target *FEATURES* list. (The point of
2907 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2908 debugging features enabled, without having to worry about propagating it
2909 into everyone's system when I do a "cvs update".) When no
2910 customize-target-features.lisp file exists, the target *FEATURES* list
2911 should be constructed the same way as before.
2912 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2913 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2914 warn about behavior which is undefined under the ANSI spec. Note: This
2915 is specified by ANSI, but it's not very popular with programmers.
2916 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2917 macro in the SBCL sources for an example of a workaround which you
2918 might use to make portable ANSI-standard code which does what you want.
2919 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2920 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2921 when it's used as a non-toplevel form. (This is required in order
2922 to implement the DEFCONSTANT-EQX macro.)
2923 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2924 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2925 funny things, due to bugs in the implementation of EVAL-WHEN
2926 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2927 non-toplevel DEFCONSTANTs are uncommon.)
2928 * The core file version number and fasl file version number have been
2929 incremented, because the old noncompliant DEFCONSTANT behavior involved
2930 calling functions which no longer exist, and because I also took the
2931 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2932 * fixed bug 1 (error handling before read-eval-print loop starts), and
2933 redid debugger restarts and related debugger commands somewhat while
2935 ** The QUIT debugger command is gone, since it did something
2936 rather different than the SB-EXT:QUIT command, and since it never
2937 worked properly outside the main toplevel read/eval/print loop.
2938 Invoking the new TOPLEVEL restart provides the same functionality.
2939 ** The GO debugger command is also gone, since you can just invoke
2940 the CONTINUE restart directly instead.
2941 ** The TOP debugger command is also gone, since it's redundant with the
2942 FRAME 0 command, and since it interfered with abbreviations for the
2944 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2946 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2947 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2948 some porting work by Martin Atzmueller)
2949 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2950 with a boatload of patches which he ported to SBCL. Now that those
2952 ** The system tries to make sure that its low-priority messages
2953 are prefixed by semicolons, to help people who like to use
2954 syntax highlighting in their ILISP buffer. (This patch
2955 was originally due to Raymond Toy.)
2956 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2957 patches originally written by Raymond Toy.
2958 ** The compiler understands coercion between single-value and
2959 multiple-VALUES type expressions better, getting rid of some very
2960 weird behavior, thanks to patches originally by Robert MacLachlan
2961 and Douglas Crosher.
2962 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2963 lambda lists, thanks to a patch originally by Pierre Mai.
2964 ** The system no longer bogusly warns about "abbreviated type
2966 ** The compiler gets less confused by inlining and RETURN-FROM,
2967 thanks to some patches originally by Tim Moore.
2968 ** The system no longer hangs when dumping circular lists to fasl
2969 files, thanks to a patch originally from Douglas Crosher.
2970 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2971 undefined function error.
2972 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2973 in structure slot initforms) for now, documented workaround instead:-|
2974 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2975 * fixed bug 5: added stubs for various Gray stream functions called
2976 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2977 installed, at least appropriate type errors are generated
2978 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2979 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2980 consistently in DEFMETHOD forms.
2981 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2982 it doesn't seem to affect SBCL after all
2983 * The C runtime system now builds with better optimization and many
2984 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2986 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2988 * The system is now under CVS at SourceForge (instead of the
2989 CVS repository on my home machine).
2990 * The new signal handling code has been tweaked to treat register
2991 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2992 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2993 downstream, e.g. in debug-int.lisp, has implicit dependencies
2994 on the unsignedness of integer representation of machine words,
2995 and that caused the system to bomb out with infinite regress
2996 when trying to recover from type errors involving signed values,
2997 e.g. (BUTLAST '(1 2 3) -1).)
2998 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2999 (This was one of the bugs Peter Van Eynde reported back in July.)
3000 * The system now uses code inspired by Colin Walters' O(N)
3001 implementation of MAP (from the cmucl-imp@cons.org mailing
3002 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3003 inline the MAP operation, and there is more than one
3004 sequence argument to the MAP call (so that it can't just
3005 do ETYPECASE once and for all based on the type of the
3006 single sequence argument). (The old non-inline implementation
3007 of the general M-argument sequence-of-length-N case required
3008 O(M*N*N) time when any of the sequence arguments were LISTs.)
3009 * The QUIT :UNIX-CODE keyword argument has been renamed to
3010 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3011 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3013 * Raymond Wiker's patches to port dynamic loading from Linux to
3014 FreeBSD have been added.
3015 * The BUGS file is now more nearly up to date, thanks in large part
3016 to Martin Atzmueller's review of it.
