1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
3 * new platform: experimental support for the Windows operating
4 system has been added. (thanks to Alastair Bridgewater)
5 * fixed several bugs in and robustified the PPC FFI (including
6 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
7 * bug fix: allow non-simple string symbol names (reported by Paul
9 * bug fix: interrupt handling on NetBSD (thanks to Richard M
11 * bug fix: saving a core corrupted callbacks on x86/x86-64
12 * optimization: faster implementation of EQUAL
13 * fixed segfaults on x86 FreeBSD 7-current (thanks to NIIMI Satoshi)
15 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
16 * minor incompatible change: (SETF CLASS-NAME) and (SETF
17 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
18 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
19 by ANSI as a generic function, and for consistency (SETF
20 GENERIC-FUNCTION-NAME) is treated likewise.
21 * fixed bug #233.b: make constraint propagation notice when a variable
22 value is changed after it is referenced but before it is used
23 * fixed bug #296: no more arbitrary behaviour from filenames with
24 odd characters as --load arguments. (reported by Adam Warner)
25 * fixed bug #390: :CHARACTER-SET pathname components now work as
26 expected. (reported by Tim Daly Jr)
27 * fixed bug #391: complicated :TYPE intersections in slot
28 definitions no longer cause an error in PCL internals.
29 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
30 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
31 returning the number of octets which would be written to the
32 file-stream. (thanks to Robert J. Macomber)
33 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
34 arguments to the internal stream functions. (thanks to David
36 * bug fix: stack exhaustion is now handled on the x86/SunOS
37 platform. (thanks to Daisuke Homma)
38 * bug fix: more accurate ROOM results on GENCGC platforms
39 * optimization: improved type inference for arithmetic-for
40 index variables in LOOP
41 * optimization: faster floating-point SQRT on x86-64
42 * fixed some bugs revealed by Paul Dietz' test suite:
43 ** DOCUMENTATION returns NIL instead of "" for method combinations
44 that don't have a docstring
46 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
47 * minor incompatible change: (SETF CLASS-NAME) and (SETF
48 GENERIC-FUNCTION-NAME) are no longer generic functions, and
49 therefore are not customizeable by user code (as seems to be at
50 least permitted and maybe required by AMOP). As a consolation,
51 however, the SBCL implementation of these functions now calls
52 REINITIALIZE-INSTANCE as specified by AMOP.
53 * bug fix: slot-definition documentation is propagated as per ANSI
54 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
55 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
56 argument, as do eslotd initialization methods. (from Pascal
57 Costanza's "Closer" project)
58 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
59 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
61 * bug fix: it is now possible to have more than one subclass of
62 STANDARD-GENERIC-FUNCTION without causing stack overflow.
63 (reported by Bruno Haible, Pascal Costanza and others)
64 * bug fix: the dependent update protocol now works for generic
65 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
67 * bug fix: condition-class instances corresponding to
68 DEFINE-CONDITION forms are now created eagerly. (reported by
69 Kalle Olavi Niemitalo on comp.lang.lisp)
70 * bug fix: floating point printing is more accurate in some
71 circumstances. (thanks to Simon Alexander)
72 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
73 merged with *DEFAULT-PATHNAME-DEFAULTS*.
74 * bug fix: callbacks on OS X now preserve stack-alignment.
75 (thanks to Cyrus Harmon)
76 * enhancement: the x86-64 disassembler is much better at
77 disassembling SSE instructions. (thanks to Lutz Euler)
78 * enhancement: CHECK-TYPE on a variable now allows type-inference
79 (thanks to Vincent Arkesteijn)
80 * optimization: improved performance of EUC-JP external format.
81 (thanks to NIIMI Satoshi)
82 * optimization: performance improvements to IO on file streams of
83 :ELEMENT-TYPE CHARACTER
84 * optimization: much faster memory allocation on x86-64
85 * optimization: faster garbage collections (latency and throughput)
87 * optimization: faster bignum subtraction and division on x86 and x86-64
88 (thanks to Lutz Euler)
89 * optimization: more accurate type-inference for loops that use a
90 floating point index variable or a negative step.
92 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
93 * bug fix: add a workaround to SBCL looping infinitely at startup on
94 Linux kernels with apparently buggy implementations of personality().
95 (thanks to Svein Ove Aas)
96 * bug fix: Unicode symbols are correctly printed in LDB backtraces
97 (thanks to David Lichteblau)
98 * bug fix: local bindings shadow global symbol macros for
99 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
100 * optimization: non-open coded uses of numeric comparison operators
101 (e.g. >) no longer cons when called with more than one parameter
102 on platforms supporting dynamic-extent allocation.
103 * enhancement: saving cores with foreign code loaded is now supported
104 on MIPS/Linux in addition to the previously supported platforms.
105 * bug fix: FIND-RESTART now tests for activity, not applicability when given
106 a restart object as identifier. (reported by Helmut Eller for CMUCL)
107 * bug fix: division by zero in sb-sprof when no samples were collected
108 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
109 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
110 (thanks to Alexander Kjeldaas)
111 * new feature: ignoring the return values of destructive functions
112 like NREVERSE signals a compile-time style-warning.
113 (thanks to Kevin Reid)
114 * new feature: the alignment of alien structure fields can be
115 explicitly specified. (thanks to Cyrus Harmon)
116 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
117 and dump core on SIGQUIT
119 ** incompatible change: threads do not inherit values of specials
120 from their parents (see manual)
121 ** bug fix: threads stacks belonging to dead threads are freed by the
122 next exiting thread, no need to gc to collect thread stacks anymore
123 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
124 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
125 ** bug fix: dynamic variable and thread start related gc lossage
126 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
128 * fixed some bugs revealed by Paul Dietz' test suite:
129 ** SUBTYPEP is slightly more accurate on heinously complicated
130 CONS types where some of the members have uncertain (in the
131 NIL, NIL sense) type relationships to each other.
133 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
134 no longer purified unless :PURIFY T is explicitly specified.
135 ** Non-purified cores are significantly smaller than before
137 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
138 * new feature: timers based on Zach Beane's excellent timer package
139 * added support for the following external formats: koi8-u,
140 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
141 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
142 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
143 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
144 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
145 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
146 * incompatible change: a threaded SBCL will no longer revert to
147 non-threaded mode on non-NPTL systems, but refuse to start entirely.
148 * bug fix: interrupts are disabled until startup is complete; no
149 more sigsegvs when receiving a signal to soon
150 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
152 * bug fix: add a workaround for the memory randomization features in
153 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
154 workaround will only be in effect on systems with the proc filesystem
156 * bug fix: printing objects of type HASH-TABLE signals a
157 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
159 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
160 have been running for over 50 days. (reported by Gilbert Baumann)
161 * bug fix: the logic for getting names of functions gets less
162 confused when confronded with alternate-metaclass
163 funcallable-instances. (reported by Cyrus Harmon)
164 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
165 consistent, even on internal alternate-metaclass objects.
166 * bug fix: flush closure information collected by physical
167 environment analysis prepass before the main pass. (bug reported
169 * bug fix: compiler pack phase does not modify a hashtable, which is
170 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
171 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
172 may be expanded. (reported by Alexander <malishev> on c.l.l)
173 * bug fix: no more style-warnings when compiling code using LAST
174 under high-SPEED low-SPACE settings. (reported by David Wallin
176 * fixed some incompatibilities between SBCL's MOP and the MOP
178 ** the METAOBJECT class is now implemented;
179 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
180 STANDARD-OBJECT, as required;
181 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
182 now compatible (as required by VALIDATE-SUPERCLASS); there
183 remains a consistency requirement over the presence or absence
184 of the FUNCTION class in the superclass list of finalized
185 classes; see the manual for more details;
186 ** the :AROUND method for COMPUTE-SLOTS on
187 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
188 requested slot ordering.
190 ** bug fix: parent thread now can be gc'ed even with a live
192 ** bug fix: binding a special with PROGV to no value is not
193 the same as that symbol not having been bound (thanks to
195 ** bug fix: binding specials is thread safe (thanks to
197 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
198 and SB-SPROF do not die with 'no handler for signal XX in
199 interrupt_handle_now(..)' anymore
200 ** bug fix: WITH-TIMEOUT works with multiple running threads
201 * fixed some bugs revealed by Paul Dietz' test suite:
202 ** ENSURE-GENERIC-FUNCTION should take a method class object for
203 the :method-class keyword argument.
205 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
206 * new port: the Solaris operating system on x86 processors is now
207 mostly supported, though some rough edges in the environment
208 remain. (thanks to Daisuke Homma)
209 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
210 space for its heap. (thanks to Thiemo Seufer)
211 * optimization: SBCL on MIPS platforms now supports dynamic-extent
212 closures. (thanks to Thiemo Seufer)
213 * minor incompatible change: eof selects abort in the debugger.
214 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
215 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
217 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
218 deprecated, and will go away in a future revision of SBCL.
219 * minor incompatible change: GC-ON and GC-OFF are no longer
220 implemented with a counter, it does not matter how many times gc
221 is switched on or off
222 * bug fix: discriminating functions for generic function classes
223 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
224 make invalid assumptions about method precedence order. (reported
226 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
227 platforms. (thanks to Thiemo Seufer)
228 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
229 * bug fix: correct pathname printing with printer escaping is on.
230 (thanks to Kevin Reid)
231 * bug fix: complex VOP definitions in "user-space" no longer trigger
232 package locks. (reported by Zach Beane)
233 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
234 for funcallable-instance functions, and is no different from
236 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
237 characters in the current readtable. (reported by Nicholas Neuss)
238 * bug fix: SUBTYPEP on various CONS types returns more a more
239 accurate acknowledgment of its certainty.
240 * optimizations: REMOVE-DUPLICATES now runs in linear time on
241 lists in some cases. This partially fixes bug 384.
242 * flush all standard streams before prompting in the REPL and the
244 * bug fix: signal handling and triggering gc do not conflict
245 directly anymore, in particular a high frequency sb-sprof does
246 not prevent gc from running
247 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
248 approximation for timezone and DST information between the
249 universal time epoch and the smallest negative 32-bit time_t.
250 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
251 year 1899 so long as the encoded time is non-negative.
252 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
253 the correct answer for non-constant rotate amounts when performing
254 an inline 32-bit rotation.
256 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
257 there is only one thread in the session
258 ** bug fix: memory leak for streams created in one thread and
259 written to in another
260 ** bug fix: lockup when compiled with gcc4
261 ** bug fix: race that allows the gc to be triggered when gc is
263 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
264 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
266 ** bug fix: finished threads are gc'ed properly
267 * fixed some bugs revealed by Paul Dietz' test suite:
268 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
269 has arguments with a different set of applicable methods from
270 the orignal arguments.
271 ** The type error thrown by MAP now has a correct expected-type
273 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
274 name a compiled function.
275 ** Three MISC tests where a large bignum was improperly coerced to
276 a float that couldn't represent that bignum during type
277 derivation were fixed.
278 ** SUBTYPEP can now handle types involving the negation of a
279 list-form FUNCTION type.
280 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
281 as (COMPLEX (AND RATIO (NOT FIXNUM))).
282 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
284 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
285 * New feature: Experimental support for bivalent streams: streams
286 opened with :element-type :default now allow character and binary
287 (unsigned-byte 8) I/O
288 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
289 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
291 * bug fix: no more highly sporadic "couldn't check whether ~S is
292 readable" when reading a stream and an interrupt hits in the middle
293 of a select system call
294 * compiler better recognizes complex arrays (reported by Eduardo
296 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
297 for variables on SBCL built with linkage-tables. (reported by Luis
299 * various error reporting improvements.
300 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
301 (Thanks to Thiemo Seufer)
302 * bug fix: nested reader invokations work correctly
303 * bug fix: it is possible to have simultaneous references to foreign
304 code and foreign data with the same name.
306 ** added x86-64 support
307 ** incompatible change: the threading api now works with thread
308 objects instead of thread ids
309 ** bug fix: threads are protected from signals and interruption when
310 starting up or going down
311 ** bug fix: a race where an exiting thread could lose its stack to gc
312 ** bug fix: don't halt on infinite error in threads if possible
313 ** fixed numerous gc deadlocks introduced in the pthread merge
314 ** bug fix: fixed thread safety issues in read and print
315 ** bug fix: debugger doesn't hang on session lock if interrupted at
316 an inappropriate moment
317 ** bug fix: run-program is now thread safe(r)
318 ** bug fix: inner with-recursive-lock no longer releases the mutex
319 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
320 * fixed some bugs revealed by Paul Dietz' test suite:
321 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
322 EXPECTED-TYPE slots filled.
323 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
325 ** MAKE-CONDITION accepts classes as type-designators.
326 ** COMPILE may never return NIL.
327 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
328 range before calling Unix time functions
330 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
331 * numerous signal handling fixes to increase stability
332 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
333 * minor incompatible change: we now correctly canonize default
334 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
335 as per the MOP, rather than the historical (INITARG INITFUNCTION
336 INITFORM). (reported by Bruno Haible)
337 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
338 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
339 for more information.
340 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
341 pathname is a directory pathname.
342 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
343 * dynamic space size on PPC has been increased to 768Mb. (thanks to
345 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
346 :METACLASS argument in addition to a class name. (reported by
347 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
348 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
349 (reported by Bruno Haible; patch from Gerd Moellmann)
350 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
352 * bug fix: more cleanups to the floating point exception handling on
353 x86-64 (thanks to James Knight)
354 * bug fix: the compiler does not try to constant fold calls of
355 COERCE and BIT-* functions when they return freshly constructed
356 objects. (reported by Nikodemus Siivola and Paul Dietz)
357 * optimization: improved the MIPS versions of generic (in the
358 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
359 * optimization: direct conversion of (unsigned-byte 32) to floats on
360 the PowerPC platform.
361 * optimization: structure instances with raw slots now use less
362 memory, and probably show better memory locality. (thanks to
364 * optimization: DYNAMIC-EXTENT declarations for lists and closures
365 are treated as requests for stack allocation on the x86-64,
366 Alpha-32, PPC and SPARC platforms.
367 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
368 the wrong file descriptor; implementation of SOCKET-OPEN-P.
369 (thanks to Tony Martinez)
371 ** gcing a dead thread can no longer lead to lockups
372 ** threads block signals until they are set up properly
373 ** errno is no longer shared by threads
374 ** interrupt-thread restores the eflags register on x86
375 ** fixed some lockups due to gc/thread interaction
376 * fixed some bugs revealed by Paul Dietz' test suite:
377 ** invalid dotted lists no longer raise a read error when
379 ** signal an error if a symbol that names a declaration is used
380 as the name of a type, or vice versa
381 ** allow using the (declare (typespec var*)) abbreviation for
382 (declare (type typespec var*)) with all type specifiers
383 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
384 ** always return NIL from PPRINT-POP when OBJECT is NIL
385 ** don't signal errors when pretty-printing malformed LABELS,
386 FLET or MACROLET forms
387 ** declarations in a DOLIST body are also in scope for the
389 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
391 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
394 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
395 * fixed cross-compiler leakages that prevented building a 32-bit
396 target with a 64-bit host compiler.
397 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
398 opened with :IF-EXISTS :OVERWRITE.
399 * fixed bug 281: error for an invalid qualifier in a short-form method
400 combination method is not signalled until the faulty method is called.
401 * bug fix: iteration variable type inferrer failed to deal with open
402 intervals. (reported by Alan Shields)
403 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
404 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
405 or more runtime options were provided to the sbcl binary.
406 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
407 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
408 * optimization: there's now a fast-path for fixnum arguments in the
409 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
410 * optimization: the code generated on x86-64 is more compact thanks
411 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
412 * bug fix: floating-point exception handling now partly works on
413 x86-64. (Thanks to James Knight)
414 * improvement to the MIPS backend from Thiemo Seufer: C-style
415 64-bit long long arguments and return values to alien functions
417 * fixed some bugs revealed by Paul Dietz' test suite:
418 ** the type-error signalled from WARN has a filled-in DATUM slot.
419 ** the type-error required when a stream is not associated with
420 a file has the stream as its datum.
421 ** type-errors on single-floats on x86-64 no longer have
422 :INVALID-OBJECT as the datum
423 ** the type-errors signalled for invalid function names now have
424 a correct expected type
425 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
426 for structures defined with a :TYPE. Documentation strings for
427 typed structures are no longer immediately discarded
428 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
429 broadcast streams and synonym streams. FILE-LENGTH now also works
430 on broadcast streams.
432 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
433 * incompatible change: the --noprogrammer option, deprecated since
434 version 0.7.5, has been removed. Please use the equivalent
435 --disable-debugger option instead.
436 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
438 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
439 inoperational for a while now) has been completely removed.
440 * null lexical environments are now printed as #<NULL-LEXENV>,
441 significantly reducing the amount of clutter in typical backtraces.
442 * documentation on weak pointers, finalization, and after GC hooks
443 has been added to the manual.
444 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
445 as other specialized array types, is much faster. SUBSEQ and
446 COPY-SEQ on such arrays have also been sped up.
447 * optimization: EQL is now more efficient when at least other argument
448 is known to be of type (OR FIXNUM (NOT NUMBER)).
449 * fixed bug: the runtime is now less vulnerable to changes in the
450 size of the SBCL object on OS X, and virtual memory is reserved for
451 all spaces that need to be at a fixed address.
452 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
453 * fixed bug: finalizers and after GC hooks that cause consing are now
455 * fixed bug: compiler error messages and summaries are now printed to
456 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
457 * fixed inference of the upper bound of an iteration variable.
458 (reported by Rajat Datta).