3017 * The debugger now flushes standard output streams before it begins
3018 its output ("debugger invoked" and so forth).
3019 * The core version number and fasl file version number have both
3020 been incremented, because of incompatible changes in the layout
3022 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3023 used to not be called for a saved Lisp image.)
3024 * Martin Atzmueller's version of a patch to fix a compiler crash,
3025 as posted on sbcl-devel 13 September 2000, has been installed.
3026 * Instead of installing Martin Atzmueller's patch for the
3027 compiler transform for SUBSEQ, I deleted the compiler transform,
3028 and transforms for some similar consing operations.
3029 * A bug in signal handling which kept TRACE from working on OpenBSD
3031 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3032 away by constant folding
3033 * The system now defines its address space constants in one place
3034 (in the Lisp sources), and propagates them automatically elsewhere
3035 (through GENESIS and the sbcl.h file). Therefore, patching the
3036 address map is less unnecessarily tedious and error-prone. The
3037 Lisp names of address space constants have also been systematized.
3038 * CVS tags like dollar-Header-dollar have been removed from
3039 the sources, because they have never saved me trouble and
3040 they've been source of trouble working with patches and other
3041 diff-related operations.
3042 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3043 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3045 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3047 * The system has been ported to OpenBSD.
3048 * The system now compiles with a simple "sh make.sh" on the systems
3049 that it's supported on. I.e., now you no longer need to tweak
3050 text in the target-features.lisp-expr and symlinks in src/runtime/
3051 by hand, the make.sh takes care of it for you.
3052 * The system is no longer so grossly inefficient when compiling code
3053 involving vectors implemented as general (not simple) vectors (VECTOR T),
3054 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3055 which dares to use the various sequence functions on non-simple
3056 vectors, takes less of a performance hit.
3057 * There is now a primitive type predicate VECTOR-T-P
3058 to test for the (VECTOR T) type, so that e.g.
3059 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3060 can now be compiled with some semblance of efficiency. (The old code
3061 turned the type declaration into a full call to %TYPEP at runtime!)
3062 * AREF on (VECTOR T) is still not fast, since it's still compiled
3063 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3064 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3065 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3066 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3067 without making full calls to SUBTYPEP at runtime.
3068 (Some analogous efficiency issues for non-simple vectors specialized to
3069 element types other than T, or for non-simple multidimensional arrays,
3070 have not been addressed. They could almost certainly be handled the
3071 same way if anyone is motivated to do so.)
3072 * The changes in array handling break binary compatibility, so
3073 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3074 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3075 instead of (VALUES T T).
3076 * By following the instructions that Dan Barlow posted to sbcl-devel
3077 on 2 July 2000, I was able to enable primitive dynamic object
3078 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3079 functionality is not implemented (since it calls ld to resolve
3080 library references automatically, requiring RUN-PROGRAM for its
3081 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3082 try to resolve library references) is now supported.
3083 * The system now flushes the standard output streams when it terminates,
3084 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3085 them at several other probably-convenient times, e.g. in each pass of
3086 the toplevel read-eval-print loop, and after evaluating a form given
3087 as an "--eval" command-line option. (These changes were motivated by a
3088 discussion of stream flushing issues on cmucl-imp in August 2000.)
3089 * The source transform for TYPEP of array types no longer assumes
3090 that an array whose element type is a not-yet-defined type
3091 is implemented as an array of T, but instead punts, so that the
3092 type will be interpreted at runtime.
3093 * There is now some support for cross-compiling in make.sh: each of
3094 the phases of make.sh has its own script. (This should be transparent
3095 to people doing ordinary, non-cross-compile builds.)
3096 * Since my laptop doesn't have hundreds of megabytes of memory like
3097 my desktop machine, I became more motivated to do some items on
3098 my to-do list in order to reduce the size of the system a little:
3099 ** Arrange for various needed-only-at-cold-init things to be
3100 uninterned after cold init. To support this, those things have
3101 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3102 symbols with such names are now uninterned after cold init).
3103 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3104 fasl files for cold load.
3105 ** Remove the old compiler structure pooling code (which used to
3106 be conditional on the target feature :SB-ALLOC) completely.
3107 ** Redo the representation of some data in cold init to be more compact.
3108 (I also looked into supporting byte compiled code at bootstrap time,
3109 which would probably reduce the size of the system a lot, but that
3110 looked too complicated, so I punted for now.)
3111 * The maximum signal nesting depth in the src/runtime/ support code has
3112 been reduced from 4096 to 256. (I don't know any reason for the very
3113 large old value. If the new smaller value turns out to break something,
3114 I'll probably just bump it back up.)
3115 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3117 * Many, many bugs reported by Peter Van Eynde have been added to
3118 the BUGS list; some have even been fixed.
3119 * While enabling dynamic object file loading, I tried to make the
3120 code easier to understand, renaming various functions and variables
3121 with less ambiguous names, and changing some function calling
3122 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3123 * While trying to figure out how to do the OpenBSD port, I tried to
3124 clean up some of the code in src/runtime/. In particular, I dropped
3125 support for non-POSIX signal handling, added various comments,
3126 tweaked the code to reduce the number of compilation warnings, and
3127 renamed some files to increase consistency.
3128 * To support the new automatic configuration functionality in make.sh,
3129 the source file target-features.lisp-expr has been replaced with the
3130 source file base-target-features.lisp-expr and the machine-generated
3131 file local-target-features.lisp-expr.
3132 * fixed a stupid quoting error in make.sh so that using CMU CL
3133 "lisp -batch" as cross-compilation host works again
3135 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3137 * DESCRIBE no longer tries to call itself recursively to describe
3138 bound/fbound values, so that it no longer fails on symbols which are
3139 bound to themselves (like keywords, T, and NIL).
3140 * DESCRIBE now works on generic functions.
3141 * The printer now prints less-screwed-up representations of closures
3142 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3143 * A private symbol is used instead of the :EMPTY keyword previously
3144 used to mark empty slots in hash tables. Thus
3145 (DEFVAR *HT* (MAKE-HASH-TABLE))
3146 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3147 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3148 now does what ANSI says that it should. (You can still get
3149 similar noncompliant behavior if bang on the hash table
3150 implementation with all the symbols you get back from
3151 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3152 This breaks binary compatibility, since tests for equality to
3153 :EMPTY are wired into things like the macroexpansion of
3154 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3156 * There's now a minimal placeholder implementation for CL:STEP,
3157 as required by ANSI.
3158 * An obscure bug in the interaction of the normal compiler, the byte
3159 compiler, inlining, and structure predicates has been patched
3160 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3161 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3163 * Missing ordinary arguments in a macro call are now detected even
3164 when the macro lambda list contains &KEY or &REST.
3165 * The debugger no longer complains about encountering the top of the
3166 stack when you type "FRAME 0" to explicitly instruct it to go to
3167 the top of the stack. And it now prints the frame you request even
3168 if it's the current frame (instead of saying "You are here.").
3169 * As specified by ANSI, the system now always prints keywords
3170 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3172 * The default initial SIZE of HASH-TABLEs is now smaller.
3173 * Type information from CLOS class dispatch is now propagated
3174 into DEFMETHOD bodies, so that e.g.
3175 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3177 is now basically equivalent to
3178 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3179 (DECLARE (TYPE SINGLE-FLOAT X))
3181 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3182 operation, without having to do run-time type dispatch.
3183 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3184 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3185 * Since I don't seem to be making big file reorganizations very often
3186 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3187 is overflowing my ability to conveniently back them up), I've finally
3188 checked the system into CVS. (The CVS repository is on my home system,
3189 not at SourceForge -- putting it on SourceForge might come later.)
3190 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3191 high-level GC-NOTIFY-FOO functions send their output. (There's
3192 still very little control of where low-level verbose GC functions
3193 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3194 less than it used to -- the GC-NOTIFY-FOO functions are now under
3195 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3196 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3197 in only one place in the source code, and propagates it automatically
3198 everywhere that it's needed. Thus e.g. when I bump the version from
3199 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3200 * The C source files now include boilerplate legalese and documentation
3201 at the head of each file (just as the Lisp source files already did).
3202 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3203 to his page will be replaced with a link to his new CLiki service.
3205 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3207 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3208 * The build process now looks for GNU make under the default name "gmake",
3209 instead of "make" as it used to. If GNU make is not available as "gmake"
3210 on your system, you can change this default behavior by setting the
3211 GNUMAKE environment variable.
3212 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3213 can build without error under CMU CL.
3215 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3217 * There is now a partial SBCL user manual (with some new text and some
3218 text cribbed from the CMU CL manual).