459 * fixed bug 211e: calling local functions with duplicated constant
460 keyword argument no longer causes a bogus style warning about an
462 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
463 ftype declaration to be disregarded. (reported by Dave Roberts)
464 * fixed bug 373: caused by erronous compilation of references to alien
465 variables in the runtime on ppc/darwin.
466 * fixed bug 376: CONJUGATE type deriver.
467 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
468 (reported by Baughn on #lisp)
469 * a cross-compiler bug on non-x86 platforms has been identified and
470 fixed. (thanks to Bruno Haible)
471 * improvements to the MIPS runtime code for increased stability.
472 (thanks to Thiemo Seufer)
473 * increased the maximimum compact environment size to allow
474 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
475 * partial workaround for bug 135: don't name the function we're
476 calling for hairy cases of EVAL, so as not to accumulate one environment
477 entry per call to EVAL. (reported by Kevin Reid)
478 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
479 * optimization: type testing for non-vector arrays should be faster.
480 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
482 * bug fix: setting 31st element of a bit vector to zero did not work
484 * bug fix: redefining a class definition which failed due to a
485 previous accessor / function clash now works (but see BUGS entry
486 #380 for more problems in this area). (thanks to Zach Beane)
487 * the long form of DEFINE-METHOD-COMBINATION disables method group
488 checking when given a single method group with pattern *. (thanks
490 * on x86 compiler supports stack allocation of results of simple
491 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
492 * support for building and running on Mac OS X 10.4 "Tiger" has been added
493 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
494 * fixed some bugs related to Unicode integration:
495 ** the restarts for recovering from input and output encoding
496 errors only appear when there is in fact such an error to
498 * fixed some bugs revealed by Paul Dietz' test suite:
499 ** MISC.549 and similar: late transformation of unsafe type
500 assertions into derived types caused unexpected code
502 ** SCALE-FLOAT type deriver is less wrong.
503 ** type derivers for EXP, LOG and similar functions compute result
504 types for complex arguments better.
505 ** (MISC.563) CONJUGATE type deriver works for very restricted
507 ** out-of-line type testers for character strings are available.
508 ** EQUAL compiler transform understands specialness of objects
510 ** accessing double-floats stored on the stack now works on x86-64.
511 ** debugger internals could sometimes create invalid lispobjs,
512 resulting in GC crashes.
513 ** MISC.548: type check weakening can convert required type into
515 ** initialization forms for bindings are not in scope of free special
518 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
519 * incompatible change: thread support for non-NPTL systems has
520 been removed - locking is buggy and unreliable. A threaded
521 SBCL build will now warn at startup and refuse to create
522 new threads, unless futex support is detected in the kernel
523 * incompatible change: the top level REPL now has only an ABORT
524 restart associated with it, not TOPLEVEL and ABORT as it used to.
525 TOP and TOPLEVEL are now available as debugger commands for
526 returning to the top level.
527 * incompatible change: forms evaluated in the REPL now use the
528 global optimization policy.
529 * incompatible change: user- and system-initialization files are
530 no longer processed with LOAD, but by READ and EVAL; hence the
531 global optimization policy, startup package, readtable, etc,
533 * The .fasl file format number has been incremented because of
534 various incompatible changes.
535 * internal entry point details and argument counts no longer appear
536 in backtraces unless explicitly requested by setting
537 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
538 * built-in and standard functions no longer have names like "top
539 level local call to FOO".
540 * fixed bug 32: functions defined in non-null lexical environments
541 now have more legible printed representation
542 * fixed bug 33: functions defined in non-null lexical environemnts
543 are now more amenable to inspection by INSPECT.
544 * workaround for bug 354: XEPs no longer appear in backtraces unless
545 explicitly requested.
546 * fixed bug: receiving the signal which results from attempting to
547 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
548 SIGBUS on other *BSDs) on architectures where the C stack is also
549 the Lisp stack (x86 and x86-64) from foreign code no longer leads
550 to debugger confusion or wild instability. (reported by Cheuksan
552 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
553 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
554 (reported by Lutz Euler)
555 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
556 compiled code. (reported by Paul Dietz)
557 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
558 specialized parameter to a method to something that is not TYPEP
559 the specializer is now possible.
560 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
561 face of package deletion.
562 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
563 pathname. (reported by David Tolpin on c.l.l)
564 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
565 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
566 STRING-STREAM and FILE-STREAM, respectively.
567 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
568 than previously; in addition, its test suite is now run on build.
569 (thanks to Luke Gorrie)
570 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
571 * optimization: added a immediate representation for single-floats
573 * fixed some bugs revealed by Paul Dietz' test suite:
574 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
575 ** attempting to create a package with a colliding nickname causes
576 correctable errors to be signalled.
577 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
578 ** several x86-64 backend bugs related to sign-extension of immediate
581 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
582 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
583 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
585 * fixed loading of multiply forward-referenced layouts.
586 (thanks to Cheuksan Wang)
587 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
588 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
589 (see COMPILE-FILE documentation for details of :PRINT option)
590 * fixed bugs 19 and 317: fixed-format floating point printing is
591 more accurate. This also fixes a bug reported by Adam Warner
592 related to the ~@F format directive.
593 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
595 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
596 dispatch table if the from-char is a dispatch macro character.
597 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
598 by the wordsize no longer ignore the last word. (reported by Lutz
600 * fixed bug in type checking of dynamic-extent variables. (reported
602 * optimization: sequence traversal functions use their freedom to
603 coerce function designators to functions.
604 * optimization: code with many calls to CLOS methods specialized on
605 CLOS classes has had redundant type checks removed.
606 * optimization: added declarations to speed up operations that access
607 the internal character database (for example STRING-UPCASE)
608 * optimization: comparison operations between floats and sufficiently small
609 fixnums no longer create extra rationals
610 * fixed some bugs related to Unicode integration:
611 ** portions of multibyte characters at the end of buffers for
612 character-based file input are correctly transferred to the
613 start of the buffer at the next read.
614 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
615 passing it through to OPEN.
616 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
617 argument given to internal calls to OPEN.
618 ** the built-in comment readers (introduced by character sequences
619 ";" and "#|") are more forgiving to encoding errors; they will
620 STYLE-WARN and then attempt to resync the stream at a character
621 boundary. (thanks to Teemu Kalvas)
622 * fixed some bugs in the x86-64 port:
623 ** Negative short int return values from c-calls are sign-extended
625 ** The stack is aligned to 16-bytes for c-calls, as required by
626 the ABI. (thanks to Cheuksan Wang)
627 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
628 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
629 for enums. (thanks to Vincent Arkesteijn)
630 ** Multiple small optimizations and bugfixes for floating point
632 * fixed some bugs revealed by Paul Dietz' test suite:
633 ** Space, Tab, Linefeed, Return and Page have the invalid
634 secondary constituent character trait.
635 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
637 ** WITH-INPUT-FROM-STRING should only update the index place on
639 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
641 ** Bit-array operations (BIT-AND and similar) worked incorrectly
642 with one-dimensional arrays with fill pointers.
643 ** TYPE-OF failed on a complex with an integer realpart and a
645 ** compiler failure during type inference for the code of form
646 (IF (EQL X (THE ...)) ...) (MISC.535).
648 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
649 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
650 platform. The port supports 61-bit fixnums, large memory spaces
651 and reloading shared object files.
652 * enhancement: saving cores with foreign code loaded is now
653 supported on ppc/Darwin in addition to the previously supported
655 * enhancement: the statistical profiler now walks deeper into the
656 call stack for more meaningful call-graphs and accrued time
657 reports (x86/x86-64 only). It also now reports time spent in
659 * enhancement: it is now possible to trace most individual methods
660 of a generic function in addition to tracing the generic function
662 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
664 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
665 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
666 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
667 produces an error. (thanks to Vincent Arkesteijn)
668 * bug fix: NAMESTRING on pathnames with :WILD components in their
669 directories works correctly. (thanks to Artem V. Andreev)
670 * fixed bug 125: compiler preserves identity of closures. (reported
672 * bug fixed: functions with &REST arguments sometimes failed with
673 "Undefined function" when compiled with (DEBUG 3). (reported by
675 * bug fix: overflow during compiling of setting element of a bit
676 vector with constant index and value. (reported by Timmy Douglas)
677 * build fix: fixed the dependence on *LOAD-PATHNAME* and
678 *COMPILE-FILE-PATHNAME* being absolute pathnames.
679 * on x86 compiler partially supports stack allocation of dynamic-extent
681 * GO and RETURN-FROM do not check the extent of their exit points
682 when compiled with SAFETY 0.
683 * fixed some bugs related to Unicode integration:
684 ** encoding and decoding errors are now much more robustly
685 handled; it should now be possible to recover even from invalid
686 input or output to the terminal. (thanks to Teemu Kalvas)
687 ** provided a first cut at implementing STRING-TO-OCTETS and
688 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
689 ** altered the SB-MD5 contributed module slightly, changing the
690 interface just enough for it to be supportable for builds where
691 lisp characters are not eight bits.
692 * fixed some bugs revealed by Paul Dietz' test suite:
693 ** the FORMATTER-generated functions for ~V[ conditionals require
694 the correct number of arguments.
695 ** READ-FROM-STRING returns the mandated second value when applied
696 to displaced strings.
697 ** the #\Rubout and #\Backspace characters are treated as invalid
698 constituent characters by the tokenizer.
700 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
701 * new feature: reloading changed shared object files with
702 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
703 * new feature: references to foreign variables and functions
704 can now be compiled and loaded before the corresponding shared
705 object file is loaded, as long as the foreign definitions are
706 available at runtime.
707 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
708 supported. (thanks to Dan Debertin)
709 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
710 just keywords. (thanks to Vincent Arkesteijn)
711 * fixed bug #331: structure-class instances corresponding to
712 DEFSTRUCT forms are now created eagerly.
713 * fixed bug #345: backtraces from calls to undefined functions work
714 on x86 as well. Related bug #61 is now also partially fixed on x86
715 (backtraces from throws to unknown catch tags.)
716 * bug fix: lambda-list parsing is now stricter vrt. order and number
717 of lambda-list keywords.
718 * bug fix: as specified by AMOP, an error is signalled if a
719 class-option appears multiple times in a DEFCLASS form. (reported
721 * bug fix: RANDOM can be compiled when the compiler derives the type
722 of its numeric argument as a disjoint set of small integers.
723 (reported by Paul Dietz)
724 * bug fix: starting a core saved with shared objects loaded when
725 those objects are not available no longer causes threaded SBCL to
726 hang. (reported by Sean Ross)
727 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
729 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
730 argument, which must always be NIL. (reported by Kalle Niemitalo)
731 * bug fix: printing 1.0d+23 no longer results in an error.
732 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
733 * bug fix: structure slot setters preserve evaluation order. (thanks
735 * fixed some bugs related to Unicode integration:
736 ** RUN-PROGRAM can allow its child to take input from a Lisp
737 stream. (reported by Stefan Scholl)
738 ** ASDF-INSTALL successfully downloads .tar.gz files in all
739 locales. (reported by Ken Causey)
740 * fixed some bugs revealed by Paul Dietz' test suite:
741 ** INCF, DECF and REMF evaluate their place form as specified in
743 ** FORMATTER expands ~{ ~} iteration directives with V or #
744 parameters correctly.
745 ** FORMATTER deals with the ~@[ ~] conditional directive where the
746 consequent uses no arguments correctly.
747 ** the system has a partial understanding of the (COMPLEX RATIO)
749 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
752 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
753 * new feature: a build-time option (controlled by the :SB-UNICODE
754 keyword feature, enabled by default) for building the system with
755 support for the entire 21-bit character space defined by the
757 * new feature: the system now has rudimentary external-format
758 support; the primary user-visible change at this time is that
759 characters with the high bit set (such as lower-case-e-acute) will
760 print correctly to a terminal in a UTF-8 environment.
761 * minor incompatible change: BASE-CHAR no longer names a class;
762 however, CHARACTER continues to do so, as required by ANSI.
763 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
764 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
765 has been moved to the SB-EXT package (temporarily re-exported from
767 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
768 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
770 * fixed bug #351: better error-handlind and reporting for malformed
772 * fixed bug #350: bignum-printing is now more memory-efficient,
773 allowing printing of very large bignums, eg. (expt 2 10000000).
774 (reported by Bruno Haible)
775 * fixed bug #302: better primitive-type selection for intersection
777 * fixed bug #308: non-graphic characters now all have names, as
778 required. (reported by Bruno Haible)
779 * bug fix: redefining a class with different superclasses now correctly
780 removes it from the direct-subclasses of its previous superclasses.
781 (reported by David Morse)
782 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
783 new value now works. (reported by Bruno Haible)
784 * bug fix: correct canonicalization of multiple non-standard slot
785 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
786 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
787 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
788 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
789 now exists, an signals an error.
790 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
791 during macroexpansion for non-iteration variables as well. (reported
792 by Bruno Haible for CMUCL)
793 * bug fix: Cyclic structures and unprintable objects in compiler
794 messages no longer cause errors. (reported by Bruno Haible)
795 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
796 and READ-FROM-STRING all now return a primary value of NIL if
797 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
798 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
799 (reported by Bruno Haible for CMUCL)
800 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
801 do not designate a condition. (reported by Bruno Haible for
803 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
804 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
805 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
806 permitting resolution in favour of any of the conflicting symbols.
807 (reported by Bruno Haible for CMUCL)
808 * bug fix: EQUAL compiler optimizations is less aggressive on
809 strings which can potentially compare true despite having distinct
810 specialized array element types.
811 * bug fix: unit enumerations can be defined without dividing by
812 zero. (thanks to Vincent Arkesteijn)
813 * FORMAT compile-time argument count checking has been enhanced.
814 (report from Bruno Haible for CMUCL)
815 * a partial workaround for the bug 262: the compiler does not try to
816 inline-expand a local function doing RETURN-FROM from a deleted
817 BLOCK. (thanks to Peter Denno for the bug report and to David
818 Wragg for the simple test case)
819 * fixed some bugs revealed by Paul Dietz' test suite:
820 ** RENAME-PACKAGE allows all package designators as new package
822 ** constraint propagation and lambda variable substitution are
823 more cautious in dealing with partially deleted code.
824 ** compiler sometimes lost reoptimization passes.
825 ** CERROR, given a condition as condition designator, passes its
826 remaining arguments to the continue format control without
828 ** Case-altering FORMAT directives work correctly on non-ASCII
830 ** The REAL type specifier handles bounds outside the floating
831 point ranges without signalling FLOATING-POINT-OVERFLOW.
832 ** Functions with IR1-transformations can create intercomponent
833 references to global functions.
834 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
836 ** FORMAT ~R treats a nil value for its first parameter correctly.
838 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
839 * enhancement: saving cores with foreign code loaded is now
840 supported on x86/NetBSD and sparc/Linux in addition to the previously
842 * bug fix: on some platforms repeated installations caused multiple
843 copies of HTML documentation to be installed -- should not happen
844 any more. (reported by Stefan Scholl)
845 * bug fix: parsing self-recursive alien record types multiple times
846 no longer causes infinite recursion. (reported by Thomas F. Burdick,
847 original patch by Helmut Eller for CMUCL)
848 * bug fix: stack-exhaustion detection works now on NetBSD as well.
849 (thanks to Richard Kreuter)
850 * bug fix: defining classes whose accessors are methods on existing
851 generic functions in other (locked) packages no longer signals
852 bogus package lock violations. (reported by François-René Rideau)
853 * bug fix: special variables as DEFMETHOD parameters no longer have
854 associated bogus type declarations. (reported by David Wragg and
856 * bug fix: read-write consistency on streams of element-type
857 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
858 * bug fix: redefiniton of the only method of a generic function with
859 no DEFGENERIC no longer emits a full WARNING. In addition,
860 redefinition of generic functions with no DEFGENERIC to an
861 incompatible lambda list now signals an error. (thanks to Zach
863 * bug fix: DEFGENERIC now works even when there's a function of the
864 same name in an enclosing lexical environment. (thanks to Zach
866 * fixed compiler failure, caused by instrumenting code during
867 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
868 * optimization: added loop analysis and improved register allocation
869 * fixed some bugs revealed by Paul Dietz' test suite:
870 ** POSITION on displaced vectors with non-zero displacement
871 returns the right answer.
872 ** (SIMPLE-STRING) is a valid type specifier for sequence
874 ** *PRINT-LEVEL* handling for slotless structures is pedantically
876 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
877 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
879 ** The justification version of the FORMAT ~< directive treats
880 non-zero minpad parameter correctly.
882 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
883 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
884 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
885 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
886 the supported interface.
887 * new feature: Single-stepping of code compiled with DEBUG 2 or
888 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
889 * new feature: saving cores with foreign code loaded is now
890 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
891 Timothy Moore's work for CMUCL)
892 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
893 parameters to * instead of NIL if no initform is supplied.
894 (reported by Johan Bockgaard)
895 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
896 correctly when FUNCALL appears as the car of the form. Note:
897 despite this FUNCALL forms are not currently subject to
898 compiler-macro expansion. (port of Raymond Toy's fix for the
899 same from CMUCL, reported by Johan Bockgaard)
900 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
901 (thanks for Teemu Kalvas)
902 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
903 case instead of terminating the iteration (thanks for Julian
904 Squires, Sean Champ and Raymond Toy)
905 * bug fix: incorrect expansion of defgeneric that caused a style
906 warning. (thanks for Zach Beane)
907 * bug fix: slot accessor effective method computation works properly
908 for classes with multiple non-standard applicable methods on
909 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
910 * on x86 compiler supports stack allocation of results of LIST and
911 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
912 CMUCL implementation by Gerd Moellmann)
913 * fixed some bugs revealed by Paul Dietz' test suite:
914 ** FORMAT strings with both the ~<~:;~> form of the justification
915 directive and pretty-printing directives cause an error.