3219 * The beginnings of a profiler have been added (starting with the
3220 CMU CL profiler and simplifying and cleaning up). Eventually the
3221 main interface should be through the TRACE macro, but for now,
3222 it's still accessed through vaguely CMU-CL-style functions and macros
3223 exported from the package SB-PROFILE.
3224 * Some problems left over from porting CMU CL to the new
3225 cross-compilation bootstrap process have been cleaned up:
3226 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3227 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3228 part when getting rid of the compiler *BACKEND* stuff.)
3229 ** The value of *NULL-TYPE* was screwed up, because it was
3230 being initialized before the type system knew the final
3231 definition of the 'NULL type. This screwed up several key
3232 optimizations in the compiler, causing inefficiency in all sorts
3233 of places. (I found it because I wanted to understand why
3234 GET-INTERNAL-RUN-TIME was consing.)
3235 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3236 PROCLAIM FTYPE information. Unfortunately this broke binary
3237 compatibility again, since now the forms output by DEFGENERIC
3238 to refer to functions which didn't exist in 0.6.3.
3239 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3240 can use the new (as of 0.6.3) transform for SOME into MAP into
3242 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3243 values to (INTEGER 0 1000000), so that the time code will no longer
3244 occasionally get blown up by Linux returning 1000000 microseconds
3245 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3246 its output conform to the ANSI spec. (Alas, this makes its output
3247 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3248 * A full call to MAP NIL with a single sequence argument no longer conses.
3249 * fixes to problems pointed out by Martin Atzmueller:
3250 * The manual page no longer talks about multiprocessing as though
3251 it were currently supported.
3252 * The ILISP support patches have been removed from the distribution,
3253 because as of version 5.10.1, ILISP now supports SBCL without us
3254 having to maintain patches.
3255 * added a modified version of Raymond Toy's recent CMU CL patch for
3256 EQUALP comparison of HASH-TABLE
3258 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3260 * The system still can't cross-compile itself with
3261 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3262 entails), but at least it can compile more of itself that way
3263 than it used to be able to, and various buglets which were uncovered
3264 by trying to cross-compile itself that way have now been fixed.
3265 * This release breaks binary compatibility again. This time
3266 at least I've incremented the FASL file format version to 2, so that the
3267 problem can be detected reliably instead of just causing weird errors.
3268 * various new style warnings:
3269 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3270 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3271 ** using the lexical binding of a variable named in the *FOO* style
3272 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3273 as specified by ANSI.
3274 * *RANDOM-STATE* is no longer automatically initialized from
3275 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3276 default behavior of the system is to repeat its behavior every time
3277 it's run. If you'd like to change this behavior, you can always
3278 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3279 old convention there was no comparably easy way to get the system to
3280 repeat its behavior every time it was run.
3281 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3282 FTYPE declarations has been removed, in favor of their ANSI
3283 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3284 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3285 the types of their sequence arguments can be determined at compile time.
3286 This is done through a new open code expansion for MAP which eliminates
3287 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3288 argument types can be determined at compile time.
3289 * The optimizer now transforms COERCE into an identity operation when it
3290 can prove that the coerced object is already of the correct type. (This
3291 can be a win for machine generated code, including the output of other
3292 optimization transforms, such as the MAP transform above.)
3293 * Credit information has been moved from source file headers into CREDITS.
3294 * Source file headers have been made more standard.
3295 * The CASE macro now compiles without complaining even when it has
3298 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3300 * (Note that the way that the PCL macroexpansions were rewritten
3301 to accommodate the change in DEFGENERIC below breaks binary
3302 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3303 not run under sbcl-0.6.2. Once we get out of alpha releases,
3304 i.e. hit release 1.0.0, we'll probably try to maintain binary
3305 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3306 and sbcl-1.4.4. Until then, however, it might be fairly common
3307 for maintenance releases to break binary compatibility.)
3308 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3310 * The reporting of unhandled signals has been changed to print some
3311 explanatory text as well as the report form. (Previously only
3312 the report form was printed.)
3313 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3314 it defines, so that the compiler no longer issues undefined function
3315 warnings for compiled-but-not-yet-loaded generic functions.
3316 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3317 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3318 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3319 section of the ANSI spec are supported.
3320 * The "" nickname for the "KEYWORD" package has been removed.