917 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
918 * incompatible change: the internal functions
919 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
920 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
921 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
922 instead of the old functions.
923 * new feature: on platforms where "dladdr" is available foreign
924 function names now appear in backtraces. (based on Helmut Eller's
926 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
927 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
929 * documentation: ASDF manual is now distributed with SBCL.
930 * bug fix: SBCL can now load its contributed modules with REQUIRE
931 even if the system-provided entries have been removed from
932 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
934 * bug fix: backtraces involving undefined functions or assembly
935 routines are more informative. (thanks to Brian Downing)
936 * bug fix: mutually referent alien structures now work correctly.
937 (reported by Rick Taube)
938 * bug fix: structures defined by WITH-ALIEN can be referred to
939 within other definitions in the same WITH-ALIEN.
940 * bug fix: division operators (MOD, TRUNCATE and the like) with
941 constant zero divisors and integer dividends no longer generate
943 * bug fix: provide default methods for INPUT-STREAM-P and
944 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
945 * bug fix: improve the detection and resolution of MOP metacycles
946 (where effective method computation depends on the generic function
947 having its effective method computed). (reported by Bruno Haible)
948 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
949 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
950 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
951 * bug fix: pathnames with relative directory components can be
952 represented relative to default pathnames.
953 * optimization: in taking the GCD of bignums, reduce the two bignums
954 to approximately the same size (using Euclid's algorithm) before
955 applying the more sophisticated binary GCD. (thanks to Juho
957 * optimization: COUNT on bitvectors now operates word-at-a-time.
958 * optimization: ASH with a positive, but not necessarily constant,
959 (leftwards) shift, when in a modular context, is compiled to a
961 * fixed some bugs revealed by Paul Dietz' test suite:
962 ** FORMAT variable parameters ("~V<char>") are defaulted properly
963 if the corresponding argument is NIL.
964 ** FORMAT directives accepting a minpad parameter treat negative
966 ** literal commas as character parameters to FORMAT directives are
968 ** literal spaces directly after ~<Newline> directives within a
969 format-logical-block (~:< ~@:>) do not induce :FILL-style
970 conditional newlines.
971 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
972 as specified: it no longer includes conditional newlines).
973 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
975 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
976 :PER-LINE-PREFIX argument does not evaluate to a string.
978 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
979 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
980 the manual for details; for now, package locks can be disabled by
981 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
982 if no major problems are found then it is likely that they will be
983 compiled in unconditionally.
984 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
985 unsupported operators on all platforms. To load a shared library
986 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
987 object file, link it into a shared library outside of SBCL and
988 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
990 * bug workaround: ROOM T and the GC were not working together
991 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
992 reporting facility aren't preserved. That reporting has been
993 disabled, so now until and unless someone figures out how to make it
994 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
995 * minor incompatible change: as threatened around sbcl-0.8.0, the
996 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
997 The symbols are also exported from SB-PCL for backwards
998 compatibility, but more so than before SB-PCL should be treated as
999 an implementation-internal package.
1000 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1002 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1003 It is known as of this release not to work on the Alpha, however.
1004 * fixed bug #167: errors signalled due to illegal syntax in method
1005 bodies are now more legible.
1006 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1007 designators and can hence be used with TYPEP.
1008 * fixed bug #333: CHECK-TYPE now ensures that the type error
1009 signalled, if any, has the right object to be accessed by
1010 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1011 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1012 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1014 * fixed bug #334: programmatic addition of slots using specialized
1015 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1016 and :ALLOCATION :CLASS slots.
1017 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1018 float exponents. (rereported by Peter Seibel)
1019 * fixed a bug: #\Space (and other whitespace characters) are no
1020 longer considered to be macro characters in standard syntax by
1021 GET-MACRO-CHARACTER.
1022 * fixed bug: initialization of condition class metaobjects no longer
1023 causes an instance of the condition to be created. (reported by
1025 * fixed bug: it is now possible to have slots such that
1026 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1027 description is neither :INSTANCE nor :CLASS.
1028 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1029 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1030 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1031 non-standard SLOT-DEFINITION-ALLOCATION values.
1032 * fixed bugs: various race conditions handling exiting threads.
1033 CL-PPCRE's thread torture test now passes.
1034 * fixed arguably-a-bug: GC time proportional to number of threads in
1035 system even when most of them are idle
1036 * optimization: improved performance of BIT and SBIT on bit-vectors.
1037 * .fasl file incompatibility: The fasl file version number has
1038 been incremented because of changes associated with package locks.
1040 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1041 * minor incompatible change: the system no longer provides
1042 optimization hints (conditions which are TYPEP
1043 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1044 mode; these hints are still emitted when higher SPEED optimization
1046 * new contrib module: a sampling profiler (profiling by statistical
1047 sampling, rather than by instrumenting functions) is available as
1048 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1049 * the behaviour of the standard function ED is now customizeable by
1050 third parties through a hook variable: see ED's documentation
1051 string for information on the protocol.
1052 * the compiler no longer emits efficiency notes for (FUNCALL X)
1053 when the type of X is uncertain under default optimization
1055 * fixed bug 276: mutating a binding of a specialized parameter to a
1056 method to something that is not TYPEP the specializer is now
1058 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1059 exhibit the required behaviour on the x86 platform. (thanks to
1060 Peter van Eynde, Eric Marsden and Bruno Haible)
1061 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1062 even for difficult arguments. (reported by Peter Graves)
1063 * fixed bug 141a: the backquote printer now descends quoted
1065 * fixed another bug in backquote printing: no more destructive
1066 modification of the form's list structure. (reported by Brian
1068 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1069 move between its address being taken and the call to
1070 interrupt_thread, fixing a crashing race condition.
1071 * the SB-POSIX contrib implementation has been adjusted so that it
1072 no longer exhibits ridiculously poor performance when constructing
1073 instances corresponding to C structs.
1075 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1076 * minor incompatible change: the sb-grovel contrib now treats C
1077 structures as alien (in the sense of SB-ALIEN) objects rather than
1078 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1079 has implications for memory management of client code
1080 (sb-grovel-returned objects must now be manually managed) and for
1081 type safety (alien objects now have full types).
1082 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1083 used to control emission of compiler diagnostics, rather than the
1084 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1085 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1086 quality should be considered deprecated.
1087 * install.sh now installs the user manual as well
1088 * (not quite a new documentable feature, but worth considering in
1089 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1090 beginnings of a semantically meaningful condition hierarchy is
1091 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1093 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1094 designator as the defaults argument.
1095 * fixed bug: Displaced arrays whose displaced-to array has become
1096 too small now cause ARRAY-DIMENSION to signal an error, providing
1097 for safer bounds-checking. (reported by Bruno Haible)
1098 * fixed bug: DEFCLASS slot definitions with identical :READER and
1099 :WRITER names now signal a reasonable error. (reported by Thomas
1101 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1103 * fixed bug: Invalid slot specification errors now print correctly.
1104 (thanks to Zach Beane)
1105 * fixed bug 320: Shared to local slot value transfers in class
1106 redefinitions now happen corrently from superclasses as
1107 well. (reported by Bruno Haible)
1108 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1110 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1111 improper lists correctly. (reported by Bruno Haible)
1112 * fixed bug 313: source-transform for <fun-name> was erroneously
1113 applied to a call of a value of a variable with name <fun-name>.
1114 (reported by Antonio Menezes Leitao)
1115 * fixed bug 307: The obsolete instance protocol ensures that
1116 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1117 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1118 remove dead unknown-values globs from the middle of the stack.
1119 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1120 easier to resume long computations after using *BREAK-ON-SIGNALS*
1121 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1122 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1123 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1124 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1125 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1126 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1128 * fixed a (fairly theoretical) bug in string printing: if
1129 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1130 does not have array-element-type equal to the most general string
1132 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1133 function name. (thanks to Nikodemus Siivola)
1134 * fixed bug: compile-time format string checker failed on
1135 non-closed ~{. (reported by Thomas F Burdick)
1136 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1137 don't warn on legal loop code involving more than one aggregate
1138 boolean. (thanks to Nikodemus Siivola)
1139 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1140 in keyword position to MAKE-PACKAGE was not being reported as
1142 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1143 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1145 * fixed bugs 280 and 312: the checking for multiple definitions in a
1146 file is less likely to become confused by uses of inline
1148 * fixed bug: the #S reader macro performs the keyword coercion
1149 specified for slot names. (reported by Kalle Niemitalo)
1150 * fixed bug: lambda lists may contain symbols whose names start with
1151 & but are not lambda-list-keywords; their occurrence triggers a
1153 * fixed bug 321: define-method-combination argument lambda lists do
1154 not cause invalid code to be generated when &optional or &aux
1155 variables have default values. (reported by Bruno Haible)
1156 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1157 have CLOS classes; this appears also to have fixed reported
1158 instability in stack exhaustion detection.
1159 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1160 MUFFLE-WARNING when no associated restart is present is now
1162 * optimization: rearranged the expansion of various defining macros
1163 so that each expands into only one top-level form in a
1164 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1166 * optimization: used a previously little-used slot in symbols to
1167 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1169 * fixed some bugs revealed by Paul Dietz' test suite:
1170 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1172 ** Fixed an optimization bug related to inheritance of initforms
1173 from local to shared slots.
1174 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1175 ** CAST optimizer forgot to flush argument derived type.
1176 ** print/read consistency on floats is now orders of magnitude
1177 more likely. (thanks also to Bruno Haible for a similar report
1179 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1180 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1181 for objects of type REAL. Make it so.
1182 ** SXHASH is specified (infelicitously) to respect similarity,
1183 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1184 it so. (thanks to Markus Ziegler)
1185 ** on the Alpha, the compiler succeeds in compiling functions
1186 returning a known number of arguments greater than 63.
1187 ** fixed handling of invalid NIL arguments in keyword position
1189 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1191 ** provided a readably-printable representation for RANDOM-STATE
1193 ** ensured that pathnames, where they have a namestring, always
1194 print using #P"..." syntax.
1196 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1197 * Support for the forthcoming 2.0 version of the NetBSD kernel
1198 running on x86 hardware has been added. (thanks to Perry
1199 E. Metzger most immediately, and others for their past work)
1200 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1201 current release version; SBCL's OpenBSD support had been broken
1202 since about the time of OpenBSD's switch to ELF binary format.)
1203 * [placeholder for DX summary]
1204 ** user code with &REST lists declared dynamic-extent, under high
1205 speed or space and low safety and debug optimization policy.
1206 * The manual has been converted to Texinfo format and the debugger
1207 chapter from the cmucl manual has been added.
1208 * A facility has been added to extract documentation strings from
1209 sbcl and store them as Texinfo-formatted snippets for inclusion in
1210 the manual (via Texinfo's @include directive)
1211 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1212 the test case to Dave Roberts)
1213 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1214 pointers. (reported by Sean Ross)
1215 * bug fix: PROFILE output is printed nicely even for large numerical
1216 values. (thanks to Zach Beane)
1217 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1218 greater than 32 handle EOF correctly.
1219 * bug fix: on X86 an immediate argument of the IMUL instruction is
1220 correctly printed in disassembly. (thanks to Lutz Euler)
1221 * bug fix: class slots in redefined classes preserve their old
1222 values. (thanks to Bruno Haible and Nikodemus Siivola)
1223 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1224 succeeds. (reported by Marco Baringer)
1225 * fixed some bugs revealed by Paul Dietz' test suite:
1226 ** READ-SEQUENCE now works on ECHO-STREAMs.
1227 ** RATIONALIZE works more according to its specification. (thanks
1230 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1231 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1232 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1233 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1234 should matter to you only if you rebind the printer control
1235 variables and then find you want different bindings in the
1236 debugger than in the ordinary execution of your program.)
1237 * The runtime build system has been tweaked to support building
1238 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1239 assembler and linker. (thanks to Nikodemus Siivola)
1240 * Unbound, undefined, undeclared variables now trigger full
1241 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1242 more often programmer error than deliberate exploitation of undefined
1244 * optimization: the hash algorithm for strings has changed to one
1245 that is less vulnerable to spurious collisions. (thanks to Juho
1247 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1248 needless bounds checking. (thanks to Juho Snellman)
1249 * optimization: implemented multiplication as a modular
1250 (UNSIGNED-BYTE 32) operation on the PPC backend.
1251 * fixed some bugs revealed by Paul Dietz' test suite:
1252 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1254 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1255 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1256 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1258 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1260 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1262 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1263 their output stream on EOF from read.
1264 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1265 have been read to end-of-file.
1266 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1268 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1269 description of determination of which consecutive characters
1271 ** Printing the "Space" character with escaping on now yields "#\\ ",
1272 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1273 ** Reading floating-point numbers with *READ-BASE* set to a number
1274 less than 10 works correctly.
1275 ** Reading floating-point numbers with *READ-BASE* set to a number
1276 more than 10 works correctly.
1277 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1278 the readtable currently in effect.
1280 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1281 * minor incompatible change: parsing of namestrings on a physical
1282 (Unix) host has changed; numbers after the final #\. in a
1283 namestring are no longer interpreted as a version field. This is
1284 intented to be largely invisible to the user, except that the
1285 meaning of the namestring "*.*.*" has changed: it now refers to a
1286 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1287 should usually be replaced by
1288 (make-pathname :name :wild :type :wild :version :wild)
1289 with the added benefit that this is more likely to be portable.
1290 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1291 option to OPEN now signals an error if the file being opened
1292 exists; this may have an impact on existing code.
1293 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1294 signals received, on the PowerPC platforms (both Linux and
1295 Darwin). (thanks to Pierre Mai for pointing out the location of
1297 * several fixes on OS X: The system now builds and runs cleanly on
1298 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1299 (thanks to Brian Mastenbrook)
1300 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1301 second-resolution: integer multiples of 1/3600 between -24 and 24.
1302 (thanks to Vincent Arkesteijn)
1303 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1304 type when called with 1 argument; PEEK-CHAR checked type of
1305 PEEK-TYPE only after having read first character from a
1306 stream. (reported by Peter Graves)
1307 * bug fix: the garbage collector now has much better locality
1308 behaviour, and in particular no longer treats all memory as being
1309 exhausted when many small objects point to each other in a deeply
1311 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1312 recognized as being TYPEP their class.
1313 * bug fix: the PUSHNEW documentation string has been corrected.
1314 (thanks to Vincent Arkesteijn)
1315 * bug fix: defaulting of the value for the last of an atypically
1316 large number of multiple values being bound was not being
1317 performed correctly on the Alpha or PPC platforms
1318 * optimization: implemented multiplication as a modular
1319 (UNSIGNED-BYTE 32) operation on the x86 backend.
1320 * optimization: SEARCH on simple-base-strings can now be open-coded.
1321 (see also contrib/compiler-extras.lisp for inspiration for
1322 teaching the compiler about the Boyer-Moore algorithm).
1323 * value, returned by MAX (and MIN) called with several EQUALP, but
1324 not EQL, arguments now does not depend on compiler settings.
1325 * fixed some bugs revealed by Paul Dietz' test suite:
1326 ** in stack analysis liveness information is propagated from
1327 non-local entry points.
1328 ** pathwise CAST removing failed when the CAST node did not start
1330 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1331 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1333 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1334 host is already defined.
1335 ** RENAME-FILE works on streams instead of signalling an internal
1337 ** PEEK-CHAR uses the current readtable when determining whether
1338 or not a character is whitespace.
1339 ** MERGE-PATHNAMES handles the case when the pathname does not
1340 specify a name while the default-pathname specifies a version.
1341 ** Pathnames now stand a better chance of respecting print/read
1343 ** Attempting to use standardized file system operators with a
1344 pathname with invalid :DIRECTORY components signals a
1346 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1348 ** DIRECTORY on logical pathnames is more correct.
1349 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1350 signal a TYPE-ERROR if their argument is not a stream.
1351 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1352 designator argument does not designate a stream.
1353 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1354 examining the synonym.
1355 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1357 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1358 element-type larger than ([UN]SIGNED-BYTE 32).
1360 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1361 * When built with the :SB-FUTEX feature, threaded builds now take
1362 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1363 for faster/more reliable mutex and condition variable support.
1364 * Incompatible change (but one you probably shouldn't have been using
1365 anyway): the interface and code for arbitrating between multiple
1366 threads in the same user session has been redesigned.
1367 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1368 error when called without an explicit environment argument.
1369 (thanks to Nikodemus Siivola)
1370 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1371 confused about their position. (thanks to Adam Warner and Gerd
1373 * bug fix: Pretty printing backquoted forms with unquotations in the
1374 argument list position of various code constructs such as LAMBDA
1375 now works correctly. (reported by Paul Dietz)
1376 * bug fix: Pretty printing unquotations no longer loses all
1377 stream position information.
1378 * optimization: performance of string output streams is now less
1379 poor for multiple small sequence writes.
1380 * optimization: performance of CSUBTYPEP in the presence of complex
1381 expressions involving CONS and NOT many times has been improved.
1382 (reported by Paul Dietz)
1383 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1385 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1386 streams for the REPL, for the debugger, and so on. (thanks to
1388 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1389 * fixed some bugs revealed by Paul Dietz' test suite:
1390 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1391 is ignored if it is too small, rather than propagating through
1392 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1393 ** extremely complex negations of CONS types were not being
1394 sufficiently canonicalized, leading to inconsistencies in
1396 ** VALUES tranformer lost derived type.
1398 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1399 * fixed a bootstrapping bug: the build process no longer assumes
1400 that the various BOOLE-related constants have the same value in
1401 host and target lisps. (noted by Paul Dietz' test suite on an
1402 SBCL binary built from CLISP)
1403 * The system can now be dynamically linked on the MIPS platform,
1404 which enables dynamic loading of foreign code from Lisp. (thanks
1405 to Ralf Baechle for discussions on the MIPS ABI)
1406 * The system now records debugging information for its own source
1407 files in a filesystem-position-independent manner, relative to
1408 the "SYS" logical host.
1409 * fixed a compiler bug: MV-LET convertion did not check references
1410 to the "max args" entry point. (reported by Brian Downing)
1411 * tweaked disassembly notes to be less confident about proclaiming
1412 some instruction as an LRA. (thanks to Brian Downing)
1413 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1414 now each have their own history, command character, and other
1415 characteristics. (thanks to David Lichteblau)
1416 * fixed some bugs revealed by Paul Dietz' test suite:
1417 ** compiler failure in compiling LOGAND expressions including a
1419 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1420 shift greater than 32.
1421 ** FLUSH-DEST did not mark blocks for type check regeneration.
1422 ** HANDLER-CASE failed to accept declarations in handler clauses
1423 in some circumstances.
1425 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1426 * New code in contrib/sb-introspect (still probably not entirely
1427 stable yet) provides some support for smart Lisp development
1428 environments like SLIME.
1429 * The conditions signalled for errors occurring when loading .fasl
1430 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1431 in a way which should help ASDF recover gracefully.
1432 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1433 stuff has been cleaned up. If you code contrib/ stuff, this might
1434 affect you, and you can look at contrib/README, contrib/STANDARDS,
1435 and/or the 0.8.4.27 diff to check.
1436 * In full calls the compiler now does not generate checks for declared
1437 argument types for all arguments.
1438 * various threading fixes
1439 ** and some experimental patches which didn't make it into
1440 the main tree for this release, but which are shipped in
1441 contrib/experimental-thread.patch as a possible fix for some
1442 failures (deadlock, spinning...) in GC-intensive multithreaded
1444 * fixed PPC build problem (source code incompatibility of different
1445 library versions): added offsetof() hackery which attempts to divine
1446 where glibc maintainers put uc_mcontext today
1447 * fixed bug 282: compiler does not trust type assertions while passing
1448 arguments to a full call.
1449 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1450 &OPTIONAL VALUES type parameter.
1451 * fix bug 214: algorithm for noting rejected templates is now more
1452 similar to that of template seletion. (also reported by rydis on
1454 * fixed bug 141b: printing backquoted information readably and prettily
1455 inserts a space where necessary.
1456 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1457 (CHARACTER 10) are now reported as errors, rather than propagated
1458 as unknown types. (reported by piso on #lisp)
1459 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1460 with values NIL and :ERROR. (thanks to Milan Zamazal)
1461 * fixed bug 191c: CLOS now does proper keyword argument checking as
1462 described in CLHS 7.6.5 and 7.6.5.1.
1463 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1464 longer raise an error; further, using a list as a for-as-arithmetic
1465 counter now raises a meaningful error.
1466 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1467 now understood by sequence creation functions such as MAKE-SEQUENCE
1469 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1470 a STRING-INPUT-STREAM.
1471 * compiler enhancement: SIGNUM is now better able to derive the type
1473 * type declarations inside WITH-SLOTS are checked. (reported by
1475 * fixed some bugs revealed by Paul Dietz' test suite:
1476 ** incorrect optimization of TRUNCATE for a positive first
1477 argument and negative second.
1478 ** compiler failure in let-convertion during flushing dead code.
1479 ** compiler failure while deriving type of TRUNCATE on an
1480 interval, containing 0.
1481 ** ASH of a negative bignum by a negative bignum count now returns
1483 ** intersection of CONS types now canonicalizes properly, fixing
1484 inconsistencies in SUBTYPEP.
1486 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1487 * incompatible change: The --disable-debugger command line
1488 option now clobbers the debugger at a more fundamental
1489 level, by redefining #'INVOKE-DEBUGGER instead of by
1490 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1491 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1492 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1493 enter the debugger REPL and then suffer recursive errors
1494 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1495 while under the new system, BREAK in --disable-debugger mode
1496 terminates the system just as an unhandled error would.
1497 * fixed compiler performance when processing loops with a step >1;
1498 * bug fix: DOCUMENTATION now retrieves generic function
1499 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1500 support has been systematized, and now supports the methods
1501 specified by ANSI, along with a default method and a method for
1502 slot documentation. (reported by Nathan Froyd)
1503 * bug fix: effective methods associated with a generic function are
1504 no longer cached over a change of that generic function's method
1505 combination. (reported by Andreas Fuchs)
1506 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1507 names a non-existent directory. (thanks to Andreas Fuchs)
1508 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1509 the PPC platform, lead to stack corruption; this has been fixed.
1510 (reported by Rainer Joswig)
1511 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1512 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1513 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1514 arithmetic in general, is now much more reliable on the PPC
1516 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1517 platform now returns the right answer.
1518 * optimization: restored some effective method precomputation in
1519 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1520 precomputation is now tunable.
1521 * optimization: compiler-internal data structure use has been
1522 reviewed, and changes have been made that should improve the
1523 performance of the compiler by about 20%.
1524 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1525 simple-base-strings and simple-bit-vectors is improved.
1526 * optimization: the optimization of 32-bit logical and arithmetic
1527 functions introduced in version 0.8.3 on the x86 has been
1528 implemented on the mips, ppc and sparc platforms; an
1529 implementation of the same facility, but for 64-bit arithmetic,
1530 has been added for the alpha.
1531 * microoptimization: the compiler is better able to make use of the
1532 x86 LEA instruction for multiplication by constants.
1533 * bug fix: in some situations compiler did not report usage of
1534 generic arithmetic in (SPEED 3) policy.
1535 * bug 145b fix: compiler used wrong type specifier while converting
1536 MEMBER-types to numeric.
1537 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1539 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1540 index argument now works properly on the Alpha platform.
1541 * bug fix: floating point exception treatment on the Alpha platform
1543 * bug fix: FILE-POSITION works much better on string input and
1544 output streams. (thanks to Nikodemus Siivola)
1545 * bug fix: many threading/garbage collection symptoms sorted.
1546 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1547 might be pseudo-atomic.
1548 * internal change: Stopping for GC is now done with signals not
1549 ptrace. GC is now done in whichever thread wanted it, instead of
1551 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1553 * bug fix: result form in DO is not contained in the implicit
1555 * incompatible change: ICR structure is changed; the value part of
1556 CONTINUATION is now called LVAR; corresponding functions are
1557 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1558 * added type deriver for ISQRT (thanks to Robert E. Brown).
1559 * bug fix: better support for loading from the command line when an
1560 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1562 * fixed some bugs revealed by Paul Dietz' test suite:
1563 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1564 ** ROUND and FROUND now give the right answer when given very
1565 small float arguments.
1566 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1568 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1569 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1570 ** LCM should return a non-negative integer.
1571 ** PARSE-INTEGER returned the index of a terminator instead of the
1572 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1573 ** PARSE-INTEGER returned an incorrect index being applied to a
1575 ** LCM with two arguments of 0 returns 0 rather than signalling
1577 ** unsigned addition of a 32-bit constant with the high bit set no
1578 longer causes an internal compiler error on the x86.
1579 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1580 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1581 with negative last argument.
1582 ** byte specifiers with zero size and position no longer cause
1583 an error during type derivation.
1584 ** bignum multiplication on the Alpha platform now returns the
1586 * porting: The system now builds on SuSE AMD64, although it still
1587 generates a 32-bit binary.
1588 * .fasl file incompatibility: The fasl file version number has
1589 been incremented (because of the changes to internal compiler
1590 data structures referred to above).
1592 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1593 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1594 more accurately, on the Darwin kernel running on PowerPC hardware.
1595 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1596 * Compiler code deletion notes now signal a condition of type
1597 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1598 an associated MUFFLE-WARNING restart.
1599 * The compiler now performs limited argument count validation of
1600 constant format strings in FORMAT, and where appropriate in ERROR,
1601 CERROR and WARN. (thanks to Gerd Moellmann)
1602 * New ASDF-INSTALL contrib can be used for automatic download and
1603 installation of third-party Lisp code from CCLAN or other sites
1605 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1606 which forces another thread to execute a function supplied by the
1608 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1609 now accept and act upon their :ELEMENT-TYPE keyword argument.
1610 (reported by Martin Atzmueller, Edi Weitz)
1611 * bug fix: FILE-POSITION now accepts position designators up to
1612 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1613 is the greater. (thanks to Patrik Nordebo)
1614 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1615 Kalle Olavi Niemitalo)
1616 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1617 printable. (reported by Eric Marsden)
1618 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1619 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1620 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1621 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1622 (thanks to Stig Sandoe)
1623 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1624 UNREAD-CHAR now correctly includes the unread character in the
1625 target sequence. (thanks to Gerd Moellmann)
1626 * bug fix in threaded builds: the system can now be suspended and
1627 resumed by shell job control with minimal disruption.
1628 * bug fixes in times and timezones >2038AD
1629 * better handling of "where is GNU make?" problem in build scripts
1630 (thanks to Nikodemus Siivola)
1631 * new optimization: inside a named function any reference to a
1632 function with the same name is considered to be a self-reference;
1633 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1634 optimization quality.
1635 * new optimization on x86: logical functions and + now have
1636 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1637 used when the result is truncated to 32 bits.
1638 * VALUES declaration is partially enabled.
1639 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1640 * fixed some bugs revealed by Paul Dietz' test suite:
1641 ** The system now obeys the constraint imposed by
1642 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1643 types form a lattice under type intersection.
1644 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1645 ** ASSOC now ignores NIL elements in an alist.
1646 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1647 and (1+ MOST-POSITIVE-FIXNUM) answers.
1648 ** The addition of a method with invalid qualifiers to a generic
1649 function does not cause an error to be signalled immediately;
1650 a warning is signalled, and the error is generated only on
1651 calling the generic function.
1652 * changed .fasl file version number, in part to add type codes for
1653 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1654 obscure ANSI requirements
1656 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1657 * fixed bug 148: failure to inline-expand a local function left
1658 garbage, confusing the compiler.
1659 * fixed bugs 3cd: structure slot readers perform type check if the
1660 slot can have an invalid value (i.e. it is either not initialized
1661 or can be written with a less specific slot writer).
1662 * bug fix: the compiler now traps array references to elements off
1663 the end of an array; previously, the bounds checking in some
1664 circumstances could go off-by-one.
1665 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1667 * type declarations for array element types now obey the description
1668 on the CLHS page "Declaration TYPE", as per discussions on
1669 sbcl-help around 2003-05-08. This means that a declaration
1670 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1671 declaration, all references to BAR will be asserted or assumed
1672 (with THE, so dependent on compiler policy) to involve objects of
1673 type FOO. Note that no such declaration is implied in
1674 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1675 * declared types of functions from the "Conditions"
1676 chapter. (reported by Paul Dietz)
1677 * bug fix: CERROR accepts a function as its first argument.
1678 * bug fix: NTH an NTHCDR accept a bignum as index
1679 arguments. (reported by Adam Warner)
1680 * optimization: character compare routines now optimize comparing
1681 against a constant character. (reported by Gilbert Baumann)
1682 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1683 argument now works properly on the MIPS platform.
1684 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1685 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1686 declared to have a functional type, does not cause an error
1688 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1689 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1690 * fixed bug: if last continuation of a deleted block has a
1691 destination, this destination should be deleted too. (reported by
1693 * fixed a bug in the bootstrap process: the host compiler's values
1694 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1695 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1696 test case from Patrik Nordebo)
1697 * improved the ability of the disassembler on the PPC platform to
1698 provide helpful disassembly notes.
1699 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1700 the class in more cases than previously.
1701 * bug fix: FILE-POSITION now understands :START and :END for
1702 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1703 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1704 8) data. (thanks to David Lichteblau)
1705 * bug fix: it is possible to add a method to a generic function
1706 without lambda list.
1707 * bug fix: reader failed to signal END-OF-FILE inside an
1708 object representation. (reported by Nikodemus Siivola)
1709 * fixed some bugs revealed by Paul Dietz' test suite:
1710 ** LAST and [N]BUTLAST should accept a bignum.
1711 ** condition slot accessors are methods.
1712 ** (VECTOR NIL) is a subtype of STRING.
1714 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1715 * minor incompatible change: some nonsensical specialized lambda
1716 lists (used in DEFMETHOD) which were previously ignored now signal
1718 * minor incompatible change: the system is now aware of the types of
1719 variables in the COMMON-LISP package, and will signal errors for
1720 most violations of these type constraints (where previously they
1721 were silently accepted).
1722 * minor incompatible change: COMPILE-FILE now uses the freedom
1723 afforded (ANSI 3.2.2.3) to use derived function types for
1724 functions defined in the same file. This also permits the system
1725 to warn on static type mismatches and function
1726 redefinition. (Currently it does not work with high DEBUG level.)
1727 * minor incompatible change: VALUES declaration is disabled.
1728 * When issuing notes, the compiler now signals a condition of type
1729 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1730 restart for use in user handlers. It is expected that the
1731 COMPILER-NOTE condition will eventually become a condition
1732 supertype to a hierarchy of note types, which will then be
1733 handleable in a similar fashion. However, at the moment, no such
1734 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1735 but it's an implementation detail, not a classification for the
1737 * Changes in type checking closed the following bugs:
1738 ** type checking of unused values (192b, 194d, 203);
1739 ** template selection based on unsafe type assertions (192c, 236);
1740 ** type checking in branches (194bc).
1741 * A short form of VALUES type specifier has ANSI meaning (it has
1742 increased the number of situations when SBCL cannot perform type
1744 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1745 keywords or constants is permissible.
1746 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1747 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1748 classes. (thanks to Antonio Martinez)
1749 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1750 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1751 argument to a pretty-print stream. Instead, it leaves any such
1752 operations to DESCRIBE-OBJECT methods.
1753 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1755 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1756 lambda lists. (thanks to David Lichteblau)
1757 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1758 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1759 (thanks to Gerd Moellmann)
1760 * fixed bug 246: increased compilation speed of long
1761 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1763 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1765 * DEFCONSTANT now throws a condition of type
1766 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1767 constant to a non-EQL value; CONTINUE and ABORT restarts
1768 respectively change and preserve the value.
1769 * fixed bug 63: The code walker, part of the implementation of CLOS,
1770 is now better at handling symbol macros.
1771 * bug fix: There is no longer an internal implementation type named
1772 CL:LENGTH. (reported by Raymond Toy)
1773 * bug fix: In macro-like defining macros/special operators the
1774 implicit block does not enclose the lambda list.
1775 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1776 enclosed in parentheses) are not suitable as type specifiers, and
1777 their use properly signals an error now.
1778 * bug fix: An argument count mismatch for a type specifier in code
1779 being compiled no longer causes an unhandled error at compile
1780 time, but signals a compile-time warning.
1781 * fixed simple vector readable printing
1782 * bug fix: DESCRIBE takes more care over whether the class
1783 precedence list slot of a class is bound before accessing it.
1784 (reported by Markus Krummenacker)
1785 * bug fix: FORMATTER can successfully compile pretty-printer format
1786 strings which use variants of the ~* directive inside.
1787 * bug fix: SEARCH now applies its TEST predicate to the elements of
1788 the arguments in the correct order. (thanks to Wolfhard Buss)
1789 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1790 in type inference. (thanks to Robert E. Brown)
1791 * bug fix: Reading in symbols with an explicit package name of ""
1792 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1793 * fixed some bugs revealed by Paul Dietz' test suite:
1794 ** NIL is now allowed as a structure slot name.
1795 ** Arbitrary numbers, not just REALs, are allowed in certain
1796 circumstances in LOOP for-as-arithmetic clauses.
1797 ** Multiple class redefinitions before slot access no longer
1798 causes a type error.
1799 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1800 association between the name and a class.
1801 ** Generic functions with non-standard method-combination and over
1802 five methods all of which return constants no longer return NIL
1803 after the first few invocations. (thanks to Gerd Moellmann)
1804 ** CALL-NEXT-METHOD with no arguments now passes the original
1805 values of the arguments, even in the presence of assignment.
1806 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1808 ** DISASSEMBLE works with closures and funcallable instances.
1809 ** ADD-METHOD now returns the generic function, not the new method.
1810 ** FIND-METHOD signals an error if the lengths of the specializers
1811 is incompatible with the generic function, even if the ERRORP
1813 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1814 which its argument is a member.
1815 ** DEFCLASS only redefines the class named by its class-name
1816 argument if that name is the proper name of the class;
1817 otherwise, it creates a new class.
1818 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1819 of the UNBOUND-SLOT condition to the name of the slot.
1820 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1821 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1822 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1823 SLOT-MAKUNBOUND in the specified fashion.
1825 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1826 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1827 cross-compilation host. As a consequence, we can now bootstrap our
1828 way up to SBCL starting with a bare gcc toolchain and human-readable
1829 source code (first the source to CLISP, then the source to SBCL).
1830 * A contributed module containing a partial implementation of the
1831 simple-streams interface has been included. (thanks to Rudi
1833 * A contributed module implementing the RFC1321 Message Digest
1834 Algorithm, known as MD5, has been included.
1835 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1836 no longer has any effect, as the code controlled by this feature
1837 has been deleted. (As far as we know, no-one has ever built using
1838 this feature, and its semantics were confused in any case).
1839 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1840 work on conditions (as required by the ANSI specification),
1841 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1842 expected behaviour on conditions. Users should note, however,
1843 that such behaviour is not required by the ANSI specification,
1844 and so use of this behaviour may render their code unportable.