3321 The reader still handles symbol tokens which begin with a package marker
3322 as keywords, but it doesn't expose its mechanism for doing so in the
3323 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3324 * The system now issues STYLE-WARNINGs for contradictory TYPE
3325 proclamations. (Warnings for contradictory FTYPE proclamations would
3326 be nice too, but those can't be done usefully unless the type system
3327 is made smarter about FUNCTION types.)
3328 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3329 systematized, so that "*target-*.lisp is supposed to exist only on the
3330 target and imply that there's a related file which exists on the
3331 host, and *host-*.lisp is supposed to exist only on the host and imply
3332 that there's a related file which exists on the target. This involves a
3333 lot of renaming. Hopefully the acute confusion caused by the renaming
3334 will be justified by the reduction in chronic confusion..
3335 ** runtime-type.lisp -> early-target-type.lisp
3336 ** target-type.lisp -> late-target-type.lisp
3337 ** early-host-format.lisp -> early-format.lisp
3338 ** late-host-format.lisp -> late-format.lisp
3339 ** host-error.lisp -> misc-error.lisp
3340 ** early-error.lisp -> early-target-error.lisp
3341 ** late-error.lisp -> late-target-error.lisp
3342 ** host-defboot.lisp -> early-defboot.lisp
3343 ** code/misc.lisp -> code/target-misc.lisp
3344 ** code/host-misc.lisp -> code/misc.lisp
3345 ** code/numbers.lisp -> code/target-numbers.lisp
3346 ** code/early-numbers.lisp -> numbers.lisp
3347 ** early-host-type.lisp -> early-type.lisp
3348 ** late-host-type.lisp -> late-type.lisp
3349 ** host-typep.lisp -> typep.lisp
3350 ** load.lisp -> target-load.lisp
3351 ** host-load.lisp -> load.lisp
3352 ** host-disassem.lisp -> disassem.lisp
3353 ** host-insts.lisp -> insts.lisp
3354 ** byte-comp.lisp -> target-byte-comp.lisp
3355 ** host-byte-comp.lisp -> byte-comp.lisp
3356 ** host-signal.lisp -> signal.lisp
3357 ** host-defstruct.lisp -> defstruct.lisp
3358 ** late-target-type.lisp -> deftypes-for-target.lisp
3359 Furthermore, several other previously target-only files foo.lisp (e.g.
3360 hash-table.lisp and random.lisp) have been split into a target-and-host
3361 foo.lisp file and a target-only target-foo.lisp file, with their key type
3362 definitions in the target-and-host part, so that the cross-compiler will
3363 know more about target types.
3364 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3365 gone away. In their place are various *BACKEND-FOO* variables
3366 corresponding to the slots of the old structure.
3367 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3368 into the target SBCL has been fixed.
3369 * The chill.lisp system for loading cold code into a running SBCL
3371 * Support for the CMU CL "scavenger hook" extension has been removed.
3372 (It was undocumented and unused in the CMU CL sources that SBCL was
3373 derived from, and stale in sbcl-0.6.1.)
3374 * Various errors in the cross-compiler type system were detected
3375 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3376 (enabling various consistency checks). Many of them were fixed,
3377 but some hard problems remain, so the compiler is back to
3378 running without *TYPE-SYSTEM-INITIALIZED* for now.
3379 * As part of the cross-compiler type system cleanup, I implemented
3380 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3381 * I have started adding UNCROSS calls throughout the type system
3382 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3383 on cross-compiler input files will be able to go away, and various
3385 * CONSTANTP now returns true for quoted forms (as explicitly required
3388 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3390 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3391 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3392 (MAKE-LIST -1). In the longer term, it ought to become true
3393 that declarations are assertions even at SAFETY 1. For now, it's not
3394 quite true even at SAFETY 3, but it's at least more nearly true..
3395 (Note that this change seems to increases the size of the system by
3396 O(5%) and to decrease the speed of the compiler by 20% or more.)
3397 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3398 for the problem of printing dozens of lines of distracting information
3399 about low-level system machinery as part of the top stack frame
3400 on entry to the debugger when an undefined function was called.
3401 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3402 is not reset to COMMON-LISP-USER.
3403 * Compilation of stuff related to dyncount.lisp has been made conditional
3404 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3405 smaller. The various dyncount-related symbols have been moved into
3406 a new "SB-DYNCOUNT" package.
3407 * tty-inspect.lisp has been renamed to inspect.lisp.
3408 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3409 feature has gone away. (When we eventually port to other flavors of
3410 libc and/or Unix, we'll try to make the differences between flavors
3411 invisible at the user level.)