1845 * fixed some bugs revealed by Paul Dietz' test suite:
1846 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1848 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1849 ** SXHASH on condition objects no longer returns NIL.
1850 ** :ALLOCATION :CLASS slots are better treated; their values are
1851 updated on class redefinition, and initforms inherited from
1852 superclasses are applied.
1853 ** REMOVE-METHOD returns its generic function argument even when
1854 no method was removed.
1855 ** SHARED-INITIALIZE now initializes the values of the requested
1856 slots, including those with :ALLOCATION :CLASS.
1857 ** ALLOCATE-INSTANCE now works on structure classes defined via
1858 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1860 ** SLOT-EXISTS-P now works on conditions, as well as structures
1862 ** MAKE-LOAD-FORM now has the required methods on
1863 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1864 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1865 keyword, and now implements the SLOT-NAMES argument.
1866 ** methods with &OPTIONAL arguments no longer allow too many
1867 arguments to be passed in the call without error.
1868 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1869 option is consistent with the required arguments of the generic
1870 function lambda list.
1871 * bug fix: REQUIRE accepts a string designator. (Thanks to
1873 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1874 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1875 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1876 * bug fix: The compiler now checks for duplicated variables in macro
1878 * bug fix: SETQ on globals returns the correct value.
1879 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1880 not a COMPILER-ERROR (followed by some other strange error on
1881 choosing the CONTINUE restart).
1882 * bug fix: make.sh and friends are now more consistent in the way that
1883 they look for GNU "make".
1885 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1886 * experimental native threads support (on x86 Linux >=2.4 only).
1887 This is not compiled in by default: you need to add :SB-THREAD to
1888 the target features. See the "Beyond ANSI" chapter of the manual
1890 * fix for longstanding nonANSIism: The old distinction between
1891 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1892 The return value from CL:FIND-CLASS is now a CLOS class, and
1893 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1894 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1895 * An interface to the MetaObject Protocol, as described in Kiczales,
1896 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1897 MIT Press, 1991, is available from the SB-MOP package.
1898 * incompatible change: the SB-PCL package should now be considered
1899 a private implementation detail, and no longer a semi-private MOP
1901 * minor incompatible change: due to rearrangement for threads, the
1902 control stack and binding stack are now allocated at arbitrary
1903 addresses instead of being hardcoded per-port. Users affected by
1904 this probably have to be doing advanced things with shared
1905 libraries, and will know who they are.
1906 * minor incompatible change: Previously, all --eval forms used were
1907 processed with READ before any of them were processed with EVAL.
1908 Now each --eval form is processed with both READ and EVAL before
1909 the next --eval form is processed. (Thus package operations like
1910 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1911 work as the user might reasonably expect.)
1912 * minor incompatible change: *STANDARD-INPUT* is now only an
1913 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1915 * minor incompatible change: Y-OR-N-P is now character-oriented,
1916 not line oriented. Also, YES-OR-NO-P now works without errors.
1917 (thanks to Antonio Martinez)
1918 * sb-aclrepl module improvements: an integrated inspector, added
1919 repl features, and a bug fix to :trace command.
1920 * Known functions, which cannot be open coded by the backend, are
1921 considered to be able to check types of their arguments. (fixing
1922 a bug report by Nathan J. Froyd)
1923 * fixed a bug in computing method discriminating functions: It is
1924 now possible to define methods specialized on classes which have
1925 forward-referenced superclasses. (thanks to Gerd Moellmann)
1926 * fixed evaluation order in optional entries (reported by Gilbert
1928 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1929 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1930 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1931 required-by-AMOP NAME argument, as well as CLASS and
1932 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1933 * fixed bug 20: DEFMETHOD can define methods using names that are
1934 not the proper names of classes to designate class specializers.
1935 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1936 Unix file descriptors, instead of blowing up. (thanks to Antonio
1938 * Garbage collection refactoring: user-visible change is that a
1939 call to the GC function during WITHOUT-GCING will not do garbage
1940 collection until the end of the WITHOUT-GCING. If you were doing
1941 this you were probably losing anyway.
1942 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1943 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1944 TYPEP the latter but not the former.
1945 * The compiler issues a full WARNING for calls to undefined functions
1946 with names from the CL package.
1947 * MAP-INTO for a vector destination is open coded. (reported by
1948 Brian Downing on c.l.l)
1949 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1950 documentation string.
1951 * fixed some bugs revealed by Paul Dietz' test suite:
1952 ** COPY-ALIST now signals an error if its argument is a dotted
1954 ** Condition slots are now accessed more correctly in the presence
1955 of multiple initargs for a given slot.
1956 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1957 exclude restarts of the same name associated with a different
1959 ** DEFCLASS of forward-referenced classes with another
1960 forward-referenced class in the superclasses list no longer
1962 ** Condition slots are now initialized once each, not multiple
1963 times. (thanks to Gerd Moellmann)
1964 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1966 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1967 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1968 arguments contain duplicated elements.
1969 ** RESTART-CASE understands local macros.
1970 ** RESTART-CASE associates exactly its own restarts with a condition.
1971 ** ENDP in safe mode checks its argument to be of type LIST.
1972 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1973 in question is unbound.
1974 ** Optimized MAKE-INSTANCE functions no longer cause internal
1975 assertion failures in the presence of duplicate initargs.
1976 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1977 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1978 propagated even in the ctor optimized implementation of
1980 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1982 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1983 * a better implementation of SXHASH on (simple) bit vectors,
1984 measured both in execution speed and in distribution of results
1985 over the positive fixnums, has been installed. Likewise, a better
1986 implementation of EQUAL for simple bit vectors is now available.
1987 * fixed CEILING optimization for a divisor of form 2^k.
1988 * fixed bug 240 (emitting extra style warnings "using the lexical
1989 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1990 by Antonio Martinez)
1991 * fixed SXHASH, giving different results for NIL depending on type
1992 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1993 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1994 variable. (found by Rolf Wester)
1995 * SBCL does not ignore type declarations for special
1996 variables. (reported by rif on c.l.l 2003-03-05)
1997 * some bug fixes in contrib/sb-aclrepl/
1998 * fixed some bugs revealed by Paul Dietz' test suite:
1999 ** a bug in the CONS type specifier, whereby the CAR and CDR
2000 types got intertwined, has been fixed;
2001 ** the type system is now able to reason about the interaction
2002 between INTEGER and RATIO types more completely;
2003 ** APPEND, [N]REVERSE and NRECONC check that those their
2004 arguments, which must be proper lists, are really so;
2005 ** An array specialized to be unable to hold elements has been
2006 implemented, as required -- yes, really -- by ANSI;
2007 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2008 on malformed property lists;
2010 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2011 * incompatible packaging change: in line with Unix convention,
2012 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2013 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2014 * REQUIRE and PROVIDE are now optionally capable of doing something
2015 useful. See the documentation string for REQUIRE.
2016 * infrastructure for a managed SBCL contrib system: contributed
2017 modules in this release include:
2018 ** the ASDF system definition facility;
2019 ** an interface to the BSD Sockets API;
2020 ** an ACL-like convenience interface to the repl;
2021 (thanks to Kevin Rosenberg)
2022 ** an implementation of ROTATE-BYTE, with efficient implementation
2024 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2025 gives a closer approximation to the right answer than previously.
2026 (thanks to Raymond Toy)
2027 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2028 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2029 optional environment arguments, as required by ANSI.
2030 * fixed bugs in other functions taking environment objects, allowing
2031 calls with an explicit NIL environment argument to be compiled
2033 * fixed bug 228: primary return values from
2034 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2035 COMPILE or FUNCTION.
2036 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2037 :TYPE will no longer signal errors on innocuous objects.
2038 * fixed bug 231b: SETQ is better at respecting type declarations in
2039 the lexical environment.
2040 * fixed a bug in DEFCLASS: classes named by symbols with no or
2041 unprintable packages can now be defined.
2042 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2043 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2044 * fixed compiler failure related to checking types of functions.
2045 (reported by Robert E. Brown)
2046 * the compiler is now much more consistent in its error-checking
2047 treatment of bounding index arguments to sequence functions: in
2048 (SAFETY 3) code, errors will be signalled in almost all cases if
2049 invalid sequence bounding indices are passed to functions defined
2050 by ANSI to operate on sequences.
2051 * fixed a bug in the build procedure: documentation of SBCL-specific
2052 packages is now preserved and available in the final Lisp image.
2053 * lifted FDEFINITION lookup out of loops in the implementation of
2054 many list operations. (thanks to Robert E. Brown)
2055 * fixed a bug in the reader: the #n# reader macro now works for
2056 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2057 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2058 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2059 E. Brown and Tony Martinez respectively)
2060 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2061 SLOT-BOUNDP: in particular, it is now able to optimize them much
2062 better, and is now not vulnerable to having packages renamed.
2063 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2064 always called when a slot is not present in an instance. (thanks
2066 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2067 longer signal an error on unreading a character following EOF on
2068 the previous constituent stream. (thanks to Tony Martinez)
2069 * fixed some bugs revealed by Paul Dietz' test suite:
2070 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2071 not just nonnegative fixnums;
2072 ** the logical bit-array operators such as BIT-AND now accept an
2073 explicit NIL for their "opt-arg" argument (to indicate a
2074 freshly-consed result bit-array);
2075 ** ELT now signals an error on an invalid sequence index in safe
2077 ** the type system is now cleverer about negations of numeric
2078 types, and consequently understands the BIGNUM and RATIO types
2080 ** the type system is now cleverer about the interaction between
2081 INTEGER and RATIO types: while bugs still remain, many more
2082 cases are accurately computed;
2083 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2084 if it is in the last clause;
2085 ** CONSTANTLY now correctly returns a side-effect-free function in
2087 ** DECLARE is no longer treated as a special-operator; in
2088 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2089 * incremented fasl file version number due to the change in the
2090 DEFSTRUCT-SLOT-DESCRIPTION structure.
2092 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2093 * minor incompatible change: code processed by the "interpreter" or
2094 EVAL now has a compilation optimization policy of (DEBUG 2)
2095 (changed from (DEBUG 1)) to improve debuggability of interactive
2096 development, and to allow the use of the debug RETURN command in
2098 * an experimental implementation of the RETURN command for the
2099 debugger has been included. (thanks to Frederik Kuivinen)
2100 * fixed bug 62: constraints were not propagated into a loop.
2101 * fixed bug in embedded calls of SORT (reported and investigated by
2103 * fixed some bugs revealed by Paul Dietz' test suite:
2104 ** printing and reading of arrays with some dimensions having
2105 length 0 (thanks to Gerd Moellmann);
2106 ** BOA constructor with &AUX argument without a default value does
2107 not cause a type error;
2108 ** CONSTANTP now returns true for all self-evaluating objects.
2110 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2111 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2112 accessors that are related by inheritance, as specified in the
2113 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2114 Valtteri Vuorikoski)
2115 * The compiler is now able to inline functions that were defined in
2116 a complex lexical environment (e.g. inside a MACROLET).
2117 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2119 * Support for the upcoming FreeBSD-5.0 release has been included.
2120 (thanks to Dag-Erling Smorgrav)
2121 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2122 effect when it is not in a toplevel context.
2123 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2124 stand a better chance of being correct. (thanks to Gerd
2126 * fixed bug in COERCE, which now signals an error on coercing a
2127 rational to a bounded real type which excludes the expected
2129 * The compiler is now able to derive types more accurately from the
2130 COERCE and COMPILE functions.
2131 * fixed bug 223: functional binding is considered to be constant
2132 only for symbols in the CL package.
2133 * fixed bug 231: SETQ did not check the type of a variable being set
2134 (reported by Robert E. Brown)
2135 * A new optimization for MAKE-INSTANCE has been included, fixing
2136 various bugs (including relating to :ALLOCATION :CLASS slots and
2137 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2139 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2140 ** As required by ANSI, LOOP now disallows anonymous collection
2141 clauses such as COLLECT I in conjunction with aggregate boolean
2142 clauses such as THEREIS (= I 1);
2143 ** LOOP now signals an error when any variable is reused in the
2144 same loop (including the potentially useful construct analogous
2145 to WITH A = 1 WITH A = (1+ A);
2146 ** IT is only a special loop symbol within the first clause of a
2147 conditional loop clause;
2148 ** LOOP with a typed iteration variable over a hashtable now
2149 signals a type error iff it should.
2150 * fixed some other bugs revealed by Paul Dietz' test suite:
2151 ** FILE-STREAM now names the class previously known as FD-STREAM;
2152 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2153 argument) no longer signals an error;
2154 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2155 of the slot symbol, rather than using the current package
2156 ((:CONC-NAME "") continues to intern the slot's name in the
2158 * incremented fasl file version number, because of the incompatible
2159 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2160 of the new implementation of DEFINE-COMPILER-MACRO.
2162 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2163 * Support for building SBCL for MIPS platforms running in
2164 little-endian mode has now been checked in, and basic
2165 functionality on said platforms verified.
2166 * minor incompatible change: PCL now records the pathname of a file
2167 in which methods and the like are defined, rather than its
2169 * minor incompatible change: TRUENAME now considers the truename of
2170 a file naming a directory to be the pathname with :DIRECTORY
2171 component indicating that directory.
2172 * minor incompatible change: a NAMED clause in the extended form of
2173 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2174 reason for the previous behaviour is unclear.
2175 * more systematization and improvement of CLOS and MOP conformance
2176 in PCL (thanks to Gerd Moellman and Pierre Mai):
2177 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2179 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2180 primary methods with no specializers;
2181 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2183 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2184 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2185 FORWARD-REFERENCED-CLASSes; error reporting on
2186 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2188 ** SXHASH on CLOS instances now uses a slot internal to the
2189 instance to return different numbers on distinct instances,
2190 while preserving the same return value through invocations of
2192 ** DEFMETHOD signals errors when methods with longer incongruent
2193 lambda lists are added to generic functions;
2194 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2195 CLASS, as specified in AMOP;
2196 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2197 on the order returned by the primary method for classes of
2198 class STANDARD-CLASS;
2199 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2200 * fixed some bugs shown by Paul Dietz' test suite:
2201 ** DOLIST puts its body in TAGBODY;
2202 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2204 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2205 value producing form;
2206 ** if more variables are given to PROGV than values, extra
2207 variables are bound and made to have no value;
2208 ** NSUBSTITUTE on list arguments gets the right answer with
2210 ** ELT signals an error of type TYPE-ERROR when the index argument
2211 is not a valid sequence index;
2212 ** LOOP signals (at macroexpansion time) an error of type
2213 PROGRAM-ERROR when duplicate variable names are found;
2214 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2215 ** FUNCALL of special-operators now cause an error of type
2217 ** PSETQ now works as required in the presence of side-effecting
2218 symbol-macro places;
2219 ** NCONC accepts any object as its last argument;
2220 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2222 ** loop-for-as-package does not require a package to be explicitely
2224 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2226 * fixed bug 166: compiler preserves "there is a way to go"
2227 invariant when deleting code.
2228 * fixed bug 172: macro lambda lists with required arguments after
2229 &REST arguments now cause an error to be signalled. (thanks to
2231 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2232 bug. (thanks to Matthew Danish)
2233 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2235 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2236 arguments are lexically rebound. (thanks to Gerd Moellmann and
2238 * fixed bug 194: error messages are now more informative when there
2239 is no primary method applicable in a call to a generic
2240 function. (thanks to Gerd Moellmann)
2241 * fixed bug in command line argument checking (thanks to Julian
2243 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2244 (thanks to Matthew Danish)
2245 * incremented fasl file version number, because of the
2246 SXHASH-related changes in the layout of CLOS data structures
2248 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2249 * minor incompatible change: The runtime (the Unix executable named
2250 "sbcl") is now much pickier about the .core files it will load.
2251 Essentially it now requires .core files to descend from the same
2252 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2253 as the runtime does. (The intent is to prevent the crashes which
2254 can occur, and which can even be reported as mysterious failures,
2255 when people patch the sources or change the build parameters
2256 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2257 sbcl and .core files.)
2258 * fixed bug: VALUES-LIST is no longer optimized away.
2259 * fixed bug 142: The FFI conversion of C string values to Lisp
2260 string values no longer conses excessively. (thanks to Nathan
2261 Froyd porting Raymond Toy's fix to CMU CL)
2262 * began to systematize and improve MOP conformance in PCL (thanks to
2263 Nathan Froyd, Gerd Moellman and Pierre Mai):
2264 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2266 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2267 ** FINALIZE-INHERITANCE is now called on class finalization;
2268 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2269 argument precedence order.
2270 * fixed bug 202: The compiler no longer fails on functions whose
2271 derived types contradict their declared type.
2272 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2273 so it can be non-toplevel.
2274 * The fasl file version number has changed (because of the new
2275 implementation of DEFMACRO).
2276 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2277 safe code, that any length requirement by their type-specifier
2278 argument is valid. The exceptions to this are described in bug
2280 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2281 functions now check the types of their inputs as required by ANSI.
2282 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2283 introduced symbol is DECLAREd to be SPECIAL.
2284 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2285 * fixed bug: :COUNT argument to sequence functions may be negative
2286 * fixed bug: body of DO-SYMBOLS may contain declarations
2287 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2288 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2289 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2290 symbol macro only once
2291 * fixed printing of call frame when argument list is unavailable
2292 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2293 * compiler no longer signals WARNING on unknown keyword
2296 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2297 * A beta-quality port to the mips architecture running Linux,
2298 based on the old CMUCL backend, has been made. It has been tested
2299 on a big-endian kernel, and works sufficiently well to be able to
2300 rebuild itself; it has not been tested in little-endian mode.
2301 * fixed an inconsistency between gencgc.c and purify.c which made
2302 dumping/loading .core files unreliable
2303 * fixed bug 120a: The compiler now deals correctly with IFs where
2304 the consequent is the same as the alternative, instead of
2305 misderiving the return type. (thanks to Alexey Dejneka)
2306 * fixed bug 113: Logical pathnames are now dumpable (the logical
2307 host is resolved at load-time, throwing an error if it is not
2309 * fixed bug 174: FORMAT's error message is slightly clearer when a
2310 non-printing character is used in a format directive.
2311 * fixed several bugs in compiler checking of type declarations, i.e.
2312 violations of the Python "declarations are assertions" principle
2313 (thanks to Alexey Dejneka)
2314 * fixed several bugs in PCL's error checking (thanks to Gerd
2316 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2318 * fixed bug in compilation of functions as first class values
2319 (thanks to Antonio Martinez-Shotton)
2320 * The compiler's handling TYPE-ERRORs which it can prove will
2321 inevitably happen at runtime has been cleaned up and corrected
2322 in several ways. (thanks to Alexey Dejneka)
2323 * improved argument type checking for various basic arithmetic
2324 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2325 lavished on them in the past that they can be compiled in many
2326 ways in different special cases
2327 * fixed bug 181: compiler checks validity of user supplied type
2329 * cleaned up code flushing in optimization: Function calls which
2330 should signal errors for safety purposes (e.g. which ANSI says
2331 should signal errors when their arguments are of incorrect type)
2332 are no longer optimized away.
2333 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2334 * incremented fasl file version number, because changes in the
2335 implementation of sequence functions like COERCE caused
2336 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2337 in old inline expansions) to become undefined. (Actually these
2338 changes were later undone, so we might very well be binary
2339 compatible with 0.7.7 after all, but leaving the version number
2340 incremented seemed like the simplest and most conservative
2343 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2344 * An alpha-quality port to the parisc architecture running Linux,
2345 based on the old CMUCL backend, has been made. This, even more so
2346 than the other backends, should be considered still a work in
2347 progress; known problems include that the Linux kernel in 64-bit
2348 mode does not propagate the correct sigcontext structure to
2349 userspace, and consequently SBCL on a parisc64 kernel will not
2351 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2352 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2353 suppressed.) Also now that the compiler is looking at declarations
2354 in the environment, it checks optimization declarations as well,
2355 and suppresses inlining when (> DEBUG SPEED).
2356 * More fixes have been made to treatment of floating point exception
2357 treatment and other Unix signals. In particular, floating point
2358 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2359 * The detection and handling of control stack exhaustion (infinite
2360 or very deeply nested recursion) has changed. Stack exhaustion
2361 detection is now done by write-protecting pages at the OS level
2362 and applies at all optimization settings; when found, a
2363 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2364 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2365 be caught using IGNORE-ERRORS.
2366 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2367 correctly when transforming recursive LABELS functions to LETs.
2368 (thanks to Alexey Dejneka)
2369 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2370 that are names of constants or global variables.
2371 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2372 alien routines with docstrings.
2373 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2374 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2376 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2377 (thanks to Eric Marsden)
2378 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2379 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2380 * Bug fix: The SPARC backend can now compile functions involving
2381 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2382 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2383 to a functions expecting a PATHNAME-DESIGNATOR.
2384 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2385 lambda lists. (thanks to Alexey Dejneka)
2386 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2387 functions so that EXPORTing the name of the class doesn't cause
2388 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2389 to Antonio Martinez for reporting this)
2390 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2391 constant-folded without failing an assertion. (thanks to Einar
2392 Floystad Dorum for reporting this)
2393 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2394 types are now dealt with more correctly by the compiler.
2395 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2396 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2397 behaviour analogous to RENAME-FILE. This puts its behaviour more
2398 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2400 * The fasl file version number has changed again. (because of the
2401 bug fix involving the names of PCL MAKE-INSTANCE functions)
2403 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2404 * bug fix: Floating point exceptions are treated much more
2405 consistently on the x86/Linux and PPC/Linux platforms.
2406 * Array initialization with :INITIAL-ELEMENT is now much faster for
2407 cases when the compiler cannot open code the array creation, but
2408 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2409 array accesses have also seen a speed increase.
2410 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2411 specified. (This was at the root of some bad interactions between
2412 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2414 * bug fix: Internal error arguments for undefined functions are now
2415 computed correctly on the PPC/Linux platform.
2416 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2417 Raymond Toy's patch for CMU CL)
2418 * Support for the Solaris 9 operating environment has been included
2419 (thanks to Daniel Merritt)
2420 * A very ugly but hopefully complete draft of the missing FFI chapter
2421 of the manual has been created by reformatting the corresponding
2422 CMU CL manual chapter into (currently very ugly and incoherent)
2423 DocBook and bringing it up to date for SBCL behavior. Thus, the
2424 manual is now essentially complete, at least by my extreme
2425 once-and-only-once standards, whereby it's acceptable to refer to
2426 the doc strings of SB-EXT functions as the primary documentation.
2427 * The fasl file version number has changed again, due to cleanup of
2428 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2429 is no longer a static symbol.)
2431 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2432 * SBCL now builds with OpenMCL (version 0.12) as the
2433 cross-compilation host; also, more progress has been made toward
2434 bootstrapping under CLISP.
2435 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2437 * bug 158 fixed: The compiler can now deal with integer loop
2438 increments different from 1; fixing this turned out also to fix
2440 * bug 169 fixed: no more bogus warnings about using lexical bindings
2441 despite the presence of perfectly good SPECIAL declarations (thanks
2442 to David Lichteblau)
2443 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2444 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2445 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2447 * bug 180 fixed: Method combination specifications no longer ignore
2448 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2449 * bug fix: Structure type predicate functions now check their argument
2450 count as they should.
2451 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2452 correctly. (thanks to Pierre Mai)
2453 * minor incompatible change: The --noprogrammer option is deprecated
2454 in favor of the new --disable-debugger option, which is very similar.
2455 (The major difference is that it takes effect at a slightly different
2456 time at startup, causing handling of errors in --sysinit and
2457 --userinit files will be affected differently.) The
2458 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2459 been added to allow this functionality to be controlled from ordinary
2460 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2461 maintainers, who might want to run non-interactive scripts to
2462 build SBCL cores which will later be used interactively.)
2463 * minor incompatible change: The LOAD function no longer, when given
2464 a wild pathname to load, loads all files matching that pathname.
2465 Instead, an error of type FILE-ERROR is signalled.
2467 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2468 * bug 147 fixed: The compiler preserves its block link/count
2469 invariants more correctly now so that it doesn't crash. (thanks
2471 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2473 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2474 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2475 no-value-for-this-character as NIL (as specified by ANSI).
2476 * HOST-NAMESTRING on physical pathnames now returns a string that is
2477 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2478 (thanks to Christophe Rhodes)
2479 * The Alpha port handles icache flushing more correctly. (thanks to
2481 * More progress has been made toward bootstrapping under CLISP. (thanks
2482 to Christophe Rhodes)
2483 * The fasl file format has changed again, because dynamic loading
2484 on OpenBSD (which has non-ELF object files) motivated some cleanups
2485 in the way that foreign symbols are transformed and passed around.
2486 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2487 is no longer treated as whitespace by the reader, but instead as
2488 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2489 |A
\7fB|, instead of A as it used to.
2491 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2492 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2493 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2494 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2495 this as of 0.7.1.45, but was left out of the previous news section
2496 (thanks to Dan Barlow)
2497 * SBCL now runs on the Solaris operating system on SPARC architectures
2498 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2499 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2500 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2501 * SPARC backend cleanups, allowing builds of cores optimized for V8
2502 and V9 SPARCS, and also emission of code targeted to a particular
2503 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2505 * SBCL is closer to bootstrapping under CLISP, thanks to various
2506 fixes by Christophe Rhodes.
2507 * The fasl file format has changed again, to allow the compiler's
2508 INFO database to support symbol macros.
2509 * The user manual (in doc/) is formatted into HTML more nicely.
2510 (thanks to coreythomas)
2511 * The system is smarter about SUBTYPEP relationships, especially
2512 those involving NOT types (including types such as ATOM which are
2513 represented internally using NOT types). Thus SUBTYPEP is less
2514 likely to return (VALUES NIL NIL) in general, and in particular
2515 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2517 * The fasl file format has changed again, because the internal
2518 representation of types now includes a new slot to support the new
2519 SUBTYPEP-of-NOT-types logic.
2520 * (not a change in the main branch of SBCL, but a related prototype
2521 which can hopefully be merged into the main branch of SBCL in the
2522 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2523 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2524 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2525 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2526 dispatch macro character combinations. (thanks to Alexey Dejneka)
2527 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2528 that superficially look like logical namestrings correctly.
2529 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2531 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2532 default supplied-p) for &optional and &key arguments. (thanks to
2535 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2536 * incompatible change: The compiler is now less aggressive about
2537 tail call optimization, doing it only when (> SPACE DEBUG) or
2538 (> SPEED DEBUG). (This is an incompatible change because there are
2539 programs which relied on the old CMU-CL-style behavior to optimize
2540 away their unbounded recursion which will now die of stack overflow.)
2541 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2542 for non-GENCGC systems has been increased to 20M (since that
2543 seems much closer to the likely performance optimum for modern
2544 systems than the old 4M value was)
2545 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2546 and *DEBUG-PRINT-LEVEL*
2547 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2548 of CMU CL's support for SPARC, and various endianness and other
2549 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2550 * new syntactic sugar for the Unix command line: --load foo.bar is now
2551 an alternate notation for --eval '(load "foo.bar")'.
2553 ** The system now detects stack overflow and handles it gracefully,
2554 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2555 optimization settings. (This is a good thing in general, and
2556 its introduction in this version should be particularly timely
2557 for anyone whose code fails because of suppression of tail
2559 ** The system now hunts for the C variable "environ" in a more
2560 devious way, to avoid segfaults when the C library version differs
2561 between compile time and run time. (thanks to Christophe Rhodes)
2562 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2563 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2564 ** The compiler no longer issues bogus style warnings for undefined
2565 classes in the same source file as the DEFCLASSes which defined
2566 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2568 ** fixes in CONDITION class precedence list for undefined function
2569 errors (thanks to Alexei Dejneka)
2570 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2571 correctly. (thanks to Dan Barlow)
2572 ** portability fixes aiming at bootstrapping under CLISP (thanks
2573 to Dave McDonald and Christophe Rhodes)
2574 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2575 ** fixes in type translation and and type inference (thanks to
2577 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2578 ** various fixes in the new ports (thanks to Dan Barlow)
2579 * several changes related to debugging:
2580 ** suppression of tail recursion, as noted above
2581 ** stack overflow detection, as noted above
2582 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2583 is now the default. (For some time encapsulation has been more
2584 reliable than the breakpoint-based :ENCAPSULATE NIL
2585 implementation, at least on X86 systems; and I just noticed that
2586 encapsulation also seems closer to the spirit of the ANSI
2589 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2591 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2592 up properly again. (There was a packaging bug in 0.7.0 which
2593 left their definitions in SB-SYS::LOAD-FOREIGN and
2594 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2595 vital for most things which interface to C-level interfaces,
2596 like extensions working with sockets or databases or
2597 Perl-compatible regexes or whatever, and the need to fix
2598 this bug is the main reason that 0.7.1 was released so
2600 ** DEFGENERIC is now choosier about the methods it redefines, so that
2601 reLOADing a previously-LOADed file containing DEFGENERICs does
2602 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2603 cycle now works as it should. (thanks to Alexey Dejneka)
2604 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2605 fixed by Christophe Rhodes. (He actually submitted this patch
2606 months ago, and I delayed until after 0.7.0.)
2607 ** Bug 111 (internal compiler confusion about runtime checks on
2608 FUNCTION types) was fixed by Alexey Dejneka.
2609 * Some internal cleanups (getting rid of variables which aren't
2610 needed now that the byte interpreter is gone) caused the fasl
2611 file format number to change again.
2613 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2614 * major incompatible change: The default fasl file extension, i.e. the
2615 default extension for files produced by COMPILE-FILE, has changed
2616 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2618 ** There are many changes in the implementation of the compiler.
2619 SBCL is now essentially a compiler-only implementation of ANSI
2620 Common Lisp. EVAL still "interprets" a few special cases, but
2621 almost all the interesting cases are handled by creating
2622 a LAMBDA expression, calling COMPILE on it, then calling
2623 FUNCALL on the result.
2624 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2625 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2626 Since the code is newer, there might still be some new bugs
2627 (though not as many as before Martin Atzmueller's fixes:-). But
2628 the new code is substantially simpler and clearer, and hopefully
2629 any remaining bugs will be simpler, less fundamental, and more
2630 fixable then the bugs in the old code.
2631 ** The revised compiler is still a little unsteady on its feet.
2633 *** The debugging information it produces (particularly the names
2634 of FUNCTION objects) is sometimes much less useful than what
2635 the old compiler produced.
2636 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2637 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2638 within a MACROLET) has been temporarily weakened.
2639 ** There are new compiler optimizations for various functions:
2640 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2641 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2642 *** the math functions TRUNCATE, FLOOR, and CEILING
2643 *** the function-of-all-trades COERCE
2644 Mostly these should be transparent, but there's one
2645 potentially-annoying problem (bug 117): when the compiler
2646 inline-expands a function and does type analysis on the result,
2647 it can create control paths which have type mismatches, and
2648 when it can't prove that those control paths aren't taken,
2649 it will issue WARNINGs about the type mismatches. This is
2650 a particular problem in practice for the new sequence functions.
2651 It's not clear how this should be fixed, and for now, a
2652 workaround is given in the entry for 117 in the BUGS file.
2653 ** (Because of the interaction between the two previous items --
2654 occasional inlining problems and new inline expansions -- some
2655 of the new sequence function optimizations won't really kick in
2656 completely until debugging information, and then inlining, are
2657 straightened out in some future version.)
2658 * minor incompatible changes:
2659 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2660 DIRECTORY no longer implicitly promotes NIL slots of its
2661 pathname argument to :WILD. In particular, when you ask for the
2662 contents of a directory (which you used to be able to do without
2663 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2664 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2665 ** changes in behavior that ANSI explicitly defines to be
2666 implementation dependent:
2667 *** The new compiler-only implementation still conforms with ANSI,
2668 but acts a little different than before. Besides the obvious
2669 changes in performance tradeoffs (that the cost per form passed
2670 to EVAL has gone up, and the cost per form executed by EVAL
2671 has gone down), the behavior of the system changes a little
2672 because there are no longer any interpreted function objects.
2673 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2674 e.g. doing COMPILE on the output of interactive DEFUN is
2676 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2678 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2679 has changed from (:CL) to NIL.
2680 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2681 e.g. #\Newline and #\Tab, don't have names specified in the
2682 ANSI Common Lisp standard, is now based on their ASCII symbolic
2683 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2684 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2685 are no longer used for output.
2686 ** changes in internal implementation constants:
2687 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2688 4 million. (If your application spends a lot of time GCing and
2689 you have a lot of RAM, you might want to experiment with
2690 increasing it even more.)
2691 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2692 However, almost all old code should still continue to work without
2693 immediate update, as SB-C-CALL is now a (deprecated) nickname
2695 ** Old operator names in the style DEF-FOO are now deprecated in
2696 favor of new corresponding names DEFINE-FOO, for consistency with
2697 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2698 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2699 internal symbols, but a few supported extensions like
2700 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2701 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2702 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2703 etc. as you get deeper into recursive calls to the debugger
2704 command loop, instead of the old "5]", "5]]", "5]]]"
2705 sequence. (I was motivated to do this when squabbles between
2706 ILISP and SBCL left me very deeply nested in the debugger. In the
2707 short term, this change will probably provoke more ILISP/SBCL
2708 squabbles, but hopefully it will be an improvement in the long run.)
2709 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2710 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2711 function, since declaiming return types involving aliens is
2712 (1) annoyingly messy to do by hand and (2) vital to efficient
2713 compilation of code which calls such functions.
2714 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2715 longer reexported by the SB-EXT package. They're solely useful
2716 for alien code, so it seems more logical that you should get
2717 them from the SB-ALIEN package, not in SB-EXT.
2718 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2719 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2720 features. Instead, the code that they used to control is always
2721 built into the system.
2722 * many other bug fixes
2723 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2724 advantage of the new EVAL-WHEN stuff and to clean them up in
2725 general, and they are now more ANSI-compliant in a number of
2726 ways. Martin Atzmueller is responsible for a lot of this.
2727 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2729 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2730 *** correct ERROR type for various file operations
2731 *** some fixes for Lisp streams
2732 *** DEFMETHOD syntax checking
2733 *** changing old weird representation of debug information as
2734 strings (which, among their other deficiencies, don't transform
2735 correctly when you rename packages, and don't change their
2736 print representation when you change things like *PACKAGE*
2737 and *PRINT-LENGTH*) to symbols and lists of symbols
2738 He also made several improvements and fixed several bugs in DESCRIBE.
2739 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2741 *** misbehavior of WRITE-STRING/WRITE-LINE
2742 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2743 and several other LOOP problems as well
2744 *** DIRECTORY when similar filenames are present
2745 *** DEFGENERIC with :METHOD options
2746 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2747 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2748 *** argument ordering in FIND with :TEST option
2749 *** mishandled package designator argument in APROPOS-LIST
2750 *** various problems in the backquote readmacro
2751 *** a bug in APROPOS
2752 *** probably some others that I'm not describing very well here,
2753 since the CVS log documents them by reference to sbcl-devel
2754 messages, and the SourceForge archives aren't working well.:-(
2755 ** Dan Barlow improved the Alpha port (and is making progress on the
2756 PPC port, for those of you who think different).
2757 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2758 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2759 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2760 continues to work on the SPARC port (for those of you in a position
2761 to look down upon our little PC-compatible boxes from a great height).