3412 * Various other *FEATURES* tags, and/or their associated conditionals,
3413 have been removed if obsolescent, or given better documentation, or
3414 sometimes given more-mnemonic names.
3416 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3418 * tidied up "make.sh" script
3419 * tidied up system directory structure
3420 * better "clean.sh" behavior
3421 * added doc/FOR-CMUCL-DEVELOPERS
3422 * many many small tweaks to output format, e.g. removing possibly-confusing
3423 trailing #\. character in DESCRIBE-INSTANCE
3424 * (EQUALP #\A 'A) no longer signals an error.
3425 * new hashing code, including EQUALP hashing
3426 * tidied up Lisp initialization and toplevel
3427 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3428 * command line argument processing
3429 * added POSIX-GETENV function to deal with Unix-ish environment variables
3430 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3431 terminating SBCL on EOF
3432 * non-verbose GC by default
3433 * There is no more "sbcl" shell script; the sbcl file is now the C
3434 runtime executable (just like CMU CL).
3435 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3437 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3438 symbol and package information as Lisp native symbol and package objects
3439 instead of strings naming symbols and strings naming packages. This way,
3440 whenever packages are renamed (as in warm init), debug information is
3441 transformed along with everything else.
3442 * tweaked the optimization policy declarations which control the building
3443 of SBCL itself. Now, among other things, the system no longer saves
3444 source location debugging information. (This helps two problems at once
3445 by reducing SBCL size and by keeping SBCL from trying to look for its
3446 sources -- which may not exist -- when reporting errors.)
3447 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3448 debugging and testing purposes
3449 * cleaned up printing, making the printer call PRINT-OBJECT for
3450 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3451 methods, giving nearly-ANSI behavior
3452 * converted almost all special variables to use *FOO* naming convention
3453 * deleted PARSE-TIME functionality, since it can be done portably
3454 * moved some files out of cold init into warm init
3455 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3457 * regularized formatting of source files
3458 * added an install.sh script
3459 * fixed ridiculous memory usage of cross-compiler by making
3460 compiler/alloc.lisp not try to do pooling unless it can hook
3461 itself into the GC of the cross-compilation host. Now the system
3462 builds nicely on my old laptop.
3463 * added :SB-ALLOC in target-features.lisp-expr
3464 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3465 was not implemented)
3466 * re-did condition handling and note reporting in the compiler. Notes
3467 are no longer handled by signalling conditions. Style warnings
3468 and warnings are handled more correctly and reported in such a way
3469 that it's easy to find one or the other in your output (so that you
3470 can e.g. figure out which of many problems caused COMPILE-FILE to
3472 * changed the severity of several compiler warnings from full WARNING
3473 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3474 compiler note reporting so that it doesn't use the condition system
3475 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3476 COMPILE-FILE command)
3477 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3478 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3479 no longer supported, since their implementation was deeply intertwingled
3480 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3481 * removed bogus "support" for compiler macros named (SETF FOO), and
3482 removed the compiler macro for SETF INFO (but only after making a fool
3483 of myself on the cmucl-imp mailing list by posting a bogus patch for
3484 DEFINE-COMPILER-MACRO..)
3485 * Compiled files containing forms which have side effects on the Lisp
3486 reader (such as DEFPACKAGE forms) are now handled more correctly.
3487 (Compiler queuing of top level lambdas has been suppressed by setting
3488 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3489 * deleted various currently-unused source files, e.g. gengc.lisp. They
3490 may be added back at some point e.g. when porting to other architectures,
3491 but until they are it's distracting to distribute them and to try to
3493 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3494 there were so many of them they're just distractions, and UNCROSS is
3495 known to be able to handle the current sources
3496 * moved PROFILE functionality into TRACE, so that it will be clear
3497 how the wrapping and unwrapping of functions when you profile them
3498 interacts with the wrapping and unwrapping of functions when you
3499 trace them. (Actually, the functionality isn't there yet, but at least
3500 the interface specification is there. Hopefully, the functionality will
3501 arrive with some future maintenance release.)
3502 * removed host-oops.lisp
3503 * changed signature of QUIT function to allow UNIX-CODE argument
3504 * fixed READ-SEQUENCE bug
3505 * tweaked verbose GC output so that it looks more like the progress
3506 output that ANSI specifies for functions like LOAD
3507 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3508 * added <http://sbcl.sourceforge.com> to the banner information printed by