2762 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2763 and uses that copy, rather than the current dynamic value, when
2764 it's trying to decide whether to truncate output. Thus e.g.
2765 (let ((*print-lines* 50))
2766 (pprint-logical-block (stream nil)
2768 (let ((*print-lines* 8))
2769 (print (aref possiblybigthings i) stream)))))
2770 should now truncate the logical block only at 50 lines, instead of
2771 often truncating it at 8 lines, as it did before.
2772 * The doc/cmucl/ directory, containing old CMU CL documentation
2773 from the time of the fork, is no longer part of the base system.
2774 SourceForge has shut down its anonymous FTP service, and with it
2775 my original plan for distributing the old CMU CL documentation
2776 there. For now, if you need these files you can download an old
2777 SBCL source release and extract them from it.
2778 * The fasl file version number changed again, for dozens of reasons,
2779 some of which are apparent above.
2781 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2782 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2783 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2784 patch, so that DISASSEMBLE output is much nicer.
2785 * The code in the SB-PROFILE package now seems reasonably stable.
2786 I still haven't decided what the final interface should look like
2787 (I'd like PROFILE to interact cleanly with TRACE, since both
2788 facilities use function encapsulation) but if you have a need
2789 for profiling now, you can probably use it successfully with
2790 the current CMU-CL-style interface.
2791 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2792 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2793 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2794 link now returns the dangling link itself, and for similar
2795 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2796 link itself. (In these cases the old code signalled an error and
2797 looped endlessly, respectively.) Thus, DIRECTORY now works even
2798 in the presence of dangling and cyclic symbolic links.
2799 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2800 is now a supported extension again, since the consensus on
2801 sbcl-devel was that it can be useful for ordinary development
2802 work, not just for debugging SBCL itself.
2803 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2804 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2805 that currently-defined functions might be redefined later with
2806 different return types.
2807 * Hash tables can be printed readably, as inspired by CMU CL code
2808 of Eric Marsden and SBCL code of Martin Atzmueller.
2809 * better error handling in CLOS method combination, thanks to
2810 Martin Atzmueller porting Pierre Mai's CMU CL patches
2811 * more overflow fixes for >16Mbyte I/O buffers
2812 * A bug in READ has been fixed, so that now a single Ctrl-D
2813 character suffices to cause end-of-file on character streams.
2814 In particular, now you only need one Ctrl-D at the command
2815 line (not two) to exit SBCL.
2816 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2818 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2819 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2820 does the right thing.
2821 * The implementation of some type tests, especially for CONDITION
2822 types, is now tidier and maybe faster, due to CMU CL code
2823 originally by Douglas Crosher, ported by Martin Atzmueller.
2824 * Some math functions have been fixed, and there are new
2825 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2826 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2827 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2828 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2829 likely to be built into sbcl-0.7.0.)
2830 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2831 CL, ported to SBCL by Martin Atzmueller)
2832 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2833 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2834 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2835 * There's a new slam.sh hack to shorten the edit/compile/debug
2836 cycle for low-level changes to SBCL itself, and a new
2837 :SB-AFTER-XC-CORE target feature to control the generation of
2838 the after-xc.core file needed by slam.sh.
2839 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2840 COMPILE-FILE is no longer supported, so that now every function
2841 gets an entry point, so that block compilation looks a little
2842 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2843 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2844 returns the number of bytes consed since the system started,
2845 rather than the number consed since the first time the function
2846 was called. (The new definition parallels ANSI functions like
2847 CL:GET-INTERNAL-RUN-TIME.)
2848 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2849 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2850 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2851 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2852 * Fasl file version numbers are now independent of the target CPU,
2853 since historically most system changes which required version
2854 number changes have affected all CPUs equally. Similarly,
2855 the byte fasl file version is now equal to the ordinary
2858 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2859 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2860 is no longer recognized. I apologize for this, because it was
2861 listed in SB-EXT as a supported extension, but I found that
2862 its existing behavior was poorly specified, as well as incorrectly
2863 specified, and it looked like too much of a mess to straighten it
2864 out. I have enough on my hands trying to get ANSI stuff to work..
2865 * many patches ported from CMU CL by Martin Atzmueller, with
2866 half a dozen bug fixes in pretty-printing and the debugger, and
2867 half a dozen others elsewhere
2868 * fixed bug 13: Floating point infinities are now supported again.
2869 They might still be a little bit flaky, but thanks to bug reports
2870 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2871 as flaky as they were.
2872 * The --noprogrammer command line option is now supported. (Its
2873 behavior is slightly different in detail from what the old man
2874 page claimed it would do, but it's still appropriate under the
2875 same circumstances that the man page talks about.)
2876 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2877 are now supported, and enabled by default. Thus, the compiler can
2878 handle many floating point and complex operations much less
2879 inefficiently. (Thus e.g. you can implement a complex FFT
2881 * The compiler now detects type mismatches between DECLAIM FTYPE
2882 and DEFUN better, and implements CHECK-TYPE more correctly, and
2883 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2884 thanks to patches from Martin Atzmueller.
2885 * various fixes to make the cross-compiler more portable to
2886 ANSI-conforming-but-different cross-compilation hosts (notably
2887 Lispworks for Windows, following bug reports from Arthur Lemmens)
2888 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2889 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2890 fixed thanks to Pierre Mai's CMU CL patches.
2891 * fixes to keep the system from overflowing internal counters when
2892 it tries to use i/o buffers larger than 16M bytes
2893 * fixed bug 45a: Various internal functions required to support
2894 complex special functions have been merged from CMU CL sources.
2895 (When I was first setting up SBCL, I misunderstood a compile-time
2896 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2897 * improved support for type intersection and union, fixing bug 12
2898 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2899 more obscure bugs as well
2900 * some steps toward byte-compiling non-performance-critical
2901 parts of the system, courtesy of patches from Martin Atzmueller
2902 * Christophe Rhodes has made some debian packages of sbcl at
2903 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2904 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2905 stable, but are nonetheless usable. When he's ready, I'd be happy
2906 to add them to the SourceForge "File Releases" section. (And if
2907 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2908 * new fasl file format version number (because of changes in
2909 internal representation of (OR ..) types to accommodate the new
2910 support for (AND ..) types, among other things)
2912 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2913 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2915 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2917 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2918 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2919 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2920 * DESCRIBE now works on structure objects again.
2921 * Most function call argument type mismatches are now handled as
2922 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2923 know whether the function will be redefined before the call is
2924 executed. (The compiler could flag local calls with full WARNINGs,
2925 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2926 it doesn't keep track of enough information to know whether calls
2927 are local in this sense.)
2928 * Compiler output is now more verbose, with messages truncated
2929 later than before. (There should be some supported way for users
2930 to override the default verbosity, but I haven't decided how to
2931 provide it yet, so this behavior is still controlled by the internal
2932 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2933 src/compiler/ir1util.lisp.)
2934 * Fasl file format version numbers have increased again, because
2935 support for the Gray streams extension changes the layout of the
2936 system's STREAM objects.
2937 * The Gray subclassable streams extension now works, thanks to a
2938 patch from Martin Atzmueller.
2939 * The full LOAD-FOREIGN extension (not just the primitive
2940 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2941 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2942 but like most other programs, it defaults to copying the Unix
2943 environment from the original process instead of starting the
2944 new process in an empty environment.
2945 * Extensions which manipulate the Unix environment now support
2946 an :ENVIRONMENT keyword option which doesn't smash case or
2947 do other bad things. The CMU-CL-style :ENV option is retained
2948 for porting convenience.
2949 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2950 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2952 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2954 * A patch from Martin Atzmueller seems to have solved the SIGINT
2955 problem, and as far as we know, signal-handling now works cleanly.
2956 (If you find any new bugs, please report them!)
2957 * The system no longer defaults Lisp source file names to types
2958 ".l", ".cl", or ".lsp", but only to ".lisp".
2959 * The compiler no longer uses special default file extensions for
2960 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2961 seems to expect a single default extension for all compiled code,
2962 and there's no compelling reason to try to stretch the standard
2963 to allow two different extensions.) Instead, byte-compiled files
2964 default to the same extension as native-compiled files.
2965 * Fasl file format version numbers have increased again, because
2966 a rearrangement of internal implementation packages made some
2967 dumped symbols in old fasl files unreadable in new cores.
2968 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2969 many fewer weird special cases.
2970 * Bug #17 (differing COMPILE-FILE behavior between logical and
2971 physical pathnames) has been fixed, and some related misbehavior too,
2972 thanks to a patch from Martin Atzmueller.
2973 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2974 by Tim Moore, ported to SBCL by Martin Atzmueller.
2975 * Martin Atzmueller fixed several filesystem-related problems,
2976 including bug #36, in part by porting CMU CL patches, which were
2977 written in part by Paul Werkowski.
2978 * More compiler warnings in src/runtime/ are gone, thanks to
2979 more patches from Martin Atzmueller.
2980 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2983 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2985 * DESCRIBE now works on CONDITION objects.
2986 * The debugger now handles errors which arise when trying to print
2987 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2989 * The build system now uses an additional file, customize-target-features.lisp,
2990 to allow local modifications to the target *FEATURES* list. (The point of
2991 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2992 debugging features enabled, without having to worry about propagating it
2993 into everyone's system when I do a "cvs update".) When no
2994 customize-target-features.lisp file exists, the target *FEATURES* list
2995 should be constructed the same way as before.
2996 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2997 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2998 warn about behavior which is undefined under the ANSI spec. Note: This
2999 is specified by ANSI, but it's not very popular with programmers.
3000 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3001 macro in the SBCL sources for an example of a workaround which you
3002 might use to make portable ANSI-standard code which does what you want.
3003 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3004 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3005 when it's used as a non-toplevel form. (This is required in order
3006 to implement the DEFCONSTANT-EQX macro.)
3007 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3008 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3009 funny things, due to bugs in the implementation of EVAL-WHEN
3010 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3011 non-toplevel DEFCONSTANTs are uncommon.)
3012 * The core file version number and fasl file version number have been
3013 incremented, because the old noncompliant DEFCONSTANT behavior involved
3014 calling functions which no longer exist, and because I also took the
3015 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3016 * fixed bug 1 (error handling before read-eval-print loop starts), and
3017 redid debugger restarts and related debugger commands somewhat while
3019 ** The QUIT debugger command is gone, since it did something
3020 rather different than the SB-EXT:QUIT command, and since it never
3021 worked properly outside the main toplevel read/eval/print loop.
3022 Invoking the new TOPLEVEL restart provides the same functionality.
3023 ** The GO debugger command is also gone, since you can just invoke
3024 the CONTINUE restart directly instead.
3025 ** The TOP debugger command is also gone, since it's redundant with the
3026 FRAME 0 command, and since it interfered with abbreviations for the
3028 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3030 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3031 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3032 some porting work by Martin Atzmueller)
3033 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3034 with a boatload of patches which he ported to SBCL. Now that those
3036 ** The system tries to make sure that its low-priority messages
3037 are prefixed by semicolons, to help people who like to use
3038 syntax highlighting in their ILISP buffer. (This patch
3039 was originally due to Raymond Toy.)
3040 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3041 patches originally written by Raymond Toy.
3042 ** The compiler understands coercion between single-value and
3043 multiple-VALUES type expressions better, getting rid of some very
3044 weird behavior, thanks to patches originally by Robert MacLachlan
3045 and Douglas Crosher.
3046 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3047 lambda lists, thanks to a patch originally by Pierre Mai.
3048 ** The system no longer bogusly warns about "abbreviated type
3050 ** The compiler gets less confused by inlining and RETURN-FROM,
3051 thanks to some patches originally by Tim Moore.
3052 ** The system no longer hangs when dumping circular lists to fasl
3053 files, thanks to a patch originally from Douglas Crosher.
3054 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3055 undefined function error.
3056 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3057 in structure slot initforms) for now, documented workaround instead:-|
3058 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3059 * fixed bug 5: added stubs for various Gray stream functions called
3060 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3061 installed, at least appropriate type errors are generated
3062 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3063 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3064 consistently in DEFMETHOD forms.
3065 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3066 it doesn't seem to affect SBCL after all
3067 * The C runtime system now builds with better optimization and many
3068 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3070 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3072 * The system is now under CVS at SourceForge (instead of the
3073 CVS repository on my home machine).
3074 * The new signal handling code has been tweaked to treat register
3075 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3076 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3077 downstream, e.g. in debug-int.lisp, has implicit dependencies
3078 on the unsignedness of integer representation of machine words,
3079 and that caused the system to bomb out with infinite regress
3080 when trying to recover from type errors involving signed values,
3081 e.g. (BUTLAST '(1 2 3) -1).)
3082 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3083 (This was one of the bugs Peter Van Eynde reported back in July.)
3084 * The system now uses code inspired by Colin Walters' O(N)
3085 implementation of MAP (from the cmucl-imp@cons.org mailing
3086 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3087 inline the MAP operation, and there is more than one
3088 sequence argument to the MAP call (so that it can't just
3089 do ETYPECASE once and for all based on the type of the
3090 single sequence argument). (The old non-inline implementation
3091 of the general M-argument sequence-of-length-N case required
3092 O(M*N*N) time when any of the sequence arguments were LISTs.)
3093 * The QUIT :UNIX-CODE keyword argument has been renamed to
3094 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3095 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3097 * Raymond Wiker's patches to port dynamic loading from Linux to
3098 FreeBSD have been added.
3099 * The BUGS file is now more nearly up to date, thanks in large part
3100 to Martin Atzmueller's review of it.
3101 * The debugger now flushes standard output streams before it begins
3102 its output ("debugger invoked" and so forth).
3103 * The core version number and fasl file version number have both
3104 been incremented, because of incompatible changes in the layout
3106 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3107 used to not be called for a saved Lisp image.)
3108 * Martin Atzmueller's version of a patch to fix a compiler crash,
3109 as posted on sbcl-devel 13 September 2000, has been installed.
3110 * Instead of installing Martin Atzmueller's patch for the
3111 compiler transform for SUBSEQ, I deleted the compiler transform,
3112 and transforms for some similar consing operations.
3113 * A bug in signal handling which kept TRACE from working on OpenBSD
3115 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3116 away by constant folding
3117 * The system now defines its address space constants in one place
3118 (in the Lisp sources), and propagates them automatically elsewhere
3119 (through GENESIS and the sbcl.h file). Therefore, patching the
3120 address map is less unnecessarily tedious and error-prone. The
3121 Lisp names of address space constants have also been systematized.
3122 * CVS tags like dollar-Header-dollar have been removed from
3123 the sources, because they have never saved me trouble and
3124 they've been source of trouble working with patches and other
3125 diff-related operations.
3126 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3127 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3129 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3131 * The system has been ported to OpenBSD.
3132 * The system now compiles with a simple "sh make.sh" on the systems
3133 that it's supported on. I.e., now you no longer need to tweak
3134 text in the target-features.lisp-expr and symlinks in src/runtime/
3135 by hand, the make.sh takes care of it for you.
3136 * The system is no longer so grossly inefficient when compiling code
3137 involving vectors implemented as general (not simple) vectors (VECTOR T),
3138 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3139 which dares to use the various sequence functions on non-simple
3140 vectors, takes less of a performance hit.
3141 * There is now a primitive type predicate VECTOR-T-P
3142 to test for the (VECTOR T) type, so that e.g.
3143 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3144 can now be compiled with some semblance of efficiency. (The old code
3145 turned the type declaration into a full call to %TYPEP at runtime!)
3146 * AREF on (VECTOR T) is still not fast, since it's still compiled
3147 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3148 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3149 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3150 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3151 without making full calls to SUBTYPEP at runtime.
3152 (Some analogous efficiency issues for non-simple vectors specialized to
3153 element types other than T, or for non-simple multidimensional arrays,
3154 have not been addressed. They could almost certainly be handled the
3155 same way if anyone is motivated to do so.)
3156 * The changes in array handling break binary compatibility, so
3157 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3158 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3159 instead of (VALUES T T).
3160 * By following the instructions that Dan Barlow posted to sbcl-devel
3161 on 2 July 2000, I was able to enable primitive dynamic object
3162 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3163 functionality is not implemented (since it calls ld to resolve
3164 library references automatically, requiring RUN-PROGRAM for its
3165 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3166 try to resolve library references) is now supported.
3167 * The system now flushes the standard output streams when it terminates,
3168 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3169 them at several other probably-convenient times, e.g. in each pass of
3170 the toplevel read-eval-print loop, and after evaluating a form given
3171 as an "--eval" command-line option. (These changes were motivated by a
3172 discussion of stream flushing issues on cmucl-imp in August 2000.)
3173 * The source transform for TYPEP of array types no longer assumes
3174 that an array whose element type is a not-yet-defined type
3175 is implemented as an array of T, but instead punts, so that the
3176 type will be interpreted at runtime.
3177 * There is now some support for cross-compiling in make.sh: each of
3178 the phases of make.sh has its own script. (This should be transparent
3179 to people doing ordinary, non-cross-compile builds.)
3180 * Since my laptop doesn't have hundreds of megabytes of memory like
3181 my desktop machine, I became more motivated to do some items on
3182 my to-do list in order to reduce the size of the system a little:
3183 ** Arrange for various needed-only-at-cold-init things to be
3184 uninterned after cold init. To support this, those things have
3185 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3186 symbols with such names are now uninterned after cold init).
3187 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3188 fasl files for cold load.
3189 ** Remove the old compiler structure pooling code (which used to
3190 be conditional on the target feature :SB-ALLOC) completely.
3191 ** Redo the representation of some data in cold init to be more compact.
3192 (I also looked into supporting byte compiled code at bootstrap time,
3193 which would probably reduce the size of the system a lot, but that
3194 looked too complicated, so I punted for now.)
3195 * The maximum signal nesting depth in the src/runtime/ support code has
3196 been reduced from 4096 to 256. (I don't know any reason for the very
3197 large old value. If the new smaller value turns out to break something,
3198 I'll probably just bump it back up.)
3199 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3201 * Many, many bugs reported by Peter Van Eynde have been added to
3202 the BUGS list; some have even been fixed.
3203 * While enabling dynamic object file loading, I tried to make the
3204 code easier to understand, renaming various functions and variables
3205 with less ambiguous names, and changing some function calling
3206 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3207 * While trying to figure out how to do the OpenBSD port, I tried to
3208 clean up some of the code in src/runtime/. In particular, I dropped
3209 support for non-POSIX signal handling, added various comments,
3210 tweaked the code to reduce the number of compilation warnings, and
3211 renamed some files to increase consistency.
3212 * To support the new automatic configuration functionality in make.sh,
3213 the source file target-features.lisp-expr has been replaced with the
3214 source file base-target-features.lisp-expr and the machine-generated
3215 file local-target-features.lisp-expr.
3216 * fixed a stupid quoting error in make.sh so that using CMU CL
3217 "lisp -batch" as cross-compilation host works again
3219 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3221 * DESCRIBE no longer tries to call itself recursively to describe
3222 bound/fbound values, so that it no longer fails on symbols which are
3223 bound to themselves (like keywords, T, and NIL).
3224 * DESCRIBE now works on generic functions.
3225 * The printer now prints less-screwed-up representations of closures
3226 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3227 * A private symbol is used instead of the :EMPTY keyword previously
3228 used to mark empty slots in hash tables. Thus
3229 (DEFVAR *HT* (MAKE-HASH-TABLE))
3230 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3231 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3232 now does what ANSI says that it should. (You can still get
3233 similar noncompliant behavior if bang on the hash table
3234 implementation with all the symbols you get back from
3235 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3236 This breaks binary compatibility, since tests for equality to
3237 :EMPTY are wired into things like the macroexpansion of
3238 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3240 * There's now a minimal placeholder implementation for CL:STEP,
3241 as required by ANSI.
3242 * An obscure bug in the interaction of the normal compiler, the byte
3243 compiler, inlining, and structure predicates has been patched
3244 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3245 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3247 * Missing ordinary arguments in a macro call are now detected even
3248 when the macro lambda list contains &KEY or &REST.
3249 * The debugger no longer complains about encountering the top of the
3250 stack when you type "FRAME 0" to explicitly instruct it to go to
3251 the top of the stack. And it now prints the frame you request even
3252 if it's the current frame (instead of saying "You are here.").
3253 * As specified by ANSI, the system now always prints keywords
3254 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3256 * The default initial SIZE of HASH-TABLEs is now smaller.
3257 * Type information from CLOS class dispatch is now propagated
3258 into DEFMETHOD bodies, so that e.g.
3259 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3261 is now basically equivalent to
3262 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3263 (DECLARE (TYPE SINGLE-FLOAT X))
3265 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3266 operation, without having to do run-time type dispatch.
3267 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3268 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3269 * Since I don't seem to be making big file reorganizations very often
3270 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3271 is overflowing my ability to conveniently back them up), I've finally
3272 checked the system into CVS. (The CVS repository is on my home system,
3273 not at SourceForge -- putting it on SourceForge might come later.)
3274 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3275 high-level GC-NOTIFY-FOO functions send their output. (There's
3276 still very little control of where low-level verbose GC functions
3277 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3278 less than it used to -- the GC-NOTIFY-FOO functions are now under
3279 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3280 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3281 in only one place in the source code, and propagates it automatically
3282 everywhere that it's needed. Thus e.g. when I bump the version from
3283 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3284 * The C source files now include boilerplate legalese and documentation
3285 at the head of each file (just as the Lisp source files already did).
3286 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3287 to his page will be replaced with a link to his new CLiki service.
3289 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3291 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3292 * The build process now looks for GNU make under the default name "gmake",
3293 instead of "make" as it used to. If GNU make is not available as "gmake"
3294 on your system, you can change this default behavior by setting the
3295 GNUMAKE environment variable.
3296 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3297 can build without error under CMU CL.
3299 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3301 * There is now a partial SBCL user manual (with some new text and some
3302 text cribbed from the CMU CL manual).
3303 * The beginnings of a profiler have been added (starting with the
3304 CMU CL profiler and simplifying and cleaning up). Eventually the
3305 main interface should be through the TRACE macro, but for now,
3306 it's still accessed through vaguely CMU-CL-style functions and macros
3307 exported from the package SB-PROFILE.
3308 * Some problems left over from porting CMU CL to the new
3309 cross-compilation bootstrap process have been cleaned up:
3310 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3311 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3312 part when getting rid of the compiler *BACKEND* stuff.)
3313 ** The value of *NULL-TYPE* was screwed up, because it was
3314 being initialized before the type system knew the final
3315 definition of the 'NULL type. This screwed up several key
3316 optimizations in the compiler, causing inefficiency in all sorts
3317 of places. (I found it because I wanted to understand why
3318 GET-INTERNAL-RUN-TIME was consing.)
3319 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3320 PROCLAIM FTYPE information. Unfortunately this broke binary
3321 compatibility again, since now the forms output by DEFGENERIC
3322 to refer to functions which didn't exist in 0.6.3.
3323 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3324 can use the new (as of 0.6.3) transform for SOME into MAP into
3326 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3327 values to (INTEGER 0 1000000), so that the time code will no longer
3328 occasionally get blown up by Linux returning 1000000 microseconds
3329 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3330 its output conform to the ANSI spec. (Alas, this makes its output
3331 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3332 * A full call to MAP NIL with a single sequence argument no longer conses.
3333 * fixes to problems pointed out by Martin Atzmueller:
3334 * The manual page no longer talks about multiprocessing as though
3335 it were currently supported.
3336 * The ILISP support patches have been removed from the distribution,
3337 because as of version 5.10.1, ILISP now supports SBCL without us
3338 having to maintain patches.
3339 * added a modified version of Raymond Toy's recent CMU CL patch for
3340 EQUALP comparison of HASH-TABLE
3342 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3344 * The system still can't cross-compile itself with
3345 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3346 entails), but at least it can compile more of itself that way
3347 than it used to be able to, and various buglets which were uncovered
3348 by trying to cross-compile itself that way have now been fixed.
3349 * This release breaks binary compatibility again. This time
3350 at least I've incremented the FASL file format version to 2, so that the
3351 problem can be detected reliably instead of just causing weird errors.
3352 * various new style warnings:
3353 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3354 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3355 ** using the lexical binding of a variable named in the *FOO* style
3356 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3357 as specified by ANSI.
3358 * *RANDOM-STATE* is no longer automatically initialized from
3359 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3360 default behavior of the system is to repeat its behavior every time
3361 it's run. If you'd like to change this behavior, you can always
3362 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3363 old convention there was no comparably easy way to get the system to
3364 repeat its behavior every time it was run.
3365 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3366 FTYPE declarations has been removed, in favor of their ANSI
3367 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3368 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3369 the types of their sequence arguments can be determined at compile time.
3370 This is done through a new open code expansion for MAP which eliminates
3371 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3372 argument types can be determined at compile time.
3373 * The optimizer now transforms COERCE into an identity operation when it
3374 can prove that the coerced object is already of the correct type. (This
3375 can be a win for machine generated code, including the output of other
3376 optimization transforms, such as the MAP transform above.)
3377 * Credit information has been moved from source file headers into CREDITS.
3378 * Source file headers have been made more standard.
3379 * The CASE macro now compiles without complaining even when it has
3382 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3384 * (Note that the way that the PCL macroexpansions were rewritten
3385 to accommodate the change in DEFGENERIC below breaks binary
3386 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3387 not run under sbcl-0.6.2. Once we get out of alpha releases,
3388 i.e. hit release 1.0.0, we'll probably try to maintain binary
3389 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3390 and sbcl-1.4.4. Until then, however, it might be fairly common
3391 for maintenance releases to break binary compatibility.)
3392 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3394 * The reporting of unhandled signals has been changed to print some
3395 explanatory text as well as the report form. (Previously only
3396 the report form was printed.)
3397 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3398 it defines, so that the compiler no longer issues undefined function
3399 warnings for compiled-but-not-yet-loaded generic functions.
3400 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3401 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3402 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3403 section of the ANSI spec are supported.
3404 * The "" nickname for the "KEYWORD" package has been removed.
3405 The reader still handles symbol tokens which begin with a package marker
3406 as keywords, but it doesn't expose its mechanism for doing so in the
3407 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3408 * The system now issues STYLE-WARNINGs for contradictory TYPE
3409 proclamations. (Warnings for contradictory FTYPE proclamations would
3410 be nice too, but those can't be done usefully unless the type system
3411 is made smarter about FUNCTION types.)
3412 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3413 systematized, so that "*target-*.lisp is supposed to exist only on the
3414 target and imply that there's a related file which exists on the
3415 host, and *host-*.lisp is supposed to exist only on the host and imply
3416 that there's a related file which exists on the target. This involves a
3417 lot of renaming. Hopefully the acute confusion caused by the renaming
3418 will be justified by the reduction in chronic confusion..
3419 ** runtime-type.lisp -> early-target-type.lisp
3420 ** target-type.lisp -> late-target-type.lisp
3421 ** early-host-format.lisp -> early-format.lisp
3422 ** late-host-format.lisp -> late-format.lisp
3423 ** host-error.lisp -> misc-error.lisp
3424 ** early-error.lisp -> early-target-error.lisp
3425 ** late-error.lisp -> late-target-error.lisp
3426 ** host-defboot.lisp -> early-defboot.lisp
3427 ** code/misc.lisp -> code/target-misc.lisp
3428 ** code/host-misc.lisp -> code/misc.lisp
3429 ** code/numbers.lisp -> code/target-numbers.lisp
3430 ** code/early-numbers.lisp -> numbers.lisp
3431 ** early-host-type.lisp -> early-type.lisp
3432 ** late-host-type.lisp -> late-type.lisp
3433 ** host-typep.lisp -> typep.lisp
3434 ** load.lisp -> target-load.lisp
3435 ** host-load.lisp -> load.lisp
3436 ** host-disassem.lisp -> disassem.lisp
3437 ** host-insts.lisp -> insts.lisp
3438 ** byte-comp.lisp -> target-byte-comp.lisp
3439 ** host-byte-comp.lisp -> byte-comp.lisp
3440 ** host-signal.lisp -> signal.lisp
3441 ** host-defstruct.lisp -> defstruct.lisp
3442 ** late-target-type.lisp -> deftypes-for-target.lisp
3443 Furthermore, several other previously target-only files foo.lisp (e.g.
3444 hash-table.lisp and random.lisp) have been split into a target-and-host
3445 foo.lisp file and a target-only target-foo.lisp file, with their key type
3446 definitions in the target-and-host part, so that the cross-compiler will
3447 know more about target types.
3448 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3449 gone away. In their place are various *BACKEND-FOO* variables
3450 corresponding to the slots of the old structure.
3451 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3452 into the target SBCL has been fixed.
3453 * The chill.lisp system for loading cold code into a running SBCL
3455 * Support for the CMU CL "scavenger hook" extension has been removed.
3456 (It was undocumented and unused in the CMU CL sources that SBCL was
3457 derived from, and stale in sbcl-0.6.1.)
3458 * Various errors in the cross-compiler type system were detected
3459 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3460 (enabling various consistency checks). Many of them were fixed,
3461 but some hard problems remain, so the compiler is back to
3462 running without *TYPE-SYSTEM-INITIALIZED* for now.
3463 * As part of the cross-compiler type system cleanup, I implemented
3464 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3465 * I have started adding UNCROSS calls throughout the type system
3466 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3467 on cross-compiler input files will be able to go away, and various
3469 * CONSTANTP now returns true for quoted forms (as explicitly required
3472 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3474 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3475 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3476 (MAKE-LIST -1). In the longer term, it ought to become true
3477 that declarations are assertions even at SAFETY 1. For now, it's not
3478 quite true even at SAFETY 3, but it's at least more nearly true..
3479 (Note that this change seems to increases the size of the system by
3480 O(5%) and to decrease the speed of the compiler by 20% or more.)
3481 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3482 for the problem of printing dozens of lines of distracting information
3483 about low-level system machinery as part of the top stack frame
3484 on entry to the debugger when an undefined function was called.
3485 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3486 is not reset to COMMON-LISP-USER.
3487 * Compilation of stuff related to dyncount.lisp has been made conditional
3488 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3489 smaller. The various dyncount-related symbols have been moved into
3490 a new "SB-DYNCOUNT" package.
3491 * tty-inspect.lisp has been renamed to inspect.lisp.
3492 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3493 feature has gone away. (When we eventually port to other flavors of
3494 libc and/or Unix, we'll try to make the differences between flavors
3495 invisible at the user level.)
3496 * Various other *FEATURES* tags, and/or their associated conditionals,
3497 have been removed if obsolescent, or given better documentation, or
3498 sometimes given more-mnemonic names.
3500 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3502 * tidied up "make.sh" script
3503 * tidied up system directory structure
3504 * better "clean.sh" behavior
3505 * added doc/FOR-CMUCL-DEVELOPERS
3506 * many many small tweaks to output format, e.g. removing possibly-confusing
3507 trailing #\. character in DESCRIBE-INSTANCE
3508 * (EQUALP #\A 'A) no longer signals an error.
3509 * new hashing code, including EQUALP hashing
3510 * tidied up Lisp initialization and toplevel
3511 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3512 * command line argument processing
3513 * added POSIX-GETENV function to deal with Unix-ish environment variables
3514 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3515 terminating SBCL on EOF
3516 * non-verbose GC by default
3517 * There is no more "sbcl" shell script; the sbcl file is now the C
3518 runtime executable (just like CMU CL).
3519 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3521 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3522 symbol and package information as Lisp native symbol and package objects
3523 instead of strings naming symbols and strings naming packages. This way,
3524 whenever packages are renamed (as in warm init), debug information is
3525 transformed along with everything else.
3526 * tweaked the optimization policy declarations which control the building
3527 of SBCL itself. Now, among other things, the system no longer saves
3528 source location debugging information. (This helps two problems at once
3529 by reducing SBCL size and by keeping SBCL from trying to look for its
3530 sources -- which may not exist -- when reporting errors.)
3531 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3532 debugging and testing purposes
3533 * cleaned up printing, making the printer call PRINT-OBJECT for
3534 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3535 methods, giving nearly-ANSI behavior
3536 * converted almost all special variables to use *FOO* naming convention
3537 * deleted PARSE-TIME functionality, since it can be done portably
3538 * moved some files out of cold init into warm init
3539 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3541 * regularized formatting of source files
3542 * added an install.sh script
3543 * fixed ridiculous memory usage of cross-compiler by making
3544 compiler/alloc.lisp not try to do pooling unless it can hook
3545 itself into the GC of the cross-compilation host. Now the system
3546 builds nicely on my old laptop.
3547 * added :SB-ALLOC in target-features.lisp-expr
3548 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3549 was not implemented)
3550 * re-did condition handling and note reporting in the compiler. Notes
3551 are no longer handled by signalling conditions. Style warnings
3552 and warnings are handled more correctly and reported in such a way
3553 that it's easy to find one or the other in your output (so that you
3554 can e.g. figure out which of many problems caused COMPILE-FILE to
3556 * changed the severity of several compiler warnings from full WARNING
3557 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3558 compiler note reporting so that it doesn't use the condition system
3559 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3560 COMPILE-FILE command)
3561 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3562 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3563 no longer supported, since their implementation was deeply intertwingled
3564 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3565 * removed bogus "support" for compiler macros named (SETF FOO), and
3566 removed the compiler macro for SETF INFO (but only after making a fool
3567 of myself on the cmucl-imp mailing list by posting a bogus patch for
3568 DEFINE-COMPILER-MACRO..)
3569 * Compiled files containing forms which have side effects on the Lisp
3570 reader (such as DEFPACKAGE forms) are now handled more correctly.
3571 (Compiler queuing of top level lambdas has been suppressed by setting
3572 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3573 * deleted various currently-unused source files, e.g. gengc.lisp. They
3574 may be added back at some point e.g. when porting to other architectures,
3575 but until they are it's distracting to distribute them and to try to
3577 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3578 there were so many of them they're just distractions, and UNCROSS is
3579 known to be able to handle the current sources
3580 * moved PROFILE functionality into TRACE, so that it will be clear
3581 how the wrapping and unwrapping of functions when you profile them
3582 interacts with the wrapping and unwrapping of functions when you
3583 trace them. (Actually, the functionality isn't there yet, but at least
3584 the interface specification is there. Hopefully, the functionality will
3585 arrive with some future maintenance release.)
3586 * removed host-oops.lisp
3587 * changed signature of QUIT function to allow UNIX-CODE argument
3588 * fixed READ-SEQUENCE bug
3589 * tweaked verbose GC output so that it looks more like the progress
3590 output that ANSI specifies for functions like LOAD
3591 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3592 * added <http://sbcl.sourceforge.com> to the banner information printed by